Neuroscience Dale Purves 3ed.

NEUROSCIENCE Third Edition NEUROSCIENCE Edited by DALE PURVES GEORGE J. AUGUSTINE DAVID FITZPATRICK WILLIAM C. HALL ANTHONY-SAMUEL LAMANTIA JAMES O. MCNAMARA S. MARK WILLIAMS Sinauer Associates, Inc. • Publishers Sunderland, Massachusetts U.S.A. THIRD EDITION THE COVER Dorsal view of the human brain. (Courtesy of S. Mark Williams.) NEUROSCIENCE: Third Edition Copyright © 2004 by Sinauer Associates, Inc. All rights reserved. This book may not be reproduced in whole or in part without permission. Address inquiries and orders to Sinauer Associates, Inc. 23 Plumtree Road Sunderland, MA 01375 U.S.A. www.sinauer.com FAX: 413-549-1118 orders@sinauer.com publish@sinauer.com Library of Congress Cataloging-in-Publication Data Neuroscience / edited by Dale Purves ... [et al.].— 3rd ed. p. ; cm. Includes bibliographical references and index. ISBN 0-87893-725-0 (casebound : alk. paper) 1. Neurosciences. [DNLM: 1. Nervous System Physiology. 2. Neurochemistry. WL 102 N50588 2004] I. Purves, Dale. QP355.2.N487 2004 612.8—dc22 2004003973 Printed in U.S.A. 5 4 3 2 1 Contributors George J. Augustine, Ph.D. Dona M. Chikaraishi, Ph.D. Michael D. Ehlers, M.D., Ph.D. Gillian Einstein, Ph.D. David Fitzpatrick, Ph.D. William C. Hall, Ph.D. Erich Jarvis, Ph.D. Lawrence C. Katz, Ph.D. Julie Kauer, Ph.D. Anthony-Samuel LaMantia, Ph.D. James O. McNamara, M.D. Richard D. Mooney, Ph.D. Miguel A. L. Nicolelis, M.D., Ph.D. Dale Purves, M.D. Peter H. Reinhart, Ph.D. Sidney A. Simon, Ph.D. J. H. Pate Skene, Ph.D. James Voyvodic, Ph.D. Leonard E. White, Ph.D. S. Mark Williams, Ph.D. UNIT EDITORS UNIT I: George J. Augustine UNIT II: David Fitzpatrick UNIT III: William C. Hall UNIT IV: Anthony-Samuel LaMantia UNIT V: Dale Purves Contents in Brief 1. Studying the Nervous Systems of Humans and Other Animals 1 UNIT I NEURAL SIGNALING 2. 3. 4. 5. 6. 7. Electrical Signals of Nerve Cells 31 Voltage-Dependent Membrane Permeability 47 Channels and Transporters 69 Synaptic Transmission 93 Neurotransmitters, Receptors, and Their Effects 129 Molecular Signaling within Neurons 165 UNIT II SENSATION AND SENSORY PROCESSING 8. 9. 10. 11. 12. 13. 14. UNIT III 15. 16. 17. 18. 19. 20. UNIT IV The Somatic Sensory System 189 Pain 209 Vision: The Eye 229 Central Visual Pathways 259 The Auditory System 283 The Vestibular System 315 The Chemical Senses 337 MOVEMENT AND ITS CENTRAL CONTROL Lower Motor Neuron Circuits and Motor Control 371 Upper Motor Neuron Control of the Brainstem and Spinal Cord 393 Modulation of Movement by the Basal Ganglia 417 Modulation of Movement by the Cerebellum 435 Eye Movements and Sensory Motor Integration 453 The Visceral Motor System 469 THE CHANGING BRAIN 21. Early Brain Development 501 22. Construction of Neural Circuits 521 23. Modification of Brain Circuits as a Result of Experience 24. Plasticity of Mature Synapses and Circuits 575 557 UNIT V COMPLEX BRAIN FUNCTIONS 25. 26. 27. 28. 29. 30. The Association Cortices 613 Language and Speech 637 Sleep and Wakefulness 659 Emotions 687 Sex, Sexuality, and the Brain 711 Memory 733 APPENDIX A THE BRAINSTEM AND CRANIAL NERVES 755 APPENDIX B VASCULAR SUPPLY, THE MENINGES, AND THE VENTRICULAR SYSTEM 763 Contents Preface xvi Acknowledgments xvii Supplements to Accompany NEUROSCIENCE xviii Chapter 1 Studying the Nervous Systems of Humans and Other Animals 1 Overview 1 Genetics, Genomics, and the Brain 1 The Cellular Components of the Nervous System Neurons 4 Neuroglial Cells 8 Cellular Diversity in the Nervous System 9 Neural Circuits 11 2 Overall Organization of the Human Nervous System 14 Neuroanatomical Terminology 16 The Subdivisions of the Central Nervous System 18 Organizational Principles of Neural Systems 20 Functional Analysis of Neural Systems 23 Analyzing Complex Behavior 24 BOX A Brain Imaging Techniques 25 Summary 26 Unit I NEURAL SIGNALING Chapter 2 Electrical Signals of Nerve Cells 31 Overview 31 Electrical Potentials across Nerve Cell Membranes 31 How Ionic Movements Produce Electrical Signals 34 The Forces That Create Membrane Potentials 36 Electrochemical Equilibrium in an Environment with More Than One Permeant Ion 38 The Ionic Basis of the Resting Membrane Potential 40 BOX A The Remarkable Giant Nerve Cells of Squid 41 The Ionic Basis of Action Potentials 43 BOX B Action Potential Form and Nomenclature 44 Summary 45 Chapter 3 Voltage-Dependent Membrane Permeability 47 Overview 47 Ionic Currents Across Nerve Cell Membranes 47 BOX A The Voltage Clamp Method 48 Two Types of Voltage-Dependent Ionic Current 49 Two Voltage-Dependent Membrane Conductances 52 Reconstruction of the Action Potential 54 Long-Distance Signaling by Means of Action Potentials 56 BOX B Threshold 57 BOX C Passive Membrane Properties 60 The Refractory Period 61 Increased Conduction Velocity as a Result of Myelination 63 Summary 65 BOX D Multiple Sclerosis 66 Contents ix Chapter 4 Channels and Transporters 69 Overview 69 Ion Channels Underlying Action Potentials Receptors 129 Overview 129 Categories of Neurotransmitters Acetylcholine 129 69 BOX A The Patch Clamp Method 70 The Diversity of Ion Channels 73 BOX B Expression of Ion Channels in Xenopus Oocytes 75 Voltage-Gated Ion Channels 76 Ligand-Gated Ion Channels 78 Stretch- and Heat-Activated Channels 78 The Molecular Structure of Ion Channels 79 BOX C Toxins That Poison Ion Channels BOX D Diseases Caused by Altered Ion Channels Chapter 6 Neurotransmitters and Their 129 BOX A Addiction 134 BOX B Neurotoxins that Act on Postsynaptic Receptors 136 Glutamate 137 BOX C Myasthenia Gravis: An Autoimmune 82 84 Active Transporters Create and Maintain Ion Gradients 86 Functional Properties of the Na+/K+ Pump 87 The Molecular Structure of the Na+/K+ Pump 89 Summary 90 Chapter 5 Synaptic Transmission 93 Overview 93 Electrical Synapses 93 Signal Transmission at Chemical Synapses 96 Properties of Neurotransmitters 96 BOX A Criteria That Define a Neurotransmitter 99 Quantal Release of Neurotransmitters 102 Release of Transmitters from Synaptic Vesicles 103 Local Recycling of Synaptic Vesicles 105 The Role of Calcium in Transmitter Secretion 107 BOX B Diseases That Affect the Presynaptic Terminal 108 Molecular Mechanisms of Transmitter Secretion 110 Neurotransmitter Receptors 113 BOX C Toxins That Affect Transmitter Release 115 Postsynaptic Membrane Permeability Changes during Synaptic Transmission 116 Excitatory and Inhibitory Postsynaptic Potentials 121 Summation of Synaptic Potentials 123 Two Families of Postsynaptic Receptors 124 Summary 126 Disease of Neuromuscular Synapses 140 GABA and Glycine 143 BOX D Excitotoxicity Following Acute Brain Injury 145 The Biogenic Amines 147 BOX E Biogenic Amine Neurotransmitters and Psychiatric Disorders 148 ATP and Other Purines 152 Peptide Neurotransmitters 153 Unconventional Neurotransmitters 157 BOX F Marijuana and the Brain 160 Summary 161 Chapter 7 Molecular Signaling within Neurons 165 Overview 165 Strategies of Molecular Signaling 165 The Activation of Signaling Pathways 167 Receptor Types 168 G-Proteins and Their Molecular Targets 170 Second Messengers 172 Second Messenger Targets: Protein Kinases and Phosphatases 175 Nuclear Signaling 178 Examples of Neuronal Signal Transduction 181 Summary 184 x Contents Unit II SENSATION AND SENSORY PROCESSING Chapter 8 The Somatic Sensory System 189 Chapter 10 Vision: The Eye 229 Overview 189 Cutaneous and Subcutaneous Somatic Sensory Receptors 189 Mechanoreceptors Specialized to Receive Tactile Information 192 Differences in Mechanosensory Discrimination across the Body Surface 193 Overview 229 Anatomy of the Eye 229 The Formation of Images on the Retina BOX A Receptive Fields and Sensory Maps in the Cricket 195 BOX B Dynamic Aspects of Somatic Sensory Receptive Fields 196 Mechanoreceptors Specialized for Proprioception 197 Active Tactile Exploration 199 The Major Afferent Pathway for Mechanosensory Information: The Dorsal Column–Medial Lemniscus System 199 The Trigeminal Portion of the Mechanosensory System 202 BOX C Dermatomes 202 The Somatic Sensory Components of the Thalamus 203 The Somatic Sensory Cortex 203 Higher-Order Cortical Representations 206 BOX D Patterns of Organization within the Sensory Cortices: Brain Modules 207 Summary 208 Chapter 9 Pain 209 Overview 209 Nociceptors 209 Transduction of Nociceptive Signals 211 BOX A Capsaicin 212 Central Pain Pathways 213 BOX B Referred Pain 215 BOX C A Dorsal Column Pathway for Visceral Pain 218 Sensitization 220 BOX D Phantom Limbs and Phantom Pain 222 Descending Control of Pain Perception 224 The Placebo Effect 224 The Physiological Basis of Pain Modulation 225 Summary 227 231 BOX A Myopia and Other Refractive Errors 232 The Retina 234 Phototransduction 236 BOX B Retinitis Pigmentosa 239 Functional Specialization of the Rod and Cone Systems 240 BOX C Macular Degeneration 243 Anatomical Distribution of Rods and Cones 244 Cones and Color Vision 245 BOX D The Importance of Context in Color Perception 247 Retinal Circuits for Detecting Luminance Change 249 BOX E The Perception of Light Intensity 250 Contribution of Retinal Circuits to Light Adaptation 254 Summary 257 Chapter 11 Central Visual Pathways 259 Overview 259 Central Projections of Retinal Ganglion Cells 259 BOX A The Blind Spot 262 The Retinotopic Representation of the Visual Field 263 Visual Field Deficits 267 The Functional Organization of the Striate Cortex 269 The Columnar Organization of the Striate Cortex 271 BOX B Random Dot Stereograms and Related Amusements 272 Division of Labor within the Primary Visual Pathway 275 BOX C Optical Imaging of Functional Domains in the Visual Cortex 276 The Functional Organization of Extrastriate Visual Areas 278 Summary 281 Chapter 12 The Auditory System 283 Overview 283 Sound 283 The Audible Spectrum 284 Contents xi A Synopsis of Auditory Function 285 BOX A Four Causes of Acquired Hearing Loss 285 BOX B Music 286 The External Ear 287 The Middle Ear 289 The Inner Ear 289 BOX C Sensorineural Hearing Loss and Cochlear Implants 290 BOX D The Sweet Sound of Distortion 295 Hair Cells and the Mechanoelectrical Transduction of Sound Waves 294 Two Kinds of Hair Cells in the Cochlea 300 Tuning and Timing in the Auditory Nerve 301 How Information from the Cochlea Reaches Targets in the Brainstem 303 Integrating Information from the Two Ears 303 Monaural Pathways from the Cochlear Nucleus to the Lateral Lemniscus 307 Integration in the Inferior Colliculus 307 The Auditory Thalamus 308 The Auditory Cortex 309 BOX E Representing Complex Sounds in the Brains of Bats and Humans 310 Summary 313 How Semicircular Canal Neurons Sense Angular Accelerations 325 BOX C Throwing Cold Water on the Vestibular System 326 Central Pathways for Stabilizing Gaze, Head, and Posture 328 Vestibular Pathways to the Thalamus and Cortex 331 BOX D Mauthner Cells in Fish 332 Summary 333 Chapter 14 The Chemical Senses 337 Overview 337 The Organization of the Olfactory System 337 Olfactory Perception in Humans 339 Physiological and Behavioral Responses to Odorants 341 The Olfactory Epithelium and Olfactory Receptor Neurons 342 BOX A Olfaction, Pheromones, and Behavior in the Hawk Moth 344 The Transduction of Olfactory Signals 345 Odorant Receptors 346 Olfactory Coding 348 The Olfactory Bulb 350 BOX B Temporal “Coding” of Olfactory Chapter 13 The Vestibular System 315 Overview 315 The Vestibular Labyrinth 315 Vestibular Hair Cells 316 The Otolith Organs: The Utricle and Saccule 317 BOX A A Primer on Vestibular Navigation 318 BOX B Adaptation and Tuning of Vestibular Hair Cells 320 How Otolith Neurons Sense Linear Forces 322 The Semicircular Canals 324 Information in Insects 350 Central Projections of the Olfactory Bulb 353 The Organization of the Taste System 354 Taste Perception in Humans 356 Idiosyncratic Responses to Tastants 357 The Organization of the Peripheral Taste System 359 Taste Receptors and the Transduction of Taste Signals 360 Neural Coding in the Taste System 362 Trigeminal Chemoreception 363 Summary 366 Unit III MOVEMENT AND ITS CENTRAL CONTROL Chapter 15 Lower Motor Neuron Circuits and Motor Control 371 Overview 371 Neural Centers Responsible for Movement 371 Motor Neuron–Muscle Relationships 373 The Motor Unit 375 The Regulation of Muscle Force 377 The Spinal Cord Circuitry Underlying Muscle Stretch Reflexes 379 xii Contents The Influence of Sensory Activity on Motor Behavior 381 Other Sensory Feedback That Affects Motor Performance 382 BOX A Locomotion in the Leech and the Lamprey 384 Flexion Reflex Pathways 387 Spinal Cord Circuitry and Locomotion 387 BOX B The Autonomy of Central Pattern Generators: Evidence from the Lobster Stomatogastric Ganglion 388 The Lower Motor Neuron Syndrome 389 BOX C Amyotrophic Lateral Sclerosis 391 Summary 391 Chapter 16 Upper Motor Neuron Control of the Brainstem and Spinal Cord 393 Overview 393 Descending Control of Spinal Cord Circuitry: General Information 393 Motor Control Centers in the Brainstem: Upper Motor Neurons That Maintain Balance and Posture 397 BOX A The Reticular Formation 398 The Corticospinal and Corticobulbar Pathways: Upper Motor Neurons That Initiate Complex Voluntary Movements 402 BOX B Descending Projections to Cranial Nerve Motor Nuclei and Their Importance in Diagnosing the Cause of Motor Deficits 404 Functional Organization of the Primary Motor Cortex 405 BOX C What Do Motor Maps Represent? 408 The Premotor Cortex 411 BOX D Sensory Motor Talents and Cortical Space 410 Damage to Descending Motor Pathways: The Upper Motor Neuron Syndrome 412 BOX E Muscle Tone 414 Summary 415 Chapter 17 Modulation of Movement by the Basal Ganglia 417 Overview 417 Projections to the Basal Ganglia 417 Projections from the Basal Ganglia to Other Brain Regions 422 Evidence from Studies of Eye Movements 423 Circuits within the Basal Ganglia System 424 BOX A Huntington’s Disease 426 BOX B Parkinson’s Disease: An Opportunity for Novel Therapeutic Approaches 429 BOX C Basal Ganglia Loops and Non-Motor Brain Functions 432 Summary 433 Chapter 18 Modulation of Movement by the Cerebellum 435 Overview 435 Organization of the Cerebellum 435 Projections to the Cerebellum 438 Projections from the Cerebellum 440 Circuits within the Cerebellum 441 BOX A Prion Diseases 444 Cerebellar Circuitry and the Coordination of Ongoing Movement 445 Futher Consequences of Cerebellar Lesions 448 Summary 449 BOX B Genetic Analysis of Cerebellar Function 450 Chapter 19 Eye Movements and Sensory Motor Integration 453 Overview 453 What Eye Movements Accomplish 453 The Actions and Innervation of Extraocular Muscles 454 BOX A The Perception of Stabilized Retinal Images 456 Types of Eye Movements and Their Functions 457 Neural Control of Saccadic Eye Movements 458 BOX B Sensory Motor Integration in the Superior Colliculus 462 Neural Control of Smooth Pursuit Movements 466 Neural Control of Vergence Movements 466 Summary 467 Chapter 20 The Visceral Motor System 469 Overview 469 Early Studies of the Visceral Motor System 469 Distinctive Features of the Visceral Motor System 470 The Sympathetic Division of the Visceral Motor System 471 The Parasympathetic Division of the Visceral Motor System 476 The Enteric Nervous System 479 Sensory Components of the Visceral Motor System 480 Contents xiii Central Control of Visceral Motor Functions 483 BOX A The Hypothalamus 484 Neurotransmission in the Visceral Motor System 487 BOX B Horner’s Syndrome 488 BOX C Obesity and the Brain 490 Visceral Motor Reflex Functions 491 Autonomic Regulation of Cardiovascular Function 491 Autonomic Regulation of the Bladder 493 Autonomic Regulation of Sexual Function 496 Summary 498 Unit IV THE CHANGING BRAIN Chapter 21 Early Brain Development 501 Overview 501 The Initial Formation of the Nervous System: Gastrulation and Neurulation 501 The Molecular Basis of Neural Induction 503 BOX A Stem Cells: Promise and Perils 504 BOX B Retinoic Acid: Teratogen and Inductive Signal 506 Formation of the Major Brain Subdivisions 510 BOX C Homeotic Genes and Human Brain Development 513 BOX D Rhombomeres 514 Genetic Abnormalities and Altered Human Brain Development 515 The Initial Differentiation of Neurons and Glia 516 BOX E Neurogenesis and Neuronal Birthdating 517 The Generation of Neuronal Diversity 518 Neuronal Migration 520 BOX F Mixing It Up: Long-Distance Neuronal Migration 524 Summary 525 Chapter 22 Construction of Neural Circuits 528 BOX A Choosing Sides: Axon Guidance at the Optic Chiasm 530 Diffusible Signals for Axon Guidance: Chemoattraction and Repulsion 534 The Formation of Topographic Maps 537 Selective Synapse Formation 539 Formation 542 Trophic Interactions and the Ultimate Size of Neuronal Populations 543 Further Competitive Interactions in the Formation of Neuronal Connections 545 Molecular Basis of Trophic Interactions 547 BOX C Why Do Neurons Have Dendrites? 548 BOX D The Discovery of BDNF and the Neurotrophin Family 552 Neurotrophin Signaling 553 Summary 554 Chapter 23 Modification of Brain Circuits as a Result of Experience 557 Overview 557 Critical Periods 557 BOX A Built-In Behaviors 558 The Development of Language: Example of a Human Critical Period 559 BOX B Birdsong 560 Critical Periods in Visual System Development 562 Effects of Visual Deprivation on Ocular Dominance 563 BOX C Transneuronal Labeling with Radioactive 527 Overview 527 The Axonal Growth Cone 527 Non-Diffusible Signals for Axon Guidance BOX B Molecular Signals That Promote Synapse Amino Acids 564 Visual Deprivation and Amblyopia in Humans 568 Mechanisms by which Neuronal Activity Affects the Development of Neural Circuits 569 Cellular and Molecular Correlates of ActivityDependent Plasticity during Critical Periods 572 Evidence for Critical Periods in Other Sensory Systems 572 Summary 573 xiv Contents Chapter 24 Plasticity of Mature Synapses and Circuits 575 Overview 575 Synaptic Plasticity Underlies Behavioral Modification in Invertebrates 575 BOX A Genetics of Learning and Memory in the Fruit Fly 581 Short-Term Synaptic Plasticity in the Mammalian Nervous System 582 Long-Term Synaptic Plasticity in the Mammalian Nervous System 583 Long-Term Potentiation of Hippocampal Synapses 584 Molecular Mechanisms Underlying LTP 587 BOX B Dendritic Spines 590 Unit V Long-Term Synaptic Depression 592 BOX C Silent Synapses 594 Changes in Gene Expression Cause Enduring Changes in Synaptic Function during LTP and LTD 597 Plasticity in the Adult Cerebral Cortex 599 BOX D Epilepsy: The Effect of Pathological Activity on Neural Circuitry 600 Recovery from Neural Injury 602 Generation of Neurons in the Adult Brain 605 BOX E Why Aren’t We More Like Fish and Frogs? 606 Summary 609 COMPLEX BRAIN FUNCTIONS Chapter 25 The Association Cortices 613 Chapter 26 Language and Speech 637 Overview 613 The Association Cortices 613 An Overview of Cortical Structure 614 Specific Features of the Association Cortices 615 Overview 637 Language Is Both Localized and Lateralized 637 Aphasias 638 BOX A A More Detailed Look at Cortical Lamination 617 Lesions of the Parietal Association Cortex: Deficits of Attention 619 Lesions of the Temporal Association Cortex: Deficits of Recognition 622 Lesions of the Frontal Association Cortex: Deficits of Planning 623 BOX B Psychosurgery 625 “Attention Neurons” in the Monkey Parietal Cortex 626 “Recognition Neurons” in the Monkey Temporal Cortex 627 “Planning Neurons” in the Monkey Frontal Cortex 630 BOX C Neuropsychological Testing 632 BOX D Brain Size and Intelligence 634 Summary 635 BOX A Speech 640 BOX B Do Other Animals Have Language? 642 BOX C Words and Meaning 645 A Dramatic Confirmation of Language Lateralization 646 Anatomical Differences between the Right and Left Hemispheres 648 Mapping Language Functions 649 BOX D Language and Handedness 650 The Role of the Right Hemisphere in Language Sign Language 655 Summary 656 654 Chapter 27 Sleep and Wakefulness 659 Overview 659 Why Do Humans (and Many Other Animals) Sleep? 659 BOX A Styles of Sleep in Different Species 661 Contents xv The Circadian Cycle of Sleep and Wakefulness Stages of Sleep 665 662 BOX B Molecular Mechanisms of Biological Clocks 666 BOX C Electroencephalography 668 Physiological Changes in Sleep States 671 The Possible Functions of REM Sleep and Dreaming 671 Neural Circuits Governing Sleep 674 BOX D Consciousness 675 Thalamocortical Interactions 679 Sleep Disorders 681 BOX E Drugs and Sleep 682 Summary 684 Chapter 28 Emotions 687 BOX A Facial Expressions: Pyramidal and Extrapyramidal Contributions 690 693 BOX B The Anatomy of the Amygdala 696 The Importance of the Amygdala 697 BOX C The Reasoning Behind an Important Discovery 698 The Relationship between Neocortex and Amygdala 701 BOX D Fear and the Human Amygdala: A Case Study 702 BOX E Affective Disorders 704 Cortical Lateralization of Emotional Functions 705 Emotion, Reason, and Social Behavior 707 Summary 708 The Limbic System Chapter 29 Sex, Sexuality, and the Brain 711 733 BOX A Phylogenetic Memory 735 The Importance of Association in Information Storage 736 Forgetting 738 Brain Systems Underlying Declarative Memory Formation 741 BOX C Clinical Cases That Reveal the Anatomical Substrate for Declarative Memories 742 Brain Systems Underlying Long-Term Storage of Declarative Memory 746 Brain Systems Underlying Nondeclarative Learning and Memory 748 Memory and Aging 749 BOX D Alzheimer’s Disease 750 Summary 753 Appendix A The Brainstem and Cranial Nerves 755 Appendix B Vascular Supply, the Meninges, and the Ventricular System 763 The Blood Supply of the Brain and Spinal Cord 763 The Blood-Brain Barrier 714 715 BOX B The Case of Bruce/Brenda 716 The Effect of Sex Hormones on Neural Circuitry 718 766 BOX A Stroke 767 768 The Ventricular System BOX A The Development of Male and Female Phenotypes Overview 733 Qualitative Categories of Human Memory Temporal Categories of Memory 734 The Meninges 711 Hormonal Influences on Sexual Dimorphism Chapter 30 Memory 733 BOX B Savant Syndrome 739 Overview 687 Physiological Changes Associated with Emotion 687 The Integration of Emotional Behavior 688 Overview 711 Sexually Dimorphic Behavior What Is Sex? 712 BOX C The Actions of Sex Hormones 718 Other Central Nervous System Dimorphisms Specifically Related to Reproductive Behaviors 720 Brain Dimorphisms Related to Cognitive Function 728 Hormone-Sensitive Brain Circuits in Adult Animals 729 Summary 731 770 Glossary Illustration Source References Index Preface Whether judged in molecular, cellular, systemic, behavioral, or cognitive terms, the human nervous system is a stupendous piece of biological machinery. Given its accomplishments—all the artifacts of human culture, for instance—there is good reason for wanting to understand how the brain and the rest of the nervous system works. The debilitating and costly effects of neurological and psychiatric disease add a further sense of urgency to this quest. The aim of this book is to highlight the intellectual challenges and excitement—as well as the uncertainties—of what many see as the last great frontier of biological science. The information presented should serve as a starting point for undergraduates, medical students, graduate students in the neurosciences, and others who want to understand how the human nervous system operates. Like any other great challenge, neuroscience should be, and is, full of debate, dissension, and considerable fun. All these ingredients have gone into the construction of the third edition of this book; we hope they will be conveyed in equal measure to readers at all levels. Acknowledgments We are grateful to numerous colleagues who provided helpful contributions, criticisms and suggestions to this and previous editions. We particularly wish to thank Ralph Adolphs, David Amaral, Eva Anton, Gary Banker, Bob Barlow, Marlene Behrmann, Ursula Bellugi, Dan Blazer, Bob Burke, Roberto Cabeza, Nell Cant, Jim Cavanaugh, John Chapin, Milt Charlton, Michael Davis, Rob Deaner, Bob Desimone, Allison Doupe, Sasha du Lac, Jen Eilers, Anne Fausto-Sterling, Howard Fields, Elizabeth Finch, Nancy Forger, Jannon Fuchs, Michela Gallagher, Dana Garcia, Steve George, the late Patricia Goldman-Rakic, Mike Haglund, Zach Hall, Kristen Harris, Bill Henson, John Heuser, Jonathan Horton, Ron Hoy, Alan Humphrey, Jon Kaas, Jagmeet Kanwal, Herb Killackey, Len Kitzes, Arthur Lander, Story Landis, Simon LeVay, Darrell Lewis, Jeff Lichtman, Alan Light, Steve Lisberger, Donald Lo, Arthur Loewy, Ron Mangun, Eve Marder, Robert McCarley, Greg McCarthy, Jim McIlwain, Chris Muly, Vic Nadler, Ron Oppenheim, Larysa Pevny, Michael Platt, Franck Polleux, Scott Pomeroy, Rodney Radtke, Louis Reichardt, Marnie Riddle, Jamie Roitman, Steve Roper, John Rubenstein, Ben Rubin, David Rubin, Josh Sanes, Cliff Saper, Lynn Selemon, Carla Shatz, Bill Snider, Larry Squire, John Staddon, Peter Strick, Warren Strittmatter, Joe Takahashi, Richard Weinberg, Jonathan Weiner, Christina Williams, Joel Winston, and Fulton Wong. It is understood, of course, that any errors are in no way attributable to our critics and advisors. We also thank the students at Duke University Medical School as well as many other students and colleagues who provided suggestions for improvement of the last edition. Finally, we owe special thanks to Robert Reynolds and Nate O’Keefe, who labored long and hard to put the third edition together, and to Andy Sinauer, Graig Donini, Carol Wigg, Christopher Small, Janice Holabird, and the rest of the staff at Sinauer Associates for their outstanding work and high standards. Supplements to Accompany NEUROSCIENCE Third Edition For the Student Sylvius for Neuroscience: A Visual Glossary of Human Neuroanatomy (CD-ROM) S. Mark Williams, Leonard E. White, and Andrew C. Mace Sylvius for Neuroscience: A Visual Glossary of Human Neuroanatomy, included in every copy of the textbook, is an interactive CD reference guide to the structure of the human nervous system. By entering a corresponding page number from the textbook, students can quickly search the CD for any neuroanatomical structure or term and view corresponding images and animations. Descriptive information is provided with all images and animations. Additionally, students can take notes on the content and share these with other Sylvius users. Sylvius is an essential study aid for learning basic human neuroanatomy. Sylvius for Neuroscience features: • Over 400 neuroanatomical structures and terms. • High-resolution images. • Animations of pathways and 3-D reconstructions. • Definitions and descriptions. • Audio pronunciations. • A searchable glossary. • Categories of anatomical structures and terms (e.g., cranial nerves, spinal cord tracts, lobes, cortical areas, etc.), that can be easily browsed. In addition, structures can be browsed by textbook chapter. Supplements xix • Images and text relevant to the textbook: Icons in the textbook indicate specific content on the CD. By entering a textbook page number, students can automatically load the relevant images and text. • A history feature that allows the student to quickly reload recently viewed structures. • A bookmark feature that adds bookmarks to structures of interest; bookmarks are automatically stored on the student’s computer. • A notes feature that allows students to type notes for any selected structure; notes are automatically saved on the student’s computer and can be shared among students and instructors (i.e., imported and exported). • A self-quiz mode that allows for testing on structure identification and functional information. • A print feature that formats images and text for printed output. • An image zoom tool. For the Instructor Instructor’s Resource CD (ISBN 0-87893-750-1) This expanded resource includes all the figures and tables from the textbook in JPEG format, reformatted and relabeled for optimal readability. Also included are ready-to-use PowerPoint® presentations of all figures and tables. In addition, new for the Third Edition, the Instructor’s Resource CD includes a set of short-answer study questions for each chapter in Microsoft® Word® format. Overhead Transparencies (ISBN 0-87893-751-X) This set includes 100 illustrations (approximately 150 transparencies), selected from throughout the textbook for teaching purposes. These are relabeled and optimized for projection in classrooms. Chapter 1 Overview Neuroscience encompasses a broad range of questions about how nervous systems are organized, and how they function to generate behavior. These questions can be explored using the analytical tools of genetics, molecular and cell biology, systems anatomy and physiology, behavioral biology, and psychology. The major challenge for a student of neuroscience is to integrate the diverse knowledge derived from these various levels of analysis into a more or less coherent understanding of brain structure and function (one has to qualify this statement because so many questions remain unanswered). Many of the issues that have been explored successfully concern how the principal cells of any nervous system—neurons and glia—perform their basic functions in anatomical, electrophysiological, and molecular terms. The varieties of neurons and supporting glial cells that have been identified are assembled into ensembles called neural circuits, and these circuits are the primary components of neural systems that process specific types of information. Neural systems comprise neurons and circuits in a number of discrete anatomical locations in the brain. These systems subserve one of three general functions. Sensory systems represent information about the state of the organism and its environment, motor systems organize and generate actions; and associational systems link the sensory and motor sides of the nervous system, providing the basis for “higher-order” functions such as perception, attention, cognition, emotions, rational thinking, and other complex brain functions that lie at the core of understanding human beings, their history and their future. Genetics, Genomics, and the Brain The recently completed sequencing of the genome in humans, mice, the fruit fly Drosophila melanogaster, and the nematode worm Caenorhabditis elegans is perhaps the logical starting point for studying the brain and the rest of the nervous system; after all, this inherited information is also the starting point of each individual organism. The relative ease of obtaining, analyzing, and correlating gene sequences with neurobiological observations has facilitated a wealth of new insights into the basic biology of the nervous system. In parallel with studies of normal nervous systems, the genetic analysis of human pedigrees with various brain diseases has led to a widespread sense that it will soon be possible to understand and treat disorders long considered beyond the reach of science and medicine. A gene consists of DNA sequences called exons that are transcribed into a messenger RNA and subsequently a protein. The set of exons that defines 1 Studying the Nervous Systems of Humans and Other Animals 2 Chapter One Figure 1.1 Estimates of the number of genes in the human genome, as well as in the genomes of the mouse, the fruit fly Drosophila melanogaster, and the nematode worm Caenorhabditis elegans. Human Mouse D. melanogaster C. elegans 0 10,000 20,000 30,000 Number of genes 40,000 50,000 the transcript of any gene is flanked by upstream (or 5′) and downstream (or 3′) regulatory sequences that control gene expression. In addition, sequences between exons—called introns—further influence transcription. Of the approximately 35,000 genes in the human genome, a majority are expressed in the developing and adult brain; the same is true in mice, flies, and worms—the species commonly used in modern genetics (and increasingly in neuroscience) (Figure 1.1). Nevertheless, very few genes are uniquely expressed in neurons, indicating that nerve cells share most of the basic structural and functional properties of other cells. Accordingly, most “brainspecific” genetic information must reside in the remainder of nucleic acid sequences—regulatory sequences and introns—that control the timing, quantity, variability and cellular specificity of gene expression. One of the most promising dividends of sequencing the human genome has been the realization that one or a few genes, when altered (mutated), can begin to explain some aspects of neurological and psychiatric diseases. Before the “postgenomic era” (which began following completion of the sequencing of the human genome), many of the most devastating brain diseases remained largely mysterious because there was little sense of how or why the normal biology of the nervous system was compromised. The identification of genes correlated with disorders such as Huntington’s disease, Parkinson’s disease, Alzheimer’s disease, major depression, and schizophrenia has provided a promising start to understanding these pathological processes in a much deeper way (and thus devising rational therapies). Genetic and genomic information alone do not completely explain how the brain normally works or how disease processes disrupt its function. To achieve these goals it is equally essential to understand the cell biology, anatomy, and physiology of the brain in health as well as disease. The Cellular Components of the Nervous System Early in the nineteenth century, the cell was recognized as the fundamental unit of all living organisms. It was not until well into the twentieth century, however, that neuroscientists agreed that nervous tissue, like all other organs, is made up of these fundamental units. The major reason was that the first generation of “modern” neurobiologists in the nineteenth century had difficulty resolving the unitary nature of nerve cells with the microscopes and cell staining techniques that were then available. This inade- Studying the Ner vous Systems of Humans and O ther Animals 3 (A) Neurons in mesencephalic nucleus of cranial nerve V (B) Retinal bipolar cell Dendrites (D) Retinal amacrine cell (C) Retinal ganglion cell Dendrites Dendrites Cell bodies Axons Cell body Cell body Axon Axon Cell body * (E) Cortical pyramidal cell (F) Cerebellar Purkinje cells * Dendrites Dendrites Cell body Cell body Axon * quacy was exacerbated by the extraordinarily complex shapes and extensive branches of individual nerve cells, which further obscured their resemblance to the geometrically simpler cells of other tissues (Figures 1.2–1.4). As a result, some biologists of that era concluded that each nerve cell was connected to its neighbors by protoplasmic links, forming a continuous nerve cell network, or reticulum. The “reticular theory” of nerve cell communication, which was championed by the Italian neuropathologist Camillo Golgi (for whom the Golgi apparatus in cells is named), eventually fell from favor and was replaced by what came to be known as the “neuron doctrine.” The major proponents of this new perspective were the Spanish neuroanatomist Santiago Ramón y Cajal and the British physiologist Charles Sherrington. The contrasting views represented by Golgi and Cajal occasioned a spirited debate in the early twentieth century that set the course of modern neuroscience. Based on light microscopic examination of nervous tissue stained with silver salts according to a method pioneered by Golgi, Cajal argued persuasively that nerve cells are discrete entities, and that they communicate Axon * Figure 1.2 Examples of the rich variety of nerve cell morphologies found in the human nervous system. Tracings are from actual nerve cells stained by impregnation with silver salts (the socalled Golgi technique, the method used in the classical studies of Golgi and Cajal). Asterisks indicate that the axon runs on much farther than shown. Note that some cells, like the retinal bipolar cell, have a very short axon, and that others, like the retinal amacrine cell, have no axon at all. The drawings are not all at the same scale. 4 Chapter One with one another by means of specialized contacts that Sherrington called “synapses.” The work that framed this debate was recognized by the award of the Nobel Prize for Physiology or Medicine in 1906 to both Golgi and Cajal ( the joint award suggests some ongoing concern about just who was correct, despite Cajal’s overwhelming evidence). The subsequent work of Sherrington and others demonstrating the transfer of electrical signals at synaptic junctions between nerve cells provided strong support of the “neuron doctrine,” but challenges to the autonomy of individual neurons remained. It was not until the advent of electron microscopy in the 1950s that any lingering doubts about the discreteness of neurons were resolved. The high-magnification, high-resolution pictures that could be obtained with the electron microscope clearly established that nerve cells are functionally independent units; such pictures also identified the specialized cellular junctions that Sherrington had named synapses (see Figures 1.3 and 1.4). The histological studies of Cajal, Golgi, and a host of successors led to the further consensus that the cells of the nervous system can be divided into two broad categories: nerve cells (or neurons), and supporting cells called neuroglia (or simply glia; see Figure 1.5). Nerve cells are specialized for electrical signaling over long distances, and understanding this process represents one of the more dramatic success stories in modern biology (and the subject of Unit I of this book). Supporting cells, in contrast, are not capable of electrical signaling; nevertheless, they have several essential functions in the developing and adult brain. Neurons Neurons and glia share the complement of organelles found in all cells, including the endoplasmic reticulum and Golgi apparatus, mitochondria, and a variety of vesicular structures. In neurons, however, these organelles are often more prominent in distinct regions of the cell. In addition to the distribution of organelles and subcellular components, neurons and glia are in some measure different from other cells in the specialized fibrillar or tubular proteins that constitute the cytoskeleton (Figures 1.3 and 1.4). Although many of these proteins—isoforms of actin, tubulin, and myosin, as well as several others—are found in other cells, their distinctive organization in neurons is critical for the stability and function of neuronal processes and synaptic junctions. The filaments, tubules, vesicular motors, and scaffolding proteins of neurons orchestrate the growth of axons and dendrites; the trafficking and appropiate positioning of membrane components, organelles, and vesicles; and the active processes of exocytosis and endocytosis that underlie synaptic communication. Understanding the ways in which these molecular components are used to insure the proper development and function of neurons and glia remains a primary focus of modern neurobiology. The basic cellular organization of neurons resembles that of other cells; however, they are clearly distinguished by specialization for intercellular communication. This attribute is apparent in their overall morphology, in the specific organization of their membrane components for electrical signaling, and in the structural and functional intricacies of the synaptic contacts between neurons (see Figures 1.3 and 1.4). The most obvious sign of neuronal specialization for communication via electrical signaling is the extensive branching of neurons. The most salient aspect of this branching for typical nerve cells is the elaborate arborization of dendrites that arise from the neuronal cell body (also called dendritic branches or dendritic processes). Dendrites are the primary target for synaptic input from other neurons and are Studying the Ner vous Systems of Humans and O ther Animals 5 (A) (B) Axon (C) Synaptic endings (terminal boutons) Endoplasmic F reticulum Mitochondrion E Nucleus Dendrite Soma Golgi apparatus (D) Myelinated axons C B Ribosomes Axons G D (E) Dendrites (F) Neuronal cell body (soma) (G) Myelinated axon and node of Ranvier Figure 1.3 The major light and electron microscopical features of neurons. (A) Diagram of nerve cells and their component parts. (B) Axon initial segment (blue) entering a myelin sheath (gold). (C) Terminal boutons (blue) loaded with synaptic vesicles (arrowheads) forming synapses (arrows) with a dendrite (purple). (D) Transverse section of axons (blue) ensheathed by the processes of oligodendrocytes (gold). (E) Apical dendrites (purple) of cortical pyramidal cells. (F) Nerve cell bodies (purple) occupied by large round nuclei. (G) Portion of a myelinated axon (blue) illustrating the intervals between adjacent segments of myelin (gold) referred to as nodes of Ranvier (arrows). (Micrographs from Peters et al., 1991.) 6 Chapter One Figure 1.4 Distinctive arrangement of cytoskeletal elements in neurons. (A) The cell body, axons, and dendrites are distinguished by the distribution of tubulin (green throughout cell) versus other cytoskeletal elements—in this case, Tau (red), a microtubule-binding protein found only in axons. (B) The strikingly distinct localization of actin (red) to the growing tips of axonal and dendritic processes is shown here in cultured neuron taken from the hippocampus. (C) In contrast, in a cultured epithelial cell, actin (red) is distributed in fibrils that occupy most of the cell body. (D) In astroglial cells in culture, actin (red) is also seen in fibrillar bundles. (E) Tubulin (green) is seen throughout the cell body and dendrites of neurons. (F) Although tubulin is a major component of dendrites, extending into spines, the head of the spine is enriched in actin (red). (G) The tubulin component of the cytoskeleton in nonneuronal cells is arrayed in filamentous networks. (H–K) Synapses have a distinct arrangement of cytoskeletal elements, receptors, and scaffold proteins. (H) Two axons (green; tubulin) from motor neurons are seen issuing two branches each to four muscle fibers. The red shows the clustering of postsynaptic receptors (in this case for the neurotransmitter acetylcholine). (I) A higher power view of a single motor neuron synapse shows the relationship between the axon (green) and the postsynaptic receptors (red). (J) The extracellular space between the axon and its target muscle is shown in green. (K) The clustering of scaffolding proteins (in this case, dystrophin) that localize receptors and link them to other cytoskeletal elements is shown in green. (A courtesy of Y. N. Jan; B courtesy of E. Dent and F. Gertler; C courtesy of D. Arneman and C. Otey; D courtesy of A. Gonzales and R. Cheney; E from Sheng, 2003; F from Matus, 2000; G courtesy of T. Salmon et al.; H–K courtesy of R. Sealock.) (A) (B) (C) (D) (E) (G) (F) (H) (J) (I) (K) Studying the Ner vous Systems of Humans and O ther Animals 7 also distinguished by their high content of ribosomes as well as specific cytoskeletal proteins that reflect their function in receiving and integrating information from other neurons. The spectrum of neuronal geometries ranges from a small minority of cells that lack dendrites altogether to neurons with dendritic arborizations that rival the complexity of a mature tree (see Figure 1.2). The number of inputs that a particular neuron receives depends on the complexity of its dendritic arbor: nerve cells that lack dendrites are innervated by (thus, receive electrical signals from) just one or a few other nerve cells, whereas those with increasingly elaborate dendrites are innervated by a commensurately larger number of other neurons. The synaptic contacts made on dendrites (and, less frequently, on neuronal cell bodies) comprise a special elaboration of the secretory apparatus found in most polarized epithelial cells. Typically, the presynaptic terminal is immediately adjacent to a postsynaptic specialization of the target cell (see Figure 1.3). For the majority of synapses, there is no physical continuity between these pre- and postsynaptic elements. Instead, pre- and postsynaptic components communicate via secretion of molecules from the presynaptic terminal that bind to receptors in the postsynaptic specialization. These molecules must traverse an interval of extracellular space between pre- and postsynaptic elements called the synaptic cleft. The synaptic cleft, however, is not simply a space to be traversed; rather, it is the site of extracellular proteins that influence the diffusion, binding, and degradation of molecules secreted by the presynaptic terminal (see Figure 1.4). The number of synaptic inputs received by each nerve cell in the human nervous system varies from 1 to about 100,000. This range reflects a fundamental purpose of nerve cells, namely to integrate information from other neurons. The number of synaptic contacts from different presynaptic neurons onto any particular cell is therefore an especially important determinant of neuronal function. The information conveyed by synapses on the neuronal dendrites is integrated and “read out” at the origin of the axon, the portion of the nerve cell specialized for signal conduction to the next site of synaptic interaction (see Figures 1.2 and 1.3). The axon is a unique extension from the neuronal cell body that may travel a few hundred micrometers (µm; usually called microns) or much farther, depending on the type of neuron and the size of the species. Moreover, the axon also has a distinct cytoskeleton whose elements are central for its functional integrity (see Figure 1.4). Many nerve cells in the human brain (as well as that of other species) have axons no more than a few millimeters long, and a few have no axons at all. Relatively short axons are a feature of local circuit neurons or interneurons throughout the brain. The axons of projection neurons, however, extend to distant targets. For example, the axons that run from the human spinal cord to the foot are about a meter long. The electrical event that carries signals over such distances is called the action potential, which is a self-regenerating wave of electrical activity that propagates from its point of initiation at the cell body (called the axon hillock) to the terminus of the axon where synaptic contacts are made. The target cells of neurons include other nerve cells in the brain, spinal cord, and autonomic ganglia, and the cells of muscles and glands throughout the body. The chemical and electrical process by which the information encoded by action potentials is passed on at synaptic contacts to the next cell in a pathway is called synaptic transmission. Presynaptic terminals (also called synaptic endings, axon terminals, or terminal boutons) and their postsynaptic specializations are typically chemical synapses, the most abundant type of 8 Chapter One synapse in the nervous system. Another type, the electrical synapse, is far more rare (see Chapter 5). The secretory organelles in the presynaptic terminal of chemical synapses are synaptic vesicles (see Figure 1.3), which are generally spherical structures filled with neurotransmitter molecules. The positioning of synaptic vesicles at the presynaptic membrane and their fusion to initiate neurotransmitter release is regulated by a number of proteins either within or associated with the vesicle. The neurotransmitters released from synaptic vesicles modify the electrical properties of the target cell by binding to neurotransmitter receptors (Figure 1.4), which are localized primarily at the postsynaptic specialization. The intricate and concerted activity of neurotransmitters, receptors, related cytoskeletal elements, and signal transduction molecules are thus the basis for nerve cells communicating with one another, and with effector cells in muscles and glands. Figure 1.5 Varieties of neuroglial cells. Tracings of an astrocyte (A), an oligodendrocyte (B), and a microglial cell (C) visualized using the Golgi method. The images are at approximately the same scale. (D) Astrocytes in tissue culture, labeled (red) with an antibody against an astrocyte-specific protein. (E) Oligodendroglial cells in tissue culture labeled with an antibody against an oligodendroglial-specific protein. (F) Peripheral axon are ensheathed by myelin (labeled red) except at a distinct region called the node of Ranvier. The green label indicates ion channels concentrated in the node; the blue label indicates a molecularly distinct region called the paranode. (G) Microglial cells from the spinal cord, labeled with a cell type-specific antibody. Inset: Higher-magnification image of a single microglial cell labeled with a macrophage-selective marker. (A–C after Jones and Cowan, 1983; D, E courtesy of A.-S. LaMantia; F courtesy of M. Bhat; G courtesy of A. Light; inset courtesy of G. Matsushima.) (D) (E) Neuroglial Cells Neuroglial cells—also referred to as glial cells or simply glia—are quite different from nerve cells. Glia are more numerous than neurons in the brain, outnumbering them by a ratio of perhaps 3 to 1. The major distinction is that glia do not participate directly in synaptic interactions and electrical signaling, although their supportive functions help define synaptic contacts and maintain the signaling abilities of neurons. Although glial cells also have complex processes extending from their cell bodies, these are generally less prominent than neuronal branches, and do not serve the same purposes as axons and dendrites (Figure 1.5). (A) Astrocyte (C) Microglial cell (B) Oligodendrocyte Cell body (F) Glial processes (G) Studying the Ner vous Systems of Humans and O ther Animals 9 The term glia (from the Greek word meaning “glue”) reflects the nineteenth-century presumption that these cells held the nervous system together in some way. The word has survived, despite the lack of any evidence that binding nerve cells together is among the many functions of glial cells. Glial roles that are well-established include maintaining the ionic milieu of nerve cells, modulating the rate of nerve signal propagation, modulating synaptic action by controlling the uptake of neurotransmitters at or near the synaptic cleft, providing a scaffold for some aspects of neural development, and aiding in (or impeding, in some instances) recovery from neural injury. There are three types of glial cells in the mature central nervous system: astrocytes, oligodendrocytes, and microglial cells (see Figure 1.5). Astrocytes, which are restricted to the brain and spinal cord, have elaborate local processes that give these cells a starlike appearance (hence the prefix “astro”). A major function of astrocytes is to maintain, in a variety of ways, an appropriate chemical environment for neuronal signaling. Oligodendrocytes, which are also restricted to the central nervous system, lay down a laminated, lipid-rich wrapping called myelin around some, but not all, axons. Myelin has important effects on the speed of the transmission of electrical signals (see Chapter 3). In the peripheral nervous system, the cells that elaborate myelin are called Schwann cells. Finally, microglial cells are derived primarily from hematopoietic precursor cells (although some may be derived directly from neural precursor cells). They share many properties with macrophages found in other tissues, and are primarily scavenger cells that remove cellular debris from sites of injury or normal cell turnover. In addition, microglia, like their macrophage counterparts, secrete signaling molecules—particularly a wide range of cytokines that are also produced by cells of the immune system—that can modulate local inflammation and influence cell survival or death. Indeed, some neurobiologists prefer to categorize microglia as a type of macrophage. Following brain damage, the number of microglia at the site of injury increases dramatically. Some of these cells proliferate from microglia resident in the brain, while others come from macrophages that migrate to the injured area and enter the brain via local disruptions in the cerebral vasculature. Cellular Diversity in the Nervous System Although the cellular constituents of the human nervous system are in many ways similar to those of other organs, they are unusual in their extraordinary numbers: the human brain is estimated to contain 100 billion neurons and several times as many supporting cells. More importantly, the nervous system has a greater range of distinct cell types—whether categorized by morphology, molecular identity, or physiological activity—than any other organ system (a fact that presumably explains why so many different genes are expressed in the nervous system; see above). The cellular diversity of any nervous system—including our own—undoubtedly underlies the the capacity of the system to form increasingly complicated networks to mediate increasingly sophisticated behaviors. For much of the twentieth century, neuroscientists relied on the same set of techniques developed by Cajal and Golgi to describe and categorize the diversity of cell types in the nervous system. From the late 1970s onward, however, new technologies made possible by the advances in cell and molecular biology provided investigators with many additional tools to discern the properties of neurons (Figure 1.6). Whereas general cell staining methods 10 Chapter One (A) (B) (C) (D) (E) (F) (G) (H) (I) (J) (K) (L) (M) (N) (O) (P) showed mainly differences in cell size and distribution, antibody stains and probes for messenger RNA added greatly to the appreciation of distinctive types of neurons and glia in various regions of the nervous system. At the same time, new tract tracing methods using a wide variety of tracing substances allowed the interconnections among specific groups of neurons to be ▲ Studying the Ner vous Systems of Humans and O ther Animals 11 Figure 1.6 Structural diversity in the nervous system demonstrated with cellular and molecular markers. First row: Cellular organization of different brain regions demonstrated with Nissl stains, which label nerve and glial cell bodies. (A) The cerebral cortex at the boundary between the primary and secondary visual areas. (B) The olfactory bulbs. (C) Differences in cell density in cerebral cortical layers. (D) Individual Nissl-stained neurons and glia at higher magnification. Second row: Classical and modern approaches to seeing individual neurons and their processes. (E) Golgi-labeled cortical pyramidal cells. (F) Golgi-labeled cerebellar Purkinje cells. (G) Cortical interneuron labeled by intracellular injection of a fluorescent dye. (H) Retinal neurons labeled by intracellular injection of fluorescent dye. Third row: Cellular and molecular approaches to seeing neural connections and systems. (I) At top, an antibody that detects synaptic proteins in the olfactory bulb; at bottom, a fluorescent label shows the location of cell bodies. (J) Synaptic zones and the location of Purkinje cell bodies in the cerebellar cortex labeled with synapse-specific antibodies (green) and a cell body marker (blue). (K) The projection from one eye to the lateral geniculate nucleus in the thalamus, traced with radioactive amino acids (the bright label shows the axon terminals from the eye in distinct layers of the nucleus). (L) The map of the body surface of a rat in the somatic sensory cortex, shown with a marker that distinguishes zones of higher synapse density and metabolic activity. Fourth row: Peripheral neurons and their projections. (M) An autonomic neuron labeled by intracellular injection of an enzyme marker. (N) Motor axons (green) and neuromuscular synapses (orange) in transgenic mice genetically engineered to express fluorescent proteins. (O) The projection of dorsal root ganglia to the spinal cord, demonstrated by an enzymatic tracer. (P) Axons of olfactory receptor neurons from the nose labeled in the olfactory bulb with a vital fluorescent dye. (G courtesy of L. C. Katz; H courtesy of C. J. Shatz; N,O courtesy of W. Snider and J. Lichtman; all others courtesy of A.-S. LaMantia and D. Purves.) explored much more fully. Tracers can be introduced into either living or fixed tissue, and are transported along nerve cell processes to reveal their origin and termination. More recently, genetic and neuroanatomical methods have been combined to visualize the expression of fluorescent or other tracer molecules under the control of regulatory sequences of neural genes. This approach, which shows individual cells in fixed or living tissue in remarkable detail, allows nerve cells to be identified by both their transcriptional state and their structure. Finally, ways of determining the molecular identity and morphology of nerve cells can be combined with measurements of their physiological activity, thus illuminating structure–function relationships. Examples of these various approaches are shown in Figure 1.6. Neural Circuits Neurons never function in isolation; they are organized into ensembles or neural circuits that process specific kinds of information and provide the foundation of sensation, perception and behavior. The synaptic connections that define such circuits are typically made in a dense tangle of dendrites, axons terminals, and glial cell processes that together constitute what is called neuropil (the suffix -pil comes from the Greek word pilos, meaning “felt”; see Figure 1.3). The neuropil is thus the region between nerve cell bodies where most synaptic connectivity occurs. Although the arrangement of neural circuits varies greatly according to the function being served, some features are characteristic of all such ensembles. Preeminent is the direction of information flow in any particular circuit, which is obviously essential to understanding its purpose. Nerve cells that 12 Chapter One Extensor muscle Sensory (afferent) axon Muscle sensory receptor 3A 2B 2A 1 Flexor muscle Hammer tap stretches tendon, which, in turn, stretches sensory receptors in leg extensor muscle Interneuron Motor (efferent) axons 4 1 3B 2 (A) Sensory neuron synapses with and excites motor neuron in the spinal cord (B) Sensory neuron also excites spinal interneuron (C) Interneuron synapse inhibits motor neuron to flexor muscles Figure 1.7 A simple reflex circuit, the knee-jerk response (more formally, the myotatic reflex), illustrates several points about the functional organization of neural circuits. Stimulation of peripheral sensors (a muscle stretch receptor in this case) initiates receptor potentials that trigger action potentials that travel centrally along the afferent axons of the sensory neurons. This information stimulates spinal motor neurons by means of synaptic contacts. The action potentials triggered by the synaptic potential in motor neurons travel peripherally in efferent axons, giving rise to muscle contraction and a behavioral response. One of the purposes of this particular reflex is to help maintain an upright posture in the face of unexpected changes. 2C 3 (A) Motor neuron conducts action potential to synapses on extensor muscle fibers, causing contraction 4 Leg extends (B) Flexor muscle relaxes because the activity of its motor neurons has been inhibited carry information toward the brain or spinal cord (or farther centrally within the spinal cord and brain) are called afferent neurons; nerve cells that carry information away from the brain or spinal cord (or away from the circuit in question) are called efferent neurons. Interneurons or local circuit neurons only participate in the local aspects of a circuit, based on the short distances over which their axons extend. These three functional classes—afferent neurons, efferent neurons, and interneurons—are the basic constituents of all neural circuits. A simple example of a neural circuit is the ensemble of cells that subserves the myotatic spinal reflex (the “knee-jerk” reflex; Figure 1.7). The afferent neurons of the reflex are sensory neurons whose cell bodies lie the dorsal root ganglia and whose peripheral axons terminate in sensory endings in skeletal muscles (the ganglia that serve this same of function for much of the head and neck are called cranial nerve ganglia; see Appendix A). The central axons of these afferent sensory neurons enter the the spinal cord where they terminate on a variety of central neurons concerned with the regualtion of muscle tone, most obviously the motor neurons that determine the activity of the related muscles. These neurons constitute the efferent neurons as well as interneurons of the circuit. One group of these efferent neurons in the ventral horn of the spinal cord projects to the flexor muscles in the limb, and the other to extensor muscles. Spinal cord interneurons are the third element of this circuit. The interneurons receive synaptic contacts from sensory afferent neurons and make synapses on the efferent motor neurons that project to the Studying the Ner vous Systems of Humans and O ther Animals 13 Hammer tap Sensory (afferent) axon Sensory neuron Motor neuron (extensor) Motor (efferent) axons Interneuron Interneuron Motor neuron (flexor) Leg extends flexor muscles; therefore they are capable of modulating the input–output linkage. The excitatory synaptic connections between the sensory afferents and the extensor efferent motor neurons cause the extensor muscles to contract; at the same time, the interneurons activated by the afferents are inhibitory, and their activation diminishes electrical activity in flexor efferent motor neurons and causes the flexor muscles to become less active (Figure 1.8). The result is a complementary activation and inactivation of the synergist and antagonist muscles that control the position of the leg. A more detailed picture of the events underlying the myotatic or any other circuit can be obtained by electrophysiological recording (Figure 1.9). There are two basic approaches to measuring the electrical activity of a nerve cell: extracellular recording (also referred to as single-unit recording), where an electrode is placed near the nerve cell of interest to detect its activity; and intracellular recording, where the electrode is placed inside the cell. Extracellular recordings primarily detect action potentials, the all-or-nothing changes in the potential across nerve cell membranes that convey information from one point to another in the nervous system. This sort of recording is particularly useful for detecting temporal patterns of action potential activity and relating those patterns to stimulation by other inputs, or to specific behavioral events. Intracellular recordings can detect the smaller, graded potential changes that trigger action potentials, and thus allow a more detailed analysis of communication between neurons within a circuit. These graded triggering potentials can arise at either sensory receptors or synapses and are called receptor potentials or synaptic potentials, respectively. For the myotatic circuit, electrical activity can be measured both extracellularly and intracellularly, thus defining the functional relationships between neurons in the circuit. The pattern of action potential activity can be measured for each element of the circuit (afferents, efferents, and interneurons) before, during, and after a stimulus (see Figure 1.8). By comparing the onset, duration, and frequency of action potential activity in each cell, a functional picture of the circuit emerges. As a result of the stimulus, the sensory neuron is triggered to fire at higher frequency (i.e., more action potentials per unit time). This increase triggers a higher frequency of action potentials in both the extensor motor neurons and the interneurons. Concurrently, the inhibitory synapses made by the interneurons onto the flexor motor neurons cause the frequency of action potentials in these cells to decline. Using intracellular recording, it is possible to observe directly the potential changes underlying the synaptic connections of the myotatic reflex circuit (see Figure 1.9). Figure 1.8 Relative frequency of action potentials (indicated by individual vertical lines) in different components of the myotatic reflex as the reflex pathway is activated. Notice the modulatory effect of the interneuron. 14 Chapter One Sensory neuron Record Interneuron Action potential Membrane potential (mV) Record (A) Sensory neuron Microelectrode to measure membrane potential Record (B) Motor neuron (extensor) Membrane potential (mV) Motor neuron (extensor) Record Action potential Synaptic potential Activate excitatory synapse (C) Interneuron Membrane potential (mV) Motor neuron (flexor) Action potential Synaptic potential Activate excitatory synapse (D) Motor neuron (flexor) Membrane potential (mV) Figure 1.9 Intracellularly recorded responses underlying the myotatic reflex. (A) Action potential measured in a sensory neuron. (B) Postsynaptic triggering potential recorded in an extensor motor neuron. (C) Postsynaptic triggering potential in an interneuron. (D) Postsynaptic inhibitory potential in a flexor motor neuron. Such intracellular recordings are the basis for understanding the cellular mechanisms of action potential generation, and the sensory receptor and synaptic potentials that trigger these conducted signals. Activate inhibitory synapse Time (ms) Overall Organization of the Human Nervous System When considered together, circuits that process similar types of information comprise neural systems that serve broader behavioral purposes. The most general functional distinction divides such collections into sensory systems that acquire and process information from the environment (e.g., the visual system or the auditory system, see Unit II), and motor systems that respond to such information by generating movements and other behavior (see Unit III). There are, however, large numbers of cells and circuits that lie between these relatively well-defined input and output systems. These are collectively referred to as associational systems, and they mediate the most complex and least well-characterized brain functions (see Unit V). In addition to these broad functional distinctions, neuroscientists and neurologists have conventionally divided the vertebrate nervous system anatomically into central and peripheral components (Figure 1.10). The central nervous system, typically referred to as the CNS, comprises the brain (cerebral hemispheres, diencephalon, cerebellum, and brainstem) and the spinal cord (see Appendix A for more information about the gross anatomical features of the CNS). The peripheral nervous system (PNS) includes the sensory neurons that link sensory receptors on the body surface or deeper within it with relevant processing circuits in the central nervous system. The motor portion of the peripheral nervous system in turn consists of two components. The motor axons that connect the brain and spinal cord to skeletal Studying the Ner vous Systems of Humans and O ther Animals 15 (A) (B) Peripheral nervous system Brain Cranial nerves Spinal cord Spinal nerves Cerebral hemispheres, diencephalon, cerebellum, brainstem, and spinal cord (analysis and integration of sensory and motor information) SENSORY COMPONENTS Sensory receptors (at surface and within the body) INTERNAL AND EXTERNAL ENVIRONMENT muscles make up the somatic motor division of the peripheral nervous system, whereas the cells and axons that innervate smooth muscles, cardiac muscle, and glands make up the visceral or autonomic motor division. Those nerve cell bodies that reside in the peripheral nervous system are located in ganglia, which are simply local accumulations of nerve cell bodies (and supporting cells). Peripheral axons are gathered into bundles called nerves, many of which are enveloped by the glial cells of the peripheral nervous system called Schwann cells. In the central nervous system, nerve cells are arranged in two different ways. Nuclei are local accumulations of neurons having roughly similar connections and functions; such collections are found throughout the cerebrum, brainstem and spinal cord. In contrast, cortex (plural, cortices) describes sheet-like arrays of nerve cells (again, consult Appendix A for additional information and illustrations). The cortices of the cerebral hemispheres and of the cerebellum provide the clearest example of this organizational principle. Axons in the central nervous system are gathered into tracts that are more or less analogous to nerves in the periphery. Tracts that cross the midline of the brain are referred to as commissures. Two gross histological terms distinguish regions rich in neuronal cell bodies versus regions rich in axons. Gray matter refers to any accumulation of cell bodies and neuropil in the brain and spinal cord (e.g., nuclei or cortices), whereas white matter, named for its relatively light appearance resulting from the lipid content of myelin, refers to axon tracts and commissures. MOTOR COMPONENTS VISCERAL MOTOR SYSTEM (sympathetic, parasympathetic, and enteric divisions) SOMATIC MOTOR SYSTEM Motor nerves Peripheral nervous system Sensory ganglia and nerves Central nervous system Central nervous system Autonomic ganglia and nerves EFFECTORS Smooth muscles, cardiac muscles, Skeletal (striated) muscles and glands Figure 1.10 The major components of the nervous system and their functional relationships. (A) The CNS (brain and spinal cord) and PNS (spinal and cranial nerves). (B) Diagram of the major components of the central and peripheral nervous systems and their functional relationships. Stimuli from the environment convey information to processing circuits within the brain and spinal cord, which in turn interpret their significance and send signals to peripheral effectors that move the body and adjust the workings of its internal organs. 16 Chapter One The organization of the visceral motor division of the peripheral nervous system is a bit more complicated (see Chapter 20). Visceral motor neurons in the brainstem and spinal cord, the so-called preganglionic neurons, form synapses with peripheral motor neurons that lie in the autonomic ganglia. The motor neurons in autonomic ganglia innervate smooth muscle, glands, and cardiac muscle, thus controlling most involuntary (visceral) behavior. In the sympathetic division of the autonomic motor system, the ganglia lie along or in front of the vertebral column and send their axons to a variety of peripheral targets. In the parasympathetic division, the ganglia are found within the organs they innervate. Another component of the visceral motor system, called the enteric system, is made up of small ganglia as well as individual neurons scattered throughout the wall of the gut. These neurons influence gastric motility and secretion. Neuroanatomical Terminology Describing the organization of any neural system requires a rudimentary understanding of anatomical terminology. The terms used to specify location in the central nervous system are the same as those used for the gross anatomical description of the rest of the body (Figure 1.11). Thus, anterior and posterior indicate front and back (head and tail); rostral and caudal, toward the head and tail; dorsal and ventral, top and bottom (back and belly); and medial and lateral, at the midline or to the side. Nevertheless, the comparison between these coordinates in the body versus the brain can be confusing. For the entire body these anatomical terms refer to the long axis, which is straight. The long axis of the central nervous system, however, has a bend in it. In humans and other bipeds, a compensatory tilting of the rostral–caudal axis for the brain is necessary to properly compare body axes to brain axes. Once this adjustment has been made, the other axes for the brain can be easily assigned. The proper assignment of the anatomical axes then dictates the standard planes for histological sections or live images (see Box A) used to study the internal anatomy of the brain (see Figure 1.11B). Horizontal sections (also referred to as axial or transverse sections) are taken parallel to the rostral– caudal axis of the brain; thus, in an individual standing upright, such sections are parallel to the ground. Sections taken in the plane dividing the two hemispheres are sagittal, and can be further categorized as midsagittal and parasagittal, according to whether the section is near the midline (midsagittal) ▲ Figure 1.11 A flexure in the long axis of the nervous system arose as humans evolved upright posture, leading to an approximately 120° angle between the long axis of the brainstem and that of the forebrain The consequences of this flexure for anatomical terminology are indicated in (A). The terms anterior, posterior, superior, and inferior refer to the long axis of the body, which is straight. Therefore, these terms indicate the same direction for both the forebrain and the brainstem. In contrast, the terms dorsal, ventral, rostral, and caudal refer to the long axis of the central nervous system. The dorsal direction is toward the back for the brainstem and spinal cord, but toward the top of the head for the forebrain. The opposite direction is ventral. The rostral direction is toward the top of the head for the brainstem and spinal cord, but toward the face for the forebrain. The opposite direction is caudal. (B) The major planes of section used in cutting or imaging the brain. (C) The subdivisions and components of the central nervous system. (Note that the position of the brackets on the left side of the figure refers to the vertebrae, not the spinal segments.) Studying the Ner vous Systems of Humans and O ther Animals 17 or more lateral (parasagittal). Sections in the plane of the face are called coronal or frontal. Different terms are usually used to refer to sections of the spinal cord. The plane of section orthogonal to the long axis of the cord is called transverse, whereas sections parallel to the long axis of the cord are called longitudinal. In a transverse section through the human spinal cord, the dorsal and ventral axes and the anterior and posterior axes indicate the same directions (see Figure 1.11). Tedious though this terminology may be, it (C) (A) Superior (above) Cerebrum Diencephalon Longitudinal axis of the forebrain Midbrain Pons Rostr al Dors al Ventr al Anterior (in front of) Cauda l sal Dor al tr Ven Posterior (behind) 2 3 4 5 6 7 8 Cervical nerves Medulla Spinal cord Cervical enlargement T1 2 Longitudinal axis of the brainstem and spinal cord Inferior (below) Cerebellum C1 3 4 Caudal 5 6 Thoracic nerves (B) Coronal 7 8 9 Sagittal Lumbar enlargement 10 11 12 Horizontal Cauda equina L1 Lumbar nerves 2 3 4 Sacral nerves Coccygeal nerve 5 S1 2 3 4 5 Coc 1 18 Chapter One is essential for understanding the basic subdivisions of the nervous system (Figure 1.11C). The Subdivisions of the Central Nervous System The central nervous system (defined as the brain and spinal cord) is usually considered to have seven basic parts: the spinal cord, the medulla, the pons, the cerebellum, the midbrain, the diencephalon, and the cerebral hemispheres (see Figures 1.10 and 1.11C). Running through all of these subdivisons are fluid-filled spaces called ventricles (a detailed account of the ventricular system can be found in Appendix B). These ventricles are the remnants of the continuous lumen initially enclosed by the neural plate as it rounded to become the neural tube during early development (see Chapter 21). Variations in the shape and size of the mature ventricular space are characteristic of each adult brain region. The medulla, pons, and midbrain are collectively called the brainstem and they surround the 4th ventricle (medulla and pons) and cerebral aqueduct (midbrain). The diencephalon and cerebral hemispheres are collectively called the forebrain, and they enclose the 3rd and lateral ventricles, respectively. Within the brainstem are the cranial nerve nuclei that either receive input from the cranial sensory ganglia mentioned earlier via the cranial sensory nerves, or give rise to axons that constitute the cranial motor nerves (see Appendix A). The brainstem is also a conduit for several major tracts in the central nervous system that relay sensory information from the spinal cord and brainstem to the forebrain, or relay motor commands from forebrain back to motor neurons in the brainstem and spinal cord. Accordingly, detailed knowledge of the consequences of damage to the brainstem provides neurologists and other clinicians an essential tool in the localization and diagnosis of brain injury. The brainstem contains numerous additional nuclei that are involved in a myriad of important functions including the control of heart rate, respiration, blood pressure, and level of consciousness. Finally, one of the most prominent features of the brainstem is the cerebellum, which extends over much of its dorsal aspect. The cerebellum is essential for the coordination and planning of movements (see Chapter 18) as well as learning motor tasks and storing that information (see Chapter 30). There are several anatomical subdivisions of the forebrain. The most obvious anatomical structures are the prominent cerebral hemispheres (Figure 1.12). In humans, the cerebral hemispheres (the outermost portions of which are continuous, highly folded sheets of cortex) are proportionally larger than in any other mammal, and are characterized by the gyri (singular, gyrus) or crests of folded cortical tissue, and sulci (singular, sulcus) the grooves that divide gyri from one another (as pictured on the cover of this book, for example). Although gyral and sulcal patterns vary from individual to individual, there are some fairly consistent landmarks that help divide the hemispheres into four lobes. The names of the lobes are derived from the cranial bones that overlie them: occipital, temporal, parietal, and frontal. A key feature of the surface anatomy of the cerebrum is the central sulcus located Figure 1.12 Gross anatomy of the forebrain (A) Cerebral hemisphere surface anatomy, showing the four lobes of the brain and the major sulci and gyri. The ventricular system and basal ganglia can also be seen in this phantom view. (B) Midsagittal view showing the location of the hippocampus, amygdala, thalamus and hypothalamus. Studying the Ner vous Systems of Humans and O ther Animals 19 (A) Precentral gyrus Central sulcus (C) Postcentral gyrus Cerebral hemisphere Parietal lobe Frontal lobe Parietooccipital sulcus Occipital lobe Temporal lobe Preoccipital notch Brainstem Spinal cord Cerebellum (B) Diencephalon Parietal lobe Frontal lobe Lateral (Sylvian) fissure Cingulate sulcus Central sulcus Occipital lobe Temporal lobe Parietooccipital sulcus Cingulate gyrus (D) Calcarine sulcus Corpus callosum Level of section shown in (E) Level of section shown in (F) Anterior commissure Midbrain Brainstem Cerebellum Pons Medulla Spinal cord (E) (F) Cerebral cortex (gray matter) Corpus callosum Internal capsule White matter Caudate Lateral ventricle Corpus callosum Thalamus Basal ganglia Caudate Internal capsule Putamen Globus pallidus Third ventricle Putamen Tail of caudate nucleus Lateral ventricle (temporal horn) Temporal lobe Anterior commissure Optic chiasm Amygdala Basal forebrain nuclei Hippocampus Mammillary body Fornix 20 Chapter One roughly halfway between the rostral and caudal poles of the hemispheres (Figure 1.12A). This prominent sulcus divides the frontal lobe at the rostral end of the hemisphere from the more caudal parietal lobe. Prominent on either side of the central sulcus are the pre- and postcentral gyri. These gyri are also functionally significant in that the precentral gyrus contains the primary motor cortex important for the control of movement, and the postcentral gyrus contains the primary somatic sensory cortex which is important for the bodily senses (see below). The remaining subdivisions of the forebrain lie deeper in the cerebral hemispheres (Figure 1.12B). The most prominent of these is the collection of deep structures involved in motor and cognitive processes collectively referred to as the basal ganglia. Other particularly important structures are the hippocampus and amygdala in the temporal lobes (these are vital substrates for memory and emotional behavior, respectively), and the olfactory bulbs (the central stations for processing chemosensory information arising from receptor neurons in the nasal cavity) on the anterior–inferior aspect of the frontal lobes. Finally, the thalamus lies in the diencephalon and is a critical relay for sensory information (although it has many other functions as well); the hypothalamus, which as the name implies lies below the thalamus, is the central organizing structure for the regulation of the body’s many homeostatic functions (e.g., feeding, drinking, thermoregulation). This rudimentary description of some prominent anatomical landmarks provides a framework for understanding how neurons resident in a number of widely distributed and distinct brain structures communicate with one another to define neural systems dedicated to encoding, processing and relaying specific sorts of information about aspects of the organism’s environment, and then initiating and coordinating appropriate behavioral responses. Organizational Principles of Neural Systems These complex perceptual and motor capacities of the brain reflect the integrated function of various neural systems. The processing of somatic sensory information (arising from receptors in the skin, subcutaneous tissues, and the musculoskeletal system that respond to physical deformation at the body surface or displacement of muscles and joints) provides a convenient example. These widely distributed structures that participate in generating somatic sensations are referred to as the somatic sensory system (Figure 1.13). The components in the peripheral nervous system include the receptors distributed throughout the skin as well as in muscles and tendons, the related neurons in dorsal root ganglia, and neurons in some cranial ganglia. The central nervous system components include neurons in the spinal cord, as well as the long tracts of their axons that originate in the spinal cord, travel through the brainstem, and ultimately terminate in distinct relay nuclei in the thalamus in the diencephalon. The still-higher targets of the thalamic neurons are the cortical areas around the postcentral gyrus that are collectively referred to as the somatic sensory cortex. Thus, the somatic sensory system includes specific populations of neurons in practically every subdivision of the nervous system. Two further principles of neural system organization are evident in the somatic sensory system: topographic organization and the prevalence of parallel pathways (see Figure 1.13). As the name implies, topography refers to a mapping function—in this case a map of the body surface that can be discerned within the various structures that constitute the somatic sensory Studying the Ner vous Systems of Humans and O ther Animals 21 (B) Cerebral cortex Somatic sensory cortex (A) Cerebral cortex Somatic sensory cortex Thalamus Brainstem Trigeminal ganglia Trigeminal ganglion Thalamus Spinal cord Cervical Brainstem Sensory receptors for face Thoracic Mechanical sensation Trigeminal ganglia Pain and temperature Lumbar Sensory receptor Sensory receptors for body Sacral Mechanical sensation Spinal cord Dorsal root ganglia Pain and temperature Dorsal root ganglia (DRG) system. Thus, adjacent areas on the body surface are mapped to adjacent regions in nuclei, in white matter tracts, and in the thalamic and cortical targets of the system. Beginning in the periphery, the cells in each dorsal root ganglion define a discrete dermatome (the area of the skin innervated by the processes of cells from a single dorsal root). In the spinal cord, from caudal to rostral, the dermatomes are represented in corresponding regions of the spinal cord from sacral (back) to lumbar (legs) to thoracic (chest) and cervical (arms and shoulders) (see Figures 1.13 and 1.11C). This so-called somatotopy is maintained in the somatic sensory tracts in spinal cord and brainstem that convey information to the relevant forebrain structures of the somatic sensory system (Figure 1.14). Parallel pathways refer to the segregation of nerve cell axons that process the distinct stimulus attributes that comprise a particular sensory, motor, or cognitive modality. For somatic sensation, the stimulus attributes relayed via parallel pathways are pain, temperature, touch, pressure, and proprioception (the sense of joint or limb position). From the dorsal root ganglia, through Peripheral nervous system Central nervous system Figure 1.13 The anatomical and functional organization of the somatic sensory system. Central nervous system components of the somatic sensory system are found in the spinal cord, brainstem, thalamus, and cerebral cortex. (A) Somatosensory information from the body surface is mapped onto dorsal root ganglia (DRG), schematically depicted here as attachments to the spinal cord. The various shades of purple indicate correspondence between regions of the body and the DRG that relay information from the body surface to the central nervous system. Information from the head and neck is relayed to the CNS via the trigeminal ganglia. (B) Somatosensory information travels from the peripheral sensory receptors via parallel pathways for mechanical sensation and for the sensation of pain and temperature. These parallel pathways relay through the spinal cord and brainstem, ultimately sending sensory information to the thalamus, from which it is relayed to the somatic sensory cortex in the postcentral gyrus (indicated in blue in the image of the whole brain; MRI courtesy of L. E. White, J. Vovoydic, and S. M. Williams). 22 Chapter One the spinal cord and brainstem, and on to the somatic sensory cortex, these submodalities are kept largely segregated. Thus anatomically, biochemically, and physiologically distinct neurons transduce, encode, and relay pain, temperature, and mechanical information. Although this information is subsequently integrated to provide unitary perception of the relevant stimuli, neurons and circuits in the somatic sensory system are clearly specialized to process discrete aspects of somatic sensation. This basic outline of the organization of the somatic system is representative of the principles pertinent to understanding any neural system. It will in every case be pertinent to consider the anatomical distribution of neural circuits dedicated to a particular function, how the function is represented or “mapped” onto the neural elements within the system, and how distinct stimulus attributes are segregated within subsets of neurons that comprise the system. Such details provide a framework for understanding how activity within the system provides a representation of relevant stimulus, the required motor response, and higher order cognitive correlates. Somatic sensory cortex Shoulder Neck Trunk Arm Head Leg Hand Feet Digits Thumb Toes Neck Eyes Nose Face Genitalia Lips Jaw Tongue Throat Lateral Medial Figure 1.14 Somatotopic organization of sensory information. (Top) The locations of primary and secondary somatosensory cortical areas on the lateral surface of the brain. (Bottom) Cortical representation of different regions of skin. Studying the Ner vous Systems of Humans and O ther Animals 23 Functional Analysis of Neural Systems A wide range of physiological methods is now available to evaluate the electrical (and metabolic) activity of the neuronal circuits that make up a neural system. Two approaches, however, have been particularly useful in defining how neural systems represent information. The most widely used method is single-cell, or single-unit electrophysiological recording with microelectrodes (see above; this method often records from several nearby cells in addition to the one selected, providing further useful information). The use of microelectrodes to record action potential activity provides a cell-by-cell analysis of the organization topographic maps (Figure 1.15), and can give specific insight into the type of stimulus to which the neuron is “tuned” (i.e., the stimulus that elicits a maximal change in action potential activity from the baseline state). Single-unit analysis is often used to define a neuron’s receptive field—the region in sensory space (e.g., the body surface, or a specialized structure such as the retina) within which a specific stimulus elicits the greatest action potential response. This approach to understanding neural systems was introduced by Stephen Kuffler and Vernon Mountcastle in the early 1950s and has now been used by several generations of neuroscientists to evaluate the relationship between stimuli and neuronal responses in both sensory and motor systems. Electrical recording techniques (A) (B) Somatic sensory cortex Activity of cortical neuron Receptive field (center) Touch in the center of receptive field increases cell firing Touch in the surround of receptive field decreases cell firing Record Receptive field (surround) Touch outside of receptive field has no effect Central sulcus Postcentral gyrus Figure 1.15 Single-unit electrophysiological recording from cortical pyramidal neuron, showing the firing pattern in response to a specific peripheral stimulus. (A) Typical experimental set-up. (B) Defining neuronal receptive fields. Period of stimulation 24 Chapter One at the single-cell level have now been extended and refined to include single and simultaneous multiple cell analysis in animals performing complex cognitive tasks, intracellular recordings in intact animals, and the use of patch electrodes to detect and monitor the activity of the individual membrane molecules that ultimately underlie neural signaling (see Unit I). The second major area in which remarkable technical advances have been made is functional brain imaging in human subjects (and to a lesser extent animals), which has revolutionized the functional understanding of neural systems over the last two decades (Box A). Unlike electrical methods of recording neural activity, which are invasive in the sense of having to expose the brain and insert electrodes into it, functional imaging is noninvasive and thus applicable to both patients and normal human subjects. Moreover, functional imaging allows the simultaneous evaluation of multiple brain structures (which is possible but obviously difficult with electrical recording methods). The tasks that can be evaluated with functional imaging permit a far more ambitious and integrative approach to studying the operations of a neural system. Over the last 20 years, these noninvasive methods have allowed neuroscientists to evaluate the representation of an enormous number of complex human behaviors, and at the same time have provided diagnostic tools that are used more and more routinely. Many of the resulting observations have confirmed inferences about functional localization and the organization of neural systems that were originally based on the study of neurological patients who exhibited altered behavior after stroke or other forms of brain injury. Others findings, however, have given new insights into the way neural systems function in the human brain. Analyzing Complex Behavior Many of the most widely heralded advances in modern neuroscience have involved reducing the complexity of the brain to more readily analyzed components—i.e., genes, molecules, or cells. Nevertheless, the brain functions as a whole, and the study of more complex (and, some might argue, more interesting) brain functions such as perception, language, emotion, memory, and consciousness remain a central challenge for contemporary neuroscientists. In recognition of this challenge, over the last 20 years or so a field called cognitive neuroscience has emerged that is specifically devoted to understanding these issues (see Unit V). This evolution has also rejuvenated the field of neuroethology (which is devoted to observing complex behaviors of animals in their native environments—for example, social communication in birds and non-human primates), and has encouraged the development of tasks to better evaluate the genesis of complex behaviors in human subjects. When used in combination with functional imaging, well designed behavioral tasks can facilitate identification of brain networks devoted to specific complex functions, including language skills, mathematical and musical ability, emotional responses, aesthetic judgments, and abstract thinking. Carefully constructed behavioral tasks can also be used to study the pathology of complex brain diseases that compromise cognition, such Alzheimer’s disease, schizophrenia, and depression. In short, new or revitalized efforts to study higher brain functions with increasingly powerful techniques offer ways of beginning to understand even the most complex aspects of human behavior. Studying the Ner vous Systems of Humans and O ther Animals 25 Box A Brain Imaging Techniques In the 1970s, computerized tomography, or CT, opened a new era in noninvasive imaging by introducing the use of computer processing technology to help probe the living brain. Prior to CT, the only brain imaging technique available was standard X-ray film, which has poor soft tissue contrast and involves relatively high radiation exposure. The CT approach uses a narrow X-ray beam and a row of very sensitive detectors placed on opposite sides of the head to probe just a small portion of tissue at a time with limited radiation exposure (see Figure A). In order to make an image, the X-ray tube and detectors rotate around the head to collect radiodensity information from every orientation around a narrow slice. Computer processing techniques then calculate the radiodensity of each point within the slice plane, producing a tomographic image (tomo means “cut” or “slice”). If the patient is slowly moved through the scanner while the Xray tube rotates in this way, a three- dimensional radiodensity matrix can be created, allowing images to be computed for any plane through the brain. CT scans can readily distinguish gray matter and white matter, differentiate the ventricles quite well, and show many other brain structures with a spatial resolution of several millimeters. Brain imaging took another large step forward in the 1980s with the development of magnetic resonance imaging (MRI). MRI is based on the fact that the nuclei of some atoms act as spinning magnets, and that if they are placed in a strong magnetic field they will line up with the field and spin at a frequency that is dependent on the field strength. If they then receive a brief radiofrequency pulse tuned to their spinning frequency they are knocked out of alignment with the field, and subsequently emit energy in an oscillatory fashion as they gradually realign themselves with the field. The strength of the emitted signal depends on how many nuclei are involved in this X-ray source X-ray detector (A) In computerized tomography, the X-ray source and detectors are moved around the patient’s head. The inset shows a horizontal CT section of a normal adult brain. process. To get spatial information in MRI, the magnetic field is distorted slightly by imposing magnetic gradients along three different spatial axes so that only nuclei at certain locations are tuned to the detector’s frequency at any given time. Almost all MRI scanners use detectors tuned to the radio frequencies of spinning hydrogen nuclei in water molecules, and thus create images based on the distribution of water in different tissues. Careful manipulation of magnetic field gradients and radiofrequency pulses make it possible to construct extraordinarily detailed images of the brain at any location and orientation with sub-millimeter resolution. The strong magnetic field and radiofrequency pulses used in MRI scanning are harmless, making this technique completely noninvasive (although metal objects in or near a scanner are a safety concern) (see Figure B). MRI is also extremely versatile because, by changing the scanning parameters, images based on a wide variety of different contrast mechanisms can be generated. For example, conventional MR images take advantage of the fact that hydrogen in different types of tissue (e.g., gray matter, white matter, cerebrospinal fluid) have slightly different realignment rates, meaning that soft tissue contrast can be manipulated simply by adjusting when the realigning hydrogen signal is measured. Different parameter settings can also be used to generate images in which gray and white matter are invisible but in which the brain vasculature stands out in sharp detail. Safety and versatility have made MRI the technique of choice for imaging brain structure in most applications. Imaging functional variations in the living brain has also become possible with the recent development of techniques for detecting small, localized (continued) 26 Chapter One Box A (continued) Brain Imaging Techniques changes in metabolism or cerebral blood flow. To conserve energy, the brain regulates its blood flow such that active neurons with relatively high metabolic demands receive more blood than relatively inactive neurons. Detecting and mapping these local changes in cerebral blood flow forms the basis for three widely used functional brain imaging techniques: positron emission tomography (PET), single-photon emission computerized tomography (SPECT), and functional magnetic resonance imaging (fMRI). In PET scanning, unstable positronemitting isotopes are incorporated into different reagents (including water, precursor molecules of specific neurotransmitters, or glucose) and injected into the bloodstream. Labeled oxygen and glucose quickly accumulate in more metabolically active areas, and labeled transmitter probes are taken up selectively by appropriate regions. As the unstable isotope decays, it results in the emission of two positrons moving in opposite directions. Gamma ray detectors placed around the head register a “hit” only when two detectors 180° apart react simultaneously. Images of tissue isotope density can then be generated (much the way CT images are calculated) showing the location of active regions with a spa- (B) In MRI scanning, the head is placed in the center of a large magnet. A radiofrequency antenna coil is placed around the head for exciting and recording the magnetic resonance signal. For fMRI, stimuli can be presented using virtual reality video goggles and stereo headphones while inside the scanner. tial resolution of about 4 mm. Depending on the probe injected, PET imaging can be used to visualize activity-dependent changes in blood flow, tissue metabolism, or biochemical activity. SPECT imaging is similar to PET in that it involves injection or inhalation of a radiolabeled compound (for example, 133Xe or 123I-labeled iodoamphetamine), which produce pho- tons that are detected by a gamma camera moving rapidly around the head. Functional MRI, a variant of MRI, currently offers the best approach for visualizing brain function based on local metabolism. fMRI is predicated on the fact that hemoglobin in blood slightly distorts the magnetic resonance properties of hydrogen nuclei in its vicinity, and Summary The brain can be studied by methods that range from genetics and molecular biology to behavioral testing of normal human subjects. In addition to an ever-increasing store of knowledge about the anatomical organization of the nervous system, many of the brightest successes of modern neuroscience have come from understanding nerve cells as the basic structural and functional unit of the nervous system. Studies of the distinct cellular architecture and molecular components of neurons and glia have revealed much about Studying the Ner vous Systems of Humans and O ther Animals 27 Right Left Tumor in progress (C) MRI images of an adult patient with a brain tumor, with fMRI activity during a hand motion task superimposed (left hand activity is shown in yellow, right hand activity in green). At right is a three-dimensional surface reconstructed view of the same data. the amount of magnetic distortion changes depending on whether the hemoglobin has oxygen bound to it. When a brain area is activated by a specific task it begins to use more oxygen and within seconds the brain microvasculature responds by increasing the flow of oxygen-rich blood to the active area. These changes in the concentration of oxygen and blood flow lead to localized blood oxygenation level-dependent (BOLD) changes in the magnetic resonance signal. Such fluctuations are detected using statistical image process- ing techniques to produce maps of taskdependent brain function (see Figure C). Because fMRI uses signals intrinsic to the brain without any radioactivity, repeated observations can be made on the same individual—a major advantage over imaging methods such as PET. The spatial resolution (2–3 mm) and temporal resolution (a few seconds) of fMRI are also superior to other functional imaging techniques. MRI has thus emerged as the technology of choice for probing both the structure and function of the living human brain. their individual functions, as well as providing a basis for understanding how nerve cells are organized into circuits, and circuits into systems that process specific types of information pertinent to perception and action. Goals that remain include understanding how basic molecular genetic phenomena are linked to cellular, circuit, and system functions; understanding how these processes go awry in neurological and psychiatric diseases; and beginning to understand the especially complex functions of the brain that make us human. References HUETTEL, S. A., A. W. SONG AND G. MCCARTHY (2004) Functional Magnetic Resonance Imaging. Sunderland, MA: Sinauer Associates. OLDENDORF, W. AND W. OLDENDORF JR. (1988) Basics of Magnetic Resonance Imaging. Boston: Kluwer Academic Publishers. RAICHLE, M. E. (1994) Images of the mind: Studies with modern imaging techniques. Ann. Rev. Psychol. 45: 333–356. SCHILD, H. (1990) MRI Made Easy (…Well, Almost). Berlin: H. Heineman. 28 Chapter One Additional Reading BRODAL, P. (1992) The Central Nervous System: Structure and Function. New York: Oxford University Press. CARPENTER, M. B. AND J. SUTIN (1983) Human Neuroanatomy, 8th Ed. Baltimore, MD: Williams and Wilkins. ENGLAND, M. A. AND J. WAKELY (1991) Color Atlas of the Brain and Spinal Cord: An Introduction to Normal Neuroanatomy. St. Louis: Mosby Yearbook. GIBSON, G. AND S. MUSE (2001) A Primer of Genome Science. Sunderland, MA: Sinauer Associates. HAINES, D. E. (1995) Neuroanatomy: An Atlas of Structures, Sections, and Systems, 2nd Ed. Baltimore: Urban and Schwarzenberg. MARTIN, J. H. (1996) Neuroanatomy: Text and Atlas, 2nd Ed. Stamford, CT: Appleton and Lange. NATURE VOL. 409, NO. 6822 (2001) Issue of February 16. Special issue on the human genome. NETTER, F. H. (1983) The CIBA Collection of Medical Illustrations, Vols. I and II. A. Brass and R. V. Dingle (eds.). Summit, NJ: CIBA Pharmaceutical Co. PETERS, A., S. L. PALAY AND H. DE F. WEBSTER (1991) The Fine Structure of the Nervous System: Neurons and Their Supporting Cells, 3rd Ed. New York: Oxford University Press. POSNER, M. I. AND M. E. RAICHLE (1997) Images of Mind, 2nd Ed. New York: W. H. Freeman & Co. RAMÓN Y CAJAL, S. (1984) The Neuron and the Glial Cell. (Transl. by J. de la Torre and W. C. Gibson.) Springfield, IL: Charles C. Thomas. RAMÓN Y CAJAL, S. (1990) New Ideas on the Structure of the Nervous System in Man and Vertebrates. (Transl. by N. Swanson and L. W. Swanson.) Cambridge, MA: MIT Press. SCIENCE VOL. 291, NO. 5507 (2001) Issue of February 16. Special issue on the human genome. SHEPHERD, G. M. (1991) Foundations of the Neuron Doctrine. History of Neuroscience Series, No. 6. Oxford: Oxford University Press. WAXMAN, S. G. AND J. DEGROOT (1995) Correlative Neuroanatomy, 22nd Ed. Norwalk, CT: Appleton and Lange. Neural Signaling I UNIT I NEURAL SIGNALING Calcium signaling in a cerebellar Purkinje neuron. An electrode was used to fill the neuron with a fluorescent calcium indicator dye. This dye revealed the release of intracellular calcium ions (color) produced by the actions of the second messenger IP3. (Courtesy of Elizabeth A. Finch and George J. Augustine.) 2 3 4 5 6 7 Electrical Signals of Nerve Cells Voltage-Dependent Membrane Permeability Channels and Transporters Synaptic Transmission Neurotransmitters, Receptors, and Their Effects Molecular Signaling within Neurons The brain is remarkably adept at acquiring, coordinating, and disseminating information about the body and its environment. Such information must be processed within milliseconds, yet it also can be stored away as memories that endure for years. Neurons within the central and peripheral nervous systems perform these functions by generating sophisticated electrical and chemical signals. This unit describes these signals and how they are produced. It explains how one type of electrical signal, the action potential, allows information to travel along the length of a nerve cell. It also explains how other types of signals—both electrical and chemical—are generated at synaptic connections between nerve cells. Synapses permit information transfer by interconnecting neurons to form the circuitry on which neural processing depends. Finally, it describes the intricate biochemical signaling events that take place within neurons. Appreciating these fundamental forms of neuronal signaling provides a foundation for appreciating the higher-level functions considered in the rest of the book. The cellular and molecular mechanisms that give neurons their unique signaling abilities are also targets for disease processes that compromise the function of the nervous system. A working knowledge of the cellular and molecular biology of neurons is therefore fundamental to understanding a variety of brain pathologies, and for developing novel approaches to diagnosing and treating these all too prevalent problems. Chapter 2 Electrical Signals of Nerve Cells Overview Nerve cells generate electrical signals that transmit information. Although neurons are not intrinsically good conductors of electricity, they have evolved elaborate mechanisms for generating these signals based on the flow of ions across their plasma membranes. Ordinarily, neurons generate a negative potential, called the resting membrane potential, that can be measured by recording the voltage between the inside and outside of nerve cells. The action potential transiently abolishes the negative resting potential and makes the transmembrane potential positive. Action potentials are propagated along the length of axons and are the fundamental signal that carries information from one place to another in the nervous system. Still other types of electrical signals are produced by the activation of synaptic contacts between neurons or by the actions of external forms of energy on sensory neurons. All of these electrical signals arise from ion fluxes brought about by nerve cell membranes being selectively permeable to different ions, and from the non-uniform distribution of these ions across the membrane. Electrical Potentials across Nerve Cell Membranes Neurons employ several different types of electrical signal to encode and transfer information. The best way to observe these signals is to use an intracellular microelectrode to measure the electrical potential across the neuronal plasma membrane. A typical microelectrode is a piece of glass tubing pulled to a very fine point (with an opening of less than 1 µm diameter) and filled with a good electrical conductor, such as a concentrated salt solution. This conductive core can then be connected to a voltmeter, such as an oscilloscope, to record the transmembrane voltage of the nerve cell. The first type of electrical phenomenon can be observed as soon as a microelectrode is inserted through the membrane of the neuron. Upon entering the cell, the microelectrode reports a negative potential, indicating that neurons have a means of generating a constant voltage across their membranes when at rest. This voltage, called the resting membrane potential, depends on the type of neuron being examined, but it is always a fraction of a volt (typically –40 to –90 mV). The electrical signals produced by neurons are caused by responses to stimuli, which then change the resting membrane potential. Receptor potentials are due to the activation of sensory neurons by external stimuli, such as light, sound, or heat. For example, touching the skin activates Pacinian corpuscles, receptor neurons that sense mechanical disturbances of the skin. These neurons respond to touch with a receptor potential that changes the resting potential for a fraction of a second (Figure 2.1A). These transient 31 32 Chapter Two Record Membrane potential (mV) (A) Receptor potential Touch skin –50 −60 Time (ms) (B) Synaptic potential Record Stimulate Membrane potential (mV) –60 Activate synapse −70 Time (ms) (C) Action potential 40 Stimulate Membrane potential (mV) Figure 2.1 Types of neuronal electrical signals. In all cases, microelectrodes are used to measure changes in the resting membrane potential during the indicated signals. (A) A brief touch causes a receptor potential in a Pacinian corpuscle in the skin. (B) Activation of a synaptic contact onto a hippocampal pyramidal neuron elicits a synaptic potential. (C) Stimulation of a spinal reflex produces an action potential in a spinal motor neuron. Record Activate motor neuron −60 Time (ms) changes in potential are the first step in generating the sensation of vibrations (or “tickles”) of the skin in the somatic sensory system (Chapter 8). Similar sorts of receptor potentials are observed in all other sensory neurons during transduction of sensory signals (Unit II). Another type of electrical signal is associated with communication between neurons at synaptic contacts. Activation of these synapses generates synaptic potentials, which allow transmission of information from one neuron to another. An example of such a signal is shown in Figure 2.1B. In this case, activation of a synaptic terminal innervating a hippocampal pyramidal neuron causes a very brief change in the resting membrane potential in the pyramidal neuron. Synaptic potentials serve as the means of exchanging information in complex neural circuits in both the central and peripheral nervous systems (Chapter 5). The use of electrical signals—as in sending electricity over wires to provide power or information—presents a series of problems in electrical engineering. A fundamental problem for neurons is that their axons, which can be quite long (remember that a spinal motor neuron can extend for a meter or more), are not good electrical conductors. Although neurons and wires Electrical Signals of Ner ve Cells 33 are both capable of passively conducting electricity, the electrical properties of neurons compare poorly to an ordinary wire. To compensate for this deficiency, neurons have evolved a “booster system” that allows them to conduct electrical signals over great distances despite their intrinsically poor electrical characteristics. The electrical signals produced by this booster system are called action potentials (which are also referred to as “spikes” or “impulses”). An example of an action potential recorded from the axon of a spinal motor neuron is shown in Figure 2.1C. One way to elicit an action potential is to pass electrical current across the membrane of the neuron. In normal circumstances, this current would be generated by receptor potentials or by synaptic potentials. In the laboratory, however, electrical current suitable for initiating an action potential can be readily produced by inserting a second microelectrode into the same neuron and then connecting the electrode to a battery (Figure 2.2A). If the current delivered in this way makes the membrane potential more negative (hyperpolarization), nothing very dramatic happens. The membrane potential simply changes in proportion to the magnitude of the injected current (central part of Figure 2.2B). Such hyperpolarizing responses do not require any unique property of neurons and are therefore called passive electrical responses. A much more interesting phenomenon is seen if current of the opposite polarity is delivered, so that the membrane potential of the nerve cell becomes more positive than the resting potential (depolarization). In this case, at a certain level of membrane potential, called the threshold potential, an action potential occurs (see right side of Figure 2.2B). The action potential, which is an active response generated by the neuron, is a brief (about 1 ms) change from negative to positive in the transmem- (A) Current (nA) (B) Stimulate Figure 2.2 Recording passive and active electrical signals in a nerve cell. (A) Two microelectrodes are inserted into a neuron; one of these measures membrane potential while the other injects current into the neuron. (B) Inserting the voltage-measuring microelectrode into the neuron reveals a negative potential, the resting membrane potential. Injecting current through the current-passing microelectrode alters the neuronal membrane potential. Hyperpolarizing current pulses produce only passive changes in the membrane potential. While small depolarizing currents also elict only passive responses, depolarizations that cause the membrane potential to meet or exceed threshold additionally evoke action potentials. Action potentials are active responses in the sense that they are generated by changes in the permeability of the neuronal membrane. +2 0 Microelectrode to inject current −2 Record Microelectrode to measure membrane potential Action potentials +40 Membrane potential (mV) Neuron Insert microelectrode 0 Depolarization Passive responses −50 Threshold −65 Resting potential −100 Hyperpolarization Time 34 Chapter Two brane potential. Importantly, the amplitude of the action potential is independent of the magnitude of the current used to evoke it; that is, larger currents do not elicit larger action potentials. The action potentials of a given neuron are therefore said to be all-or-none, because they occur fully or not at all. If the amplitude or duration of the stimulus current is increased sufficiently, multiple action potentials occur, as can be seen in the responses to the three different current intensities shown in Figure 2.2B (right side). It follows, therefore, that the intensity of a stimulus is encoded in the frequency of action potentials rather than in their amplitude. This arrangement differs dramatically from receptor potentials, whose amplitudes are graded in proportion to the magnitude of the sensory stimulus, or synaptic potentials, whose amplitude varies according to the number of synapses activated and the previous amount of synaptic activity. Because electrical signals are the basis of information transfer in the nervous system, it is essential to understand how these signals arise. Remarkably, all of the neuronal electrical signals described above are produced by similar mechanisms that rely upon the movement of ions across the neuronal membrane. The remainder of this chapter addresses the question of how nerve cells use ions to generate electrical potentials. Chapter 3 explores more specifically the means by which action potentials are produced and how these signals solve the problem of long-distance electrical conduction within nerve cells. Chapter 4 examines the properties of membrane molecules responsible for electrical signaling. Finally, Chapters 5–7 consider how electrical signals are transmitted from one nerve cell to another at synaptic contacts. How Ionic Movements Produce Electrical Signals Electrical potentials are generated across the membranes of neurons—and, indeed, all cells—because (1) there are differences in the concentrations of specific ions across nerve cell membranes, and (2) the membranes are selectively permeable to some of these ions. These two facts depend in turn on two different kinds of proteins in the cell membrane (Figure 2.3). The ion concentration gradients are established by proteins known as active transporters, which, as their name suggests, actively move ions into or out of cells against their concentration gradients. The selective permeability of membranes is ION TRANSPORTERS ION CHANNELS Ions Outside Figure 2.3 Ion transporters and ion channels are responsible for ionic movements across neuronal membranes. Transporters create ion concentration differences by actively transporting ions against their chemical gradients. Channels take advantage of these concentration gradients, allowing selected ions to move, via diffusion, down their chemical gradients. Neuronal Neuronal membrane membrane Inside 1 Ion binds 2 Ion transported across membrane Ion transporters −Actively move ions against concentration gradient −Create ion concentration gradients Ion diffuses through channel Ion channels −Allow ions to diffuse down concentration gradient −Cause selective permeability to certain ions Electrical Signals of Ner ve Cells 35 due largely to ion channels, proteins that allow only certain kinds of ions to cross the membrane in the direction of their concentration gradients. Thus, channels and transporters basically work against each other, and in so doing they generate the resting membrane potential, action potentials, and the synaptic potentials and receptor potentials that trigger action potentials. The structure and function of these channels and transporters are described in Chapter 4. To appreciate the role of ion gradients and selective permeability in generating a membrane potential, consider a simple system in which an artificial membrane separates two compartments containing solutions of ions. In such a system, it is possible to determine the composition of the two solutions and, thereby, control the ion gradients across the membrane. For example, take the case of a membrane that is permeable only to potassium ions (K+). If the concentration of K+ on each side of this membrane is equal, then no electrical potential will be measured across it (Figure 2.4A). However, if the concentration of K+ is not the same on the two sides, then an electrical potential will be generated. For instance, if the concentration of K+ on one side of the membrane (compartment 1) is 10 times higher than the K+ concentration on the other side (compartment 2), then the electrical potential of compartment 1 will be negative relative to compartment 2 (Figure 2.4B). This difference in electrical potential is generated because the potassium ions flow down their concentration gradient and take their electrical charge (one positive charge per ion) with them as they go. Because neuronal membranes contain pumps that accumulate K+ in the cell cytoplasm, and because potassium-permeable channels in the plasma membrane allow a transmembrane flow of K+, an analogous situation exists in living nerve cells. A continual resting efflux of K+ is therefore responsible for the resting membrane potential. In the hypothetical case just described, an equilibrium will quickly be reached. As K+ moves from compartment 1 to compartment 2 (the initial conditions on the left of Figure 2.4B), a potential is generated that tends to impede further flow of K+. This impediment results from the fact that the (B) Initial conditions Initially V=0 Voltmeter V=0 1 2 1 mM KCl 1 mM KCl At equilibrium V1−2 =−58 mV [K+]1 (mM) 100 1 2 10 mM KCl 1 mM KCl – + – + – + – + – + Net flux of K+ from 1 to 2 10 1 2 10 mM KCl 1 mM KCl Flux of K+ from 1 to 2 balanced by opposing membrane potential 1 0 −58 Slope = 58 mV per tenfold change in K+ gradient −116 Permeable to K+ No net flux of K+ (C) Membrane potential V1−2 (mV) (A) Figure 2.4 Electrochemical equilibrium. (A) A membrane permeable only to K+ (yellow spheres) separates compartments 1 and 2, which contain the indicated concentrations of KCl. (B) Increasing the KCl concentration in compartment 1 to 10 mM initially causes a small movement of K+ into compartment 2 (initial conditions) until the electromotive force acting on K+ balances the concentration gradient, and the net movement of K+ becomes zero (at equilibrium). (C) The relationship between the transmembrane concentration gradient ([K+]2/[K+]1) and the membrane potential. As predicted by the Nernst equation, this relationship is linear when plotted on semi-logarithmic coordinates, with a slope of 58 mV per tenfold difference in the concentration gradient. −2 −1 + log [K+]2 [K ]1 0 36 Chapter Two potential gradient across the membrane tends to repel the positive potassium ions that would otherwise move across the membrane. Thus, as compartment 2 becomes positive relative to compartment 1, the increasing positivity makes compartment 2 less attractive to the positively charged K+. The net movement (or flux) of K+ will stop at the point (at equilibrium on the right of Figure 2.4B) where the potential change across the membrane (the relative positivity of compartment 2) exactly offsets the concentration gradient (the tenfold excess of K+ in compartment 1). At this electrochemical equilibrium, there is an exact balance between two opposing forces: (1) the concentration gradient that causes K+ to move from compartment 1 to compartment 2, taking along positive charge, and (2) an opposing electrical gradient that increasingly tends to stop K+ from moving across the membrane (Figure 2.4B). The number of ions that needs to flow to generate this electrical potential is very small (approximately 10–12 moles of K+ per cm2 of membrane, or 1012 K+ ions). This last fact is significant in two ways. First, it means that the concentrations of permeant ions on each side of the membrane remain essentially constant, even after the flow of ions has generated the potential. Second, the tiny fluxes of ions required to establish the membrane potential do not disrupt chemical electroneutrality because each ion has an oppositely charged counter-ion (chloride ions in the example shown in Figure 2.4) to maintain the neutrality of the solutions on each side of the membrane. The concentration of K+ remains equal to the concentration of Cl– in the solutions in compartments 1 and 2, meaning that the separation of charge that creates the potential difference is restricted to the immediate vicinity of the membrane. The Forces That Create Membrane Potentials The electrical potential generated across the membrane at electrochemical equilibrium, the equilibrium potential, can be predicted by a simple formula called the Nernst equation. This relationship is generally expressed as EX = RT [X] 2 ln zF [X] 1 where EX is the equilibrium potential for any ion X, R is the gas constant, T is the absolute temperature (in degrees on the Kelvin scale), z is the valence (electrical charge) of the permeant ion, and F is the Faraday constant (the amount of electrical charge contained in one mole of a univalent ion). The brackets indicate the concentrations of ion X on each side of the membrane and the symbol ln indicates the natural logarithm of the concentration gradient. Because it is easier to perform calculations using base 10 logarithms and to perform experiments at room temperature, this relationship is usually simplified to [X] 2 EX = 58 z log X [ ]1 where log indicates the base 10 logarithm of the concentration ratio. Thus, for the example in Figure 2.4B, the potential across the membrane at electrochemical equilibrium is [K ] 1 2 EK = 58 z log K = 58 log 10 = − 58 mV [ ]1 The equilibrium potential is conventionally defined in terms of the potential difference between the reference compartment, side 2 in Figure 2.4, and the other side. This approach is also applied to biological systems. In this case, Electrical Signals of Ner ve Cells 37 the outside of the cell is the conventional reference point (defined as zero potential). Thus, when the concentration of K+ is higher inside than out, an inside-negative potential is measured across the K+-permeable neuronal membrane. For a simple hypothetical system with only one permeant ion species, the Nernst equation allows the electrical potential across the membrane at equilibrium to be predicted exactly. For example, if the concentration of K+ on side 1 is increased to 100 mM, the membrane potential will be –116 mV. More generally, if the membrane potential is plotted against the logarithm of the K+ concentration gradient ([K]2/[K]1), the Nernst equation predicts a linear relationship with a slope of 58 mV (actually 58/z) per tenfold change in the K+ gradient (Figure 2.4C). To reinforce and extend the concept of electrochemical equilibrium, consider some additional experiments on the influence of ionic species and ionic permeability that could be performed on the simple model system in Figure 2.4. What would happen to the electrical potential across the membrane (the potential of side 1 relative to side 2) if the potassium on side 2 were replaced with 10 mM sodium (Na+) and the K+ in compartment 1 were replaced by 1 mM Na+? No potential would be generated, because no Na+ could flow across the membrane (which was defined as being permeable only to K+). However, if under these ionic conditions (10 times more Na+ in compartment 2) the K+-permeable membrane were to be magically replaced by a membrane permeable only to Na+, a potential of +58 mV would be measured at equilibrium. If 10 mM calcium (Ca2+) were present in compartment 2 and 1 mM Ca2+ in compartment 1, and a Ca2+-selective membrane separated the two sides, what would happen to the membrane potential? A potential of +29 mV would develop, because the valence of calcium is +2. Finally, what would happen to the membrane potential if 10 mM Cl– were present in compartment 1 and 1 mM Cl– were present in compartment 2, with the two sides separated by a Cl–-permeable membrane? Because the valence of this anion is –1, the potential would again be +58 mV. The balance of chemical and electrical forces at equilibrium means that the electrical potential can determine ionic fluxes across the membrane, just as the ionic gradient can determine the membrane potential. To examine the influence of membrane potential on ionic flux, imagine connecting a battery across the two sides of the membrane to control the electrical potential across the membrane without changing the distribution of ions on the two sides (Figure 2.5). As long as the battery is off, things will be just as in Figure 2.4, with the flow of K+ from compartment 1 to compartment 2 causing a negative membrane potential (Figure 2.5A, left). However, if the battery is used to make compartment 1 initially more negative relative to compartment 2, there will be less K+ flux, because the negative potential will tend to keep K+ in compartment 1. How negative will side 1 need to be before there is no net flux of K+? The answer is –58 mV, the voltage needed to counter the tenfold difference in K+ concentrations on the two sides of the membrane (Figure 2.5A, center). If compartment 1 is initially made more negative than –58 mV, then K+ will actually flow from compartment 2 into compartment 1, because the positive ions will be attracted to the more negative potential of compartment 1 (Figure 2.5A, right). This example demonstrates that both the direction and magnitude of ion flux depend on the membrane potential. Thus, in some circumstances the electrical potential can overcome an ionic concentration gradient. The ability to alter ion flux experimentally by changing either the potential imposed on the membrane (Figure 2.5B) or the transmembrane concen- 38 Chapter Two (A) Battery off Battery on Battery on V1−2 = 0 mV V1−2 = −58 mV V1−2 = −116 mV − Battery + (B) Battery Net flux of K+ from 1 to 2 − + − + − + − + 0 1 2 10 mM KCl 1 mM KCl Net flux of K+ from 1 to 2 1 2 10 mM KCl −116 1 2 1 mM KCl 10 mM KCl 1 mM KCl No net flux of K+ Figure 2.5 Membrane potential influences ion fluxes. (A) Connecting a battery across the K+-permeable membrane allows direct control of membrane potential. When the battery is turned off (left), K+ ions (yellow) flow simply according to their concentration gradient. Setting the initial membrane potential (V1–2) at the equilibrium potential for K+ (center) yields no net flux of K+, while making the membrane potential more negative than the K+ equilibrium potential (right) causes K+ to flow against its concentration gradient. (B) Relationship between membrane potential and direction of K+ flux. 0 −58 + 2 − Net flux of K+ 1 1 2 Battery − + Net flux of K+ from 2 to 1 No net flux of K+ Membrane potential V1−2 (mV) Net flux of K+ from 2 to 1 tration gradient for an ion (see Figure 2.4C) provides convenient tools for studying ion fluxes across the plasma membranes of neurons, as will be evident in many of the experiments described in the following chapters. Electrochemical Equilibrium in an Environment with More Than One Permeant Ion Now consider a somewhat more complex situation in which Na+ and K+ are unequally distributed across the membrane, as in Figure 2.6A. What would happen if 10 mM K+ and 1 mM Na+ were present in compartment 1, and 1 mM K+ and 10 mM Na+ in compartment 2? If the membrane were permeable only to K+, the membrane potential would be –58 mV; if the membrane were permeable only to Na+, the potential would be +58 mV. But what would the potential be if the membrane were permeable to both K+ and Na+? In this case, the potential would depend on the relative permeability of the membrane to K+ and Na+. If it were more permeable to K+, the potential would approach –58 mV, and if it were more permeable to Na+, the potential would be closer to +58 mV. Because there is no permeability term in the Nernst equation, which only considers the simple case of a single permeant ion species, a more elaborate equation is needed that takes into account both the concentration gradients of the permeant ions and the relative permeability of the membrane to each permeant species. Such an equation was developed by David Goldman in 1943. For the case most relevant to neurons, in which K+, Na+, and Cl– are the primary permeant ions, the Goldman equation is written V = 58 log PK [K ] 2 + PNa [ Na] 2 + PCl [Cl] 1 PK [K ] 1 + PNa [ Na] 1 + PCl [Cl] 2 where V is the voltage across the membrane (again, compartment 1 relative to the reference compartment 2) and P indicates the permeability of the Electrical Signals of Ner ve Cells 39 (A) Voltmeter (B) Na+ permeable K+ permeable PNa>> PK Membrane potential ENa 1 10 mM KCl 1 mM NaCl 2 1 mM KCl 10 mM NaCl Variable permeability to Na+ and K+ PNa 0 Resting potential PNa Action potential PK>>PNa Repolarization PK>>PNa EK Time membrane to each ion of interest. The Goldman equation is thus an extended version of the Nernst equation that takes into account the relative permeabilities of each of the ions involved. The relationship between the two equations becomes obvious in the situation where the membrane is permeable only to one ion, say, K+; in this case, the Goldman expression collapses back to the simpler Nernst equation. In this context, it is important to note that the valence factor (z) in the Nernst equation has been eliminated; this is why the concentrations of negatively charged chloride ions, Cl–, have been inverted relative to the concentrations of the positively charged ions [remember that –log (A/B) = log (B/A)]. If the membrane in Figure 2.6A is permeable to K+ and Na+ only, the terms involving Cl– drop out because PCl is 0. In this case, solution of the Goldman equation yields a potential of –58 mV when only K+ is permeant, +58 mV when only Na+ is permeant, and some intermediate value if both ions are permeant. For example, if K+ and Na+ were equally permeant, then the potential would be 0 mV. With respect to neural signaling, it is particularly pertinent to ask what would happen if the membrane started out being permeable to K+, and then temporarily switched to become most permeable to Na+. In this circumstance, the membrane potential would start out at a negative level, become positive while the Na+ permeability remained high, and then fall back to a negative level as the Na+ permeability decreased again. As it turns out, this last case essentially describes what goes on in a neuron during the generation of an action potential. In the resting state, PK of the neuronal plasma membrane is much higher than PNa; since, as a result of the action of ion transporters, there is always more K+ inside the cell than outside (Table 2.1), the resting potential is negative (Figure 2.6B). As the membrane potential is depolarized (by synaptic action, for example), PNa increases. The transient increase in Na+ permeability causes the membrane potential to become even more positive (red region in Figure 2.6B), because Na+ rushes in (there is much more Na+ outside a neuron than inside, again as a result of ion pumps). Because of this positive feedback loop, an action potential occurs. The rise in Na+ permeability during the action potential is transient, however; as the membrane permeability to K+ is restored, the membrane potential quickly returns to its resting level. Figure 2.6 Resting and action potentials entail permeabilities to different ions. (A) Hypothetical situation in which a membrane variably permeable to Na+ (red) and K+ (yellow) separates two compartments that contain both ions. For simplicity, Cl– ions are not shown in the diagram. (B) Schematic representation of the membrane ionic permeabilities associated with resting and action potentials. At rest, neuronal membranes are more permeable to K+ (yellow) than to Na+ (red); accordingly, the resting membrane potential is negative and approaches the equilibrium potential for K+, EK. During an action potential, the membrane becomes very permeable to Na+ (red); thus the membrane potential becomes positive and approaches the equilibrium potential for Na+, ENa. The rise in Na+ permeability is transient, however, so that the membrane again becomes primarily permeable to K+ (yellow), causing the potential to return to its negative resting value. Notice that at the equilibrium potential for a given ion, there is no net flux of that ion across the membrane. 40 Chapter Two TABLE 2.1 Extracellular and Intracellular Ion Concentrations Concentration (mM) Ion Squid neuron Potassium (K+) Sodium (Na+) Chloride (Cl–) Calcium (Ca2+) Mammalian neuron Potassium (K+) Sodium (Na+) Chloride (Cl–) Calcium (Ca2+) Intracellular Extracellular 400 50 40–150 0.0001 20 440 560 10 140 5–15 4–30 0.0001 5 145 110 1–2 Armed with an appreciation of these simple electrochemical principles, it will be much easier to understand the following, more detailed account of how neurons generate resting and action potentials. The Ionic Basis of the Resting Membrane Potential The action of ion transporters creates substantial transmembrane gradients for most ions. Table 2.1 summarizes the ion concentrations measured directly in an exceptionally large nerve cell found in the nervous system of the squid (Box A). Such measurements are the basis for stating that there is much more K+ inside the neuron than out, and much more Na+ outside than in. Similar concentration gradients occur in the neurons of most animals, including humans. However, because the ionic strength of mammalian blood is lower than that of sea-dwelling animals such as squid, in mammals the concentrations of each ion are several times lower. These transporterdependent concentration gradients are, indirectly, the source of the resting neuronal membrane potential and the action potential. Once the ion concentration gradients across various neuronal membranes are known, the Nernst equation can be used to calculate the equilibrium potential for K+ and other major ions. Since the resting membrane potential of the squid neuron is approximately –65 mV, K+ is the ion that is closest to being in electrochemical equilibrium when the cell is at rest. This fact implies that the resting membrane is more permeable to K+ than to the other ions listed in Table 2.1, and that this permeability is the source of resting potentials. It is possible to test this guess, as Alan Hodgkin and Bernard Katz did in 1949, by asking what happens to the resting membrane potential if the concentration of K+ outside the neuron is altered. If the resting membrane were permeable only to K+, then the Goldman equation (or even the simpler Nernst equation) predicts that the membrane potential will vary in proportion to the logarithm of the K+ concentration gradient across the membrane. Assuming that the internal K+ concentration is unchanged during the experiment, a plot of membrane potential against the logarithm of the external K+ concentration should yield a straight line with a slope of 58 mV per tenfold change in external K+ concentration at room temperature (see Figure 2.4C). (The slope becomes about 61 mV at mammalian body temperatures.) Electrical Signals of Ner ve Cells 41 Box A The Remarkable Giant Nerve Cells of Squid Many of the initial insights into how ion concentration gradients and changes in membrane permeability produce electrical signals came from experiments performed on the extraordinarily large nerve cells of the squid. The axons of these nerve cells can be up to 1 mm in diameter—100 to 1000 times larger than mammalian axons. Thus, squid axons are large enough to allow experiments that would be impossible on most other nerve cells. For example, it is not difficult to insert simple wire electrodes inside these giant axons and make reliable electrical measurements. The relative ease of this approach yielded the first intracellular recordings of action potentials from nerve cells and, as discussed in the next chapter, the first experimental measure- ments of the ion currents that produce action potentials. It also is practical to extrude the cytoplasm from giant axons and measure its ionic composition (see Table 2.1). In addition, some giant nerve cells form synaptic contacts with other giant nerve cells, producing very large synapses that have been extraordinarily valuable in understanding the fundamental mechanisms of synaptic transmission (see Chapter 5). Giant neurons evidently evolved in squid because they enhanced survival. These neurons participate in a simple neural circuit that activates the contraction of the mantle muscle, producing a jet propulsion effect that allows the squid to move away from predators at a remarkably fast speed. As discussed in Chapter 3, larger axonal diameter allows faster conduction of action potentials. Thus, presumably these huge nerve cells help squid escape more successfully from their numerous enemies. Today—nearly 70 years after their discovery by John Z. Young at University College London—the giant nerve cells of squid remain useful experimental systems for probing basic neuronal functions. References LLINÁS, R. (1999) The Squid Synapse: A Model for Chemical Transmission. Oxford: Oxford University Press. YOUNG, J. Z. (1939) Fused neurons and synaptic contacts in the giant nerve fibres of cephalopods. Phil. Trans. R. Soc. Lond. 229(B): 465–503. (A) Diagram of a squid, showing the location of its giant nerve cells. Different colors indicate the neuronal components of the escape circuitry. The first- and second-level neurons originate in the brain, while the third-level neurons are in the stellate ganglion and innervate muscle cells of the mantle. (B) Giant synapses within the stellate ganglion. The second-level neuron forms a series of fingerlike processes, each of which makes an extraordinarily large synapse with a single third-level neuron. (C) Structure of a giant axon of a third-level neuron lying within its nerve. The enormous difference in the diameters of a squid giant axon and a mammalian axon are shown below. Giant axon Brain Presynaptic (2nd level) 1st-level neuron Stellate nerve Smaller axons 2nd-level neuron Stellate ganglion 3rd-level neuron Stellate nerve with giant axon (A) (B) Postsynaptic (3rd level) Cross section 1 mm 1 mm (C) Squid giant axon = 800 µm diameter Mammalian axon = 2 µm diameter 42 Chapter Two Resting membrane potential (mV) (A) 200 mM K+ 0 −20 −40 3.5 mM K+ −60 −80 10 mM K+ 0 20 mM K+ 50 mM K+ 450 mM K+ 5 Time (min) 10 Resting membrane potential (mV) (B) 0 −20 −40 Slope = 58 mV per tenfold change in K+ gradient −60 −80 2 5 10 20 50 100 200 500 [K+]out (mM) Figure 2.7 Experimental evidence that the resting membrane potential of a squid giant axon is determined by the K+ concentration gradient across the membrane. (A) Increasing the external K+ concentration makes the resting membrane potential more positive. (B) Relationship between resting membrane potential and external K+ concentration, plotted on a semi-logarithmic scale. The straight line represents a slope of 58 mV per tenfold change in concentration, as given by the Nernst equation. (After Hodgkin and Katz, 1949.) When Hodgkin and Katz carried out this experiment on a living squid neuron, they found that the resting membrane potential did indeed change when the external K+ concentration was modified, becoming less negative as external K+ concentration was raised (Figure 2.7A). When the external K+ concentration was raised high enough to equal the concentration of K+ inside the neuron, thus making the K+ equilibrium potential 0 mV, the resting membrane potential was also approximately 0 mV. In short, the resting membrane potential varied as predicted with the logarithm of the K+ concentration, with a slope that approached 58 mV per tenfold change in K+ concentration (Figure 2.7B). The value obtained was not exactly 58 mV because other ions, such as Cl– and Na+, are also slightly permeable, and thus influence the resting potential to a small degree. The contribution of these other ions is particularly evident at low external K+ levels, again as predicted by the Goldman equation. In general, however, manipulation of the external concentrations of these other ions has only a small effect, emphasizing that K+ permeability is indeed the primary source of the resting membrane potential. In summary, Hodgkin and Katz showed that the inside-negative resting potential arises because (1) the membrane of the resting neuron is more permeable to K+ than to any of the other ions present, and (2) there is more K+ inside the neuron than outside. The selective permeability to K+ is caused by K+-permeable membrane channels that are open in resting neurons, and the Electrical Signals of Ner ve Cells 43 large K+ concentration gradient is, as noted, produced by membrane transporters that selectively accumulate K+ within neurons. Many subsequent studies have confirmed the general validity of these principles. The Ionic Basis of Action Potentials What causes the membrane potential of a neuron to depolarize during an action potential? Although a general answer to this question has been given (increased permeability to Na+), it is well worth examining some of the experimental support for this concept. Given the data presented in Table 2.1, one can use the Nernst equation to calculate that the equilibrium potential for Na+ (ENa) in neurons, and indeed in most cells, is positive. Thus, if the membrane were to become highly permeable to Na+, the membrane potential would approach ENa. Based on these considerations, Hodgkin and Katz hypothesized that the action potential arises because the neuronal membrane becomes temporarily permeable to Na+. Taking advantage of the same style of ion substitution experiment they used to assess the resting potential, Hodgkin and Katz tested the role of Na+ in generating the action potential by asking what happens to the action potential when Na+ is removed from the external medium. They found that lowering the external Na+ concentration reduces both the rate of rise of the action potential and its peak amplitude (Figure 2.8A–C). Indeed, when they examined this Na+ dependence quantitatively, they found a more-or-less linear relationship between the amplitude of the action potential and the logarithm of the external Na+ concentration (Figure 2.8D). The slope of this rela(D) 100 +40 Control 0 −40 −80 0 1 2 Time (ms) 3 Action potential amplitude (mV) Membrane potential (mV) (A) 80 60 Slope = 58 mV per tenfold change in Na+ gradient 40 20 50 +40 Low [Na+] 0 0 1 2 Time (ms) 3 Membrane potential (mV) (C) +40 Recovery 1000 −40 0 1 2 Time (ms) −20 −40 −60 −80 0 −80 200 500 [Na+]out (mM) 0 −40 −80 100 (E) Resting membrane potential (mV) Membrane potential (mV) (B) 3 50 100 500 200 [Na+]out (mM) 1000 Figure 2.8 The role of sodium in the generation of an action potential in a squid giant axon. (A) An action potential evoked with the normal ion concentrations inside and outside the cell. (B) The amplitude and rate of rise of the action potential diminish when external sodium concentration is reduced to onethird of normal, but (C) recover when the Na+ is replaced. (D) While the amplitude of the action potential is quite sensitive to the external concentration of Na+, the resting membrane potential (E) is little affected by changing the concentration of this ion. (After Hodgkin and Katz, 1949.) 44 Chapter Two Box B Action Potential Form and Nomenclature The action potential of the squid giant axon has a characteristic shape, or waveform, with a number of different phases (Figure A). During the rising phase, the membrane potential rapidly depolarizes. In fact, action potentials cause the membrane potential to depolarize so much that the membrane potential transiently becomes positive with respect to the external medium, producing an overshoot. The overshoot of the action potential gives way to a falling phase in which the membrane potential rapidly repolarizes. Repolarization takes the membrane potential to levels even more negative than the resting membrane potential for a short time; this brief period of hyperpolarization is called the undershoot. Although the waveform of the squid action potential is typical, the details of the action potential form vary widely from neuron to neuron in different animals. In myelinated axons of vertebrate motor neurons (Figure B), the action potential is virtually indistinguishable from that of the squid axon. However, the action potential recorded in the cell body of this same motor neuron (Figure C) looks rather different. Thus, the action potential waveform can vary even within the same neuron. More complex action potentials are seen in other central neurons. For example, action potentials recorded from the cell bodies of neurons in the mammalian inferior olive (a region of the brainstem involved in motor control) last tens of milliseconds (Figure D). These action potentials exhibit a pronounced plateau during their falling phase, and their undershoot lasts even longer than that of the motor neuron. One of the most dramatic types of action potentials occurs in the cell bodies of cerebellar Purkinje neurons (Figure E). These potentials have several complex phases that result from the summation of multiple, discrete action potentials. The variety of action potential waveforms could mean that each type of neuron has a different mechanism of action potential production. Fortunately, however, these diverse waveforms all result from relatively minor variations in the scheme used by the squid giant axon. For example, plateaus in the repolarization phase result from the presence of ion channels that are permeable to Ca2+, and long-lasting undershoots result from the presence of additional types of membrane K+ channels. The complex action potential of the Purkinje cell results from these extra features plus the fact that different types of action potentials are generated in various parts of the Purkinje neuron—cell body, dendrites, and axons—and are summed together in recordings from the cell body. Thus, the lessons learned from the squid axon are applicable to, and indeed essential for, understanding action potential generation in all neurons. References BARRETT, E. F. AND J. N. BARRETT (1976) Separation of two voltage-sensitive potassium currents, and demonstration of a tetrodotoxin-resistant calcium current in frog motoneurones. J. Physiol. (Lond.) 255: 737–774. DODGE, F. A. AND B. FRANKENHAEUSER (1958) Membrane currents in isolated frog nerve fibre under voltage clamp conditions. J. Physiol. (Lond.) 143: 76–90. HODGKIN, A. L. AND A. F. HUXLEY (1939) Action potentials recorded from inside a nerve fibre. Nature 144: 710–711. LLINÁS, R. AND M. SUGIMORI (1980) Electrophysiological properties of in vitro Purkinje cell dendrites in mammalian cerebellar slices. J. Physiol. (Lond.) 305: 197–213. LLINÁS, R. AND Y. YAROM (1981) Electrophysiology of mammalian inferior olivary neurones in vitro. Different types of voltagedependent ionic conductances. J. Physiol. (Lond.) 315: 549–567. (A) The phases of an action potential of the squid giant axon. (B) Action potential recorded from a myelinated axon of a frog motor neuron. (C) Action potential recorded from the cell body of a frog motor neuron. The action potential is smaller and the undershoot prolonged in comparison to the action potential recorded from the axon of this same neuron (B). (D) Action potential recorded from the cell body of a neuron from the inferior olive of a guinea pig. This action potential has a pronounced plateau during its falling phase. (E) Action potential recorded from the cell body of a Purkinje neuron in the cerebellum of a guinea pig. (A after Hodgkin and Huxley, 1939; B after Dodge and Frankenhaeuser, 1958; C after Barrett and Barrett, 1976; D after Llinás and Yarom, 1981; E after Llinás and Sugimori, 1980.) Membrane potential (mV) (A) (B) +40 0 Rising phase (D) (E) Overshoot phase Falling phase Undershoot phase −40 0 (C) 2 4 6 8 0 1 2 3 4 0 2 Time (ms) 4 6 8 0 10 20 30 40 0 50 100 150 Electrical Signals of Ner ve Cells 45 tionship approached a value of 58 mV per tenfold change in Na+ concentration, as expected for a membrane selectively permeable to Na+. In contrast, lowering Na+ concentration had very little effect on the resting membrane potential (Figure 2.8E). Thus, while the resting neuronal membrane is only slightly permeable to Na+, the membrane becomes extraordinarily permeable to Na+ during the rising phase and overshoot phase of the action potential (see Box B for an explanation of action potential nomenclature). This temporary increase in Na+ permeability results from the opening of Na+-selective channels that are essentially closed in the resting state. Membrane pumps maintain a large electrochemical gradient for Na+, which is in much higher concentration outside the neuron than inside. When the Na+ channels open, Na+ flows into the neuron, causing the membrane potential to depolarize and approach ENa. The time that the membrane potential lingers near ENa (about +58 mV) during the overshoot phase of an action potential is brief because the increased membrane permeability to Na+ itself is short-lived. The membrane potential rapidly repolarizes to resting levels and is actually followed by a transient undershoot. As will be described in Chapter 3, these latter events in the action potential are due to an inactivation of the Na+ permeability and an increase in the K+ permeability of the membrane. During the undershoot, the membrane potential is transiently hyperpolarized because K+ permeability becomes even greater than it is at rest. The action potential ends when this phase of enhanced K+ permeability subsides, and the membrane potential thus returns to its normal resting level. The ion substitution experiments carried out by Hodgkin and Katz provided convincing evidence that the resting membrane potential results from a high resting membrane permeability to K+, and that depolarization during an action potential results from a transient rise in membrane Na+ permeability. Although these experiments identified the ions that flow during an action potential, they did not establish how the neuronal membrane is able to change its ionic permeability to generate the action potential, or what mechanisms trigger this critical change. The next chapter addresses these issues, documenting the surprising conclusion that the neuronal membrane potential itself affects membrane permeability. Summary Nerve cells generate electrical signals to convey information over substantial distances and to transmit it to other cells by means of synaptic connections. These signals ultimately depend on changes in the resting electrical potential across the neuronal membrane. A resting potential occurs because nerve cell membranes are permeable to one or more ion species subject to an electrochemical gradient. More specifically, a negative membrane potential at rest results from a net efflux of K+ across neuronal membranes that are predominantly permeable to K+. In contrast, an action potential occurs when a transient rise in Na+ permeability allows a net flow of Na+ in the opposite direction across the membrane that is now predominantly permeable to Na+. The brief rise in membrane Na+ permeability is followed by a secondary, transient rise in membrane K+ permeability that repolarizes the neuronal membrane and produces a brief undershoot of the action potential. As a result of these processes, the membrane is depolarized in an all-or-none fashion during an action potential. When these active permeability changes subside, the membrane potential returns to its resting level because of the high resting membrane permeability to K+. 46 Chapter Two Additional Reading Reviews HODGKIN, A. L. (1951) The ionic basis of electrical activity in nerve and muscle. Biol. Rev. 26: 339–409. HODGKIN, A. L. (1958) The Croonian Lecture: Ionic movements and electrical activity in giant nerve fibres. Proc. R. Soc. Lond. (B) 148: 1–37. Important Original Papers BAKER, P. F., A. L. HODGKIN AND T. I. SHAW (1962) Replacement of the axoplasm of giant nerve fibres with artificial solutions. J. Physiol. (London) 164: 330–354. COLE, K. S. AND H. J. CURTIS (1939) Electric impedence of the squid giant axon during activity. J. Gen. Physiol. 22: 649–670. GOLDMAN, D. E. (1943) Potential, impedence, and rectification in membranes. J. Gen. Physiol. 27: 37–60. HODGKIN, A. L. AND P. HOROWICZ (1959) The influence of potassium and chloride ions on the membrane potential of single muscle fibres. J. Physiol. (London) 148: 127–160. HODGKIN, A. L. AND B. KATZ (1949) The effect of sodium ions on the electrical activity of the giant axon of the squid. J. Physiol. (London) 108: 37–77. HODGKIN, A. L. AND R. D. KEYNES (1953) The mobility and diffusion coefficient of potassium in giant axons from Sepia. J. Physiol. (London) 119: 513–528. KEYNES, R. D. (1951) The ionic movements during nervous activity. J. Physiol. (London) 114: 119–150. Books HODGKIN, A. L. (1967) The Conduction of the Nervous Impulse. Springfield, IL: Charles C. Thomas. HODGKIN, A. L. (1992) Chance and Design. Cambridge: Cambridge University Press. JUNGE, D. (1992) Nerve and Muscle Excitation, 3rd Ed. Sunderland, MA: Sinauer Associates. KATZ, B. (1966) Nerve, Muscle, and Synapse. New York: McGraw-Hill. Chapter 3 Overview The action potential, the primary electrical signal generated by nerve cells, reflects changes in membrane permeability to specific ions. Present understanding of these changes in ionic permeability is based on evidence obtained by the voltage clamp technique, which permits detailed characterization of permeability changes as a function of membrane potential and time. For most types of axons, these changes consist of a rapid and transient rise in sodium (Na+) permeability, followed by a slower but more prolonged rise in potassium (K+) permeability. Both permeabilities are voltage-dependent, increasing as the membrane potential depolarizes. The kinetics and voltage dependence of Na+ and K+ permeabilities provide a complete explanation of action potential generation. Depolarizing the membrane potential to the threshold level causes a rapid, self-sustaining increase in Na+ permeability that produces the rising phase of the action potential; however, the Na+ permeability increase is short-lived and is followed by a slower increase in K+ permeability that restores the membrane potential to its usual negative resting level. A mathematical model that describes the behavior of these ionic permeabilities predicts virtually all of the observed properties of action potentials. Importantly, this same ionic mechanism permits action potentials to be propagated along the length of neuronal axons, explaining how electrical signals are conveyed throughout the nervous system. Ionic Currents Across Nerve Cell Membranes The previous chapter introduced the idea that nerve cells generate electrical signals by virtue of a membrane that is differentially permeable to various ion species. In particular, a transient increase in the permeability of the neuronal membrane to Na+ initiates the action potential. This chapter considers exactly how this increase in Na+ permeability occurs. A key to understanding this phenomenon is the observation that action potentials are initiated only when the neuronal membrane potential becomes more positive than a threshold level. This observation suggests that the mechanism responsible for the increase in Na+ permeability is sensitive to the membrane potential. Therefore, if one could understand how a change in membrane potential activates Na+ permeability, it should be possible to explain how action potentials are generated. The fact that the Na+ permeability that generates the membrane potential change is itself sensitive to the membrane potential presents both conceptual and practical obstacles to studying the mechanism of the action potential. A practical problem is the difficulty of systematically varying the membrane 47 VoltageDependent Membrane Permeability 48 Chapter Three Box A The Voltage Clamp Method Breakthroughs in scientific research often rely on the development of new technologies. In the case of the action potential, detailed understanding came only after of the invention of the voltage clamp technique by Kenneth Cole in the 1940s. This device is called a voltage clamp because it controls, or clamps, membrane potential (or voltage) at any level desired by the experimenter. The method measures the membrane potential with a microelectrode (or other type of electrode) placed inside the cell (1), and electronically compares this voltage to the voltage to be maintained (called the command voltage) (2). The clamp circuitry then passes a current back into the cell though another intracellular elec- trode (3). This electronic feedback circuit holds the membrane potential at the desired level, even in the face of permeability changes that would normally alter the membrane potential (such as those generated during the action potential). Most importantly, the device permits the simultaneous measurement of the current needed to keep the cell at a given voltage (4). This current is exactly equal to the amount of current flowing across the neuronal membrane, allowing direct measurement of these membrane currents. Therefore, the voltage clamp technique can indicate how membrane potential influences ionic current flow across the membrane. This information gave Hodgkin and Huxley the key 1 One internal electrode measures membrane potential (Vm) and is connected to the voltage clamp amplifier 2 Voltage clamp amplifier compares membrane potential to the desired (command) potential Measure Vm Reference electrode − Command voltage insights that led to their model for action potential generation. Today, the voltage clamp method remains widely used to study ionic currents in neurons and other cells. The most popular contemporary version of this approach is the patch clamp technique, a method that can be applied to virtually any cell and has a resolution high enough to measure the minute electrical currents flowing through single ion channels (see Box A in Chapter 4). References COLE, K. S. (1968) Membranes, Ions and Impulses: A Chapter of Classical Biophysics. Berkeley, CA: University of California Press. 3 When Vm is different from the command potential, the clamp amplifier injects current into the axon through a second electrode. This feedback arrangement causes the membrane potential to become the same as the command potential + Voltage clamp amplifier Measure current 4 The current flowing back into the axon, and thus across its membrane, can be measured here Saline solution Squid axon Recording electrode Currentpassing electrode Voltage clamp technique for studying membrane currents of a squid axon. potential to study the permeability change, because such changes in membrane potential will produce an action potential, which causes further, uncontrolled changes in the membrane potential. Historically, then, it was not really possible to understand action potentials until a technique was developed that allowed experimenters to control membrane potential and simultaneously measure the underlying permeability changes. This tech- Voltage-Dependent Membrane Permeability 49 nique, the voltage clamp method (Box A), provides the information needed to define the ionic permeability of the membrane at any level of membrane potential. In the late 1940s, Alan Hodgkin and Andrew Huxley working at the University of Cambridge used the voltage clamp technique to work out the permeability changes underlying the action potential. They again chose to use the giant neuron of the squid because its large size (up to 1 mm in diameter; see Box A in Chapter 2) allowed insertion of the electrodes necessary for voltage clamping. They were the first investigators to test directly the hypothesis that potential-sensitive Na+ and K+ permeability changes are both necessary and sufficient for the production of action potentials. Hodgkin and Huxley’s first goal was to determine whether neuronal membranes do, in fact, have voltage-dependent permeabilities. To address this issue, they asked whether ionic currents flow across the membrane when its potential is changed. The result of one such experiment is shown in Figure 3.1. Figure 3.1A illustrates the currents produced by a squid axon when its membrane potential, Vm, is hyperpolarized from the resting level of –65 mV to –130 mV. The initial response of the axon results from the redistribution of charge across the axonal membrane. This capacitive current is nearly instantaneous, ending within a fraction of a millisecond. Aside from this brief event, very little current flows when the membrane is hyperpolarized. However, when the membrane potential is depolarized from –65 mV to 0 mV, the response is quite different (Figure 3.1B). Following the capacitive current, the axon produces a rapidly rising inward ionic current (inward refers to a positive charge entering the cell—that is, cations in or anions out), which gives way to a more slowly rising, delayed outward current. The fact that membrane depolarization elicits these ionic currents establishes that the membrane permeability of axons is indeed voltage-dependent. Two Types of Voltage-Dependent Ionic Current The results shown in Figure 3.1 demonstrate that the ionic permeability of neuronal membranes is voltage-sensitive, but the experiments do not identify how many types of permeability exist, or which ions are involved. As discussed in Chapter 2 (see Figure 2.5), varying the potential across a membrane makes it possible to deduce the equilibrium potential for the ionic fluxes through the membrane, and thus to identify the ions that are flowing. (B) Membrane potential (mV) (A) Membrane current (mA/cm2) Figure 3.1 Current flow across a squid axon membrane during a voltage clamp experiment. (A) A 65 mV hyperpolarization of the membrane potential produces only a very brief capacitive current. (B) A 65 mV depolarization of the membrane potential also produces a brief capacitive current, which is followed by a longer lasting but transient phase of inward current and a delayed but sustained outward current. (After Hodgkin et al., 1952.) 0 0 −65 −65 65 mV Depolarization 65 mV Hyperpolarization −130 −130 +1 Outward +1 Capacitive current Outward 0 Delayed outward current 0 Capacitive current Inward −1 0 1 2 Time (ms) Inward −1 3 4 Transient inward current 0 1 2 Time (ms) 3 4 Membrane potential (mV) 50 Chapter Three +65 75 50 +52 +26 25 0 0 −26 −25 −50 Membrane current (mA/cm2) 7 6 4 2 0 −2 0 2 4 6 8 0 2 4 6 8 0 2 4 6 Time (ms) 8 0 2 4 6 8 0 2 4 6 8 Figure 3.2 Current produced by membrane depolarizations to several different potentials. The early current first increases, then decreases in magnitude as the depolarization increases; note that this current is actually reversed in polarity at potentials more positive than about +55 mV. In contrast, the late current increases monotonically with increasing depolarization. (After Hodgkin et al., 1952.) Membrane current (mA/cm2) 3.0 Because the voltage clamp method allows the membrane potential to be changed while ionic currents are being measured, it was a straightforward matter for Hodgkin and Huxley to determine ionic permeability by examining how the properties of the early inward and late outward currents changed as the membrane potential was varied (Figure 3.2). As already noted, no appreciable ionic currents flow at membrane potentials more negative than the resting potential. At more positive potentials, however, the currents not only flow but change in magnitude. The early current has a Ushaped dependence on membrane potential, increasing over a range of depolarizations up to approximately 0 mV but decreasing as the potential is depolarized further. In contrast, the late current increases monotonically with increasingly positive membrane potentials. These different responses to membrane potential can be seen more clearly when the magnitudes of the two current components are plotted as a function of membrane potential, as in Figure 3.3. The voltage sensitivity of the early inward current gives an important clue about the nature of the ions carrying the current, namely, that no current flows when the membrane potential is clamped at +52 mV. For the squid neurons studied by Hodgkin and Huxley, the external Na+ concentration is 440 mM, and the internal Na+ concentration is 50 mM. For this concentration gradient, the Nernst equation predicts that the equilibrium poten- Late 2.0 1.0 0 Early 0 −100 −50 50 Membrane potential (mV) Figure 3.3 Relationship between current amplitude and membrane potential, taken from experiments such as the one shown in Figure 3.2. Whereas the late outward current increases steeply with increasing depolarization, the early inward current first increases in magnitude, but then decreases and reverses to outward current at about +55 mV (the sodium equilibrium potential). (After Hodgkin et al., 1952.) Figure 3.4 Dependence of the early inward current on sodium. In the presence of normal external concentrations of Na+, depolarization of a squid axon to 0 mV produces an inward initial current. However, removal of external Na+ causes the initial inward current to become outward, an effect that is reversed by restoration of external Na+. (After Hodgkin and Huxley, 1952a.) Membrane potential (mV) Voltage-Dependent Membrane Permeability 51 25 0 −25 −50 −75 +1 + tial for Na should be +55 mV. Recall further from Chapter 2 that at the Na equilibrium potential there is no net flux of Na+ across the membrane, even if the membrane is highly permeable to Na+. Thus, the experimental observation that no current flows at the membrane potential where Na+ cannot flow is a strong indication that the early inward current is carried by entry of Na+ into the axon. An even more demanding way to test whether Na+ carries the early inward current is to examine the behavior of this current after removing external Na+ . Removing the Na+ outside the axon makes ENa negative; if the permeability to Na+ is increased under these conditions, current should flow outward as Na+ leaves the neuron, due to the reversed electrochemical gradient. When Hodgkin and Huxley performed this experiment, they obtained the result shown in Figure 3.4. Removing external Na+ caused the early inward current to reverse its polarity and become an outward current at a membrane potential that gave rise to an inward current when external Na+ was present. This result demonstrates convincingly that the early inward current measured when Na+ is present in the external medium must be due to Na+ entering the neuron. Notice that removal of external Na+ in the experiment shown in Figure 3.4 has little effect on the outward current that flows after the neuron has been kept at a depolarized membrane voltage for several milliseconds. This further result shows that the late outward current must be due to the flow of an ion other than Na+. Several lines of evidence presented by Hodgkin, Huxley, and others showed that this late outward current is caused by K+ exiting the neuron. Perhaps the most compelling demonstration of K+ involvement is that the amount of K+ efflux from the neuron, measured by loading the neuron with radioactive K+, is closely correlated with the magnitude of the late outward current. Taken together, these experiments using the voltage clamp show that changing the membrane potential to a level more positive than the resting potential produces two effects: an early influx of Na+ into the neuron, followed by a delayed efflux of K+. The early influx of Na+ produces a transient inward current, whereas the delayed efflux of K+ produces a sustained outward current. The differences in the time course and ionic selectivity of the two fluxes suggest that two different ionic permeability mechanisms are activated by changes in membrane potential. Confirmation that there are indeed two distinct mechanisms has come from pharmacological studies of drugs that specifically affect these two currents (Figure 3.5). Tetrodotoxin, an alkaloid neurotoxin found in certain puffer fish, tropical frogs, and salamanders, blocks the Na+ current without affecting the K+ current. Conversely, tetraethylammonium ions block K+ currents without affecting Na+ currents. The differential sensitivity of Na+ and K+ currents to these drugs provides strong additional evidence that Na+ and K+ flow through independent permeability pathways. As discussed in Chapter 4, it is now known that these pathways are ion channels that are selectively permeable to either Na+ or K+. In fact, tetrodotoxin, tetraethylammonium, and other drugs that interact with spe- 460 mM Na+ 0 Membrane current (mA/cm2) + Early current is inward −1 +1 Na+-free 0 Early current is outward −1 +1 460 mM Na+ 0 −1 Early current is inward again 0 2 4 Time (ms) 6 8 Membrane current (mA/cm2 ) Figure 3.5 Pharmacological separation of Na+ and K+ currents into sodium and potassium components. Panel (1) shows the current that flows when the membrane potential of a squid axon is depolarized to 0 mV in control conditions. (2) Treatment with tetrodotoxin causes the early Na+ currents to disappear but spares the late K+ currents. (3) Addition of tetraethylammonium blocks the K+ currents without affecting the Na+ currents. (After Moore et al., 1967 and Armstrong and Binstock, 1965.) Membrane potential (mV) 52 Chapter Three 25 0 −25 −50 −75 +1 (1) K+ current blocked (3) Add tetraethylammonium 0 −1 0 5 Time (ms) 10 0 5 Time (ms) 10 Add tetrodotoxin +1 (2) 0 Na+ current blocked −1 0 5 Time (ms) 10 cific types of ion channels have been extraordinarily useful tools in characterizing these channel molecules (see Chapter 4). Two Voltage-Dependent Membrane Conductances The next goal Hodgkin and Huxley set for themselves was to describe Na+ and K+ permeability changes mathematically. To do this, they assumed that the ionic currents are due to a change in membrane conductance, defined as the reciprocal of the membrane resistance. Membrane conductance is thus closely related, although not identical, to membrane permeability. When evaluating ionic movements from an electrical standpoint, it is convenient to describe them in terms of ionic conductances rather than ionic permeabilities. For present purposes, permeability and conductance can be considered synonymous. If membrane conductance (g) obeys Ohm’s Law (which states that voltage is equal to the product of current and resistance), then the ionic current that flows during an increase in membrane conductance is given by Iion = gion (Vm – Eion) where Iion is the ionic current, Vm is the membrane potential, and Eion is the equilibrium potential for the ion flowing through the conductance, gion. The difference between Vm and Eion is the electrochemical driving force acting on the ion. Hodgkin and Huxley used this simple relationship to calculate the dependence of Na+ and K+ conductances on time and membrane potential. They knew Vm, which was set by their voltage clamp device (Figure 3.6A), and could determine ENa and EK from the ionic concentrations on the two sides Voltage-Dependent Membrane Permeability 53 Membrane potential (mV) (A) 50 25 0 −25 −50 −75 + 44 + 23 −2 −27 −39 Na+ conductance mSiemens/cm2 Membrane current mA/cm2 (B) 6 4 2 0 −2 30 20 10 0 60 K+ conductance mSiemens/cm2 (C) (D) 40 20 0 0 2 4 6 8 0 2 4 6 8 0 2 4 6 Time (ms) 8 0 of the axonal membrane (see Table 2.1). The currents carried by Na+ and K+—INa and IK—could be determined separately from recordings of the membrane currents resulting from depolarization (Figure 3.6B) by measuring the difference between currents recorded in the presence and absence of external Na+ (as shown in Figure 3.4). From these measurements, Hodgkin and Huxley were able to calculate gNa and gK (Figure 3.6C,D), from which they drew two fundamental conclusions. The first conclusion is that the Na+ and K+ conductances change over time. For example, both Na+ and K+ conductances require some time to activate, or turn on. In particular, the K+ conductance has a pronounced delay, requiring several milliseconds to reach its maximum (Figure 3.6D), whereas the Na+ conductance reaches its maximum more rapidly (Figure 3.6C). The more rapid activation of the Na+ conductance allows the resulting inward Na+ current to precede the delayed outward K+ current (see Figure 3.6B). Although the Na+ conductance rises rapidly, it quickly declines, even though the membrane potential is kept at a depolarized level. This fact shows that depolarization not only causes the Na+ conductance to activate, but also causes it to decrease over time, or inactivate. The K+ conductance of the squid axon does not inactivate in this way; thus, while the Na+ and K+ conductances share the property of time-dependent activation, only the Na+ conductance inactivates. (Inactivating K+ conductances have since been discovered in other types of nerve cells; see Chapter 4.) The time courses of the Na+ and K+ conductances are voltage- 2 4 6 8 0 2 4 6 8 Figure 3.6 Membrane conductance changes underlying the action potential are time- and voltage-dependent. Depolarizations to various membrane potentials (A) elicit different membrane currents (B). Below are shown the Na+ (C) and K+ (D) conductances calculated from these currents. Both peak Na+ conductance and steady-state K+ conductance increase as the membrane potential becomes more positive. In addition, the activation of both conductances, as well as the rate of inactivation of the Na+ conductance, occur more rapidly with larger depolarizations. (After Hodgkin and Huxley, 1952b.) 54 Chapter Three 40 15 10 Na+ 5 Conductance (mSiemens/cm2 ) 20 Conductance (mSiemens/cm2 ) Figure 3.7 Depolarization increases Na+ and K+ conductances of the squid giant axon. The peak magnitude of Na+ conductance and steady-state value of K+ conductance both increase steeply as the membrane potential is depolarized. (After Hodgkin and Huxley, 1952b.) 30 20 K+ 10 0 −80 −60 −40 −20 0 20 40 Membrane potential (mV) 0 −80 −60 −40 −20 0 20 40 Membrane potential (mV) dependent, with the speed of both activation and inactivation increasing at more depolarized potentials. This finding accounts for more rapid time courses of membrane currents measured at more depolarized potentials. The second conclusion derived from Hodgkin and Huxley’s calculations is that both the Na+ and K+ conductances are voltage-dependent—that is, both conductances increase progressively as the neuron is depolarized. Figure 3.7 illustrates this by plotting the relationship between peak value of the conductances (from Figure 3.6C,D) against the membrane potential. Note the similar voltage dependence for each conductance; both conductances are quite small at negative potentials, maximal at very positive potentials, and exquisitely dependent on membrane voltage at intermediate potentials. The observation that these conductances are sensitive to changes in membrane potential shows that the mechanism underlying the conductances somehow “senses” the voltage across the membrane. All told, the voltage clamp experiments carried out by Hodgkin and Huxley showed that the ionic currents that flow when the neuronal membrane is depolarized are due to three different voltage-sensitive processes: (1) activation of Na+ conductance, (2) activation of K+ conductance, and (3) inactivation of Na+ conductance. Reconstruction of the Action Potential From their experimental measurements, Hodgkin and Huxley were able to construct a detailed mathematical model of the Na+ and K+ conductance changes. The goal of these modeling efforts was to determine whether the Na+ and K+ conductances alone are sufficient to produce an action potential. Using this information, they could in fact generate the form and time course of the action potential with remarkable accuracy (Figure 3.8A). Further, the Hodgkin-Huxley model predicted other features of action potential behavior in the squid axon, such as how the delay before action potential generation changes in response to stimulating currents of different intensities (Figure 3.8B,C). The model also predicted that the axon membrane would become refractory to further excitation for a brief period following an action potential, as was experimentally observed. The Hodgkin-Huxley model also provided many insights into how action potentials are generated. Figure 3.8A shows a reconstructed action potential, together with the time courses of the underlying Na+ and K+ conductances. The coincidence of the initial increase in Na+ conductance with the rapid rising phase of the action potential demonstrates that a selective increase in Voltage-Dependent Membrane Permeability 55 Figure 3.8 Mathematical reconstruction of the action potential. (A) Reconstruction of an action potential (black curve) together with the underlying changes in Na+ (red curve) and K+ (yellow curve) conductance. The size and time course of the action potential were calculated using only the properties of gNa and gK measured in voltage clamp experiments. Real action potentials evoked by brief current pulses of different intensities (B) are remarkably similar to those generated by the mathematical model (C). The reconstructed action potentials shown in (A) and (C) differ in duration because (A) simulates an action potential at 19°C, whereas (C) simulates an action potential at 6°C. (After Hodgkin and Huxley, 1952d.) (A) Conductance mSiemens/cm2 Membrane potential (mV) +40 +20 0 −20 −40 −60 −80 30 Na+ 20 10 K+ 0 0 (B) 2 Time (ms) 3 4 Stimulus current Membrane potential (mV) ACTION POTENTIALS OF SQUID AXON 75 50 25 0 50 25 0 −25 −50 −75 (C) Membrane potential (mV) 1 50 25 0 −25 −50 −75 0 1 2 3 4 0 1 2 3 Time (ms) 4 0 1 2 3 4 3 4 MATHEMATICAL MODEL BASED ON Na+ AND K+ CONDUCTANCES 0 1 2 3 4 0 2 3 1 Time (ms) 4 0 1 2 Na+ conductance is responsible for action potential initiation. The increase in Na+ conductance causes Na+ to enter the neuron, thus depolarizing the membrane potential, which approaches ENa. The rate of depolarization subsequently falls both because the electrochemical driving force on Na+ decreases and because the Na+ conductance inactivates. At the same time, depolarization slowly activates the voltage-dependent K+ conductance, causing K+ to leave the cell and repolarizing the membrane potential toward EK. Because the K+ conductance becomes temporarily higher than it is in the resting condition, the membrane potential actually becomes briefly more negative than the normal resting potential (the undershoot). The hyperpolarization of the membrane potential causes the voltage-dependent K+ conductance (and any Na+ conductance not inactivated) to turn off, allowing the membrane potential to return to its resting level. 56 Chapter Three more rizes Increase Na+ current ola ep FAST POSITIVE CYCLE Open Na+ channels D Hyperpo lar ize Depolarize membrane potential s SLOW NEGATIVE CYCLE Increase K+ current Open K+ channels Figure 3.9 Feedback cycles responsible for membrane potential changes during an action potential. Membrane depolarization rapidly activates a positive feedback cycle fueled by the voltage-dependent activation of Na+ conductance. This phenomenon is followed by the slower activation of a negative feedback loop as depolarization activates a K+ conductance, which helps to repolarize the membrane potential and terminate the action potential. This mechanism of action potential generation represents a positive feedback loop: Activating the voltage-dependent Na+ conductance increases Na+ entry into the neuron, which makes the membrane potential depolarize, which leads to the activation of still more Na+ conductance, more Na+ entry, and still further depolarization (Figure 3.9). Positive feedback continues unabated until Na+ conductance inactivation and K+ conductance activation restore the membrane potential to the resting level. Because this positive feedback loop, once initiated, is sustained by the intrinsic properties of the neuron—namely, the voltage dependence of the ionic conductances—the action potential is self-supporting, or regenerative. This regenerative quality explains why action potentials exhibit all-or-none behavior (see Figure 2.1), and why they have a threshold (Box B). The delayed activation of the K+ conductance represents a negative feedback loop that eventually restores the membrane to its resting state. Hodgkin and Huxley’s reconstruction of the action potential and all its features shows that the properties of the voltage-sensitive Na+ and K+ conductances, together with the electrochemical driving forces created by ion transporters, are sufficient to explain action potentials. Their use of both empirical and theoretical methods brought an unprecedented level of rigor to a long-standing problem, setting a standard of proof that is achieved only rarely in biological research. Long-Distance Signaling by Means of Action Potentials The voltage-dependent mechanisms of action potential generation also explain the long-distance transmission of these electrical signals. Recall from Chapter 2 that neurons are relatively poor conductors of electricity, at least compared to a wire. Current conduction by wires, and by neurons in the absence of action potentials, is called passive current flow (Box C). The passive electrical properties of a nerve cell axon can be determined by measuring the voltage change resulting from a current pulse passed across the axonal membrane (Figure 3.10A). If this current pulse is not large enough to generate action potentials, the magnitude of the resulting potential change decays exponentially with increasing distance from the site of current injection (Figure 3.10B). Typically, the potential falls to a small fraction of its initial value at a distance of no more than a couple of millimeters away from the site of injection (Figure 3.10C). The progressive decrease in the amplitude of the induced potential change occurs because the injected current leaks out across the axonal membrane; accordingly, less current is available to change the membrane potential farther along the axon. Thus, the leakiness of the axonal membrane prevents effective passive transmission of electrical signals in all but the shortest axons (those 1 mm or less in length). Likewise, the leakiness of the membrane slows the time course of the responses measured at increasing distances from the site where current was injected (Figure 3.10D). Voltage-Dependent Membrane Permeability 57 Box B Threshold An important—and potentially puzzling—property of the action potential is its initiation at a particular membrane potential, called threshold. Indeed, action potentials never occur without a depolarizing stimulus that brings the membrane to this level. The depolarizing “trigger” can be one of several events: a synaptic input, a receptor potential generated by specialized receptor organs, the endogenous pacemaker activity of cells that generate action potentials spontaneously, or the local current that mediates the spread of the action potential down the axon. Why the action potential “takes off” at a particular level of depolarization can be understood by comparing the underlying events to a chemical explosion (Figure A). Exogenous heat (analogous to the initial depolarization of the membrane potential) stimulates an exothermic chemical reaction, which produces more heat, which further enhances the reaction (Figure B). As a result of this positive feedback loop, the rate of the reaction builds up exponentially—the definition of an explosion. In any such (A) process, however, there is a threshold, that is, a point up to which heat can be supplied without resulting in an explosion. The threshold for the chemical explosion diagrammed here is the point at which the amount of heat supplied exogenously is just equal to the amount of heat that can be dissipated by the circumstances of the reaction (such as escape of heat from the beaker). The threshold of action potential initiation is, in principle, similar (Figure C). There is a range of “subthreshold” depolarization, within which the rate of increased sodium entry is less than the rate of potassium exit (remember that the membrane at rest is highly permeable to K+, which therefore flows out as the membrane is depolarized). The point at which Na+ inflow just equals K+ outflow represents an unstable equilibrium analogous to the ignition point of an explosive mixture. The behavior of the membrane at threshold reflects this instability: The membrane potential may linger at the threshold level for a variable period before either returning to the resting level or flaring up into a full-blown (C) (B) Some heat escapes action potential. In theory at least, if there is a net internal gain of a single Na+ ion, an action potential occurs; conversely, the net loss of a single K+ ion leads to repolarization. A more precise definition of threshold, therefore, is that value of membrane potential, in depolarizing from the resting potential, at which the current carried by Na+ entering the neuron is exactly equal to the K+ current that is flowing out. Once the triggering event depolarizes the membrane beyond this point, the positive feedback loop of Na+ entry on membrane potential closes and the action potential “fires.” Because the Na+ and K+ conductances change dynamically over time, the threshold potential for producing an action potential also varies as a consequence of the previous activity of the neuron. For example, following an action potential, the membrane becomes temporarily refractory to further excitation because the threshold for firing an action potential transiently rises. There is, therefore, no specific value of membrane potential that defines the threshold for a given nerve cell in all circumstances. Increase in reaction rate Na+ entry Increase in Na+ permeability Exothermic reaction Additional heat produced CHEMICAL EXPLOSION ACTION POTENTIAL Depolarization of membrane Heat Heat source Heat escape slows reaction A positive feedback loop underlying the action potential explains the phenomenon of threshold. K+ loss repolarizes membrane potential 58 Chapter Three (A) Stimulate Current injection electrode 1 mm Axon Potential recording electrodes (B) Membrane potential (mV) Record Record Record Record Record Record −62 −65 (C) −50 Membrane potential (mV) 0 10 20 30 40 −55 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 Time (ms) 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 2.0 2.5 Threshold −60 −65 Resting potential 0 −0.5 0.5 1.0 1.5 Distance along axon (mm) (D) Current (nA) Record −59 +1 −0 −1 Distance from current injection (mm) −60 Membrane potential (mV) 0 0.5 1.0 1.5 2.0 −65 0 10 2.5 20 Time (ms) 30 40 Figure 3.10 Passive current flow in an axon. (A) Experimental arrangement for examining the local flow of electrical current in an axon. A current-passing electrode produces a subthreshold change in membrane potential, which spreads passively along the axon. (B) Potential responses recorded at the positions indicated by microelectrodes. With increasing distance from the site of current injection, the amplitude of the potential change is attenuated. (C) Relationship between the amplitude of potential responses and distance. (D) Superimposed responses (from B) to current pulse, measured at indicated distances along axon. Note that the responses develop more slowly at greater distances from the site of current injection, for reasons explained in Box C. (After Hodgkin and Rushton, 1938.) If the experiment shown in Figure 3.10 is repeated with a depolarizing current pulse large enough to produce an action potential, the result is dramatically different (Figure 3.11A). In this case, an action potential occurs without decrement along the entire length of the axon, which in humans Voltage-Dependent Membrane Permeability 59 (A) Stimulate Current injection electrode Axon Membrane potential (mV) (B) Potential recording electrodes Record 1 mm Record Record Record Record Record Record 0 −50 −65 0 2 4 6 8 ms 0 2 4 6 8 ms 0 2 4 6 ms 8 0 2 4 ms 6 8 0 2 4 6 ms 8 0 2 4 ms 6 8 0 2 4 ms 6 8 (C) Membrane potential (mV) 25 0 −25 −50 Threshold −65 −0.5 Resting potential 0 0.5 1.0 1.5 Distance along axon (mm) may be a distance of a meter or more (Figure 3.11B). Thus, action potentials somehow circumvent the inherent leakiness of neurons. How, then, do action potentials traverse great distances along such a poor passive conductor? The answer is in part provided by the observation that the amplitude of the action potentials recorded at different distances is constant. This all-or-none behavior indicates that more than simple passive flow of current must be involved in action potential propagation. A second clue comes from examination of the time of occurrence of the action potentials recorded at different distances from the site of stimulation: Action potentials occur later and later at greater distances along the axon (Figure 3.11B). Thus, the action potential has a measurable rate of transmission, called the conduction velocity. The delay in the arrival of the action potential at successively more distant points along the axon differs from the case shown in Figure 3.10, in which the electrical changes produced by passive current flow occur at more or less the same time at successive points. The mechanism of action potential propagation is easy to grasp once one understands how action potentials are generated and how current passively flows along an axon (Figure 3.12). A depolarizing stimulus—a synaptic potential or a receptor potential in an intact neuron, or an injected current pulse in an experiment—locally depolarizes the axon, thus opening the voltage-sensitive Na+ channels in that region. The opening of Na+ channels causes inward movement of Na+, and the resultant depolarization of the membrane potential generates an action potential at that site. Some of the local current generated by the action potential will then flow passively down 2.0 2.5 Figure 3.11 Propagation of an action potential. (A) In this experimental arrangement, an electrode evokes an action potential by injecting a suprathreshold current. (B) Potential responses recorded at the positions indicated by microelectrodes. The amplitude of the action potential is constant along the length of the axon, although the time of appearance of the action potential is delayed with increasing distance. (C) The constant amplitude of an action potential (solid black line) measured at different distances. 60 Chapter Three Box C Passive Membrane Properties The passive flow of electrical current plays a central role in action potential propagation, synaptic transmission, and all other forms of electrical signaling in nerve cells. Therefore, it is worthwhile understanding in quantitative terms how passive current flow varies with distance along a neuron. For the case of a cylindrical axon, such as the one depicted in Figure 3.10, subthreshold current injected into one part of the axon spreads passively along the axon until the current is dissipated by leakage out across the axon membrane. The decrement in the current flow with distance (Figure A) is described by a simple exponential function: plasma membrane (rm), the intracellular axoplasm (ri), and the extracellular medium (r0 ). The relationship between these parameters is: λ= Hence, to improve the passive flow of current along an axon, the resistance of the plasma membrane should be as high as possible and the resistances of the axoplasm and extracellular medium should be low. 1.0 Vx = V0 e–x/λ VX = V0e−x/λ 0.8 VX/V0 where Vx is the voltage response at any distance x along the axon, V0 is the voltage change at the point where current is injected into the axon, e is the base of natural logarithms (approximately 2.7), and λ is the length constant of the axon. As evident in this relationship, the length constant is the distance where the initial voltage response (V0 ) decays to 1/e (or 37%) of its value. The length constant is thus a way to characterize how far passive current flow spreads before it leaks out of the axon, with leakier axons having shorter length constants. The length constant depends upon the physical properties of the axon, in particular the relative resistances of the rm r0 + ri Another important consequence of the passive properties of neurons is that currents flowing across a membrane do not immediately change the membrane potential. For example, when a rectangular current pulse is injected into the axon shown in the experiment illustrated in Figure 3.10A, the membrane potential depolarizes slowly over a few milliseconds and then repolarizes over a similar time course when the current pulse ends (see Figure 3.10D). These delays in changing the membrane potential are due to the fact that the plasma mem- 0.6 0.4 37% 0.2 0.0 −5 −4 −3 −2 −1 0 1 2 3 4 5 λ λ Distance from current injection (mm) (A) Spatial decay of membrane potential along a cylindrical axon. A current pulse injected at one point in the axon (0 mm) produces voltage responses (Vx) that decay exponentially with distance. The distance where the voltage response is 1/e of its initial value (V0) is the length constant, λ. the axon, in the same way that subthreshold currents spread along the axon (see Figure 3.10). Note that this passive current flow does not require the movement of Na+ along the axon but, instead, occurs by a shuttling of charge, somewhat similar to what happens when wires passively conduct electricity by transmission of electron charge. This passive current flow depolarizes the membrane potential in the adjacent region of the axon, thus opening the Na+ channels in the neighboring membrane. The local depolarization triggers an action potential in this region, which then spreads again in a continuing cycle until the end of the axon is reached. Thus, action potential propagation requires the coordinated action of two forms of current Current (nA) Voltage-Dependent Membrane Permeability 61 tial returns to 1/e of V∞ at a time equal to t. For cells with more complex geometries than the axon in Figure 3.10, the time courses of the changes in membrane potential are not simple exponentials, but nonetheless depend on the membrane time constant. Thus, the time constant characterizes how rapidly current flow changes the membrane potential. The membrane time constant also depends on the physical properties of the nerve cell, specifically on the resistance (rm) and capacitance (cm) of the plasma membrane such that: +1 −0 −1 1.0 Vt = V∞(1 – e−t/τ) Vt = V∞e−t/τ 0.80 V∞/Vτ 63% 0.60 0.40 37% τ = rmcm 0.20 0.00 0 5 10 15 20 τ 25 30 35 40 τ Time (ms) (B) Time course of potential changes produced in a spatially uniform cell by a current pulse. The rise and fall of the membrane potential (Vt) can be described as exponential functions, with the time constant τ defining the time required for the response to rise to 1 – (1/e) of the steady-state value (V∞), or to decline to 1/e of V∞. brane behaves as a capacitor, storing the initial charge that flows at the beginning and end of the current pulse. For the case of a cell whose membrane potential is spatially uniform, the change in the membrane potential at any time, Vt , after beginning the current pulse (Figure B) can also be described by an exponential relationship: membrane potential change, t is the time after the current pulse begins, and τ is the membrane time constant. The time constant is thus defined as the time when the voltage response (Vt ) rises to 1 − (1/e) (or 63%) of V∞. After the current pulse ends, the membrane potential change also declines exponentially according to the relationship Vt = V∞(1 − e−t/τ) Vt = V∞ e−t/τ where V∞ is the steady-state value of the During this decay, the membrane poten- flow—the passive flow of current as well as active currents flowing through voltage-dependent ion channels. The regenerative properties of Na+ channel opening allow action potentials to propagate in an all-or-none fashion by acting as a booster at each point along the axon, thus ensuring the long-distance transmission of electrical signals. The Refractory Period Recall that the depolarization that produces Na+ channel opening also causes delayed activation of K+ channels and Na+ channel inactivation, lead- The values of rm and cm depend, in part, on the size of the neuron, with larger cells having lower resistances and larger capacitances. In general, small nerve cells tend to have long time constants and large cells brief time constants. References HODGKIN, A. L. AND W. A. H. RUSHTON (1938) The electrical constants of a crustacean nerve fibre. Proc. R. Soc. Lond. 133: 444–479. JOHNSTON, D. AND S. M.-S. WU (1995) Foundations of Cellular Neurophysiology. Cambridge, MA: MIT Press. RALL, W. (1977) Core conductor theory and cable properties of neurons. In Handbook of Physiology, Section 1: The Nervous System, Vol. 1: Cellular Biology of Neurons. E. R. Kandel (ed.). Bethesda, MD: American Physiological Society, pp. 39–98. 62 Chapter Three 1 Na+ channels locally open in response to stimulus, generating an action potential here Stimulate 2 Some depolarizing current passively flows down axon Na+ channel K+ channel Na+ Membrane t=1 Na+ Axon Na+ Point A Point B Point C 3 Local depolarization causes neighboring Na+ channels to open and generates an action potential here K+ Na+ t=2 K+ Na+ K+ Point A Na+ Point B Point C 4 Upstream Na+ channels inactivate, while K+ channels open. Membrane potential repolarizes and axon is refractory here K+ 5 The process is repeated, propagating the action potential along the axon Na+ t=3 K+ Point A t=1 t=2 t=3 0 mV Point A Threshold −65 Resting potential 0 Point B Threshold −65 Resting potential 0 Point C Threshold −65 Resting potential Point B Na+ Na+ K+ Point C Figure 3.12 Action potential conduction requires both active and passive current flow. Depolarization opens Na+ channels locally and produces an action potential at point A of the axon (time t = 1). The resulting inward current flows passively along the axon, depolarizing the adjacent region (point B) of the axon. At a later time (t = 2), the depolarization of the adjacent membrane has opened Na+ channels at point B, resulting in the initiation of the action potential at this site and additional inward current that again spreads passively to an adjacent point (point C) farther along the axon. At a still later time (t = 3), the action potential has propagated even farther. This cycle continues along the full length of the axon. Note that as the action potential spreads, the membrane potential repolarizes due to K+ channel opening and Na+ channel inactivation, leaving a “wake” of refractoriness behind the action potential that prevents its backward propagaPurves Neuroscience 3E tion (panel 4). The panel to the left of this figure legend shows the time Pyramis Studios course of membrane potential changes at the points indicated. P3_312 Voltage-Dependent Membrane Permeability 63 ing to repolarization of the membrane potential as the action potential sweeps along the length of an axon (see Figure 3.12). In its wake, the action potential leaves the Na+ channels inactivated and K+ channels activated for a brief time. These transitory changes make it harder for the axon to produce subsequent action potentials during this interval, which is called the refractory period. Thus, the refractory period limits the number of action potentials that a given nerve cell can produce per unit time. As might be expected, different types of neurons have different maximum rates of action potential firing due to different types and densities of ion channels. The refractoriness of the membrane in the wake of the action potential also explains why action potentials do not propagate back toward the point of their initiation as they travel along an axon. Increased Conduction Velocity as a Result of Myelination The rate of action potential conduction limits the flow of information within the nervous system. It is not surprising, then, that various mechanisms have evolved to optimize the propagation of action potentials along axons. Because action potential conduction requires passive and active flow of current (see Figure 3.12), the rate of action potential propagation is determined by both of these phenomena. One way of improving passive current flow is to increase the diameter of an axon, which effectively decreases the internal resistance to passive current flow (see Box C). The consequent increase in action potential conduction velocity presumably explains why giant axons evolved in invertebrates such as squid, and why rapidly conducting axons in all animals tend to be larger than slowly conducting ones. Another strategy to improve the passive flow of electrical current is to insulate the axonal membrane, reducing the ability of current to leak out of the axon and thus increasing the distance along the axon that a given local current can flow passively (see Box C). This strategy is evident in the myelination of axons, a process by which oligodendrocytes in the central nervous system (and Schwann cells in the peripheral nervous system) wrap the axon in myelin, which consists of multiple layers of closely opposed glial membranes (Figure 3.13; see also Chapter 1). By acting as an electrical insulator, myelin greatly speeds up action potential conduction (Figure 3.14). For example, whereas unmyelinated axon conduction velocities range from about 0.5 to 10 m/s, myelinated axons can conduct at velocities of up to 150 m/s. The major reason underlying this marked increase in speed is that the time-consuming process of action potential generation occurs only at specific points along the axon, called nodes of Ranvier, where there is a gap in the myelin wrapping (see Figure 1.4F). If the entire surface of an axon were insulated, there would be no place for current to flow out of the axon and action potentials could not be generated. As it happens, an action potential generated at one node of Ranvier elicits current that flows passively within the myelinated segment until the next node is reached. This local current flow then generates an action potential in the neighboring segment, and the cycle is repeated along the length of the axon. Because current flows across the neuronal membrane only at the nodes (see Figure 3.13), this type of propagation is called saltatory, meaning that the action potential jumps from node to node. Not surprisingly, loss of myelin, as occurs in diseases such as multiple sclerosis, causes a variety of serious neurological problems (Box D). (A) Myelinated axon Node of Ranvier Oligodendrocyte Myelin sheath (B) Action potential propagation t=1 Na+ Axon Na+ Point A Point B Point C Na+ K+ t = 1.5 K+ Na+ Na+ Point A Point B Point C K+ t=2 K+ Na+ Na+ K+ Point A Point B Point C t = 1 t = 1.5 t = 2 0 mV Point A Threshold −65 Resting potential 0 Point B Threshold −65 Resting potential 0 Point C Threshold −65 Resting potential Figure 3.13 Saltatory action potential conduction along a myelinated axon. (A) Diagram of a myelinated axon. (B) Local current in response to action potential initiation at a particular site flows locally, as described in Figure 3.12. However, the presence of myelin prevents the local current from leaking across the internodal membrane; it therefore flows farther along the axon than it would in the absence of myelin. Moreover, voltage-gated Na+ channels are present only at the nodes of Ranvier (K+ channels are present at the nodes of some neurons, but not others). This arrangement means that the generation of active, voltage-gated Na+ currents need only occur at these unmyelinated regions. The result is a greatly enhanced velocity Purves 3E legof action potential conduction. The panel to theNeuroscience left of this figure Pyramis Studios end shows the time course of membrane potential changes at the points indicated. P3_313 120103 Voltage-Dependent Membrane Permeability 65 t=1 Unmyelinated axon Myelinated axon t=2 t=3 Summary The action potential and all its complex properties can be explained by timeand voltage-dependent changes in the Na+ and K+ permeabilities of neuronal membranes. This conclusion derives primarily from evidence obtained by a device called the voltage clamp. The voltage clamp technique is an electronic feedback method that allows control of neuronal membrane potential Figure 3.14 Comparison of speed of action potential conduction in unmyelinated (upper) and myelinated (lower) axons. 66 Chapter Three Box D Multiple Sclerosis Multiple sclerosis (MS) is a disease of the central nervous system characterized by a variety of clinical problems arising from multiple regions of demyelination and inflammation along axonal pathways. The disorder commonly begins between ages 20 and 40, characterized by the abrupt onset of neurological deficits that typically persist for days or weeks and then remit. The clinical course ranges from patients with no persistent neurological loss, some of whom experience only occasional later exacerbations, to others who progressively deteriorate as a result of extensive and relentless central nervous system involvement. The signs and symptoms of MS are determined by the location of the affected regions. Particularly common are monocular blindness (due to lesions of the optic nerve), motor weakness or paralysis (due to lesions of the corticospinal tracts), abnormal somatic sensations (due to lesions of somatic sensory pathways, often in the posterior columns), double vision (due to lesions of medial longitudinal fasciculus), and dizziness (due to lesions of vestibular pathways). Abnormalities are often apparent in the cerebrospinal fluid, which usually contains an abnormal number of cells associated with inflammation and an increased content of antibodies (a sign of an altered immune response). The diagnosis of MS generally relies on the presence of a neurological problem that remits and then returns at an unrelated site. Confirmation can sometimes be obtained from magnetic resonance imaging (MRI), or functional evidence of lesions in a particular pathway by abnormal evoked potentials. The histological hallmark of MS at postmortem exam is multiple lesions at different sites showing loss of myelin associated with infiltration of inflammatory cells and, in some instances, loss of axons themselves. The concept of MS as a demyelinating disease is deeply embedded in the clinical literature, although precisely how the demyelination translates into functional deficits is poorly understood. The loss of the myelin sheath surrounding many axons clearly compromises action potential conduction, and the abnormal patterns of nerve conduction that result presumably produce most of the clinical deficits in the disease. However, MS may have effects that extend beyond loss of the myelin sheath. It is clear that some axons are actually destroyed, probably as a result of inflammatory processes in the overlying myelin and/or loss of trophic support of the axon by oligodendrocytes. Thus, axon loss also contributes to the functional deficits in MS, especially in the chronic, progressive forms of the disease. The ultimate cause of MS remains unclear. The immune system undoubtedly contributes to the damage and new immunoregulatory therapies provide substantial benefits to many patients. Precisely how the immune system is activated to cause the injury is not known. The most popular hypothesis is that MS is an autoimmune disease (i.e., a disease in which the immune system attacks the body’s proper constituents). The fact that immunization of experimental animals with any one of several molecular constituents of the myelin sheath can induce a demyelinating disease (called experimental allergic encephalomyelitis) shows that an autoimmune attack on the myelin membrane is sufficient to produce a picture similar to MS. A possible explanation of the human disease is that a genetically susceptible individual becomes transiently infected (by a minor viral illness, for example) with a microorganism that expresses a molecule struc- turally similar to a component of myelin. An immune response to this antigen is mounted to attack the invader, but the failure of the immune system to discriminate between the foreign protein and self results in destruction of otherwise normal myelin, a scenario occurring in mice infected with Theiler’s virus. An alternative hypothesis is that MS is caused by a persistent infection by a virus or other microorganism. In this interpretation, the immune system’s ongoing efforts to get rid of the pathogen cause the damage to myelin. Tropical spastic paraparesis (TSP) provides a precedent for this idea. TSP is a disease characterized by the gradual progression of weakness of the legs and impaired control of bladder function associated with increased deep tendon reflexes and a positive Babinski sign (see Chapter 16). This clinical picture is similar to that of rapidly advancing MS. TSP is known to be caused by persistent infection with a retrovirus (human T lymphotropic virus-1). This precedent notwithstanding, proving the persistent viral infection hypothesis for MS requires unambiguous demonstration of the presence of a virus. Despite periodic reports of a virus associated with MS, convincing evidence has not been forthcoming. In sum, MS remains a daunting clinical challenge. References ADAMS, R. D. AND M. VICTOR (2001) Principles of Neurology, 7th Ed. New York: McGrawHill, pp. 954–982. MILLER, D. H. AND 9 OTHERS. (2003) A controlled trial of natalizumab for relapsing multiple sclerosis. N. Engl. J. Med. 348: 15–23. ZANVIL, S. S. AND L. STEINMAN (2003) Diverse targets for intervention during inflammatory and neurodegenerative phases of multiple sclerosis. Neuron 38: 685–688. Voltage-Dependent Membrane Permeability 67 and, simultaneously, direct measurement of the voltage-dependent fluxes of Na+ and K+ that produce the action potential. Voltage clamp experiments show that a transient rise in Na+ conductance activates rapidly and then inactivates during a sustained depolarization of the membrane potential. Such experiments also demonstrate a rise in K+ conductance that activates in a delayed fashion and, in contrast to the Na+ conductance, does not inactivate. Mathematical modeling of the properties of these conductances indicates that they, and they alone, are responsible for the production of all-ornone action potentials in the squid axon. Action potentials propagate along the nerve cell axons initiated by the voltage gradient between the active and inactive regions of the axon by virtue of the local current flow. In this way, action potentials compensate for the relatively poor passive electrical properties of nerve cells and enable neural signaling over long distances. These classical electrophysiological findings provide a solid basis for considering the functional and ultimately molecular variations on neural signaling taken up in the next chapter. Additional Reading Reviews ARMSTRONG, C. M. AND B. HILLE (1998) Voltage-gated ion channels and electrical excitability. Neuron 20: 371–80. NEHER, E. (1992) Ion channels for communication between and within cells. Science 256: 498–502. Important Original Papers ARMSTRONG, C. M. AND L. BINSTOCK (1965) Anomalous rectification in the squid giant axon injected with tetraethylammonium chloride. J. Gen. Physiol. 48: 859–872. HODGKIN, A. L. AND A. F. HUXLEY (1952a) Currents carried by sodium and potassium ions through the membrane of the giant axon of Loligo. J. Physiol. 116: 449–472. HODGKIN, A. L. AND A. F. HUXLEY (1952b) The components of membrane conductance in the giant axon of Loligo. J. Physiol. 116: 473–496. HODGKIN, A. L. AND A. F. HUXLEY (1952c) Thedual effect of membrane potential on sodium conductance in the giant axon of Loligo. J. Physiol. 116: 497–506. HODGKIN, A. L. AND A. F. HUXLEY (1952d) A quantitative description of membrane current and its application to conduction and excitation in nerve. J. Physiol. 116: 507–544. HODGKIN, A. L. AND W. A. H. RUSHTON (1938) The electrical constants of a crustacean nerve fibre. Proc. R. Soc. Lond. 133: 444–479. HODGKIN, A. L., A. F. HUXLEY AND B. KATZ (1952) Measurements of current–voltage relations in the membrane of the giant axon of Loligo. J. Physiol. 116: 424–448. MOORE, J. W., M. P. BLAUSTEIN, N. C. ANDERSON AND T. NARAHASHI (1967) Basis of tetrodotoxin’s selectivity in blockage of squid axons. J. Gen. Physiol. 50: 1401–1411. Books AIDLEY, D. J. AND P. R. STANFIELD (1996) Ion Channels: Molecules in Action. Cambridge: Cambridge University Press. HILLE, B. (2001) Ion Channels of Excitable Membranes, 3rd Ed. Sunderland, MA: Sinauer Associates. JOHNSTON, D. AND S. M.-S. WU (1995) Foundations of Cellular Neurophysiology. Cambridge, MA: MIT Press. JUNGE, D. (1992) Nerve and Muscle Excitation, 3rd Ed. Sunderland, MA: Sinauer Associates. Chapter 4 Channels and Transporters Overview The generation of electrical signals in neurons requires that plasma membranes establish concentration gradients for specific ions and that these membranes undergo rapid and selective changes in the membrane permeability to these ions. The membrane proteins that create and maintain ion gradients are called active transporters, whereas other proteins called ion channels give rise to selective ion permeability changes. As their name implies, ion channels are transmembrane proteins that contain a specialized structure, called a pore, that permits particular ions to cross the neuronal membrane. Some of these channels also contain other structures that are able to sense the electrical potential across the membrane. Such voltage-gated channels open or close in response to the magnitude of the membrane potential, allowing the membrane permeability to be regulated by changes in this potential. Other types of ion channels are gated by extracellular chemical signals such as neurotransmitters, and some by intracellular signals such as second messengers. Still others respond to mechanical stimuli, temperature changes, or a combination of such effects. Many types of ion channels have now been characterized at both the gene and protein level, resulting in the identification of a large number of ion channel subtypes that are expressed differentially in neuronal and non-neuronal cells. The specific expression pattern of ion channels in each cell type can generate a wide spectrum of electrical characteristics. In contrast to ion channels, active transporters are membrane proteins that produce and maintain ion concentration gradients. The most important of these is the Na+ pump, which hydrolyzes ATP to regulate the intracellular concentrations of both Na+ and K+. Other active transporters produce concentration gradients for the full range of physiologically important ions, including Cl–, Ca2+, and H+. From the perspective of electrical signaling, active transporters and ion channels are complementary: Transporters create the concentration gradients that help drive ion fluxes through open ion channels, thus generating electrical signals. Ion Channels Underlying Action Potentials Although Hodgkin and Huxley had no knowledge of the physical nature of the conductance mechanisms underlying action potentials, they nonetheless proposed that nerve cell membranes have channels that allow ions to pass selectively from one side of the membrane to the other (see Chapter 3). Based on the ionic conductances and currents measured in voltage clamp experiments, the postulated channels had to have several properties. First, because the ionic currents are quite large, the channels had to be capable of allowing ions to move across the membrane at high rates. Second, because 69 70 Chapter Four Box A Cell-attached recording Recording pipette The Patch Clamp Method A wealth of new information about ion channels resulted from the invention of the patch clamp method in the 1970s. This technique is based on a very simple idea. A glass pipette with a very small opening is used to make tight contact with a tiny area, or patch, of neuronal membrane. After the application of a small amount of suction to the back of the pipette, the seal between pipette and membrane becomes so tight that no ions can flow between the pipette and the membrane. Thus, all the ions that flow when a single ion channel opens must flow into the pipette. The resulting electrical current, though small, can be measured with an ultrasensitive electronic amplifier connected to the pipette. Based on the geometry involved, this arrangement usually is called the cell-attached patch clamp recording method. As with the conventional voltage clamp method, the patch clamp method allows experimental control of the membrane potential to characterize the voltage dependence of membrane currents. Although the ability to record currents flowing through single ion channels is an important advantage of the cell-attached patch clamp method, minor technical modifications yield still other advantages. For example, if the membrane patch within the pipette is disrupted by briefly applying strong suction, the interior of the pipette becomes continuous with the cytoplasm of the cell. This arrangement allows measurements of electrical potentials and currents from the entire cell and is therefore called the whole-cell recording method. The whole-cell configuration also allows diffusional exchange between the pipette and the cytoplasm, producing a convenient way to inject substances into the interior of a “patched” cell. Two other variants of the patch clamp method originate from the finding that once a tight seal has formed between the Mild suction Tight contact between pipette and membrane Whole-cell recording Strong pulse of suction Cytoplasm is continuous with pipette interior Inside-out recording Expose to air Cytoplasmic domain accessible Outside-out recording Retract pipette Ends of membrane anneal Extracellular domain accessible Four configurations in patch clamp measurements of ionic currents. membrane and the glass pipette, small pieces of membrane can be pulled away from the cell without disrupting the seal; this yields a preparation that is free of the complications imposed by the rest of the cell. Simply retracting a pipette that is in the cell-attached configuration causes a small vesicle of membrane to remain attached to the pipette. By exposing the tip of the pipette to air, the vesicle opens to yield a small patch of membrane with its (former) intracellular sur- Channels and Transpor ters 71 face exposed. This arrangement, called the inside-out patch recording configuration, allows the measurement of singlechannel currents with the added benefit of making it possible to change the medium to which the intracellular surface of the membrane is exposed. Thus, the inside-out configuration is particularly valuable when studying the influence of intracellular molecules on ion channel function. Alternatively, if the pipette is retracted while it is in the whole-cell configuration, a membrane patch is produced that has its extracellular surface exposed. This arrangement, called the outside-out recording configuration, is optimal for studying how channel activity is influenced by extracellular chemical signals, such as neurotransmitters (see Chapter 5). This range of possible configurations makes the patch clamp method an unusually versatile technique for studies of ion channel function. the ionic currents depend on the electrochemical gradient across the membrane, the channels had to make use of these gradients. Third, because Na+ and K+ flow across the membrane independently of each other, different channel types had to be capable of discriminating between Na+ and K+, allowing only one of these ions to flow across the membrane under the relevant conditions. Finally, given that the conductances are voltage-dependent, the channels had to be able to sense the voltage drop across the membrane, opening only when the voltage reached appropriate levels. While this concept of channels was highly speculative in the 1950s, later experimental work established beyond any doubt that transmembrane proteins called voltage-sensitive ion channels indeed exist and are responsible for all of the ionic conductance phenomena described in Chapter 3. The first direct evidence for the presence of voltage-sensitive, ion-selective channels in nerve cell membranes came from measurements of the ionic currents flowing through individual ion channels. The voltage-clamp apparatus used by Hodgkin and Huxley could only resolve the aggregate current resulting from the flow of ions through many thousands of channels. A technique capable of measuring the currents flowing through single channels was devised in 1976 by Erwin Neher and Bert Sakmann at the Max Planck Institute in Goettingen. This remarkable approach, called patch clamping (Box A), revolutionized the study of membrane currents. In particular, the patch clamp method provided the means to test directly Hodgkin and Huxley’s proposals about the characteristics of ion channels. Currents flowing through Na+ channels are best examined in experimental circumstances that prevent the flow of current through other types of channels that are present in the membrane (e.g., K+ channels). Under such conditions, depolarizing a patch of membrane from a squid giant axon causes tiny inward currents to flow, but only occasionally (Figure 4.1). The size of these currents is minuscule—approximately l–2 pA (i.e., 10–12 ampere), which is orders of magnitude smaller than the Na+ currents measured by voltage clamping the entire axon. The currents flowing through single channels are called microscopic currents to distinguish them from the macroscopic currents flowing through a large number of channels distributed over a much more extensive region of surface membrane. Although microscopic currents are certainly small, a current of 1 pA nonetheless reflects the flow of thousands of ions per millisecond. Thus, as predicted, a single channel can let many ions pass through the membrane in a very short time. References HAMILL, O. P., A. MARTY, E. NEHER, B. SAKF. J. SIGWORTH (1981) Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pflügers Arch. 391: 85–100. LEOIS, R. A. AND J. L. RAE (1998) Low-noise patch-clamp techniques. Meth. Enzym. 293: 218–266. SIGWORTH, F. J. (1986) The patch clamp is more useful than anyone had expected. Fed. Proc. 45: 2673–2677. MANN AND 72 Chapter Four Membrane potential (mV) (A) 0 −40 −80 0 5 10 Time (ms) 15 (B) Microscopic INa Closed Open 2 pA Summed microscopic INa (pA) 0 15 5 10 Time (ms) 15 5 10 Time (ms) 15 (C) 0.4 0 −0.4 −0.8 0 200 Macroscopic INa (pA) 5 10 Time (ms) (D) 0 −200 −400 −600 −800 Probability of Na+ channel opening 0.8 0 (E) 0.6 0.4 0.2 0 −80 −60 −40 –20 0 20 40 60 Membrane potential (mV) Figure 4.1 Patch clamp measurements of ionic currents flowing through single Na+ channels in a squid giant axon. In these experiments, Cs+ was applied to the axon to block voltage-gated K+ channels. Depolarizing voltage pulses (A) applied to a patch of membrane containing a single Na+ channel result in brief currents (B, downward deflections) in the seven successive recordings of membrane current (INa). (C) The sum of many such current records shows that most channels open in the initial 1–2 ms following depolarization of the membrane, after which the probability of channel openings diminishes because of channel inactivation. (D) A macroscopic current measured from another axon shows the close correlation between the time courses of microscopic and macroscopic Na+ currents. (E) The probability of an Na+ channel opening depends on the membrane potential, increasing as the membrane is depolarized. (B,C after Bezanilla and Correa, 1995; D after Vandenburg and Bezanilla, 1991; E after Correa and Bezanilla, 1994.) Several observations further proved that the microscopic currents in Figure 4.1B are due to the opening of single, voltage-activated Na+ channels. First, the currents are carried by Na+; thus, they are directed inward when the membrane potential is more negative than ENa, reverse their polarity at ENa, are outward at more positive potentials, and are reduced in size when the Na+ concentration of the external medium is decreased. This behavior exactly parallels that of the macroscopic Na+ currents described in Chapter 3. Second, the channels have a time course of opening, closing, and inactivating that matches the kinetics of macroscopic Na+ currents. This correspondence is difficult to appreciate in the measurement of microscopic currents flowing through a single open channel, because individual channels open and close in a stochastic (random) manner, as can be seen by examining the individual traces in Figure 4.1B. However, repeated depolarization of the membrane potential causes each Na+ channel to open and close many times. When the current responses to a large number of such stimuli are averaged together, the collective response has a time course that looks much like the macroscopic Na+ current (Figure 4.1C). In particular, the channels open mostly at the beginning of a prolonged depolarization, showing that they subsequently inactivate, as predicted from the macroscopic Na+ current (compare Figures 4.1C and 4.1D). Third, both the opening and closing of the channels are voltage-dependent; thus, the channels are closed at –80 mV but open when the membrane potential is depolarized. In fact, the probability that any given channel will be open varies with membrane potential (Figure 4.1E), again as predicted from the macroscopic Na+ conductance (see Figure 3.7). Finally, tetrodotoxin, which blocks the macroscopic Na+ current (see Box C), also blocks microscopic Na+ currents. Taken together, these results show that the macroscopic Na+ current measured by Hodgkin and Huxley does indeed arise from the aggregate effect of many thousands of microscopic Na+ currents, each representing the opening of a single voltage-sensitive Na+ channel. Patch clamp experiments have also revealed the properties of the channels responsible for the macroscopic K+ currents associated with action potentials. When the membrane potential is depolarized (Figure 4.2A), microscopic outward currents (Figure 4.2B) can be observed under conditions that block Na+ channels. The microscopic outward currents exhibit all the features expected for currents flowing through action-potential-related K+ channels. Thus, the microscopic currents (Figure 4.2C), like their macroscopic counterparts (Figure 4.2D), fail to inactivate during brief depolarizations. Moreover, these single-channel currents are sensitive to ionic manipu- The Diversity of Ion Channels Molecular genetic studies, in conjunction with the patch clamp method and other techniques, have led to many additional advances in understanding ion channels. Genes encoding Na+ and K+ channels, as well as many other channel types, have now been identified and cloned. A surprising fact that has emerged from these molecular studies is the diversity of genes that code for ion channels. Well over 100 ion channel genes have now been discovered, a number that could not have been anticipated from early studies of ion channel function. To understand the functional significance of this multitude of ion channel genes, the channels can be selectively expressed in well- 50 0 −50 −100 0 10 20 30 Time (ms) 40 10 20 30 Time (ms) 40 10 20 30 Time (ms) 40 10 20 30 Time (ms) 40 (B) Microscopic IK Open Closed 2 pA 0 Summed microscopic IK (pA) (C) 1 0 0 3 Macroscopic IK (mA/cm2) lations and drugs that affect the macroscopic K+ currents and, like the macroscopic K+ currents, are voltage-dependent (Figure 4.2E). This and other evidence shows that macroscopic K+ currents associated with action potentials arise from the opening of many voltage-sensitive K+ channels. In summary, patch clamping has allowed direct observation of microscopic ionic currents flowing through single ion channels, confirming that voltage sensitive Na+ and K+ channels are responsible for the macroscopic conductances and currents that underlie the action potential. Measurements of the behavior of single ion channels has also provided some insight into the molecular attributes of these channels. For example, single channel studies show that the membrane of the squid axon contains at least two types of channels—one selectively permeable to Na+ and a second selectively permeable to K+. Both channel types are voltage-gated, meaning that their opening is influenced by membrane potential (Figure 4.3). For each channel, depolarization increases the probability of channel opening, whereas hyperpolarization closes them (see Figures 4.1E and 4.2E). Thus, both channel types must have a voltage sensor that detects the potential across the membrane (Figure 4.3). However, these channels differ in important respects. In addition to their different ion selectivities, depolarization also inactivates the Na+ channel but not the K+ channel, causing Na+ channels to pass into a nonconducting state. The Na+ channel must therefore have an additional molecular mechanism responsible for inactivation. And, as expected from the macroscopic behavior of the Na+ and K+ currents described in Chapter 3, the kinetic properties of the gating of the two channels differs. This information about the physiology of single channels set the stage for subsequent studies of the molecular diversity of ion channels in various cell types, and of their detailed functional characteristics. (A) (D) 2 1 0 0 (E) 0.6 Probability of K+ channel opening Figure 4.2 Patch clamp measurements of ionic currents flowing through single K+ channels in a squid giant axon. In these experiments, tetrodotoxin was applied to the axon to block voltage-gated Na+ channels. Depolarizing voltage pulses (A) applied to a patch of membrane containing a single K+ channel results in brief currents (B, upward deflections) whenever the channel opens. (C) The sum of such current records shows that most channels open with a delay, but remain open for the duration of the depolarization. (D) A macroscopic current measured from another axon shows the correlation between the time courses of microscopic and macroscopic K+ currents. (E) The probability of a K+ channel opening depends on the membrane potential, increasing as the membrane is depolarized. (B and C after Augustine and Bezanilla, in Hille 1992; D after Augustine and Bezanilla, 1990; E after Perozo et al., 1991.) Membrane potential (mV) Channels and Transpor ters 73 0.4 0.2 0 −80 −60 −40 −20 0 20 40 Membrane potential (mV) 60 Figure 4.3 Functional states of voltagegated Na+ and K+ channels. The gates of both channels are closed when the membrane potential is hyperpolarized. When the potential is depolarized, voltage sensors (indicated by +) allow the channel gates to open—first the Na+ channels and then the K+ channels. Na+ channels also inactivate during prolonged depolarization, whereas many types of K+ channels do not. Membrane potential (mV) 74 Chapter Four 50 0 −50 −100 0 5 Time (ms) 10 15 Na+ CHANNEL Na+ + Na+ + + + Closed + Open Inactivating Inactivated Closed K+ CHANNEL + + + + K+ Closed + Closed K+ Open Open Closed defined experimental systems, such as in cultured cells or frog oocytes (Box B), and then studied with patch clamping and other physiological techniques. Such studies have found many voltage-gated channels that respond to membrane potential in much the same way as the Na+ and K+ channels that underlie the action potential. Other channels, however, are gated by chemical signals that bind to extracellular or intracellular domains on these proteins and are insensitive to membrane voltage. Still others are sensitive to mechanical displacement, or to changes in temperature. Further magnifying this diversity of ion channels are a number of mechanisms that can produce functionally different types of ion channels from a single gene. Ion channel genes contain a large number of coding regions that can be spliced together in different ways, giving rise to channel proteins that can have dramatically different functional properties. RNAs encoding ion channels also can be edited, modifying their base composition after transcription from the gene. For example, editing the RNA encoding of some receptors for the neurotransmitter glutamate (Chapter 6) changes a single amino acid within the receptor, which in turn gives rise to channels that differ in their selectivity for cations and in their conductance. Channel proteins can also undergo posttranslational modifications, such as phosphorylation by protein kinases (see Chapter 7), which can further change their functional characteristics. Thus, although the basic electrical signals of the nervous system are relatively stereotyped, the proteins responsible for generating these signals are remarkably diverse, conferring specialized signaling properties to many of the neuronal cell types that populate the nervous system. These channels also are involved in a broad range of neurological diseases. Channels and Transpor ters 75 Box B Expression of Ion Channels in Xenopus Oocytes The ability to combine molecular and physiological methods in a single cell system has made Xenopus oocytes a powerful experimental tool. Indeed, this system has been as valuable to contemporary studies of voltage-gated ion channels as the squid axon was to such studies in the 1950s and 1960s. (A) References GUNDERSEN, C. B., R. MILEDI AND I. PARKER (1984) Slowly inactivating potassium channels induced in Xenopus oocytes by messenger ribonucleic acid from Torpedo brain. J. Physiol. (Lond.) 353: 231–248. GURDON, J. B., C. D. LANE, H. R. WOODLAND AND G. MARBAIX (1971) Use of frog eggs and oocytes for the study of messenger RNA and its translation in living cells. Nature 233: 177–182. STÜHMER, W. (1998) Electrophysiological recordings from Xenopus oocytes. Meth. Enzym. 293: 280–300. SUMIKAWA, K., M. HOUGHTON, J. S. EMTAGE, B. M. RICHARDS AND E. A. BARNARD (1981) Active multi-subunit ACh receptor assembled by translation of heterologous mRNA in Xenopus oocytes. Nature 292: 862– 864. (B) PHOTO: PU04BXCB.tif as in 2/e Membrane potential (mV) (C) (A) The clawed African frog, Xenopus laevis. (B) Several oocytes from Xenopus highlighting the dark coloration of the original pole and the lighter coloration of the vegetal pole. (Courtesy of P. Reinhart.) (C) Results of a voltage clamp experiment showing K+ currents produced following injection of K+ channel mRNA into an oocyte. (After Gundersen et al., 1984.) 50 0 −50 −100 4 K+ current (µA) Bridging the gap between the sequence of an ion channel gene and understanding channel function is a challenge. To meet this challenge, it is essential to have an experimental system in which the gene product can be expressed efficiently, and in which the function of the resulting channel can be studied with methods such as the patch clamp technique. Ideally, the vehicle for expression should be readily available, have few endogenous channels, and be large enough to permit mRNA and DNA to be microinjected with ease. Oocytes (immature eggs) from the clawed African frog, Xenopus laevis (Figure A), fulfill all these demands. These huge cells (approximately 1 mm in diameter; Figure B) are easily harvested from the female Xenopus. Work performed in the 1970s by John Gurdon, a developmental biologist, showed that injection of exogenous mRNA into frog oocytes causes them to synthesize foreign protein in prodigious quantities. In the early 1980s, Ricardo Miledi, Eric Barnard, and other neurobiologists demonstrated that Xenopus oocytes could express exogenous ion channels, and that physiological methods could be used to study the ionic currents generated by the newly-synthesized channels (Figure C). As a result of these pioneering studies, heterologous expression experiments have now become a standard way of studying ion channels. The approach has been especially valuable in deciphering the relationship between channel structure and function. In such experiments, defined mutations (often affecting a single nucleotide) are made in the part of the channel gene that encodes a structure of interest; the resulting channel proteins are then expressed in oocytes to assess the functional consequences of the mutation. 3 2 1 0 0 0.2 0.4 0.6 0.8 1.0 Time (s) 76 Chapter Four Voltage-Gated Ion Channels Voltage-gated ion channels that are selectively permeable to each of the major physiological ions—Na+, K+, Ca2+, and Cl–—have now been discovered (Figure 4.4 A–D). Indeed, many different genes have been discovered for each type of voltage-gated ion channel. An example is the identification of 10 human Na+ channel genes. This finding was unexpected because Na+ channels from many different cell types have similar functional properties, consistent with their origin from a single gene. It is now clear, however, that all of these Na+ channel genes (called SCN genes) produce proteins that differ in their structure, function, and distribution in specific tissues. For instance, in addition to the rapidly inactivating Na+ channels discovered by Hodgkin and Huxley in squid axon, a voltage-sensitive Na+ channel that does not inactivate has been identified in mammalian axons. As might be expected, this channel gives rise to action potentials of long duration and is a target of local anesthetics such as benzocaine and lidocaine. Other electrical responses in neurons entail the activation of voltage-gated Ca2+ channels (Figure 4.4B). In some neurons, voltage-gated Ca2+ channels give rise to action potentials in much the same way as voltage-sensitive Na+ channels. In other neurons, Ca2+ channels control the shape of action potentials generated primarily by Na+ conductance changes. More generally, by affecting intracellular Ca2+ concentrations, the activity of Ca2+ channels regulates an enormous range of biochemical processes within cells (see Chapter 7). Perhaps the most important of the processes regulated by voltage-sensitive Ca2+ channels is the release of neurotransmitters at synapses (see Chapter 5). Given these crucial functions, it is perhaps not surprising that 16 different Ca2+ channel genes (called CACNA genes) have been identified. Like Na+ channels, Ca2+ channels differ in their activation and inactivation properties, allowing subtle variations in both electrical and chemical signaling processes mediated by Ca2+. As a result, drugs that block voltage-gated Ca2+ channels are especially valuable in treating a variety of conditions ranging from heart disease to anxiety disorders. By far the largest and most diverse class of voltage-gated ion channels are the K+ channels (Figure 4.4C). Nearly 100 K+ channel genes are now known, and these fall into several distinct groups that differ substantially in their activation, gating, and inactivation properties. Some take minutes to inactivate, as in the case of squid axon K+ channels studied by Hodgkin and Huxley (Figure 4.5A). Others inactivate within milliseconds, as is typical of most voltage-gated Na+ channels (Figure 4.5B). These properties influence the Figure 4.4 Types of voltage-gated ion channels. Examples of voltage-gated channels include those selectively permeable to Na+ (A), Ca2+ (B), K+ (C), and Cl– (D). Ligand-gated ion channels include those activated by the extracellular presence of neurotransmitters, such as glutamate (E). Other ligandgated channels are activated by intracellular second messengers, such as Ca2+ (F) or the cyclic nucleotides, cAMP and cGMP (G). VOLTAGE-GATED CHANNELS (A) Na+ channel Na+ (B) Ca2+ channel LIGAND-GATED CHANNELS (C) K+ channel Ca2+ (D) Cl− channel Cl− (E) Neurotransmitter receptor Na+ (F) Ca2+-activated K+ channel (G) Cyclic nucleotide gated channel Na+ Glutamate Outside + + + + cAMP Voltage sensor K+ K+ Ca2+ K+ Inside cGMP cAMP K+ 50 30 0 − 30 − 60 − 90 − 120 100 200 Time (ms) (A) KV2.1 300 K+ current (µA) +50 mV −120 mV (B) KV4.1 +50 mV K+ current (µA) K+ conductance 0 K+ conductance Membrane potential (mV) Channels and Transpor ters 77 K+ conductance K+ current (µA) +50 mV K+ conductance K+ conductance K+ current (µA) 10µM Ca2+ +50 mV 1µM Ca2+ +50 mV 1 0 1 0 −100 0 100 Membrane potential (mV) 1 0 1 10 µM Ca2+ 1 µM Ca2+ 0 −100 0 100 Membrane potential (mV) −120 mV pH 8 pH 6 0 −100 0 100 Membrane potential (mV) −100 0 100 Membrane potential (mV) −120 mV 100 200 Time (ms) 300 K+ conductance (F) 2-pore +50 mV Shaw K+ current (µA) (E) Ca2+activated K+ current (µA) −120 mV (D) Inward rectifier 0 −100 0 100 Membrane potential (mV) −120 mV (C) HERG 1 1 0 6 7 pH 8 Figure 4.5 Diverse properties of K+ channels. Different types of K+ channels were expressed in Xenopus oocytes (see Box B), and the voltage clamp method was used to change the membrane potential (top) and measure the resulting currents flowing through each type of channel. These K+ channels vary markedly in their gating properties, as evident in their currents (left) and conductances (right). (A) KV2.1 channels show little inactivation and are closely related to the delayed rectifier K+ channels involved in action potential repolarization. (B) KV4.1 channels inactivate during a depolarization. (C) HERG channels inactivate so rapidly that current flows only when inactivation is rapidly removed at the end of a depolarization. (D) Inward rectifying K+ channels allow more K+ current to flow at hyperpolarized potentials than at depolarized potentials. (E) Ca2+-activated K+ channels open in response to intracellular Ca2+ ions and, in some cases, membrane depolarization. (F) K+ channels with two pores usually respond to chemical signals, such as pH, rather than changes in membrane potential. 78 Chapter Four duration and rate of action potential firing, with important consequences for axonal conduction and synaptic transmission. Perhaps the most important function of K+ channels is the role they play in generating the resting membrane potential (see Chapter 2). At least two families of K+ channels that are open at substantially negative membrane voltage levels contribute to setting the resting membrane potential (Figure 4.5D). Finally, several types of voltage-gated Cl– channel have been identified (see Figure 4.4D). These channels are present in every type of neuron, where they control excitability, contribute to the resting membrane potential, and help regulate cell volume. Ligand-Gated Ion Channels Many types of ion channels respond to chemical signals (ligands) rather than to changes in the membrane potential (Figure 4.4E–G). The most important of these ligand-gated ion channels in the nervous system is the class activated by binding neurotransmitters (Figure 4.4E). These channels are essential for synaptic transmission and other forms of cell-cell signaling phenomena discussed in Chapters 5–7. Whereas the voltage-gated ion channels underlying the action potential typically allow only one type of ion to permeate, channels activated by extracellular ligands are usually less selective, allowing two or more types of ions to pass through the channel pore. Other ligand-gated channels are sensitive to chemical signals arising within the cytoplasm of neurons (see Chapter 7), and can be selective for specific ions such as K+ or Cl–, or permeable to all physiological cations. Such channels are distinguised by ligand-binding domains on their intracellular surfaces that interact with second messengers such as Ca2+, the cyclic nucleotides cAMP and cGMP, or protons. Examples of channels that respond to intracellular cues include Ca2+-activated K+ channels (Figure 4.4.F), the cyclic nucleotide gated cation channel (Figure 4.4G), or acid-sensing ion channels (ASICs). The main function of these channels is to convert intracellular chemical signals into electrical information. This process is particularly important in sensory transduction, where channels gated by cyclic nucleotides convert odors and light, for example, into electrical signals. Although many of these ligand-gated ion channels are located in the cell surface membrane, others are in membranes of intracellular organelles such as mitochondria or the endoplasmic reticulum . Some of these latter channels are selectively permeable to Ca2+ and regulate the release of Ca2+ from the lumen of the endoplasmic reticulum into the cytoplasm, where this second messenger can then trigger a spectrum of cellular responses such as described in Chapter 7. Stretch- and Heat-Activated Channels Still other ion channels respond to heat or membrane deformation. Heatactivated ion channels, such as some members of the transient receptor potential (TRP) gene family, contribute to the sensations of pain and temperature and help mediate inflammation (see Chapter 9). These channels are often specialized to detect specific temperature ranges, and some are even activated by cold. Other ion channels respond to mechanical distortion of the plasma membrane and are the basis of stretch receptors and neuromuscular stretch reflexes (see Chapters 8, 15 and 16). A specialized form of these channels enables hearing by allowing auditory hair cells to respond to sound waves (see Chapter 12). Channels and Transpor ters 79 In summary, this tremendous variety of ion channels allows neurons to generate electrical signals in response to changes in membrane potential, synaptic input, intracellular second messengers, light, odors, heat, sound, touch, and many other stimuli. The Molecular Structure of Ion Channels Understanding the physical structure of ion channels is obviously the key to sorting out how they actually work. Until recently, most information about channel structure was derived indirectly from studies of the amino acid composition and physiological properties of these proteins. For example, a great deal has been learned by exploring the functions of particular amino acids within the proteins using mutagenesis and the expression of such channels in Xenopus oocytes (see Box B). Such studies have discovered a general transmembrane architecture common to all the major ion channel families. Thus, these molecules are all integral membrane proteins that span the plasma membrane repeatedly. Na+ (and Ca2+) channel proteins, consist of repeating motifs of 6 membrane-spanning regions that are repeated 4 times, for a total of 24 transmembrane regions (Figure 4.6A,B). Na+ (or Ca2+) channels can be produced by just one of these proteins, although other accessory proteins, called β subunits, can regulate the function of these channels. K+ channel proteins typically span the membrane six times (Figure 4.6C), (A) Na+ CHANNEL Figure 4.6 Topology of the principal subunits of voltage-gated Na+, Ca2+, K+, and Cl– channels. Repeating motifs of Na+ (A) and Ca2+ (B) channels are labeled I, II, III, and IV; (C–F) K+ channels are more diverse. In all cases, four subunits combine to form a functional channel. (G) Chloride channels are structurally distinct from all other voltage-gated channels. (B) Ca2+ CHANNEL I II III IV I β subunit N II III IV N C C β subunit C N C N β subunit N C K+ CHANNELS (G) Cl− CHANNEL (C) Kv and HERG (D) Inward rectifier (E) Ca2+-activated (F) 2-pore N N C N C N C N C C 80 Chapter Four Pore closed Pore open Depolarize Ion flux Hyperpolarize Figure 4.7 A charged voltage sensor permits voltage-dependent gating of ion channels. The process of voltage activation may involve the rotation of a positively charged transmembrane domain. This movement causes a change in the conformation of the pore loop, enabling the channel to conduct specific ions. Membrane depolarization causes charged helix to rotate though there are some K+ channels, such as a bacterial channel and some mammalian channels, that span the membrane only twice (Figure 4.6D), and others that span the membrane four times (Figure 4.6F) or seven times (Figure 4.6E). Each of these K+ channel proteins serves as a channel subunit, with 4 of these subunits typically aggregating to form a single functional ion channel. Other imaginative mutagenesis experiments have provided information about how these proteins function. Two membrane-spanning domains of all ion channels appear to form a central pore through which ions can diffuse, and one of these domains contains a protein loop that confers an ability to selectivity allow certain ions to diffuse through the channel pore (Figure 4.7). As might be expected, the amino acid composition of the pore loop differs among channels that conduct different ions. These distinct structural features of channel proteins also provide unique binding sites for drugs and for various neurotoxins known to block specific subclasses of ion channels (Box C). Furthermore, many voltage gated ion channels contain a distinct type of transmembrane helix containing a number of positively charged amino acids along one face of the helix (Figures 4.6 and 4.7). This structure evidently serves as a sensor that detects changes in the electrical potential across the membrane. Membrane depolarization influences the charged amino acids such that the helix undergoes a conformational change, which in turn allows the channel pore to open. One suggestion is that the helix rotates to cause the pore to open (Figure 4.7). Other types of mutagenesis experiments have demonstrated that one end of certain K+ channels plays a key role in channel inactivation. This intracellular structure (labeled “N” in Figure 4.6C) can plug the channel pore during prolonged depolarization. More recently, very direct information about the structural underpinnings of ion channel function has come from X-ray crystallography studies of bacterial K+ channels (Figure 4.8). This molecule was chosen for analysis because the large quantity of channel protein needed for crystallography could be obtained by growing large numbers of bacteria expressing this molecule. The results of such studies showed that the channel is formed by subunits that each cross the plasma membrane twice; between these two membrane-spanning structures is a loop that inserts into the plasma membrane (Figure 4.8A). Four of these subunits are assembled together to form a chan- Channels and Transpor ters 81 (A) (B) SIDE VIEW TOP VIEW Selectivity filter Pore loop K+ ion in pore Pore helix Outer helix Outer helix Inner helix Inner helix (C) Selectivity filter K+ ions Water-filled cavity Pore Figure 4.8 Structure of a simple bacterial K+ channel determined by crystallography. (A) Structure of one subunit of the channel, which consists of two membrane-spanning domains and a pore loop that inserts into the membrane. (B) Three-dimensional arrangement of four subunits (each in a different color) to form a K+ channel. The top view illustrates a K+ ion (green) within the channel pore. (C) The permeation pathway of the K+ channel consists of a large aqueous cavity connected to a narrow selectivity filter. Helical domains of the channel point negative charges (red) toward this cavity, allowing K+ ions (green) to become dehydrated and then move through the selectivity filter. (A, B from Doyle et al., 1998; C after Doyle et al., 1998.) Negatively charged pore helix nel (Figure 4.8B). In the center of the assembled channel is a narrow opening through the protein that allows K+ to flow across the membrane. This opening is the channel pore and is formed by the protein loop, as well as by the membrane-spanning domains. The structure of the pore is well suited for conducting K+ ions (Figure 4.8C). The narrowest part is near the outside mouth of the channel and is so constricted that only a non-hydrated K+ ion can fit through the bottleneck. Larger cations, such as Cs+, cannot traverse this region of the pore, and smaller cations such as Na+ cannot enter the pore because the “walls” of the pore are too far apart to stabilize a dehydrated Na+ ion. This part of the channel complex is responsible for the selective permeability to K+ and is therefore called the selectivity filter. The sequence of amino acids making up part of this selectivity filter is often referred to as the K+ channel “signature sequence”. Deeper within the channel is a water-filled cavity that connects to the interior of the cell. This cavity evidently collects K+ from the cytoplasm and, utilizing negative charges from the protein, 82 Chapter Four Box C Toxins That Poison Ion Channels thereby scrambling information flow within the nervous system of the soonto-be-devoured victim. Other peptides in scorpion venom, called b-toxins, shift the voltage dependence of Na+ channel activation (Figure B). These toxins cause Na+ channels to open at potentials much more negative than normal, disrupting action potential generation. Some alkaloid toxins combine these actions, both removing inactivation and shifting activation of Na+ channels. One such toxin is batrachotoxin, produced by a species of frog; some tribes of South American Indians use this poison on their arrow tips. A number of plants produce similar toxins, including aconitine, from buttercups; veratridine, from lilies; and a number of insecticidal toxins produced by plants such as chrysanthemums and rhododendrons. Potassium channels have also been targeted by toxin-producing organisms. (1) References CAHALAN, M. (1975) Modification of sodium channel gating in frog myelinated nerve fibers by Centruroides sculpturatus scorpion venom. J. Physiol. (Lond.) 244: 511–534. NARAHASHI, T. (2000) Neuroreceptors and ion channels as the basis for drug action: Present and future. J. Pharmacol. Exptl. Therapeutics 294: 1–26. SCHMIDT, O. AND H. SCHMIDT (1972) Influence of calcium ions on the ionic currents of nodes of Ranvier treated with scorpion venom. Pflügers Arch. 333: 51–61. (B) Treated with scorpion toxin Control 0 −40 −80 conductance Normalized Na+ Membrane potential (mV) (A) Peptide toxins affecting K+ channels include dendrotoxin, from wasps; apamin, from bees; and charybdotoxin, yet another toxin produced by scorpions. All of these toxins block K+ channels as their primary action; no toxin is known to affect the activation or inactivation of these channels, although such agents may simply be awaiting discovery. Na+ current (nA/cm2 ) Given the importance of Na+ and K+ channels for neuronal excitation, it is not surprising that a number of organisms have evolved channel-specific toxins as mechanisms for self-defense or for capturing prey. A rich collection of natural toxins selectively target the ion channels of neurons and other cells. These toxins are valuable not only for survival, but for studying the function of cellular ion channels. The best-known channel toxin is tetrodotoxin, which is produced by certain puffer fish and other animals. Tetrodotoxin produces a potent and specific obstruction of the Na+ channels responsible for action potential generation, thereby paralyzing the animals unfortunate enough to ingest it. Saxitoxin, a chemical homologue of tetrodotoxin produced by dinoflagellates, has a similar action on Na+ channels. The potentially lethal effects of eating shellfish that have ingested these “red tide” dinoflagellates are due to the potent neuronal actions of saxitoxin. Scorpions paralyze their prey by injecting a potent mix of peptide toxins that also affect ion channels. Among these are the a-toxins, which slow the inactivation of Na+ channels (Figure A1); exposure of neurons to these toxins prolongs the action potential (Figure A2), 0 20 (2) 40 60 80 0 20 Time (ms) 40 60 80 Treated with scorpion toxin Control +50 Membrane potential (mV) (A) Effects of toxin treatment on frog axons. (1) α-Toxin from the scorpion Leiurus quinquestriatus prolongs Na+ currents recorded with the voltage clamp method. (2) As a result of the increased Na+ current, αtoxin greatly prolongs the duration of the axonal action potential. Note the change in timescale after treating with toxin. (B) Treatment of a frog axon with β-toxin from another scorpion, Centruroides sculpturatus, shifts the activation of Na+ channels, so that Na+ conductance begins to increase at potentials much more negative than usual. (A after Schmidt and Schmidt, 1972; B after Cahalan, 1975.) +25 0 −25 −50 −75 0 2 4 Time (ms) 6 0 4 8 Time (s) 10 Control Treated with scorpion toxin −120 −80 −40 0 +40 Membrane potential (mV) Channels and Transpor ters 83 allows K+ ions to become dehydrated so they can enter the selectivity filter. These “naked” ions are then able to move through four K+ binding sites within the selectivity filter to eventually reach the extracellular space (recall that the normal concentration gradient drives K+ out of cells). On average, two K+ ions reside within the selectivity filter at any moment, with electrostatic repulsion between the two ions helping to speed their transit through the selectivity filter, thereby permitting rapid ion flux through the channel. Crystallographic studies have also determined the structure of the voltage sensor in another type of bacterial K+ channel. Such studies indicate that the sensor is at the interface between proteins and lipid on the cytoplasmic surface of the channel, leading to the suggestion that the sensor is a paddle-like structure that moves through the membrane to gate the opening of the channel pore (Figure 4.9A), rather than being a rotating helix buried within the ion channel protein (as in Figure 4.7). Crystallographic work has also revealed the molecular basis of the rapid transitions between the closed and the open state of the channel during channel gating. By comparing data from K+ channels crystallized in what is believed to be closed and open conformations (Figure 4.9B), it appears that channels gate by a conformational change in one of the transmembrane helices lining the channel pore. Producing a “kink” in one of these helices increases the opening from the central water-filled pore to the intracellular space, thereby permitting ion fluxes. (A) (A) Depolarize Hyperpolarize Closed Open (B) Closed Open Figure 4.9 Structural features of K+ channel gating. (A) Voltage sensing may involve paddle-like structures of the channel. These paddles reside within the lipid bilayer of the plasma membrane and may respond to changes in membrane potential by moving through the membrane. The gating charges that sense membrane potential are indicated by red “plus” signs. (B) Structure of K+ channels in closed (left) and open (right) conformations. Three of the four channel subunits are shown. Opening of the pore of the channel involves kinking of a transmembrane domain at the point indicated in red, which then dilates the pore. (A after Jiang et al., 2003; B after MacKinnon, 2003). 84 Chapter Four Box D Diseases Caused by Altered Ion Channels Several genetic diseases, collectively called channelopathies, result from small but critical alterations in ion channel genes. The best-characterized of these diseases are those that affect skeletal muscle cells. In these disorders, alterations in ion channel proteins produce either myotonia (muscle stiffness due to excessive electrical excitability) or paralysis (due to insufficient muscle excitability). Other disorders arise from ion channel defects in heart, kidney, and the inner ear. Channelopathies associated with ion channels localized in brain are much more difficult to study. Nonetheless, voltage-gated Ca2+ channels have recently been implicated in a range of neurological diseases. These include episodic ataxia, spinocerebellar degeneration, night blindness, and migraine headaches. Familial hemiplegic migraine (FHM) is characterized by migraine attacks that typically last one to three days. During such episodes, patients experience severe headaches and vomiting. Several mutations in a human Ca2+ channel have been identified in families with FHM, each having different clinical symptoms. For example, a mutation in the pore-forming region of the channel produces hemiplegic migraine with progressive cerebellar ataxia, whereas other mutations cause only the usual FHM symptoms. How these altered Ca2+ channel properties lead to migraine attacks is not known. Episodic ataxia type 2 (EA2) is a neurological disorder in which affected individuals suffer recurrent attacks of abnormal limb movements and severe ataxia. These problems are sometimes accompaGenetic mutations in (A) Ca2+ channels, (B) Na+ channels, (C) K+ channels, and (D) Cl– channels that result in diseases. Red regions indicate the sites of these mutations; the red circles indicate mutations. (After LehmannHorn and Jurkat-Kott, 1999.) (A) Ca2+ CHANNEL I II III IV C N FHM EA2 CSNB Paralysis (B) Na+ CHANNEL I II III IV N N C C β subunit C N GEFS Myotonia Paralysis (C) K+ CHANNEL (D) Cl− CHANNEL C N EA1 BFNC Myotonia C Membrane potential (mV) Channels and Transpor ters 85 40 0 −40 −80 Na+ current (nA) Wild type Na+ channel mutants 0 5 Time (ms) 10 Mutations in Na+ channels slow the rate of inactivation of Na+ currents. (After Barchi, 1995.) nied by vertigo, nausea, and headache. Usually, attacks are precipitated by emotional stress, exercise, or alcohol and last for a few hours. The mutations in EA2 cause Ca2+ channels to be truncated at various sites, which may cause the clinical manifestations of the disease by preventing the normal assembly of Ca2+ channels in the membrane. X-linked congenital stationary night blindness (CSNB) is a recessive retinal disorder that causes night blindness, decreased visual acuity, myopia, nystagmus, and strabismus. Complete CSNB causes retinal rod photoreceptors to be nonfunctional. Incomplete CSNB causes subnormal (but measurable) functioning of both rod and cone photoreceptors. Like EA2, the incomplete type of CSNB is caused by mutations producing truncated Ca2+ channels. Abnormal retinal function may arise from decreased Ca2+ currents and neurotransmitter release from photoreceptors (see Chapter 11). A defect in brain Na+ channels causes generalized epilepsy with febrile seizures (GEFS) that begins in infancy and usually continues through early puberty. This defect has been mapped to two mutations: one on chromosome 2 that encodes an α subunit for a voltage-gated Na+ channel, and the other on chromosome 19 that encodes a Na+ channel β subunit. These mutations cause a slowing of Na+ channel inactivation (see figure above), which may explain the neuronal hyperexcitability underlying GEFS. Another type of seizure, benign familial neonatal convulsion (BFNC), is due to K+ channel mutations. This disease is characterized by frequent brief seizures commencing within the first week of life and disappearing spontaneously within a few months. The mutation has been mapped to at least two voltage-gated K+ channel genes. A reduction in K+ current flow through the mutated channels probably accounts for the hyperexcitability associated with this defect. A related disease, episodic ataxia type 1 (EA1), has been linked to a defect in another type of voltage-gated K+ channel. EA1 is characterized by brief episodes of ataxia. Mu- In short, ion channels are integral membrane proteins with characteristic features that allow them to assemble into multimolecular aggregates. Collectively, these structures allow channels to conduct ions, sense the transmembrane potential, to inactivate, and to bind to various neurotoxins. A combination of physiological, molecular biological and crystallographic studies has begun to provide a detailed physical picture of K+ channels. This work has now provided considerable insight into how ions are conducted from one side of the plasma membrane to the other, how a channel can be selectively permeable to a single type of ion, how they are able to sense changes in membrane voltage, and how they gate the opening of their pores. It is likely that other types of ion channels will be similar in their functional architecture. Finally, this sort of work has illuminated how mutations in ion channel genes can lead to a variety of neurological disorders (Box D). tant channels inhibit the function of other, non-mutant K+ channels and may produce clinical symptoms by impairing action potential repolarization. Mutations in the K+ channels of cardiac muscle are responsible for the irregular heartbeat of patients with long Q-T syndrome. Numerous genetic disorders affect the voltage-gated channels of skeletal muscle and are responsible for a host of muscle diseases that either cause muscle weakness (paralysis) or muscle contraction (myotonia). References BARCHI, R. L. (1995) Molecular pathology of the skeletal muscle sodium channel. Ann. Rev. Physiol. 57: 355–385. BERKOVIC, S. F. AND I. E. SCHEFFER (1997) Epilepsies with single gene inheritance. Brain Develop. 19 :13–28. COOPER, E. C. AND L. Y. JAN (1999) Ion channel genes and human neurological disease: Recent progress, prospects, and challenges. Proc. Natl. Acad. Sci. USA 96: 4759–4766. DAVIES, N. P. AND M. G. HANNA (1999) Neurological channelopathies: Diagnosis and therapy in the new millennium. Ann. Med. 31: 406–420. JEN, J. (1999) Calcium channelopathies in the central nervous system. Curr. Op. Neurobiol. 9: 274–280. LEHMANN-HORN, F. AND K. JURKAT-ROTT (1999) Voltage-gated ion channels and hereditary disease. Physiol. Rev. 79: 1317–1372. OPHOFF, R. A., G. M. TERWINDT, R. R. FRANTS AND M. D. FERRARI (1998) P/Q-type Ca2+ channel defects in migraine, ataxia and epilepsy. Trends Pharm. Sci. 19: 121–127. 86 Chapter Four Active Transporters Create and Maintain Ion Gradients Up to this point, the discussion of the molecular basis of electrical signaling has taken for granted the fact that nerve cells maintain ion concentration gradients across their surface membranes. However, none of the ions of physiological importance (Na+, K+, Cl–, and Ca2+) are in electrochemical equilibrium. Because channels produce electrical effects by allowing one or more of these ions to diffuse down their electrochemical gradients, there would be a gradual dissipation of these concentration gradients unless nerve cells could restore ions displaced during the current flow that occurs as a result of both neural signaling and the continual ionic leakage that occurs at rest. The work of generating and maintaining ionic concentration gradients for particular ions is carried out by a group of plasma membrane proteins known as active transporters. Active transporters carry out this task by forming complexes with the ions that they are translocating. The process of ion binding and unbinding for transport typically requires several milliseconds. As a result, ion translocation by active transporters is much slower than ion movement through channels: Recall that ion channels can conduct thousands of ions across a membrane each millisecond. In short, active transporters gradually store energy in the form of ion concentration gradients, whereas the opening of ion channels rapidly dissipates this stored energy during relatively brief electrical signaling events. Several types of active transporter have now been identified (Figure 4.10). Although the specific jobs of these transporters differ, all must translocate ions against their electrochemical gradients. Moving ions uphill requires the consumption of energy, and neuronal transporters fall into two classes based on their energy sources. Some transporters acquire energy directly from the hydrolysis of ATP and are called ATPase pumps (Figure 4.10, left). The most prominent example of an ATPase pump is the Na+ pump (or, more properly, the Na+/K+ ATPase pump), which is responsible for maintaining transmembrane concentration gradients for both Na+ and K+ (Figure 4.10A). Another is the Ca2+ pump, which provides one of the main mechanisms for removing Ca2+ from cells (Figure 4.10B). The second class of active transporter does not use ATP directly, but depends instead on the electrochemical gradients of other ions as an energy source. This type of transporter carries one or more ions up its electrochemical gradient while simultaneously taking another ion (most often Na+) down its gradient. Because at least two species of ions are Figure 4.10 Examples of ion transporters found in cell membranes. (A,B) Some transporters are powered by the hydrolysis of ATP (ATPase pumps), whereas others (C–F) use the electrochemical gradients of co-transported ions as a source of energy (ion exchangers). ATPase PUMPS (A) Na+/K+ pump ION EXCHANGERS (C) Na+/Ca2+ (D) Cl−/HCO3− (E) Na+/H+ (F) Na+/neurotransmitter exchanger exchanger exchanger transporter (B) Ca2+ pump Na+ H+ K+ Na+ Na+ HCO3− Outside Inside Na+ ADP ATP Ca2+ Ca2+ ADP ATP Cl− H+ GABA, Dopamine Channels and Transpor ters 87 involved in such transactions, these transporters are usually called ion exchangers (Figure 4.10, right). An example of such a transporter is the Na+/Ca2+ exchanger, which shares with the Ca2+ pump the important job of keeping intracellular Ca2+ concentrations low (Figure 4.10C). Another exchanger in this category regulates both intracellular Cl– concentration and pH by swapping intracellular Cl– for another extracellular anion, bicarbonate (Figure 4.10D). Other ion exchangers, such as the Na+/H+ exchanger (Figure 4.10E), also regulate intracellular pH, in this case by acting directly on the concentration of H+. Yet other ion exchangers are involved in transporting neurotransmitters into synaptic terminals (Figure 4.10F), as described in Chapter 6. Although the electrochemical gradient of Na+ (or other counter ions) is the proximate source of energy for ion exchangers, these gradients ultimately depend on the hydrolysis of ATP by ATPase pumps, such as the Na+/K+ ATPase pump. Functional Properties of the Na+/K+ Pump Of these various transporters, the best understood is the Na+/K+ pump. The activity of this pump is estimated to account for 20–40% of the brain’s energy consumption, indicating its importance for brain function. The Na+ pump was first discovered in neurons in the 1950s, when Richard Keynes at Cambridge University used radioactive Na+ to demonstrate the energydependent efflux of Na+ from squid giant axons. Keynes and his collaborators found that this efflux ceased when the supply of ATP in the axon was interrupted by treatment with metabolic poisons (Figure 4.11A, point 4). Other conditions that lower intracellular ATP also prevent Na+ efflux. These experiments showed that removing intracellular Na+ requires cellular metabolism. Further studies with radioactive K+ demonstrated that Na+ efflux is associated with simultaneous, ATP-dependent influx of K+. These opposing fluxes of Na+ and K+ are operationally inseparable: Removal of external K+ greatly reduces Na+ efflux (Figure 4.11, point 2), and vice versa. These energy-dependent movements of Na+ and K+ implicated an ATPhydrolyzing Na+/K+ pump in the generation of the transmembrane gradients of both Na+ and K+. The exact mechanism responsible for these fluxes of Na+ and K+ is still not entirely clear, but the pump is thought to alternately shuttle these ions across the membranes in a cycle fueled by the transfer of a phosphate group from ATP to the pump protein (Figure 4.11B). Additional quantitative studies of the movements of Na+ and K+ indicate that the two ions are not pumped at identical rates: The K+ influx is only about two-thirds the Na+ efflux. Thus, the pump apparently transports two K+ into the cell for every three Na+ that are removed (see Figure 4.11B). This stoichiometry causes a net loss of one positively charged ion from inside of the cell during each round of pumping, meaning that the pump generates an electrical current that can hyperpolarize the membrane potential. For this reason, the Na+/K+ pump is said to be electrogenic. Because pumps act much more slowly than ion channels, the current produced by the Na+/K+ pump is quite small. For example, in the squid axon, the net current generated by the pump is less than 1% of the current flowing through voltagegated Na+ channels and affects the resting membrane potential by only a millivolt or less. Although the electrical current generated by the activity of the Na+/K+ pump is small, under special circumstances the pump can significantly influence the membrane potential. For instance, prolonged stimulation of 88 Chapter Four (A) 1 Efflux of Na+ 2 Na+ efflux reduced by removal of external K+ 0 (B) 3 Recovery when K+ is restored 5 Recovery when ATP is restored 4 Efflux decreased by metabolic inhibitors, such as dinitrophenol, which block ATP synthesis Na+ efflux (logarithmic scale) Figure 4.11 Ionic movements due to the Na+/K+ pump. (A) Measurement of radioactive Na+ efflux from a squid giant axon. This efflux depends on external K+ and intracellular ATP. (B) A model for the movement of ions by the Na+/K+ pump. Uphill movements of Na+ and K+ are driven by ATP, which phosphorylates the pump. These fluxes are asymmetrical, with three Na+ carried out for every two K+ brought in. (A after Hodgkin and Keynes, 1955; B after Lingrel et al., 1994.) 50 100 150 Time (min) 200 250 300 2. Phosphorylation ADP ATP 3. Conformational change causes Na+ release and K+ binding Pi 1. Na+ binding Outside K+ Na+ Pi Inside Na+ 4. Dephosphorylationinduced conformational change leads to K+ release Pi K+ Channels and Transpor ters 89 Individual action potentials mV Figure 4.12 The electrogenic transport of ions by the Na+/K+ pump can influence membrane potential. Measurements of the membrane potential of a small unmyelinated axon show that a train of action potentials is followed by a long-lasting hyperpolarization. This hyperpolarization is blocked by ouabain, indicating that it results from the activity of the Na+/K+ pump. (After Rang and Ritchie, 1968.) − 20 − 40 − 60 − 80 0 − 60 Ouabain − 80 0 The Molecular Structure of the Na+/K+ Pump These observations imply that the Na+ and K+ pump must exhibit several molecular properties: (1) It must bind both Na+ and K+; (2) it must possess sites that bind ATP and receive a phosphate group from this ATP; and (3) it must bind ouabain, the toxin that blocks this pump (Figure 4.13A). A variety of studies have now identified the aspects of the protein that account for these properties of the Na+/K+ pump. This pump is a large, integral membrane protein made up of at least two subunits, called α and β. The primary sequence shows that the α subunit spans the membrane 10 times, with most of the molecule found on the cytoplasmic side, whereas the β subunit spans the membrane once and is predominantly extracellular. Although a detailed account of the functional domains of the Na+/K+ pump is not yet available, some parts of the amino acid sequence have identified functions (Figure 4.13B). One intracellular domain of the protein is required for ATP binding (A) (B) Outside Ouabain binding site Ouabain binding site 2 K+ 10 Time (min) 20 Poststimulus hyperpolarization blocked Figure 4.13 Molecular structure of the Na+/K+ pump. (A) General features of the pump. (B) The molecule spans the membrane 10 times. Amino acid residues thought to be important for binding of ATP, K+, and ouabain are highlighted. (After Lingrel et al., 1994.) C Na+ and K+ binding Outside Membrane Membrane Inside Inside C N ATP ADP + Pi 5 Trains of action potentials Membrane potential (mV) small unmyelinated axons produces a substantial hyperpolarization (Figure 4.12). During the period of stimulation, Na+ enters through voltage-gated channels and accumulates within the axons. As the pump removes this extra Na+, the resulting current generates a long-lasting hyperpolarization. Support for this interpretation comes from the observation that conditions that block the Na+/K+ pump—for example, treatment with ouabain, a plant glycoside that specifically inhibits the pump—prevent the hyperpolarization. The electrical contribution of the Na+/K+ pump is particularly significant in these small-diameter axons because their large surface-to-volume ratio causes intracellular Na+ concentration to rise to higher levels than it would in other cells. Nonetheless, it is important to emphasize that, in most circumstances, the Na+/K+ pump plays no part in generating the action potential and has very little direct effect on the resting potential. Time (s) 3 Na+ α subunit Phosphorylation site ATP binding site β subunit N 90 Chapter Four and hydrolysis, and the amino acid phosphorylated by ATP has been identified. Another extracellular domain may represent the binding site for ouabain. However, the sites involved in the most critical function of the pump—the movement of Na+ and K+—have not yet been defined. Nonetheless, altering certain membrane-spanning domains (red in Figure 4.13B) impairs ion translocation; moreover, kinetic studies indicate that both ions bind to the pump at the same site. Because these ions move across the membrane, it is likely that this site traverses the plasma membrane; it is also likely that the site has a negative charge, since both Na+ and K+ are positively charged. The observation that removing negatively charged residues in a membrane-spanning domain of the protein (pale yellow in Figure 4.13B) greatly reduces Na+ and K+ binding provides at least a hint about the iontranslocating domain of the transporter molecule. Summary Ion transporters and channels have complementary functions. The primary purpose of transporters is to generate transmembrane concentration gradients, which are then exploited by ion channels to generate electrical signals. Ion channels are responsible for the voltage-dependent conductances of nerve cell membranes. The channels underlying the action potential are integral membrane proteins that open or close ion-selective pores in response to the membrane potential, allowing specific ions to diffuse across the membrane. The flow of ions through single open channels can be detected as tiny electrical currents, and the synchronous opening of many such channels generates the macroscopic currents that produce action potentials. Molecular studies show that such voltage-gated channels have highly conserved structures that are responsible for features such as ion permeation and voltage sensing, as well as the features that specify ion selectivity and toxin sensitivity. Other types of channels are sensitive to chemical signals, such as neurotransmitters or second messengers, or to heat or membrane deformation. A large number of ion channel genes create channels with a correspondingly wide range of functional characteristics, thus allowing different types of neurons to have a remarkable spectrum of electrical properties. Ion transporter proteins are quite different in both structure and function. The energy needed for ion movement against a concentration gradient (e.g., in maintaining the resting potential) is provided either by the hydrolysis of ATP or by the electrochemical gradient of co-transported ions. The Na+/K+ pump produces and maintains the transmembrane gradients of Na+ and K+, while other transporters are responsible for the electrochemical gradients for other physiologically important ions, such as Cl–, Ca2+, and H+. Together, transporters and channels provide a reasonably comprehensive molecular explanation for the ability of neurons to generate electrical signals. Channels and Transpor ters 91 Additional Reading Reviews ARMSTRONG, C. M. AND B. HILLE (1998) Voltage-gated ion channels and electrical excitability. Neuron 20: 371–380. BEZANILLA, F. AND A. M. CORREA (1995) Singlechannel properties and gating of Na+ and K+ channels in the squid giant axon. In Cephalopod Neurobiology, N. J. Abbott, R. Williamson and L. Maddock (eds.). New York: Oxford University Press, pp. 131–151. CATTERALL, W. A. (1988) Structure and function of voltage-sensitive ion channels. Science 242: 50–61. ISOM, L. L., K. S. DE JONGH AND W. A. CATTERALL (1994) Auxiliary subunits of voltage-gated ion channels. Neuron 12: 1183–1194. JAN, L. Y. AND Y. N. JAN (1997) Voltage-gated and inwardly rectifying potassium channels. J. Physiol. 505: 267–282. JENTSCH, T. J., T. FRIEDRICH, A. SCHRIEVER AND H. YAMADA (1999) The CLC chloride channel family. Pflügers Archiv. 437: 783–795. KAPLAN, J. H. (2002) Biochemistry of Na,KATPase. Annu. Rev. Biochem. 71: 511–535. KRISHTAL, O. (2003). The ASICs: Signaling molecules? Modulators? Trends Neurosci, 26: 477–483. LINGREL, J. B., J. VAN HUYSSE, W. O’BRIEN, E. JEWELL-MOTZ, R. ASKEW AND P. SCHULTHEIS (1994) Structure-function studies of the Na, KATPase. Kidney Internat. 45: S32–S39. MACKINNON, R. (2003) Potassium channels. FEBS Lett. 555: 62–65. NEHER, E. (1992) Nobel lecture: Ion channels for communication between and within cells. Neuron 8: 605–612. PATAPOUTIAN, A., A. M. PEIER, G. M. STORY AND V. VISWANATH (2003). ThermoTRP channels and beyond: Mechanisms of temperature sensation. Nat. Rev. Neurosci. 4: 529–539. SEEBURG, P. H. (2002). A-to-I editing: New and old sites, functions and speculations. Neuron 35: 17–20. SKOU, J. C. (1988) Overview: The Na,K pump. Meth. Enzymol. 156: 1–25. Important Original Papers ANTZ, C. AND 7 OTHERS (1997) NMR structure of inactivation gates from mammalian voltage-dependent potassium channels. Nature 385: 272–275. BEZANILLA, F., E. PEROZO, D. M. PAPAZIAN AND E. STEFANI (1991) Molecular basis of gating charge immobilization in Shaker potassium channels. Science 254: 679–683. BOULTER, J. AND 6 OTHERS (1990) Molecular cloning and functional expression of glutamate receptor subunit genes. Science 249: 1033–1037. CATERINA, M. J., M. A. SCHUMACHER, M. TOMINAGA, T. A. ROSEN, J. D. LEVINE AND D. JULIUS (1997) The capsaicin receptor: A heat-activated ion channel in the pain pathway. Nature 389: 816–824. CHA, A., G. E. SNYDER, P. R. SELVIN AND F. BEZANILLA (1999) Atomic scale movement of the voltage-sensing region in a potassium channel measured via spectroscopy. Nature 402: 809–813. DOYLE, D. A. AND 7 OTHERS (1998) The structure of the potassium channel: Molecular basis of K+ conduction and selectivity. Science 280: 69–77. FAHLKE, C., H. T. YU, C. L. BECK, T. H. RHODES AND A. L. GEORGE JR. (1997) Pore-forming segments in voltage-gated chloride channels. Nature 390: 529–532. HO, K. AND 6 OTHERS (1993) Cloning and expression of an inwardly rectifying ATP-regulated potassium channel. Nature 362: 31–38. HODGKIN, A. L. AND R. D. KEYNES (1955) Active transport of cations in giant axons from Sepia and Loligo. J. Physiol. 128: 28–60. HOSHI, T., W. N. ZAGOTTA AND R. W. ALDRICH (1990) Biophysical and molecular mechanisms of Shaker potassium channel inactivation. Science 250: 533–538. JIANG, Y. AND 6 OTHERS (2003) X-ray structure of a voltage-dependent K+ channel. Nature 423: 33–41. LLANO, I., C. K. WEBB AND F. BEZANILLA (1988) Potassium conductance of squid giant axon. Single-channel studies. J. Gen. Physiol. 92: 179–196. MIKAMI, A. AND 7 OTHERS (1989) Primary structure and functional expression of the cardiac dihydropyridine-sensitive calcium channel. Nature 340: 230–233. NODA, M. AND 6 OTHERS (1986) Expression of functional sodium channels from cloned cDNA. Nature 322: 826–828. NOWYCKY, M. C., A. P. FOX AND R. W. TSIEN (1985) Three types of neuronal calcium channel with different calcium agonist sensitivity. Nature 316: 440–443. PAPAZIAN, D. M., T. L. SCHWARZ, B. L. TEMPEL, Y. N. JAN AND L. Y. JAN (1987) Cloning of genomic and complementary DNA from Shaker, a putative potassium channel gene from Drosophila. Science 237: 749–753. RANG, H. P. AND J. M. RITCHIE (1968) On the electrogenic sodium pump in mammalian non-myelinated nerve fibres and its activation by various external cations. J. Physiol. 196: 183–221. SIGWORTH, F. J. AND E. NEHER (1980) Single Na+ channel currents observed in cultured rat muscle cells. Nature 287: 447–449. THOMAS, R. C. (1969) Membrane current and intracellular sodium changes in a snail neurone during extrusion of injected sodium. J. Physiol. 201: 495–514. TOYOSHIMA, C., M. NAKASAKO, H. NOMURA AND H. OGAWA (2000) Crystal structure of the calcium pump of sarcoplasmic reticulum at 2.6 Å resolution. Nature 405: 647–655. VANDERBERG, C. A. AND F. BEZANILLA (1991) A sodium channel model based on single channel, macroscopic ionic, and gating currents in the squid giant axon. Biophys. J. 60: 1511–1533. WALDMANN, R., G. CHAMPIGNY, F. BASSILANA, C. HEURTEAUX AND M. LAZDUNSKI (1997) A proton-gated cation channel involved in acidsensing. Nature 386: 173–177. WEI, A. M., A. COVARRUBIAS, A. BUTLER, K. BAKER, M. PAK AND L. SALKOFF (1990) K+ current diversity is produced by an extended gene family conserved in Drosophila and mouse. Science 248: 599–603. YANG, N., A. L. GEORGE JR. AND R. HORN (1996) Molecular basis of charge movement in voltage-gated sodium channels. Neuron 16: 113–22. Books AIDLEY, D. J. AND P. R. STANFIELD (1996) Ion Channels: Molecules in Action. Cambridge: Cambridge University Press. ASHCROFT, F. M. (2000) Ion Channels and Disease. Boston: Academic Press. HILLE, B. (2001) Ion Channels of Excitable Membranes, 3rd Ed. Sunderland, MA: Sinauer Associates. JUNGE, D. (1992) Nerve and Muscle Excitation, 3rd Ed. Sunderland, MA: Sinauer Associates. NICHOLLS, D. G. (1994) Proteins, Transmitters and Synapses. Oxford: Blackwell Scientific SIEGEL, G. J., B. W. AGRANOFF, R. W. ALBERS, S. K. FISHER AND M. D. UHLER (1999) Basic Neurochemistry. Philadelphia: Lippincott-Raven. Chapter 5 Synaptic Transmission Overview The human brain contains at least 100 billion neurons, each with the ability to influence many other cells. Clearly, sophisticated and highly efficient mechanisms are needed to enable communication among this astronomical number of elements. Such communication is made possible by synapses, the functional contacts between neurons. Two different types of synapse—electrical and chemical—can be distinguished on the basis of their mechanism of transmission. At electrical synapses, current flows through gap junctions, which are specialized membrane channels that connect two cells. In contrast, chemical synapses enable cell-to-cell communication via the secretion of neurotransmitters; these chemical agents released by the presynaptic neurons produce secondary current flow in postsynaptic neurons by activating specific receptor molecules. The total number of neurotransmitters is not known, but is well over 100. Virtually all neurotransmitters undergo a similar cycle of use: synthesis and packaging into synaptic vesicles; release from the presynaptic cell; binding to postsynaptic receptors; and, finally, rapid removal and/or degradation. The secretion of neurotransmitters is triggered by the influx of Ca2+ through voltage-gated channels, which gives rise to a transient increase in Ca2+ concentration within the presynaptic terminal. The rise in Ca2+ concentration causes synaptic vesicles to fuse with the presynaptic plasma membrane and release their contents into the space between the pre- and postsynaptic cells. Although it is not yet understood exactly how Ca2+ triggers exocytosis, specific proteins on the surface of the synaptic vesicle and elsewhere in the presynaptic terminal mediate this process. Neurotransmitters evoke postsynaptic electrical responses by binding to members of a diverse group of neurotransmitter receptors. There are two major classes of receptors: those in which the receptor molecule is also an ion channel, and those in which the receptor and ion channel are separate molecules. These receptors give rise to electrical signals by transmitter-induced opening or closing of the ion channels. Whether the postsynaptic actions of a particular neurotransmitter are excitatory or inhibitory is determined by the ionic permeability of the ion channel affected by the transmitter, and by the concentration of permeant ions inside and outside the cell. Electrical Synapses Although there are many kinds of synapses within the human brain, they can be divided into two general classes: electrical synapses and chemical synapses. Although they are a distinct minority, electrical synapses are found in all nervous systems, permitting direct, passive flow of electrical current from one neuron to another. 93 94 Chapter Five (A) ELECTRONIC SYNAPSE (B) CHEMICAL SYNAPSE Microtubule Presynaptic neuron Cytoplasm Presynaptic neuron Synaptic vesicle Mitochondrion Gap junction Presynaptic membrane Postsynaptic neuron Ions flow through gap junction channels Postsynaptic neuron Synaptic vesicle fusing Neurotransmitter released Presynaptic membrane Synaptic cleft Postsynaptic membrane Gap junction channels Figure 5.1 Electrical and chemical synapses differ fundamentally in their P N mechanisms. i 3E(A) At electritransmission cal synapses, gap junctions between preand postsynaptic membranes permit current to flow passively through intercellular channels (blowup). This current flow changes the postsynaptic membrane potential, initiating (or in some instances inhibiting) the generation of postsynaptic action potentials. (B) At chemical synapses, there is no intercellular continuity, and thus no direct flow of current from pre- to postsynaptic cell. Synaptic current flows across the postsynaptic membrane only in response to the secretion of neurotransmitters, which open or close postsynaptic ion channels after binding to receptor molecules (blowup). Postsynaptic neurotransmitter receptor Ions flow through postsynaptic channels Postsynaptic membrane The structure of an electrical synapse is shown schematically in Figure 5.1A. The “upstream” neuron, which is the source of current, is called the presynaptic element, and the “downstream” neuron into which this current flows is termed postsynaptic. The membranes of the two communicating neurons come extremely close at the synapse and are actually linked together by an intercellular specialization called a gap junction. Gap junctions contain precisely aligned, paired channels in the membrane of the preand postsynaptic neurons, such that each channel pair forms a pore (see Figure 5.2A). The pore of a gap junction channel is much larger than the pores of the voltage-gated ion channels described in the previous chapter. As a result, a variety of substances can simply diffuse between the cytoplasm of the pre- and postsynaptic neurons. In addition to ions, substances that diffuse through gap junction pores include molecules with molecular weights as great as several hundred daltons. This permits ATP and other important intracellular metabolites, such as second messengers (see Chapter 7), to be transferred between neurons. Electrical synapses thus work by allowing ionic current to flow passively through the gap junction pores from one neuron to another. The usual source of this current is the potential difference generated locally by the action potential (see Chapter 3). This arrangement has a number of interesting consequences. One is that transmission can be bidirectional; that is, current can flow in either direction across the gap junction, depending on which member of the coupled pair is invaded by an action potential (although Synaptic Transmission 95 some types of gap junctions have special features that render their transmission unidirectional). Another important feature of the electrical synapse is that transmission is extraordinarily fast: because passive current flow across the gap junction is virtually instantaneous, communication can occur without the delay that is characteristic of chemical synapses. These features are apparent in the operation of the first electrical synapse to be discovered, which resides in the crayfish nervous system. A postsynaptic electrical signal is observed at this synapse within a fraction of a millisecond after the generation of a presynaptic action potential (Figure 5.2). In fact, at least part of this brief synaptic delay is caused by propagation of the action potential into the presynaptic terminal, so that there may be essentially no delay at all in the transmission of electrical signals across the synapse. Such synapses interconnect many of the neurons within the circuit that allows the crayfish to escape from its predators, thus minimizing the time between the presence of a threatening stimulus and a potentially life-saving motor response. A more general purpose of electrical synapses is to synchronize electrical activity among populations of neurons. For example, the brainstem neurons that generate rhythmic electrical activity underlying breathing are synchronized by electrical synapses, as are populations of interneurons in the cerebral cortex, thalamus, cerebellum, and other brain regions. Electrical transmission between certain hormone-secreting neurons within the mammalian hypothalamus ensures that all cells fire action potentials at about the same time, thus facilitating a burst of hormone secretion into the circulation. The fact that gap junction pores are large enough to allow molecules such as ATP and second messengers to diffuse intercellularly also permits electrical synapses to coordinate the intracellular signaling and metabolism of coupled cells. This property may be particularly important for glial cells, which form large intracellular signaling networks via their gap junctions. Presynaptic cell membrane (A) Figure 5.2 Structure and function of gap junctions at electrical synapses. (A) Gap junctions consist of hexameric complexes formed by the coming together of subunits called connexons, which are present in both the pre- and postsynaptic membranes. The pores of the channels connect to one another, creating electrical continuity between the two cells. (B) Rapid transmission of signals at an electrical synapse in the crayfish. An action potential in the presynaptic neuron causes the postsynaptic neuron to be depolarized within a fraction of a millisecond. (B after Furshpan and Potter, 1959.) (B) Presynaptic neuron 25 Connexons Postsynaptic cell membrane 3.5 3.5 nm nm Pores connecting cytoplasm of two neurons Membrane potential (mV) 0 20 20 nm nm −25 −50 25 Postsynaptic neuron 0 −25 −50 Brief (~0.1 ms) synaptic delay 0 1 2 Time (ms) 3 4 96 Chapter Five Signal Transmission at Chemical Synapses The general structure of a chemical synapse is shown schematically in Figure 5.1B. The space between the pre- and postsynaptic neurons is substantially greater at chemical synapses than at electrical synapses and is called the synaptic cleft. However, the key feature of all chemical synapses is the presence of small, membrane-bounded organelles called synaptic vesicles within the presynaptic terminal. These spherical organelles are filled with one or more neurotransmitters, the chemical signals secreted from the presynaptic neuron, and it is these chemical agents acting as messengers between the communicating neurons that gives this type of synapse its name. Transmission at chemical synapses is based on the elaborate sequence of events depicted in Figure 5.3. The process is initiated when an action potential invades the terminal of the presynaptic neuron. The change in membrane potential caused by the arrival of the action potential leads to the opening of voltage-gated calcium channels in the presynaptic membrane. Because of the steep concentration gradient of Ca2+ across the presynaptic membrane (the external Ca2+ concentration is approximately 10–3 M, whereas the internal Ca2+ concentration is approximately 10–7 M), the opening of these channels causes a rapid influx of Ca2+ into the presynaptic terminal, with the result that the Ca2+ concentration of the cytoplasm in the terminal transiently rises to a much higher value. Elevation of the presynaptic Ca2+ concentration, in turn, allows synaptic vesicles to fuse with the plasma membrane of the presynaptic neuron. The Ca2+-dependent fusion of synaptic vesicles with the terminal membrane causes their contents, most importantly neurotransmitters, to be released into the synaptic cleft. Following exocytosis, transmitters diffuse across the synaptic cleft and bind to specific receptors on the membrane of the postsynaptic neuron. The binding of neurotransmitter to the receptors causes channels in the postsynaptic membrane to open (or sometimes to close), thus changing the ability of ions to flow into (or out of) the postsynaptic cells. The resulting neurotransmitter-induced current flow alters the conductance and (usually) the membrane potential of the postsynaptic neuron, increasing or decreasing the probability that the neuron will fire an action potential. In this way, information is transmitted from one neuron to another. Properties of Neurotransmitters The notion that electrical information can be transferred from one neuron to the next by means of chemical signaling was the subject of intense debate through the first half of the twentieth century. A key experiment that supported this idea was performed in 1926 by German physiologist Otto Loewi. Acting on an idea that allegedly came to him in the middle of the night, Loewi proved that electrical stimulation of the vagus nerve slows the heartbeat by releasing a chemical signal. He isolated and perfused the hearts of two frogs, monitoring the rates at which they were beating (Figure 5.4). His experiment collected the perfusate flowing through the stimulated heart and transferred this solution to the second heart. When the vagus nerve to the first heart was stimulated, the beat of this heart slowed. Remarkably, even though the vagus nerve of the second heart had not been stimulated, its beat also slowed when exposed to the perfusate from the first heart. This result showed that the vagus nerve regulates the heart rate by releasing a chemical that accumulates in the perfusate. Originally referred to as “vagus substance,” the agent was later shown to be acetylcholine (ACh). ACh is now known to be a neurotransmitter that acts not only in the heart but at a vari- Synaptic Transmission 97 Myelin 2 An action potential invades the presynaptic terminal 3 Depolarization of presynaptic terminal causes opening of voltage-gated Ca2+ channels 1 Transmitter is synthesized and then stored in vesicles 4 Influx of Ca2+ through channels Synaptic vesicle 5 Ca2+ causes vesicles to fuse with presynaptic membrane Transmitter molecules 10 Retrieval of vesicular membrane from plasma membrane Ca 2+ 6 Transmitter is released into synaptic cleft via exocytosis Across dendrite Transmitter molecules Ions 9 Postsynaptic current causes excitatory or inhibitory postsynaptic potential that changes the excitability of the postsynaptic cell Transmitter receptor 8 Opening or closing of postsynaptic channels Postsynaptic current flow 7 Transmitter binds to receptor molecules in postsynaptic membrane ety of postsynaptic targets in the central and peripheral nervous systems, preeminently at the neuromuscular junction of striated muscles and in the visceral motor system (see Chapters 6 and 20). Over the years, a number of formal criteria have emerged that definitively identify a substance as a neurotransmitter (Box A). These have led to the identification of more than 100 different neurotransmitters, which can be Figure 5.3 Sequence of events involved in transmission at a typical chemical synapse. 98 Chapter Five (A) (B) Stimulate vagus nerve of heart 1 Stimulate Solution transferred to heart 2 Heart 1 Contraction force Vagus nerve Heartbeat slowed Time (s) Heart 1 Heart 2 Contraction force Inhibitory effect of vagus transferred Heart 2 Time (s) Figure 5.4 Loewi’s experiment demonstrating chemical neurotransmission. (A) Diagram of experimental setup. (B) Where the vagus nerve of an isolated frog’s heart was stimulated, the heart rate decreased (upper panel). If the perfusion fluid from the stimulated heart was transferred to a second heart, its rate decreased as well (lower panel). classified into two broad categories: small-molecule neurotransmitters and neuropeptides (Chapter 6). Having more than one transmitter diversifies the physiological repertoire of synapses. Multiple neurotransmitters can produce different types of responses on individual postsynaptic cells. For example, a neuron can be excited by one type of neurotransmitter and inhibited by another type of neurotransmitter. The speed of postsynaptic responses produced by different transmitters also differs, allowing control of electrical signaling over different time scales. In general, small-molecule neurotransmitters mediate rapid synaptic actions, whereas neuropeptides tend to modulate slower, ongoing synaptic functions. Until relatively recently, it was believed that a given neuron produced only a single type of neurotransmitter. It is now clear, however, that many types of neurons synthesize and release two or more different neurotransmitters. When more than one transmitter is present within a nerve terminal, the molecules are called co-transmitters. Because different types of transmitters can be packaged in different populations of synaptic vesicles, co-transmitters need not be released simultaneously. When peptide and small-molecule neurotransmitters act as co-transmitters at the same synapse, they are differentially released according to the pattern of synaptic activity: low-frequency activity often releases only small neurotransmitters, whereas highfrequency activity is required to release neuropeptides from the same presynaptic terminals. As a result, the chemical signaling properties of such synapses change according to the rate of activity. Effective synaptic transmission requires close control of the concentration of neurotransmitters within the synaptic cleft. Neurons have therefore developed a sophisticated ability to regulate the synthesis, packaging, release, and Synaptic Transmission 99 Box A Criteria That Define a Neurotransmitter Three primary criteria have been used to confirm that a molecule acts as a neurotransmitter at a given chemical synapse. 1. The substance must be present within the presynaptic neuron. Clearly, a chemical cannot be secreted from a presynaptic neuron unless it is present there. Because elaborate biochemical pathways are required to produce neurotransmitters, showing that the enzymes and precursors required to synthesize the substance are present in presynaptic neurons provides additional evidence that the substance is used as a transmitter. Note, however, that since the transmitters glutamate, glycine, and aspartate are also needed for protein synthesis and other metabolic reactions in all neurons, their presence is not sufficient evidence to establish them as neurotransmitters. 2. The substance must be released in response to presynaptic depolarization, and the release must be Ca2+-dependent. Another essential criterion for identifying a neurotransmitter is to demonstrate that it is released from the presynaptic neuron in response to presynaptic electrical activity, and that this release requires Ca2+ influx into the presynaptic terminal. Meeting this criterion is technically challenging, not only because it may be difficult to selectively stimulate the presynaptic neurons, but also because enzymes and transporters efficiently remove the secreted neurotransmitters. 3. Specific receptors for the substance must be present on the postsynaptic cell. A neurotransmitter cannot act on its target unless specific receptors for the transmitter are present in the postsynaptic membrane. One way to demonstrate receptors is to show that application of exogenous transmitter mimics the post- synaptic effect of presynaptic stimulation. A more rigorous demonstration is to show that agonists and antagonists that alter the normal postsynaptic response have the same effect when the substance in question is applied exogenously. High-resolution histological methods can also be used to show that specific receptors are present in the postsynaptic membrane (by detection of radioactively labeled receptor antibodies, for example). Fulfilling these criteria establishes unambiguously that a substance is used as a transmitter at a given synapse. Practical difficulties, however, have prevented these standards from being applied at many types of synapses. It is for this reason that so many substances must be referred to as “putative” neurotransmitters. Demonstrating the identity of a neurotransmitter at a synapse requires showing (1) its presence, (2) its release, and (3) the postsynaptic presence of specific receptors. (1) (2) (3) Action potential 1 Neurotransmitter present Presynaptic terminal Ca2+ Postsynaptic cell Application of transmitter, agonists, or antagonists Ca2+ 2 Neurotransmitter released 3 Neurotransmitter receptors activated 100 Chapter Five degradation (or removal) of neurotransmitters to achieve the desired levels of transmitter molecules. The synthesis of small-molecule neurotransmitters occurs locally within presynaptic terminals (Figure 5.5A). The enzymes needed to synthesize these transmitters are produced in the neuronal cell body and transported to the nerve terminal cytoplasm at 0.5–5 millimeters a day by a mechanism called slow axonal transport. The precursor molecules required to make new molecules of neurotransmitter are usually taken into the nerve terminal by transporters found in the plasma membrane of the terminal. The enzymes synthesize neurotransmitters in the cytoplasm of the presynaptic terminal and the transmitters are then loaded into synaptic vesicles via transporters in the vesicular membrane (see Chapter 4). For some small-molecule neurotransmitters, the final steps of synthesis occur inside the synaptic vesicles. Most small-molecule neurotransmitters are packaged in vesicles 40 to 60 nm in diameter, the centers of which appear clear in electron micrographs; accordingly, these vesicles are referred to as small clearcore vesicles (Figure 5.5B). Neuropeptides are synthesized in the cell body of a neuron, meaning that the peptide is produced a long distance away from its site of secretion (Figure 5.5C). To solve this problem, peptide-filled vesicles are transported along an axon and down to the synaptic terminal via fast axonal transport. This process carries vesicles at rates up to 400 mm/day along cytoskeletal elements called microtubules (in contrast to the slow axonal transport of the enzymes that synthesize small-molecule transmitters). Microtubules are long, cylindrical filaments, 25 nm in diameter, present throughout neurons and other cells. Peptide-containing vesicles are moved along these microtubule “tracks” by ATP-requiring “motor” proteins such as kinesin. Neuropeptides are packaged into synaptic vesicles that range from 90 to 250 nm in diameter. These vesicles are electron-dense in electron micrographs—hence they are referred to as large dense-core vesicles (Figure 5.5D). After a neurotransmitter has been secreted into the synaptic cleft, it must be removed to enable the postsynaptic cell to engage in another cycle of syn- ▲ Figure 5.5 Metabolism of small-molecule and peptide transmitters. (A) Small-molecule neurotransmitters are synthesized at nerve terminals. The enzymes necessary for neurotransmitter synthesis are made in the cell body of the presynaptic cell (1) and are transported down the axon by slow axonal transport (2). Precursors are taken up into the terminals by specific transporters, and neurotransmitter synthesis and packaging take place within the nerve endings (3). After vesicle fusion and release (4), the neurotransmitter may be enzymatically degraded. The reuptake of the neurotransmitter (or its metabolites) starts another cycle of synthesis, packaging, release, and removal (5). (B) Small clear-core vesicles at a synapse between an axon terminal (AT) and a dendritic spine (Den) in the central nervous system. Such vesicles typically contain small-molecule neurotransmitters. (C) Peptide neurotransmitters, as well as the enzymes that modify their precursors, are synthesized in the cell body (1). Enzymes and propeptides are packaged into vesicles in the Golgi apparatus. During fast axonal transport of these vesicles to the nerve terminals (2), the enzymes modify the propeptides to produce one or more neurotransmitter peptides (3). After vesicle fusion and exocytosis, the peptides diffuse away and are degraded by proteolytic enzymes (4). (D) Large dense-core vesicles in a central axon terminal (AT) synapsing onto a dendrite (Den). Such vesicles typically contain neuropeptides or, in some cases, biogenic amines. (B and D from Peters, Palay, and Webster, 1991.) (C) (A) 1 Nucleus 1 Synthesis of enzymes in cell body Synthesis of neurotransmitter precursors and enzymes RER Golgi apparatus 2 Microtubules 2 Transport of enzymes and peptide precursors down microtubule tracks Slow axonal transport of enzymes Axon 5 Transport of precursors into terminal Terminal 3 Precursor Enzymes 4 Release and diffusion of neuro– transmitter 4 Neurotransmitter diffuses away and is degraded by proteolytic enzymes Synthesis and packaging of neurotransmitter 3 Enzymes modify precursors to produce peptide neurotransmitter Neurotransmitter Diffusion and degradation (B) (D) AT AT Den Den 0.5 mm 102 Chapter Five aptic transmission. The removal of neurotransmitters involves diffusion away from the postsynaptic receptors, in combination with reuptake into nerve terminals or surrounding glial cells, degradation by specific enzymes, or a combination of these mechanisms. Specific transporter proteins remove most small-molecule neurotransmitters (or their metabolites) from the synaptic cleft, ultimately delivering them back to the presynaptic terminal for reuse. (A) Stimulate Stimulate axon Record Record postsynaptic membrane potential Axon Muscle cell Postsynaptic membrane potential (mV) (B) Stimulate motor axon +50 Action potential 0 Threshold −50 −100 0 End plate potential (EPP) 2 4 6 Time (ms) Postsynaptic membrane potential (mV) (C) 1 mV MEPP 0 200 400 Time (ms) Postsynaptic membrane potential (mV) (D) Stimulate motor axon Subthreshold EPP 1 mV 0 Spontaneous MEPP 20 40 60 80 100 Time (ms) Quantal Release of Neurotransmitters Much of the evidence leading to the present understanding of chemical synaptic transmission was obtained from experiments examining the release of ACh at neuromuscular junctions. These synapses between spinal motor neurons and skeletal muscle cells are simple, large, and peripherally located, making them particularly amenable to experimental analysis. Such synapses occur at specializations called end plates because of the saucer-like appearance of the site on the muscle fiber where the presynaptic axon elaborates its terminals (Figure 5.6A). Most of the pioneering work on neuromuscular transmission was performed by Bernard Katz and his collaborators at University College London during the 1950s and 1960s, and Katz has been widely recognized for his remarkable contributions to understanding synaptic transmission. Though he worked primarily on the frog neuromuscular junction, numerous subsequent experiments have confirmed the applicability of his observations to transmission at chemical synapses throughout the nervous system. When an intracellular microelectrode is used to record the membrane potential of a muscle cell, an action potential in the presynaptic motor neuron can be seen to elicit a transient depolarization of the postsynaptic muscle fiber. This change in membrane potential, called an end plate potential (EPP), is normally large enough to bring the membrane potential of the muscle cell well above the threshold for producing a postsynaptic action potential (Figure 5.6B). The postsynaptic action potential triggered by the EPP causes the muscle fiber to contract. Unlike the case for electrical synapses, there is a pronounced delay between the time that the presynaptic motor neuron is stimulated and when the EPP occurs in the postsynaptic muscle cell. Such a delay is characteristic of all chemical synapses. One of Katz’s seminal findings, in studies carried out with Paul Fatt in 1951, was that spontaneous changes in muscle cell membrane potential occur even in the absence of stimulation of the presynaptic motor neuron (Figure 5.6C). These changes have the same shape as EPPs but are much Figure 5.6 Synaptic transmission at the neuromuscular junction. (A) Experimental arrangement, typically using the muscle of a frog or rat. The axon of the motor neuron innervating the muscle fiber is stimulated with an extracellular electrode, while an intracellular microelectrode is inserted into the postsynaptic muscle cell to record its electrical responses. (B) End plate potentials (EPPs) evoked by stimulation of a motor neuron are normally above threshold and therefore produce an action potential in the postsynaptic muscle cell. (C) Spontaneous miniature EPPs (MEPPs) occur in the absence of presynaptic stimulation. (D) When the neuromuscular junction is bathed in a solution that has a low concentration of Ca2+, stimulating the motor neuron evokes EPPs whose amplitudes are reduced to about the size of MEPPs. (After Fatt and Katz, 1952.) Synaptic Transmission 103 smaller (typically less than 1 mV in amplitude, compared to an EPP of perhaps 40 or 50 mV). Both EPPs and these small, spontaneous events are sensitive to pharmacological agents that block postsynaptic acetylcholine receptors, such as curare (see Box B in Chapter 6). These and other parallels between EPPs and the spontaneously occurring depolarizations led Katz and his colleagues to call these spontaneous events miniature end plate potentials, or MEPPs. The relationship between the full-blown end plate potential and MEPPs was clarified by careful analysis of the EPPs. The magnitude of the EPP provides a convenient electrical assay of neurotransmitter secretion from a motor neuron terminal; however, measuring it is complicated by the need to prevent muscle contraction from dislodging the microelectrode. The usual means of eliminating muscle contractions is either to lower Ca2+ concentration in the extracellular medium or to partially block the postsynaptic ACh receptors with the drug curare. As expected from the scheme illustrated in Figure 5.3, lowering the Ca2+ concentration reduces neurotransmitter secretion, thus reducing the magnitude of the EPP below the threshold for postsynaptic action potential production and allowing it to be measured more precisely. Under such conditions, stimulation of the motor neuron produces very small EPPs that fluctuate in amplitude from trial to trial (Figure 5.6D). These fluctuations give considerable insight into the mechanisms responsible for neurotransmitter release. In particular, the variable evoked response in low Ca2+ is now known to result from the release of unit amounts of ACh by the presynaptic nerve terminal. Indeed, the amplitude of the smallest evoked response is strikingly similar to the size of single MEPPs (compare Figure 5.6C and D). Further supporting this similarity, increments in the EPP response (Figure 5.7A) occur in units about the size of single MEPPs (Figure 5.7B). These “quantal” fluctuations in the amplitude of EPPs indicated to Katz and colleagues that EPPs are made up of individual units, each equivalent to a MEPP. The idea that EPPs represent the simultaneous release of many MEPP-like units can be tested statistically. A method of statistical analysis based on the independent occurrence of unitary events (called Poisson statistics) predicts what the distribution of EPP amplitudes would look like during a large number of trials of motor neuron stimulation, under the assumption that EPPs are built up from unitary events like MEPPs (see Figure 5.7B). The distribution of EPP amplitudes determined experimentally was found to be just that expected if transmitter release from the motor neuron is indeed quantal (the red curve in Figure 5.7A). Such analyses confirmed the idea that release of acetylcholine does indeed occur in discrete packets, each equivalent to a MEPP. In short, a presynaptic action potential causes a postsynaptic EPP because it synchronizes the release of many transmitter quanta. Release of Transmitters from Synaptic Vesicles The discovery of the quantal release of packets of neurotransmitter immediately raised the question of how such quanta are formed and discharged into the synaptic cleft. At about the time Katz and his colleagues were using physiological methods to discover quantal release of neurotransmitter, electron microscopy revealed, for the first time, the presence of synaptic vesicles in presynaptic terminals. Putting these two discoveries together, Katz and others proposed that synaptic vesicles loaded with transmitter are the source of the quanta. Subsequent biochemical studies confirmed that synaptic vesi- 104 Chapter Five Figure 5.7 Quantized distribution of EPP amplitudes evoked in a low Ca2+ solution. Peaks of EPP amplitudes (A) tend to occur in integer multiples of the mean amplitude of MEPPs, whose amplitude distribution is shown in (B). The leftmost bar in the EPP amplitude distribution shows trials in which presynaptic stimulation failed to elicit an EPP in the muscle cell. The red curve indicates the prediction of a statistical model based on the assumption that the EPPs result from the independent release of multiple MEPP-like quanta. The observed match, including the predicted number of failures, supports this interpretation. (After Boyd and Martin, 1955.) (A) No EPP in response to stimulation Number of EPPs 20 Prediction of statistical model 15 10 5 0 0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 EPP amplitude (mV) (B) Number of MEPPs 30 20 10 0 0 0.4 0.8 MEPP amplitude (mV) cles are the repositories of transmitters. These studies have shown that ACh is highly concentrated in the synaptic vesicles of motor neurons, where it is present at a concentration of about 100 mM. Given the diameter of a small, clear-core synaptic vesicle (∼50 nm), approximately 10,000 molecules of neurotransmitter are contained in a single vesicle. This number corresponds quite nicely to the amount of ACh that must be applied to a neuromuscular junction to mimic a MEPP, providing further support for the idea that quanta arise from discharge of the contents of single synaptic vesicles. To prove that quanta are caused by the fusion of individual synaptic vesicles with the plasma membrane, it is necessary to show that each fused vesicle causes a single quantal event to be recorded postsynaptically. This challenge was met in the late 1970s, when John Heuser, Tom Reese, and colleagues correlated measurements of vesicle fusion with the quantal content of EPPs at the neuromuscular junction. In their experiments, the number of vesicles that fused with the presynaptic plasma membrane was measured by electron microscopy in terminals that had been treated with a drug (4aminopyridine, or 4-AP) that enhances the number of vesicle fusion events produced by single action potentials (Figure 5.8A). Parallel electrical measurements were made of the quantal content of the EPPs elicited in this way. A comparison of the number of synaptic vesicle fusions observed with the electron microscope and the number of quanta released at the synapse showed a good correlation between these two measures (Figure 5.8B). These results remain one of the strongest lines of support for the idea that a quantum of transmitter release is due to a synaptic vesicle fusing with the presynaptic membrane. Subsequent evidence, based on other means of measuring vesicle fusion, has left no doubt about the validity of this general interpretation of chemical synaptic transmission. Very recent work has identified structures within the presynaptic terminal that connect vesicles to the plasma membrane and may be involved in membrane fusion (Figure 5.8C). Synaptic Transmission 105 (A) (B) Number of vesicles fusing 6000 5000 4-AP concentration: 10−3M 4000 3000 10−4M 2000 1000 0 10−5M 0 1000 2000 3000 4000 5000 6000 Number of quanta released (C) Local Recycling of Synaptic Vesicles The fusion of synaptic vesicles causes new membrane to be added to the plasma membrane of the presynaptic terminal, but the addition is not permanent. Although a bout of exocytosis can dramatically increase the surface area of presynaptic terminals, this extra membrane is removed within a few minutes. Heuser and Reese performed another important set of experiments showing that the fused vesicle membrane is actually retrieved and taken back into the cytoplasm of the nerve terminal (a process called endocytosis). The experiments, again carried out at the frog neuromuscular junction, were based on filling the synaptic cleft with horseradish peroxidase (HRP), an enzyme that can be made to produce a dense reaction product that is visible in an electron microscope. Under appropriate experimental conditions, endocytosis could then be visualized by the uptake of HRP into the nerve terminal (Figure 5.9). To activate endocytosis, the presynaptic terminal was stimulated with a train of action potentials, and the subsequent fate of the HRP was followed by electron microscopy. Immediately follow- Figure 5.8 Relationship of synaptic vesicle exocytosis and quantal transmitter release. (A) A special electron microscopical technique called freeze-fracture microscopy was used to visualize the fusion of synaptic vesicles in presynaptic terminals of frog motor neurons. Left: Image of the plasma membrane of an unstimulated presynaptic terminal. Right: Image of the plasma membrane of a terminal stimulated by an action potential. Stimulation causes the appearance of dimple-like structures that represent the fusion of synaptic vesicles with the presynaptic membrane. The view is as if looking down on the release sites from outside the presynaptic terminal. (B) Comparison of the number of observed vesicle fusions to the number of quanta released by a presynaptic action potential. Transmitter release was varied by using a drug (4AP) that affects the duration of the presynaptic action potential, thus changing the amount of calcium that enters during the action potential. The diagonal line is the 1:1 relationship expected if each vesicle that opened released a single quantum of transmitter. (C) Fine structure of vesicle fusion sites of frog presynaptic terminals. Synaptic vesicles are arranged in rows and are connected to each other and to the plasma membrane by a variety of proteinaceous structures (yellow). Green structures in the presynaptic membrane, corresponding to the rows of particles seen in (A), are thought to be Ca2+ channels. (A and B from Heuser et al., 1979; C after Harlow et al., 2001) 106 Chapter Five (A) (B) (C) Wash away extracellular HRP; wait 5 minutes Briefly stimulate presynaptic terminal 2 Coated pits and coated vesicles contain HRP 1 Synaptic vesicles fuse Horseradish peroxidase (HRP) 1 hour later 3 Endosome contains HRP 4 Synaptic vesicles contain HRP (E) Endosome Endocytosis Budding 1 min Budding Docking Priming (D) Fusion Exocytosis Ca2+ 10–20 sec Figure 5.9 Local recycling of synaptic vesicles in presynaptic terminals. (A) Horseradish peroxidase (HRP) introduced into the synaptic cleft is used to follow the fate of membrane retrieved from the presynaptic plasma membrane. Stimulation of endocytosis by presynaptic action potentials causes HRP to be taken up into the presynaptic terminals via a pathway that includes (B) coated vesicles and (C) endosomes. (D) Eventually, the HRP is found in newly formed synaptic vesicles. (E) Interpretation of the results shown in A–D. Calcium-regulated fusion of vesicles with the presynaptic membrane is followed by endocytotic retrieval of vesicular membrane via coated vesicles and endosomes, and subsequent re-formation of new synaptic vesicles. (After Heuser and Reese, 1973.) 1 msec ing stimulation, the HRP was found within special endocytotic organelles called coated vesicles (Figure 5.9A,B). A few minutes later, however, the coated vesicles had disappeared and the HRP was found in a different organelle, the endosome (Figure 5.9C). Finally, within an hour after stimulating the terminal, the HRP reaction product appeared inside synaptic vesicles (Figure 5.9D). These observations indicate that synaptic vesicle membrane is recycled within the presynaptic terminal via the sequence summarized in Figure 5.9E. In this process, called the synaptic vesicle cycle, the retrieved vesicular membrane passes through a number of intracellular compartments—such as coated vesicles and endosomes—and is eventually used to make new synaptic vesicles. After synaptic vesicles are re-formed, they are stored in a reserve pool within the cytoplasm until they need to participate again in neurotransmitter release. These vesicles are mobilized from the reserve pool, docked at the presynaptic plasma membrane, and primed to participate in exocytosis once again. More recent experiments, employing a fluorescent label rather than HRP, have determined the time course of synaptic vesicle recycling. These studies indicate that the entire vesicle cycle requires approximately 1 minute, with membrane budding during endocytosis requiring 10–20 sec- Synaptic Transmission 107 onds of this time. As can be seen from the 1-millisecond delay in transmission following excitation of the presynaptic terminal (see Figure 5.6B), membrane fusion during exocytosis is much more rapid than budding during endocytosis. Thus, all of the recycling steps interspersed between membrane budding and subsequent refusion of a vesicle are completed in less than a minute. The precursors to synaptic vesicles originally are produced in the endoplasmic reticulum and Golgi apparatus in the neuronal cell body. Because of the long distance between the cell body and the presynaptic terminal in most neurons, transport of vesicles from the soma would not permit rapid replenishment of synaptic vesicles during continuous neural activity. Thus, local recycling is well suited to the peculiar anatomy of neurons, giving nerve terminals the means to provide a continual supply of synaptic vesicles. As might be expected, defects in synaptic vesicle recycling can cause severe neurological disorders, some of which are described in Box B. The Role of Calcium in Transmitter Secretion As was apparent in the experiments of Katz and others described in the preceding sections, lowering the concentration of Ca2+ outside a presynaptic motor nerve terminal reduces the size of the EPP (compare Figure 5.6B and D). Moreover, measurement of the number of transmitter quanta released under such conditions shows that the reason the EPP gets smaller is that lowering Ca2+ concentration decreases the number of vesicles that fuse with the plasma membrane of the terminal. An important insight into how Ca2+ regulates the fusion of synaptic vesicles was the discovery that presynaptic terminals have voltage-sensitive Ca2+ channels in their plasma membranes (see Chapter 4). The first indication of presynaptic Ca2+ channels was provided by Katz and Ricardo Miledi. They observed that presynaptic terminals treated with tetrodotoxin (which blocks Na+ channels; see Chapter 3) could still produce a peculiarly prolonged type of action potential. The explanation for this surprising finding was that current was still flowing through Ca2+ channels, substituting for the current ordinarily carried by the blocked Na+ channels. Subsequent voltage clamp experiments, performed by Rodolfo Llinás and others at a giant presynaptic terminal of the squid (Figure 5.10A), confirmed (A) Record Postsynaptic membrane potential Presynaptic neuron Vpre Voltage clamp Postsynaptic Ipre neuron (B) 0 Presynaptic –25 membrane –50 potential (mV) –75 Presynaptic calcium current (µA/cm2) Figure 5.10 The entry of Ca2+ through the specific voltage-dependent calcium channels in the presynaptic terminals causes transmitter release. (A) Experimental setup using an extraordinarily large synapse in the squid. The voltage clamp method detects currents flowing across the presynaptic membrane when the membrane potential is depolarized. (B) Pharmacological agents that block currents flowing through Na+ and K+ channels reveal a remaining inward current flowing through Ca2+ channels. This influx of calcium triggers transmitter secretion, as indicated by a change in the postsynaptic membrane potential. Treatment of the same presynaptic terminal with cadmium, a calcium channel blocker, eliminates both the presynaptic calcium current and the postsynaptic response. (After Augustine and Eckert, 1984.) CADMIUM ADDED CONTROL 0 200 0 Postsynaptic –25 membrane potential (mV) –50 –75 –3 0 3 6 9 12 –3 0 Time (ms) 3 6 9 12 108 Chapter Five Box B Diseases That Affect the Presynaptic Terminal Various steps in the exocytosis and endocytosis of synaptic vesicles are targets of a number of rare but debilitating neurological diseases. Many of these are myasthenic syndromes, in which abnormal transmission at neuromuscular synapses leads to weakness and fatigability of skeletal muscles (see Box B in Chapter 7). One of the best-understood examples of such disorders is the Lambert-Eaton myasthenic syndrome (LEMS), an occasional complication in patients with certain kinds of cancers. Biopsies of muscle tissue removed from LEMS patients allow intracellular recordings identical to those shown in Figure 5.6. Such recordings have shown that when a motor neuron is stimulated, the number of quanta contained in individual EPPs is greatly reduced, although the amplitude of spontaneous MEPPs is normal. Thus, LEMS impairs evoked neurotransmitter release, but does not affect the size of individual quanta. Several lines of evidence indicate that this reduction in neurotransmitter release is due to a loss of voltage-gated Ca2+ channels in the presynaptic terminal of motor neurons (see figure). Thus, the defect in neuromuscular transmission can be overcome by increasing the extracellular concentration of Ca2+, and anatomical studies indicate a lower density of Ca2+ channel proteins in the presynaptic plasma membrane. The loss of presynaptic Ca2+ channels in LEMS apparently arises from a defect in the immune system. The blood of LEMS patients has a very high concentration of antibodies that bind to Ca2+ channels, and it seems likely that these antibodies are the primary cause of LEMS. For example, removal of Ca2+ channel antibodies from the blood of LEMS patients by plasma exchange reduces muscle weakness. Similarly, immunosuppressant drugs also can alleviate LEMS symptoms. Perhaps most telling, injecting these antibodies into experimental animals elicits muscle weakness and abnormal neuromuscular transmission. Why the immune system generates antibodies against Ca2+ channels is not clear. Most LEMS patients have small-cell carcinoma, a form of lung cancer that may somehow initiate the immune response to Ca2+ channels. Whatever the origin, the binding of antibodies to Ca2+ channels causes a reduction in Ca2+ channel currents. It is this antibody-induced defect in presynaptic Ca2+ entry that accounts for the muscle weakness associated with LEMS. Congenital myasthenic syndromes are genetic disorders that also cause muscle weakness by affecting neuromuscular transmission. Some of these syndromes affect the acetylcholinesterase that degrades acetylcholine in the synaptic cleft, whereas others arise from autoimmune attack of acetylcholine receptors (see Box C in Chapter 6). However, a number of congenital myasthenic syndromes arise from defects in acetylcholine release due to altered synaptic vesicle traffic within the motor neuron terminal. Neuromuscular synapses in some of these patients have EPPs with reduced quantal content, a deficit that is especially prominent when the synapse is activated repeatedly. Electron microscopy shows that presynaptic motor nerve terminals have a greatly reduced number of synaptic vesicles. The defect in neurotransmitter release evidently results from an inadequate number of synaptic vesicles available for release during sustained presynaptic activity. The origins of this shortage of synaptic vesicles is not clear, but could result either from an impairment in endocytosis in the nerve terminal (see figure) or from a reduced supply of vesicles from the motor neuron cell body. Still other patients suffering from familial infantile myasthenia appear to have neuromuscular weakness that arises from reductions in the size of individual quanta, rather than the number of quanta released. Motor nerve terminals from these patients have synaptic vesicles that are normal in number, but smaller in diameter. This finding suggests a different type of genetic lesion that somehow alters formation of new synaptic vesicles following endocytosis, thereby leading to less acetylcholine in each vesicle. Another disorder of synaptic transmitter release results from poisoning by anaerobic Clostridium bacteria. This genus of microorganisms produces some Impaired endocytosis in congenital myasthenic syndromes Endosome Budding Budding Docking Fusion Priming LEMS attacks presynaptic Ca2+ channels Ca2+ Botulinum and tetanus toxins affect SNARE proteins involved in vesicle fusion Presynaptic targets of several neurological disorders. Synaptic Transmission 109 of the most potent toxins known, including several botulinum toxins and tetanus toxin. Both botulism and tetanus are potentially deadly disorders. Botulism can occur by consuming food containing Clostridium bacteria or by infection of wounds with the spores of these ubiquitous organisms. In either case, the presence of the toxin can cause paralysis of peripheral neuromuscular synapses due to abolition of neurotransmitter release. This interference with neuromuscular transmission causes skeletal muscle weakness, in extreme cases producing respiratory failure due to paralysis of the diaphragm and other muscles required for breathing. Botulinum toxins also block synapses innervating the smooth muscles of several organs, giving rise to visceral motor dysfunction. Tetanus typically results from the contamination of puncture wounds by Clostridium bacteria that produce tetanus toxin. In contrast to botulism, tetanus poisoning blocks the release of inhibitory transmitters from interneurons in the spinal cord. This effect causes a loss of synaptic inhibition on spinal motor neurons, producing hyperexcitation of skeletal muscle and tetanic contractions in affected muscles (hence the name of the disease). Although their clinical consequences are dramatically different, clostridial toxins have a common mechanism of action (see figure). Tetanus toxin and botulinum toxins work by cleaving the SNARE proteins involved in fusion of synaptic vesicles with the presynaptic plasma membrane (see Box C). This proteolytic action presumably accounts for the block of transmitter release at the afflicted synapses. The different actions of these toxins on synaptic transmission at excitatory motor versus inhibitory synapses appar- the presence of voltage-gated Ca2+ channels in the presynaptic terminal (Figure 5.10B). Such experiments showed that the amount of neurotransmitter released is very sensitive to the exact amount of Ca2+ that enters. Further, blockade of these Ca2+ channels with drugs also inhibits transmitter release (Figure 5.10B, right). These observations all confirm that the voltage-gated Ca2+ channels are directly involved in neurotransmission. Thus, presynaptic action potentials open voltage-gated Ca2+ channels, with a resulting influx of Ca2+. That Ca2+ entry into presynaptic terminals causes a rise in the concentration of Ca2+ within the terminal has been documented by microscopic imaging of terminals filled with Ca2+-sensitive fluorescent dyes (Figure 5.11A). The consequences of the rise in presynaptic Ca2+ concentration for neurotransmitter release has been directly shown in two ways. First, microinjection of Ca2+ into presynaptic terminals triggers transmitter release in the absence of presynaptic action potentials (Figure 5.11B). Second, presynaptic microinjection of calcium chelators (chemicals that bind Ca2+ and keep its concentration buffered at low levels) prevents presynaptic action potentials from causing transmitter secretion (Figure 5.11C). These results prove beyond any doubt that a rise in presynaptic Ca2+ concentration is both necessary and sufficient for neurotransmitter release. Thus, as is the case for many other forms of neuronal signaling (see Chapter 7), Ca2+ serves as a second messenger during transmitter release. While Ca2+ is a universal trigger for transmitter release, not all transmitters are released with the same speed. For example, while secretion of ACh ently results from the fact that these toxins are taken up by different types of neurons: Whereas the botulinum toxins are taken up by motor neurons, tetanus toxin is preferentially targeted to interneurons. The basis for this differential uptake of toxins is not known, but presumably arises from the presence of different types of toxin receptors on the two types of neurons. References ENGEL, A. G. (1991) Review of evidence for loss of motor nerve terminal calcium channels in Lambert-Eaton myasthenic syndrome. Ann. N.Y. Acad. Sci. 635: 246–258. ENGEL, A. G. (1994) Congenital myasthenic syndromes. Neurol. Clin. 12: 401–437. LANG, B. AND A. VINCENT (2003) Autoantibodies to ion channels at the neuromuscular junction. Autoimmun. Rev. 2: 94–100. MASELLI, R. A. (1998) Pathogenesis of human botulism. Ann. N.Y. Acad. Sci. 841: 122–139. 110 Chapter Five (A) (B) Ca2+ PHOTO with line overlay Postsynaptic membrane potential (mV) Ca2+ injection 2 5 0µ m − 64 − 65 0 (C) Postsynaptic membrane Presynaptic membrane potential (mV) potential (mV) Figure 5.11 Evidence that a rise in presynaptic Ca2+ concentration triggers transmitter release from presynaptic terminals. (A) Fluorescence microscopy measurements of presynaptic Ca2+ concentration at the squid giant synapse (see Figure 5.8A). A train of presynaptic action potentials causes a rise in Ca2+ concentration, as revealed by a dye (called fura-2) that fluoresces more strongly when the Ca2+ concentration increases. (B) Microinjection of Ca2+ into a squid giant presynaptic terminal triggers transmitter release, measured as a depolarization of the postsynaptic membrane potential. (C) Microinjection of BAPTA, a Ca2+ chelator, into a squid giant presynaptic terminal prevents transmitter release. (A from Smith et al., 1993; B after Miledi, 1971; C after Adler et al., 1991.) 25 1 2 Time (s) 3 4 INJECT Ca2+ BUFFER CONTROL 0 −25 −50 −75 25 0 −25 −50 −75 0 1 2 3 4 5 0 Time (ms) 1 2 3 4 5 from motor neurons requires only a fraction of a millisecond (see Figure 5.6), release of neuropeptides require high-frequency bursts of action potentials for many seconds. These differences in the rate of release probably arise from differences in the spatial arrangement of vesicles relative to presynaptic Ca2+ channels. This perhaps is most evident in cases where small molecules and peptides serve as co-transmitters (Figure 5.12). Whereas the small, clearcore vesicles containing small-molecule transmitters are typically docked at the plasma membrane in advance of Ca2+ entry, large dense core vesicles containing peptide transmitters are farther away from the plasma membrane (see Figure 5.5D). At low firing frequencies, the concentration of Ca2+ may increase only locally at the presynaptic plasma membrane, in the vicinity of open Ca2+ channels, limiting release to small-molecule transmitters from the docked small, clear-core vesicles. Prolonged high-frequency stimulation increases the Ca2+ concentration throughout the presynaptic terminal, thereby inducing the slower release of neuropeptides. Molecular Mechanisms of Transmitter Secretion Precisely how an increase in presynaptic Ca2+ concentration goes on to trigger vesicle fusion and neurotransmitter release is not understood. However, many important clues have come from molecular studies that have identified and characterized the proteins found on synaptic vesicles and their binding Synaptic Transmission 111 Small-molecule neurotransmitter in small clearcore vesicles Neuropeptide in large densecore vesicles Localized increase in Ca2+ concentration Low-frequency stimulation Preferential release of smallmolecule neurotransmitter More diffuse increase in Ca2+ concentration High-frequency stimulation Release of both types of transmitter partners on the presynaptic plasma membrane and cytoplasm (Figure 5.13). Most, if not all, of these proteins act at one or more steps in the synaptic vesicle cycle. Although a complete molecular picture of neurotransmitter release is still lacking, the roles of several proteins involved in vesicle fusion have been deduced. Several of the proteins important for neurotransmitter release are also involved in other types of membrane fusion events common to all cells. For example, two proteins originally found to be important for the fusion of vesicles with membranes of the Golgi apparatus, the ATPase NSF (NEM-sensitive fusion protein) and SNAPs (soluble NSF-attachment proteins), are also involved in priming synaptic vesicles for fusion. These two proteins work by regulating the assembly of other proteins that are called SNAREs (SNAP receptors). One of these SNARE proteins, synaptobrevin, is in the membrane of synaptic vesicles, while two other SNARE proteins called syntaxin and SNAP-25 are found primarily on the plasma membrane. These SNARE proteins can form a macromolecular complex that spans the two membranes, thus bringing them into close apposition (Figure 5.14A). Such an arrangement is well suited to promote the fusion of the two membranes, and several lines of evidence suggest that this is what actually occurs. One important observation is that toxins that cleave the SNARE proteins block neurotransmitter release (Box C). In addition, putting SNARE proteins into artificial lipid membranes and allowing these proteins to form complexes with each other causes the membranes to fuse. Many other proteins, such as Figure 5.12 Differential release of neuropeptide and small-molecule co-transmitters. Low-frequency stimulation preferentially raises the Ca2+ concentration close to the membrane, favoring the release of transmitter from small clearcore vesicles docked at presynaptic specializations. High-frequency stimulation leads to a more general increase in Ca2+, causing the release of peptide neurotransmitters from large dense-core vesicles, as well as small-molecule neurotransmitters from small clear-core vesicles. 112 Chapter Five Ca2+-binding proteins Proteins that form channels, transporters, or receptors SNARE-associated proteins GTP-binding proteins Proteins involved in endocytosis Miscellaneous important proteins Cysteine string protein Synaptobrevin Synaptic vesicle Synaptic vesicle membrane Ca2+/CaM dependent protein kinase II Synaptotagmin SV2 Synapsin Synaptophysin Rab 3 Dynamin Snapin Rabphilin SNAP Syndapin RIM Complexin Amphiphysin AP180 Synaptojanin Clathrin NSF AP–2 DOC2 Tomosyn nSec1 Hsc70 Auxilin Syntaphilin Ca2+ channel Syntaxin SNAP–25 Plasma membrane of presynaptic terminal Neurexin I CLI Cytoplasm Synaptic cleft Figure 5.13 Presynaptic proteins implicated in neurotransmitter release. Structures adapted from Brunger (2001) and Brodsky et al. (2001). Synaptic Transmission 113 complexin, nSec-1, snapin, syntaphilin, and tomosyn, bind to the SNAREs and presumably regulate the formation or disassembly of this complex. Because the SNARE proteins do not bind Ca2+, still other molecules must be responsible for Ca2+ regulation of neurotransmitter release. Several presynaptic proteins, including calmodulin, CAPS, and munc-13, are capable of binding Ca2+. However, the leading candidate for Ca2+ regulation of neurotransmitter release is synaptotagmin, a protein found in the membrane of synaptic vesicles. Synaptotagmin binds Ca2+ at concentrations similar to those required to trigger vesicle fusion within the presynaptic terminal. It may act as a Ca2+ sensor, signaling the elevation of Ca2+ within the terminal and thus triggering vesicle fusion. In support of this idea, alterations of the properties of synaptotagmin in the presynaptic terminals of mice, fruit flies, squid, and other experimental animals impair Ca2+-dependent neurotransmitter release. In fact, deletion of only one of the 19 synaptotagmin genes of mice is a lethal mutation, causing the mice to die soon after birth. How Ca2+ binding to synaptotagmin could lead to exocytosis is not yet clear. It is known that Ca2+ changes the chemical properties of synaptotagmin, allowing it to insert into membranes and to bind to other proteins, including the SNAREs. A plausible model is that the SNARE proteins bring the two membranes close together, and that Ca2+-induced changes in synaptotagmin then produce the final fusion of these membranes (Figure 5.14B). Still other proteins appear to be involved at subsequent steps of the synaptic vesicle cycle (Figure 5.14C). For example, the protein clathrin is involved in endocytotic budding of vesicles from the plasma membrane. Clathrin forms structures that resemble geodesic domes (Figure 5.14D); these structures form coated pits that initiate membrane budding. Assembly of individual clathrin triskelia (so named because of their 3-legged appearance) into coats is aided by several other accessory proteins, such as AP2, AP180 and amphiphysin. The coats increase the curvature of the budding membrane until it forms a coated vesicle-like structure. Another protein, called dynamin, is at least partly responsible for the final pinching-off of membrane to convert the coated pits into coated vesicles. The coats are then removed by an ATPase, Hsc70, with another protein called auxilin serving as a co-factor. Other proteins, such as synaptojanin, are also important for vesicle uncoating. Several lines of evidence indicate that the protein synapsin, which reversibly binds to synaptic vesicles, may cross-link newly formed vesicles to the cytoskeleton to keep the vesicles tethered within the reserve pool. Mobilization of these reserve pool vesicles is caused by phosphorylation of synapsin by proteins kinases (Chapter 7), which allows synapsin to dissociate from the vesicles, thus freeing the vesicles to make their way to the plasma membrane. In summary, a complex cascade of proteins, acting in a defined temporal and spatial order, allows neurons to secrete transmitters. Although the detailed mechanisms responsible for transmitter secretion are not completely clear, rapid progress is being made toward this goal. Neurotransmitter Receptors The generation of postsynaptic electrical signals is also understood in considerable depth. Such studies began in 1907, when the British physiologist John N. Langley introduced the concept of receptor molecules to explain the specific and potent actions of certain chemicals on muscle and nerve cells. Much subsequent work has shown that receptor molecules do indeed account for the ability of neurotransmitters, hormones, and drugs to alter the 114 Chapter Five (A) (B) Synaptic vesicle membrane (1) Vesicle docks Synaptotagmin Synaptotagmin Synaptobrevin Synaptobrevin Syntaxin Vesicle Ca2+ channel SNAP-25 (2) SNARE complexes form to pull membranes together SNAP-25 Syntaxin Presynaptic plasma membrane (C) (3) Entering Ca2+ binds to synaptotagmin Hsc 70 Auxilin Synaptojanin Endosome Ca2+ Dynamin Budding Uncoating (4) Ca2+-bound synaptotagmin catalyzes membrane fusion Synapsin Budding Docking Fusion Priming NSF SNAPs SNAREs Ca2+ Clathrin Synaptotagmin Figure 5.14 Molecular mechanisms of neurotransmitter release. (A) Structure of the SNARE complex. The vesicular SNARE, synaptobrevin (blue), forms a helical complex with the plasma membrane SNAREs syntaxin (red) and SNAP-25 (green). Also shown is the structure of synaptotagmin, a vesicular Ca2+-binding protein. (B) A model for Ca2+-triggered vesicle fusion. SNARE proteins on the synaptic vesicle and plasma membranes form a complex (as in A) that brings together the two membranes. Ca2+ then binds to synaptotagmin, causing the cytoplasmic region of this protein to insert into the plasma membrane, bind to SNAREs and catalyze membrane fusion. (C) Roles of presynaptic proteins in synaptic vesicle cycling. (D) Individual clathrin triskelia (left) assemble together to form membrane coats (right) involved in membrane budding during endocytosis. (A after Sutton et al., 1998; C after Sudhof, 1995; D after Marsh and McMahon, 2001.) (D) Clathrin triskelion Clathrin coat Synaptic Transmission 115 Box C Toxins That Affect Transmitter Release Several important insights about the molecular basis of neurotransmitter secretion have come from analyzing the actions of a series of biological toxins produced by a fascinating variety of organisms. One family of such agents is the clostridial toxins responsible for botulism and tetanus (see Box B). Clever and patient biochemical work has shown that these toxins are highly specific proteases that cleave presynaptic SNARE proteins (see figure). Tetanus toxin and botulinum toxin (types B, D, F, and G) specifically cleave the vesicle SNARE protein, synaptobrevin. Other botulinum toxins are proteases that cleave syntaxin (type C) and SNAP-25 (types A and E), SNARE proteins found on the presynaptic plasma membrane. Destruction of these presynaptic proteins is the basis for the actions of the toxins on neurotransmitter release. The evidence described in the text also implies that these three syn- aptic SNARE proteins are somehow important in the process of vesicle–plasma membrane fusion. Another toxin that targets neurotransmitter release is α-latrotoxin, a protein found in the venom of the female black widow spider. Application of this molecule to neuromuscular synapses causes a massive discharge of synaptic vesicles, even when Ca2+ is absent from the extracellular medium. While it is not yet clear how this toxin triggers Ca2+-independent exocytosis, α-latrotoxin binds to two different types of presynaptic proteins that may mediate its actions. One group of binding partners for α-latrotoxin is the neurexins, a group of integral membrane proteins found in presynaptic terminals (see Figure 5.13). Several lines of evidence implicate binding to neurexins in at least some of the actions of α-latrotoxin. Because the neurexins bind to synaptotagmin, a vesicular Ca2+-binding Synaptic vesicle membrane Synaptobrevin BoTX−G TeTX BoTX−B BoTX−D BoTX−F BoTX−A protein that is known to be important in exocytosis, this interaction may allow αlatrotoxin to bypass the usual Ca2+ requirement for triggering vesicle fusion. Another type of presynaptic protein that can bind to α-latrotoxin is called CL1 (based on its previous names, Ca2+-independent receptor for latrotoxin and latrophilin-1). CL1 is a relative of the G-protein-coupled receptors that mediate the actions of neurotransmitters and other extracellular chemical signals (see Chapter 7). Thus, the binding of α-latrotoxin to CL1 is thought to activate an intracellular signal transduction cascade that may be involved in the Ca2+-independent actions of α-latrotoxin. While more work is needed to establish the roles of neurexins and CL1 in the actions of αlatrotoxin definitively, effects on these two proteins probably account for the potent presynaptic actions of this toxin. Still other toxins produced by snakes, snails, spiders, and other predatory animals are known to affect transmitter release, but their sites of action have yet to be identified. Based on the precedents described here, it is likely that these biological poisons will continue to provide valuable tools for elucidating the molecular basis of neurotransmitter release, just as they will continue to enable the predators to feast on their prey. References SNAP-25 BoTX−C BoTX−E Syntaxin Presynaptic plasma membrane Cleavage of SNARE proteins by clostridial toxins. Indicated are the sites of proteolysis by tetanus toxin (TeTX) and various types of botulinum toxin (BoTX). (After Sutton et al., 1998.) KRASNOPEROV, V. G. AND 10 OTHERS (1997) αLatrotoxin stimulates exocytosis by the interaction with a neuronal G-protein-coupled receptor. Neuron 18: 925–937. MONTECUCCO, C. AND G. SCHIAVO (1994) Mechanism of action of tetanus and botulinum neurotoxins. Mol. Microbiol. 13: 1–8. SCHIAVO, G., M. MATTEOLI AND C. MONTECUCCO (2000) Neurotoxins affecting neuroexocytosis. Physiol. Rev. 80: 717–766. SUGITA, S., M. KHVOCHTEV AND T. C. SUDHOF (1999) Neurexins are functional α-latrotoxin receptors. Neuron 22: 489–496. 116 Chapter Five functional properties of neurons. While it has been clear since Langley’s day that receptors are important for synaptic transmission, their identity and detailed mechanism of action remained a mystery until quite recently. It is now known that neurotransmitter receptors are proteins embedded in the plasma membrane of postsynaptic cells. Domains of receptor molecules that extend into the synaptic cleft bind neurotransmitters that are released into this space by the presynaptic neuron. The binding of neurotransmitters, either directly or indirectly, causes ion channels in the postsynaptic membrane to open or close. Typically, the resulting ion fluxes change the membrane potential of the postsynaptic cell, thus mediating the transfer of information across the synapse. Postsynaptic Membrane Permeability Changes during Synaptic Transmission Just as studies of the neuromuscular synapse paved the way for understanding neurotransmitter release mechanisms, this peripheral synapse has been equally valuable for understanding the mechanisms that allow neurotransmitter receptors to generate postsynaptic signals. The binding of ACh to postsynaptic receptors opens ion channels in the muscle fiber membrane. This effect can be demonstrated directly by using the patch clamp method (see Box A in Chapter 4) to measure the minute postsynaptic currents that flow when two molecules of individual ACh bind to receptors, as Erwin Neher and Bert Sakmann first did in 1976. Exposure of the extracellular surface of a patch of postsynaptic membrane to ACh causes single-channel currents to flow for a few milliseconds (Figure 5.15A). This shows that ACh binding to its receptors opens ligand-gated ion channels, much in the way that changes in membrane potential open voltage-gated ion channels (Chapter 4). The electrical actions of ACh are greatly multiplied when an action potential in a presynaptic motor neuron causes the release of millions of molecules of ACh into the synaptic cleft. In this more physiological case, the transmitter molecules bind to many thousands of ACh receptors packed in a dense array on the postsynaptic membrane, transiently opening a very large number of postsynaptic ion channels. Although individual ACh receptors only open briefly, (Figure 5.15B1), the opening of a large number of channels is synchronized by the brief duration during which ACh is secreted from presynaptic terminals (Figure 5.15B2,3). The macroscopic current resulting from the summed opening of many ion channels is called the end plate current, or EPC. Because the current flowing during the EPC is normally inward, it causes the postsynaptic membrane potential to depolarize. This depolarizing change in potential is the EPP (Figure 5.15C), which typically triggers a postsynaptic action potential by opening voltage-gated Na+ and K+ channels (see Figure 5.6B). The identity of the ions that flow during the EPC can be determined via the same approaches used to identify the roles of Na+ and K+ fluxes in the currents underlying action potentials (Chapter 3). Key to such an analysis is identifying the membrane potential at which no current flows during transmitter action. When the potential of the postsynaptic muscle cell is controlled by the voltage clamp method (Figure 5.16A), the magnitude of the membrane potential clearly affects the amplitude and polarity of EPCs (Figure 5.16B). Thus, when the postsynaptic membrane potential is made more negative than the resting potential, the amplitude of the EPC becomes larger, whereas this current is reduced when the membrane potential is made more positive. At approximately 0 mV, no EPC is detected, and at even more positive poten- Synaptic Transmission 117 tials, the current reverses its polarity, becoming outward rather than inward (Figure 5.16C). The potential where the EPC reverses, about 0 mV in the case of the neuromuscular junction, is called the reversal potential. As was the case for currents flowing through voltage-gated ion channels (see Chapter 3), the magnitude of the EPC at any membrane potential is given by the product of the ionic conductance activated by ACh (gACh) and the electrochemical driving force on the ions flowing through ligand-gated channels. Thus, the value of the EPC is given by the relationship EPC = gACh(Vm – Erev) where Erev is the reversal potential for the EPC. This relationship predicts that the EPC will be an inward current at potentials more negative than Erev because the electrochemical driving force, Vm – Erev, is a negative number. Further, the EPC will become smaller at potentials approaching Erev because the driving force is reduced. At potentials more positive than Erev, the EPC is outward because the driving force is reversed in direction (that is, positive). Because the channels opened by ACh are largely insensitive to membrane voltage, gACh will depend only on the number of channels opened by ACh, which depends in turn on the concentration of ACh in the synaptic cleft. (A) Patch clamp measurement of single ACh receptor current (B) Currents produced by: Micropipette ACh 2 2 4 (2) FEW OPEN CHANNELS Number of open channels 2 µM Acetylcholine (ACh) 0 Channel closed 2 Channel open 0 2 4 6 8 Time (ms) 10 0 1 ACh receptor Na+ I (pA) 0 12 0 0 10 20 (3) ALL CHANNELS OPEN Number of open channels 0 300,000 Figure 5.15 Activation of ACh receptors at neuromuscular syn–2 0 2 4 6 8 10 12 14 apses. (A) Outside-out patch clamp measurement of single ACh Time (ms) receptor currents from a patch of membrane removed from the (C) Postsynaptic potential change (EPP) produced by EPC postsynaptic muscle cell. When ACh is applied to the extracellu−70 lar surface of the membrane clamped at negative voltages, the Membrane −80 repeated brief opening of a single channel can be seen as downpotential ward deflections corresponding to inward current (i.e., positive −90 (mV) ions flowing into the cell). (B) Synchronized opening of many −100 ACh-activated channels at a synapse being voltage-clamped at –2 0 2 4 6 8 10 12 14 negative voltages. (1) If a single channel is examined during the Time (ms) release of ACh from the presynaptic terminal, the channel opens transiently. (2) If a number of channels are examined together, ACh release opens the channels almost synchronously. (3) The opening of a very large number of postsynaptic channels produces a macroscopic EPC. (C) In a normal muscle cell (i.e., not being voltage-clamped), the inward EPC depolarizes the postsynaptic muscle cell, giving rise to an EPP. Typically, this depolarization generates an action potential (not shown). Membranre current (pA) Number of open channels Outside-out membrane patch (1) SINGLE OPEN CHANNEL ACh release by stimulating motor neuron 0 200,000 600,000 118 Chapter Five Thus, the magnitude and polarity of the postsynaptic membrane potential determines the direction and amplitude of the EPC solely by altering the driving force on ions flowing through the receptor channels opened by ACh. When Vm is at the reversal potential, Vm – Erev is equal to 0 and there is no net driving force on the ions that can permeate the receptor-activated channel. As a result, the identity of the ions that flow during the EPC can be deduced by observing how the reversal potential of the EPC compares to the equilibrium potential for various ion species (Figure 5.17). For example, if ACh were to open an ion channel permeable only to K+, then the reversal Figure 5.16 The influence of the postsynaptic membrane potential on end plate currents. (A) A postsynaptic muscle fiber is voltage clamped using two electrodes, while the presynaptic neuron is electrically stimulated to cause the release of ACh from presynaptic terminals. This experimental arrangement allows the recording of macroscopic EPCs produced by ACh. (B) Amplitude and time course of EPCs generated by stimulating the presynaptic motor neuron while the postsynaptic cell is voltage clamped at four different membrane potentials. (C) The relationship between the peak amplitude of EPCs and postsynaptic membrane potential is nearly linear, with a reversal potential (the voltage at which the direction of the current changes from inward to outward) close to 0 mV. Also indicated on this graph are the equilibrium potentials of Na+, K+, and Cl– ions. (D) Lowering the external Na+ concentration causes EPCs to reverse at more negative potentials. (E) Raising the external K+ concentration makes the reversal potential more positive. (After Takeuchi and Takeuchi, 1960.) (A) Scheme for voltage clamping postsynaptic muscle fiber Stimulate Axon of presynaptic motor neuron Voltage clamp amplifier Current-passing electrode Postsynaptic muscle fiber Voltage-measuring electrode Presynaptic terminals (B) Effect of membrane voltage on postsynaptic end plate currents −110 mV −60 mV Stimulate presynaptic axon Stimulate presynaptic axon 200 EPC (nA) 100 0 mV +70 mV 0 −100 Stimulate presynaptic axon Stimulate presynaptic axon −200 −300 0 2 6 4 0 2 (C) (D) EPC amplitude (nA) EK ECl ENa 200 4 6 0 Time (ms) 2 Lower external [Na+] shifts reversal potential to left 4 (E) 6 0 2 4 6 Higher external [K+] shifts reversal potential to right Reversal potential 100 0 −100 −200 −300 −110 −60 0 +70 −110 −60 0 +70 Postsynaptic membrane potential (mV) −110 −60 0 +70 Synaptic Transmission 119 EPC amplitude (nA) 300 200 K+ efflux 100 0 Erev = EK −100 −200 K+ influx −300 −150 −100 −50 0 50 Membrane potential 100 Only Na+ selective channel open (B) EPC amplitude (nA) 300 200 100 0 −100 Na+ efflux Erev = ENa Na+ influx −200 −300 −150 −100 −50 0 50 Membrane potential 100 Only Cl− selective channel open (C) 300 EPC amplitude (nA) potential of the EPC would be at the equilibrium potential for K+, which for a muscle cell is close to –100 mV (Figure 5.17A). If the ACh-activated channels were permeable only to Na+, then the reversal potential of the current would be approximately +70 mV, the Na+ equilibrium potential of muscle cells (Figure 5.17B); if these channels were permeable only to Cl–, then the reversal potential would be approximately –50 mV (Figure 5.17C). By this reasoning, ACh-activated channels cannot be permeable to only one of these ions, because the reversal potential of the EPC is not near the equilibrium potential for any of them (see Figure 5.16C). However, if these channels were permeable to both Na+ and K+, then the reversal potential of the EPC would be between +70 mV and –100 mV (Figure 5.17D). The fact that EPCs reverse at approximately 0 mV is therefore consistent with the idea that ACh-activated ion channels are almost equally permeable to both Na+ and K+. This was tested in 1960, by the husband and wife team of Akira and Noriko Takeuchi, by altering the extracellular concentration of these two ions. As predicted, the magnitude and reversal potential of the EPC was changed by altering the concentration gradient of each ion. Lowering the external Na+ concentration, which makes ENa more negative, produces a negative shift in Erev (Figure 5.16D), whereas elevating external K+ concentration, which makes EK more positive, causes Erev to shift to a more positive potential (Figure 5.16E). Such experiments confirm that the AChactivated ion channels are in fact permeable to both Na+ and K+. Even though the channels opened by the binding of ACh to its receptors are permeable to both Na+ and K+, at the resting membrane potential the EPC is generated primarily by Na+ influx (Figure 5.18). If the membrane potential is kept at EK, the EPC arises entirely from an influx of Na+ because at this potential there is no driving force on K+ (Figure 5.18A). At the usual muscle fiber resting membrane potential of –90 mV, there is a small driving force on K+, but a much greater one on Na+. Thus, during the EPC, much more Na+ flows into the muscle cell than K+ flows out (Figure 5.18B); it is the net influx of positively charged Na+ that constitutes the inward current measured as the EPC. At the reversal potential of about 0 mV, Na+ influx and K+ efflux are exactly balanced, so no current flows during the opening of channels by ACh binding (Figure 5.18C). At potentials more positive than Erev the balance reverses; for example, at ENa there is no influx of Na+ and a large efflux of K+ because of the large driving force on Na+ (Figure 5.18D). Even more positive potentials cause efflux of both Na+ and K+ and produce an even larger outward EPC. Were it possible to measure the EPP at the same time as the EPC (of course, the voltage clamp technique prevents this by keeping membrane potential constant), the EPP would be seen to vary in parallel with the amplitude and polarity of the EPC (Figures 5.18E,F). At the usual postsynaptic resting membrane potential of –90 mV, the large inward EPC causes the postsynaptic membrane potential to become more depolarized (see Figure Only K+ selective channel open (A) 200 100 Erev = ECl Cl− influx 0 −100 −200 −300 Cl− efflux −150 −100 −50 0 50 Membrane potential (D) 100 Cation non-selective channel open Cation efflux 300 EPC amplitude (nA) Figure 5.17 The effect of ion channel selectivity on the reversal potential. Voltage clamping a postsynaptic cell while activating presynaptic neurotransmitter release reveals the identity of the ions permeating the postsynaptic receptors being activated. (A) The activation of postsynaptic channels permeable only to K+ results in currents reversing at EK, near –100 mV. (B) The activation of postsynaptic Na+ channels results in currents reversing at ENa, near +70 mV. (C) Cl–-selective currents reverse at ECl, near –50 mV. (D) Ligand-gated channels that are about equally permeable to both K+ and Na+ show a reversal potential near 0 mV. 200 100 Erev = 0 0 −100 −200 Cation influx −300 −150 −100 −50 0 50 Membrane potential 100 Postsynaptic membrane potential Outside cell Na+ EPCs EPPs ACh −100 mV (EK) AChactivated channel Inside cell (B) Na+ −90 mV K+ (C) Na+ 0 mV (Erev) K+ (D) +70 mV (ENa) K+ (E) (F) Outward ENa EK Inward −100 −90 0 +70 Postsynaptic membrane potential EPP peak amplitude (mV) Figure 5.18 Na+ and K+ movements during EPCs and EPPs. (A–D) Each of the postsynaptic potentials (Vpost) indicated at the left results in different relative fluxes of net Na+ and K+ (ion fluxes). These ion fluxes determine the amplitude and polarity of the EPCs, which in turn determine the EPPs. Note that at about 0 mV the Na+ flux is exactly balanced by an opposite K+ flux, resulting in no net current flow, and hence no change in the membrane potential. (E) EPCs are inward currents at potentials more negative than Erev and outward currents at potentials more positive than Erev. (F) EPPs depolarize the postsynaptic cell at potentials more negative than Erev. At potentials more positive than Erev, EPPs hyperpolarize the cell. NET ION FLUXES (A) EPC peak amplitude (nA) 120 Chapter Five Depolarizing EK ENa Hyper− polarizing −100 −90 0 +70 Postsynaptic membrane potential 5.18F). However, at 0 mV, the EPP reverses its polarity, and at more positive potentials, the EPP is hyperpolarizing. Thus, the polarity and magnitude of the EPC depend on the electrochemical driving force, which in turn determines the polarity and magnitude of the EPP. EPPs will depolarize when the membrane potential is more negative than Erev, and hyperpolarize when the membrane potential is more positive than Erev. The general rule, then, is that Synaptic Transmission 121 the action of a transmitter drives the postsynaptic membrane potential toward Erev for the particular ion channels being activated. Although this discussion has focused on the neuromuscular junction, similar mechanisms generate postsynaptic responses at all chemical synapses. The general principle is that transmitter binding to postsynaptic receptors produces a postsynaptic conductance change as ion channels are opened (or sometimes closed). The postsynaptic conductance is increased if—as at the neuromuscular junction—channels are opened, and decreased if channels are closed. This conductance change typically generates an electrical current, the postsynaptic current (PSC), which in turn changes the postsynaptic membrane potential to produce a postsynaptic potential (PSP). As in the specific case of the EPP at the neuromuscular junction, PSPs are depolarizing if their reversal potential is more positive than the postsynaptic membrane potential and hyperpolarizing if their reversal potential is more negative. The conductance changes and the PSPs that typically accompany them are the ultimate outcome of most chemical synaptic transmission, concluding a sequence of electrical and chemical events that begins with the invasion of an action potential into the terminals of a presynaptic neuron. In many ways, the events that produce PSPs at synapses are similar to those that generate action potentials in axons; in both cases, conductance changes produced by ion channels lead to ionic current flow that changes the membrane potential (see Figure 5.18). Excitatory and Inhibitory Postsynaptic Potentials PSPs ultimately alter the probability that an action potential will be produced in the postsynaptic cell. At the neuromuscular junction, synaptic action increases the probability that an action potential will occur in the postsynaptic muscle cell; indeed, the large amplitude of the EPP ensures that an action potential always is triggered. At many other synapses, PSPs similarly increase the probability of firing a postsynaptic action potential. However, still other synapses actually decrease the probability that the postsynaptic cell will generate an action potential. PSPs are called excitatory (or EPSPs) if they increase the likelihood of a postsynaptic action potential occurring, and inhibitory (or IPSPs) if they decrease this likelihood. Given that most neurons receive inputs from both excitatory and inhibitory synapses, it is important to understand more precisely the mechanisms that determine whether a particular synapse excites or inhibits its postsynaptic partner. The principles of excitation just described for the neuromuscular junction are pertinent to all excitatory synapses. The principles of postsynaptic inhibition are much the same as for excitation, and are also quite general. In both cases, neurotransmitters binding to receptors open or close ion channels in the postsynaptic cell. Whether a postsynaptic response is an EPSP or an IPSP depends on the type of channel that is coupled to the receptor, and on the concentration of permeant ions inside and outside the cell. In fact, the only distinction between postsynaptic excitation and inhibition is the reversal potential of the PSP in relation to the threshold voltage for generating action potentials in the postsynaptic cell. Consider, for example, a neuronal synapse that uses glutamate as the transmitter. Many such synapses have receptors that, like the ACh receptors at neuromuscular synapses, open ion channels that are nonselectively permeable to cations (see Chapter 6). When these glutamate receptors are activated, both Na+ and K+ flow across the postsynaptic membrane, yielding an Erev of approximately 0 mV for the resulting postsynaptic current. If the rest- 122 Chapter Five ENa +50 (A) (D) (C) (B) Action potential 0 mV Erev Threshold −40 −50 Vrest −60 −70 Erev > threshold = excitatory Activate GABA synapse EPSP Erev < threshold = inhibitory IPSP IPSP Erev Activate glutamate synapse EK −110 Erev 0 1 2 Figure 5.19 Reversal potentials and threshold potentials determine postsynaptic excitation and inhibition. (A) If the reversal potential for a PSP (0 mV) is more positive than the action potential threshold (–40 mV), the effect of a transmitter is excitatory, and it generates EPSPs. (B) If the reversal potential for a PSP is more negative than the action potential threshold, the transmitter is inhibitory and generate IPSPs. (C) IPSPs can nonetheless depolarize the postsynaptic cell if their reversal potential is between the resting potential and the action potential threshold. (D) The general rule of postsynaptic action is: If the reversal potential is more positive than threshold, excitation results; inhibition occurs if the reversal potential is more negative than threshold. 3 4 0 Activate GABA synapse 1 2 3 4 0 Time (ms) 1 2 3 4 ing potential of the postsynaptic neuron is –60 mV, the resulting EPSP will depolarize by bringing the postsynaptic membrane potential toward 0 mV. For the hypothetical neuron shown in Figure 5.19A, the action potential threshold voltage is –40 mV. Thus, a glutamate-induced EPSP will increase the probability that this neuron produces an action potential, defining the synapse as excitatory. As an example of inhibitory postsynaptic action, consider a neuronal synapse that uses GABA as its transmitter. At such synapses, the GABA receptors typically open channels that are selectively permeable to Cl– and the action of GABA causes Cl– to flow across the postsynaptic membrane. Consider a case where ECl is –70 mV, as is typical for many neurons, so that the postsynaptic resting potential of –60 mV is less negative than ECl. The resulting positive electrochemical driving force (Vm – Erev) will cause negatively charged Cl– to flow into the cell and produce a hyperpolarizing IPSP (Figure 5.19B). This hyperpolarizing IPSP will take the postsynaptic membrane away from the action potential threshold of –40 mV, clearly inhibiting the postsynaptic cell. Surprisingly, inhibitory synapses need not produce hyperpolarizing IPSPs. For instance, if ECl were –50 mV instead of –70 mV, then the negative electrochemical driving force would cause Cl– to flow out of the cell and produce a depolarizing IPSP (Figure 5.19C). However, the synapse would still be inhibitory: Given that the reversal potential of the IPSP still is more negative than the action potential threshold (–40 mV), the depolarizing IPSP would inhibit because the postsynaptic membrane potential would be kept more negative than the threshold for action potential initiation. Another way to think about this peculiar situation is that if another excitatory input onto this neuron brought the cell’s membrane potential to –41 mV, just below threshold for firing an action potential, the IPSP would then hyperpolarize the membrane potential toward –50 mV, bringing the potential away from the action potential threshold. Thus, while EPSPs depolarize the postsynaptic cell, IPSPs can hyperpolarize or depolarize; indeed, an inhibitory conductance change may produce no potential change at all and still exert an inhibitory effect by making it more difficult for an EPSP to evoke an action potential in the postsynaptic cell. Although the particulars of postsynaptic action can be complex, a simple rule distinguishes postsynaptic excitation from inhibition: An EPSP has a reversal potential more positive than the action potential threshold, whereas Synaptic Transmission 123 an IPSP has a reversal potential more negative than threshold (Figure 5.19D). Intuitively, this rule can be understood by realizing that an EPSP will tend to depolarize the membrane potential so that it exceeds threshold, whereas an IPSP will always act to keep the membrane potential more negative than the threshold potential. Summation of Synaptic Potentials The PSPs produced at most synapses in the brain are much smaller than those at the neuromuscular junction; indeed, EPSPs produced by individual excitatory synapses may be only a fraction of a millivolt and are usually well below the threshold for generating postsynaptic action potentials. How, then, can such synapses transmit information if their PSPs are subthreshold? The answer is that neurons in the central nervous system are typically innervated by thousands of synapses, and the PSPs produced by each active synapse can sum together—in space and in time—to determine the behavior of the postsynaptic neuron. Consider the highly simplified case of a neuron that is innervated by two excitatory synapses, each generating a subthreshold EPSP, and an inhibitory synapse that produces an IPSP (Figure 5.20A). While activation of either one of the excitatory synapses alone (E1 or E2 in Figure 5.20B) produces a sub(A) Excitatory (E1) Record Inhibitory (I) Postsynaptic membrane potential Excitatory (E2) Cell body Dendrites Axon (B) EPSP (Synapse E1 or E2) Summed EPSPs (Synapses E1 + E2) IPSP (Synapse I) Postsynaptic membrane potential (mV) +20 0 −20 −40 −60 Threshold Vrest Time (ms) Summed Summed EPSP + IPSP EPSPs + IPSP (Synapses (Synapses E1 + I) E1 + I +E2) Figure 5.20 Summation of postsynaptic potentials. (A) A microelectrode records the postsynaptic potentials produced by the activity of two excitatory synapses (E1 and E2) and an inhibitory synapse (I). (B) Electrical responses to synaptic activation. Stimulating either excitatory synapse (E1 or E2) produces a subthreshold EPSP, whereas stimulating both synapses at the same time (E1 + E2) produces a suprathreshold EPSP that evokes a postsynaptic action potential (shown in blue). Activation of the inhibitory synapse alone (I) results in a hyperpolarizing IPSP. Summing this IPSP (dashed red line) with the EPSP (dashed yellow line) produced by one excitatory synapse (E1 + I) reduces the amplitude of the EPSP (orange line), while summing it with the suprathreshold EPSP produced by activating synapses E1 and E2 keeps the postsynaptic neuron below threshold, so that no action potential is evoked. 124 Chapter Five Neurotransmitter release Receptor binding Ion channels open or close Conductance change causes current flow Postsynaptic potential changes Postsynaptic cells excited or inhibited Summation determines whether or not an action potential occurs Figure 5.21 Events from neurotransmitter release to postsynaptic excitation or inhibition. Neurotransmitter release at all presynaptic terminals on a cell results in receptor binding, which causes the opening or closing of specific ion channels. The resulting conductance change causes current to flow, which may change the membrane potential. The postsynaptic cell sums (or integrates) all of the EPSPs and IPSPs, resulting in moment-to-moment control of action potential generation. threshold EPSP, activation of both excitatory synapses at about the same time causes the two EPSPs to sum together. If the sum of the two EPSPs (E1 + E2) depolarizes the postsynaptic neuron sufficiently to reach the threshold potential, a postsynaptic action potential results. Summation thus allows subthreshold EPSPs to influence action potential production. Likewise, an IPSP generated by an inhibitory synapse (I) can sum (algebraically speaking) with a subthreshold EPSP to reduce its amplitude (E1 + I) or can sum with suprathreshold EPSPs to prevent the postsynaptic neuron from reaching threshold (E1 + I + E2). In short, the summation of EPSPs and IPSPs by a postsynaptic neuron permits a neuron to integrate the electrical information provided by all the inhibitory and excitatory synapses acting on it at any moment. Whether the sum of active synaptic inputs results in the production of an action potential depends on the balance between excitation and inhibition. If the sum of all EPSPs and IPSPs results in a depolarization of sufficient amplitude to raise the membrane potential above threshold, then the postsynaptic cell will produce an action potential. Conversely, if inhibition prevails, then the postsynaptic cell will remain silent. Normally, the balance between EPSPs and IPSPs changes continually over time, depending on the number of excitatory and inhibitory synapses active at a given moment and the magnitude of the current at each active synapse. Summation is therefore a neurotransmitterinduced tug-of-war between all excitatory and inhibitory postsynaptic currents; the outcome of the contest determines whether or not a postsynaptic neuron fires an action potential and, thereby, becomes an active element in the neural circuits to which it belongs (Figure 5.21). Two Families of Postsynaptic Receptors The opening or closing of postsynaptic ion channels is accomplished in different ways by two broad families of receptor proteins. The receptors in one family—called ionotropic receptors—are linked directly to ion channels (the Greek tropos means to move in response to a stimulus). These receptors contain two functional domains: an extracellular site that binds neurotransmitters, and a membrane-spanning domain that forms an ion channel (Figure 5.22A). Thus ionotropic receptors combine transmitter-binding and channel functions into a single molecular entity (they are also called ligand-gated ion channels to reflect this concatenation). Such receptors are multimers made up of at least four or five individual protein subunits, each of which contributes to the pore of the ion channel. The second family of neurotransmitter receptors are the metabotropic receptors, so called because the eventual movement of ions through a channel depends on one or more metabolic steps. These receptors do not have ion channels as part of their structure; instead, they affect channels by the activation of intermediate molecules called G-proteins (Figure 5.22B). For this reason, metabotropic receptors are also called G-protein-coupled receptors. Metabotropic receptors are monomeric proteins with an extracellular domain that contains a neurotransmitter binding site and an intracellular domain that binds to G-proteins. Neurotransmitter binding to metabotropic receptors activates G-proteins, which then dissociate from the receptor and interact directly with ion channels or bind to other effector proteins, such as enzymes, that make intracellular messengers that open or close ion channels. Thus, G-proteins can be thought of as transducers that couple neurotransmitter binding to the regulation of postsynaptic ion channels. The postsynaptic signaling events initiated by metabotropic receptors are taken up in detail in Chapter 7. Synaptic Transmission 125 (A) Ligand-gated ion channels Ions (B) G-protein-coupled receptors 1 Neurotransmitter binds 1 Neurotransmitter binds Neurotransmitter 2 Channel opens Receptor Neurotransmitter Effector protein Outside cell 5 Ions flow across membrane Inside cell α β γ α G-protein 3 Ions flow across membrane 2 G-protein is activated I n tr a c e ll u l a m e ss e n g e r s 3 G-protein subunits or intracellular messengers modulate ion channels Figure 5.22 A neurotransmitter can affect the activity of a postsynaptic cell via two different types of receptor proteins: ionotropic or ligand-gated ion channels, and metabotropic or G-protein-coupled receptors. (A) Ligand-gated ion channels combine receptor and channel functions in a single protein complex. (B) Metabotropic receptors usually activate G-proteins, which modulate ion channels directly or indirectly through intracellular effector enzymes and second messengers. These two families of postsynaptic receptors give rise to PSPs with very different time courses, producing postsynaptic actions that range from less than a millisecond to minutes, hours, or even days. Ionotropic receptors generally mediate rapid postsynaptic effects. Examples are the EPP produced at neuromuscular synapses by ACh (see Figure 5.15), EPSPs produced at certain glutamatergic synapses (Figure 5.19A), and IPSPs produced at certain GABAergic synapses (Figure 5.19B). In all three cases, the PSPs arise within a millisecond or two of an action potential invading the presynaptic terminal and last for only a few tens of milliseconds or less. In contrast, the activation of metabotropic receptors typically produces much slower responses, ranging from hundreds of milliseconds to minutes or even longer. The comparative slowness of metabotropic receptor actions reflects the fact that multiple proteins need to bind to each other sequentially in order to produce the final physiological response. Importantly, a given transmitter may activate both ionotropic and metabotropic receptors to produce both fast and slow PSPs at the same synapse. Perhaps the most important principle to keep in mind is that the response elicited at a given synapse depends upon the neurotransmitter released and the postsynaptic complement of receptors and associated channels. The molecular mechanisms that allow neurotransmitters and their receptors to generate synaptic responses are considered in the next chapter. r 4 Ion channel opens Ions 126 Chapter Five Summary Synapses communicate the information carried by action potentials from one neuron to the next in neural circuits. The cellular mechanisms that underlie synaptic transmission are closely related to the mechanisms that generate other types of neuronal electrical signals, namely ion flow through membrane channels. In the case of electrical synapses, these channels are gap junctions; direct but passive flow of current through the gap junctions is the basis for transmission. In the case of chemical synapses, channels with smaller and more selective pores are activated by the binding of neurotransmitters to postsynaptic receptors after release from the presynaptic terminal. The large number of neurotransmitters in the nervous system can be divided into two broad classes: small-molecule transmitters and neuropeptides. Neurotransmitters are synthesized from defined precursors by regulated enzymatic pathways, packaged into one of several types of synaptic vesicle, and released into the synaptic cleft in a Ca2+-dependent manner. Many synapses release more than one type of neurotransmitter, and multiple transmitters can even be packaged within the same synaptic vesicle. Transmitter agents are released presynaptically in units or quanta, reflecting their storage within synaptic vesicles. Vesicles discharge their contents into the synaptic cleft when the presynaptic depolarization generated by the invasion of an action potential opens voltage-gated calcium channels, allowing Ca2+ to enter the presynaptic terminal. How calcium triggers neurotransmitter release is not yet established, but synaptotagmin, SNAREs, and a number of other proteins found within the presynaptic terminal are clearly involved. Postsynaptic receptors are a diverse group of proteins that transduce binding of neurotransmitters into electrical signals by opening or closing postsynaptic ion channels. The postsynaptic currents produced by the synchronous opening or closing of ion channels changes the conductance of the postsynaptic cell, thus increasing or decreasing its excitability. Conductance changes that increase the probability of firing an action potential are excitatory, whereas those that decrease the probability of generating an action potential are inhibitory. Because postsynaptic neurons are usually innervated by many different inputs, the integrated effect of the conductance changes underlying all EPSPs and IPSPs produced in a postsynaptic cell at any moment determines whether or not the cell fires an action potential. Two broadly different families of neurotransmitter receptors have evolved to carry out the postsynaptic signaling actions of neurotransmitters. The postsynaptic effects of neurotransmitters are terminated by the degradation of the transmitter in the synaptic cleft, by transport of the transmitter back into cells, or by diffusion out of the synaptic cleft. Synaptic Transmission 127 Additional Reading Reviews AUGUSTINE, G. J. (2001) How does calcium trigger neurotransmitter release? Curr. Opin. Neurobiol. 11: 320–326. BENNETT, M. V. L. (2000) Electrical synapses, a personal perspective (or history). Brain Res. Rev. 32: 16–28. BRODSKY, F. M., C. Y. CHEN, C. KNUEHL, M. C. TOWLER AND D. E. WAKEHAM (2001) Biological basket weaving: Formation and function of clathrin-coated vesicles. Annu. Rev. Cell. Dev. Biol. 17: 517–568. BRUNGER, A. T. (2001) Structure of proteins involved in synaptic vesicle fusion in neurons. Annu. Rev. Biophys. Biomol. Struct. 30: 157–171. CARLSSON, A. (1987) Perspectives on the discovery of central monoaminergic neurotransmission. Annu. Rev. Neurosci. 10: 19–40. CHANGEUX, J.-P. (1993) Chemical signaling in the brain. Sci. Am. 269 (May): 58–62. EMSON, P. C. (1979) Peptides as neurotransmitter candidates in the CNS. Prog. Neurobiol. 13: 61–116. GALARRETA, M. AND S. HESTRIN (2001) Electrical synapses between GABA-releasing interneurons. Nature Rev. Neurosci. 2: 425–433. JAHN, R., T. LANG AND T. C. SÜDHOF (2003) Membrane fusion. Cell 112: 519–533. KUPFERMANN, I. (1991) Functional studies of cotransmission. Physiol. Rev. 71: 683–732. MARSH, M. AND H. T. MCMAHON (1999) The structural era of endocytosis. Science 285: 215–220. MURTHY, V. N. AND P. DE CAMILLI (2003) Cell biology of the presynaptic terminal. Annu. Rev. Neurosci. 26: 701–728. ROTHMAN, J. E. (1994) Mechanisms of intracellular protein transport. Nature 372: 55–63. SÜDHOF, T. C. (1995) The synaptic vesicle cycle: A cascade of protein-protein interactions. Nature 375: 645–653. TUCKER, W. C. AND E. R. CHAPMAN (2002) Role of synaptotagmin in Ca2+ triggered exocytosis. Biochem. J. 366: 1–13. Important Original Papers ADLER, E., M. ADLER, G. J. AUGUSTINE, M. P. CHARLTON AND S. N. DUFFY (1991) Alien intracellular calcium chelators attenuate neurotransmitter release at the squid giant synapse. J. Neurosci. 11: 1496–1507. AUGUSTINE, G. J. AND R. ECKERT (1984) Divalent cations differentially support transmitter release at the squid giant synapse. J. Physiol. (Lond.) 346: 257–271. BOYD, I. A. AND A. R. MARTIN (1955) The endplate potential in mammalian muscle. J. Physiol. (Lond.) 132: 74–91. CURTIS, D. R., J. W. PHILLIS AND J. C. WATKINS (1959) Chemical excitation of spinal neurons. Nature 183: 611–612. DALE, H. H., W. FELDBERG AND M. VOGT (1936) Release of acetylcholine at voluntary motor nerve endings. J. Physiol. 86: 353–380. DEL CASTILLO, J. AND B. KATZ (1954) Quantal components of the end plate potential. J. Physiol. (Lond.) 124: 560–573. FATT, P. AND B. KATZ (1951) An analysis of the end plate potential recorded with an intracellular electrode. J. Physiol. (Lond.) 115: 320–370. FATT, P. AND B. KATZ (1952) Spontaneous subthreshold activity at motor nerve endings. J. Physiol. (Lond.) 117: 109–128. FURSHPAN, E. J. AND D. D. POTTER (1959) Transmission at the giant motor synapses of the crayfish. J. Physiol. (Lond.) 145: 289–325. GEPPERT, M. AND 6 OTHERS (1994) Synaptotagmin I: A major Ca2+ sensor for transmitter release at a central synapse. Cell 79: 717–727. GIBSON, J. R., M. BEIERLEIN AND B. W. CONNORS. (1999) Two networks of electrically coupled inhibtory neurons in neocortex. Nature 402: 75–79. HARRIS, B. A., J. D. ROBISHAW, S. M. MUMBY AND A. G. GILMAN (1985) Molecular cloning of complementary DNA for the alpha subunit of the G protein that stimulates adenylate cyclase. Science 229: 1274–1277. HEUSER, J. E. AND 5 OTHERS (1979) Synaptic vesicle exocytosis captured by quick freezing and correlated with quantal transmitter release. J. Cell Biol. 81: 275–300. HEUSER, J. E. AND T. S. REESE (1973) Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction. J. Cell Biol. 57: 315–344. HÖKFELT, T., O. JOHANSSON, A. LJUNGDAHL, J. M. LUNDBERG AND M. SCHULTZBERG (1980) Peptidergic neurons. Nature 284: 515–521. JONAS, P., J. BISCHOFBERGER AND J. SANDKUHLER (1998) Corelease of two fast neurotransmitters at a central synapse. Science 281: 419–424. LOEWI, O. (1921) Über humorale übertragbarheit der herznervenwirkung. Pflügers Arch. 189: 239–242. MILEDI, R. (1973) Transmitter release induced by injection of calcium ions into nerve terminals. Proc. R. Soc. Lond. B 183: 421–425. NEHER, E. AND B. SAKMANN (1976) Singlechannel currents recorded from membrane of denervated frog muscle fibres. Nature 260:799-802. REKLING, J. C., X. M. SHAO AND J. L. FELDMAN (2000) Electrical coupling and excitatory synaptic transmission between rhythmogenic respiratory neurons in the preBotzinger complex. J. Neurosci. 20: RC113: 1–5. SMITH, S. J., J. BUCHANAN, L. R. OSSES, M. P. CHARLTON AND G. J. AUGUSTINE (1993) The spatial distribution of calcium signals in squid presynaptic terminals. J. Physiol. (Lond.) 472: 573–593. SOSSIN, W. S., A. SWEET-CORDERO AND R. H. SCHELLER (1990) Dale’s hypothesis revisited: Different neuropeptides derived from a common prohormone are targeted to different processes. Proc. Natl. Acad. Sci. U.S.A. 87: 4845–4548. SUTTON, R. B., D. FASSHAUER, R. JAHN AND A. T. BRÜNGER (1998) Crystal structure of a SNARE complex involved in synaptic exocytosis at 2.4 Å resolution. Nature 395: 347–353. TAKEUCHI, A. AND N. TAKEUCHI (1960) One the permeability of end-plate membrane during the action of transmitter. J. Physiol. (Lond.) 154: 52–67. WICKMAN, K. AND 7 OTHERS (1994) Recombinant Gβγ activates the muscarinic-gated atrial potassium channel IKACh. Nature 368: 255– 257. Books BRADFORD, H. F. (1986) Chemical Neurobiology. New York: W. H. Freeman. COOPER, J. R., F. E. BLOOM AND R. H. ROTH (1991) The Biochemical Basis of Neuropharmacology. New York: Oxford University Press. HALL, Z. (1992) An Introduction to Molecular Neurobiology. Sunderland, MA: Sinauer Associates. KATZ, B. (1966) Nerve, Muscle, and Synapse. New York: McGraw-Hill. KATZ, B. (1969) The Release of Neural Transmitter Substances. Liverpool: Liverpool University Press. LLINÁS, R. R. (1999) The Squid Giant Synapse: A Model for Chemical Synaptic Transmission. Oxford: Oxford University Press. NICHOLLS, D. G. (1994) Proteins, Transmitters, and Synapses. Oxford: Blackwell. PETERS, A., S. L. PALAY AND H. DEF. WEBSTER (1991) The Fine Structure of the Nervous System: Neurons and their Supporting Cells. 3rd edition. Oxford: Oxford University Press. Chapter 6 Neurotransmitters and Their Receptors Overview For the most part, neurons in the human brain communicate with one another by releasing chemical messengers called neurotransmitters. A large number of neurotransmitters are now known and more remain to be discovered. Neurotransmitters evoke postsynaptic electrical responses by binding to members of a diverse group of proteins called neurotransmitter receptors. There are two major classes of receptors: those in which the receptor molecule is also an ion channel, and those in which the receptor and ion channel are separate molecules. The former are called ionotropic receptors or ligandgated ion channels, and give rise to fast postsynaptic responses that typically last only a few milliseconds. The latter are called metabotropic receptors, and they produce slower postsynaptic effects that may endure much longer. Abnormalities in the function of neurotransmitter systems contribute to a wide range of neurological and psychiatric disorders. As a result, many neuropharmacological therapies are based on drugs that affect neurotransmitter release, binding, and/or removal. Categories of Neurotransmitters More than 100 different agents are known to serve as neurotransmitters. This large number of transmitters allows for tremendous diversity in chemical signaling between neurons. It is useful to separate this panoply of transmitters into two broad categories based simply on size (Figure 6.1). Neuropeptides are relatively large transmitter molecules composed of 3 to 36 amino acids. Individual amino acids, such as glutamate and GABA, as well as the transmitters acetylcholine, serotonin, and histamine, are much smaller than neuropeptides and have therefore come to be called small-molecule neurotransmitters. Within the category of small-molecule neurotransmitters, the biogenic amines (dopamine, norepinephrine, epinephrine, serotonin, and histamine) are often discussed separately because of their similar chemical properties and postsynaptic actions. The particulars of synthesis, packaging, release, and removal differ for each neurotransmitter (Table 6.1). This chapter will describe some of the main features of these transmitters and their postsynaptic receptors. Acetylcholine As mentioned in the previous chapter, acetylcholine (ACh) was the first substance identified as a neurotransmitter. In addition to the action of ACh as the neurotransmitter at skeletal neuromuscular junctions (see Chapter 5), as well as the neuromuscular synapse between the vagus nerve and cardiac 129 130 Chapter Six SMALL-MOLECULE NEUROTRANSMITTERS O + Acetylcholine (CH3)3N AMINO ACIDS H3N CH2 O C CH3 CATECHOLAMINES COO− + CH2 NH3 CH2 NH3 CH2 NH2 CH2 Dopamine H C + Glutamate CH2 BIOGENIC AMINES HO CH2 OH CH2 COOH H C + H3N Aspartate OH CH2 Norepinephrine COO− + HO CH2 OH COOH OH + H3N GABA CH2 CH2 COO− CH2 CH2 Epinephrine CH3 HO OH H C + H3N Glycine COO− INDOLEAMINE H HO Serotonin (5-HT) PURINES O− CH2 CH2 NH3 N NH2 O P CH2 N ATP O− + O O P O− N O O P O− IMIDAZOLEAMINE O CH2 O N N CH2 Histamine HN H H + N OH OH PEPTIDE NEUROTRANSMITTERS (more than 100 peptides, usually 3−30 amino acids long) Example: Methionine enkephalin (Tyr–Gly–Gly–Phe–Met) H3N H C CH2 C O O O + H N H C H C H N H C H C O H N H C CH2 C O H N H C C CH2 CH2 S CH3 OH Tyr Gly Gly Phe Met O− + NH3 ▲ Neurotransmitters and Their Receptors 131 Figure 6.1 Examples of small-molecule and peptide neurotransmitters. Small-molecule transmitters can be subdivided into acetylcholine, the amino acids, purines, and biogenic amines. The catcholamines, so named because they all share the catechol moiety (i.e., a hydroxylated benzene ring), make up a distinctive subgroup within the biogenic amines. Serotonin and histamine contain an indole ring and an imidazole ring, respectively. Size differences between the small-molecule neurotransmitters and the peptide neurotransmitters are indicated by the space-filling models for glycine, norepinephrine, and methionine enkephalin. (Carbon atoms are black, nitrogen atoms blue, and oxygen atoms red.) muscle fibers, ACh serves as a transmitter at synapses in the ganglia of the visceral motor system, and at a variety of sites within the central nervous system. Whereas a great deal is known about the function of cholinergic transmission at neuromuscular junctions and ganglionic synapses, the actions of ACh in the central nervous system are not as well understood. Acetylcholine is synthesized in nerve terminals from the precursors acetyl coenzyme A (acetyl CoA, which is synthesized from glucose) and choline, in a reaction catalyzed by choline acetyltransferase (CAT; Figure 6.2). Choline is present in plasma at a high concentration (about 10 mM) and is taken up into cholinergic neurons by a high-affinity Na+/choline transporter. After synthesis in the cytoplasm of the neuron, a vesicular ACh TABLE 6.1 Functional Features of the Major Neurotransmitters Neurotransmitter Postsynaptic effecta ACh Excitatory Glutamate GABA Glycine Catecholamines (epinephrine, norepinephrine, dopamine) Excitatory Inhibitory Inhibitory Excitatory Choline + acetyl CoA Glutamine Glutamate Serine Tyrosine Serotonin (5-HT) Excitatory Tryptophan Histamine Excitatory Histidine ATP Excitatory ADP Neuropeptides Excitatory and inhibitory Inhibits inhibition Excitatory and inhibitory Amino acids (protein synthesis) Membrane lipids Endocannabinoids Nitric oxide Precursor(s) Arginine Rate-limiting step in synthesis Removal mechanism Type of vesicle CAT AChEase Small, clear Glutaminase GAD Phosphoserine Tyrosine hydroxylase Transporters Transporters Transporters Transporters, MAO, COMT Tryptophan hydroxylase Histidine decarboxylase Mitochondrial oxidative phosphorylation; glycolysis Synthesis and transport Enzymatic modification of lipids Nitric oxide synthase Transporters, MAO Transporters Small, clear Small, clear Small, clear Small densecore, or large irregular dense-core Large, dense-core Large, dense-core Small, clear Hydrolysis to AMP and adenosine Proteases Hydrolasis by FAAH Spontaneous oxidation Large, dense-core None None The most common postsynaptic effect is indicated; the same transmitter can elicit postsynaptic excitation or inhibition depending on the nature of the ion channels affected by transmitter binding (see Chapter 7). a 132 Chapter Six Figure 6.2 Acetylcholine metabolism in cholinergic nerve terminals. The synthesis of acetylcholine from choline and acetyl CoA requires choline acetyltransferase. Acetyl CoA is derived from pyruvate generated by glycolysis, while choline is transported into the terminals via a Na+-dependent transporter. Acetylcholine is loaded into synaptic vesicles via a vesicular transporter. After release, acetylcholine is rapidly metabolized by acetylcholinesterase, and choline is transported back into the terminal. Glucose Pyruvate Na+/choline transporter Presynaptic terminal Acetyl CoA Choline Choline O CoA S CH3 + HO C Acetylcholine CH2 CH2 + N (CH3)3 Choline acetyltransferase O CH3 C O CH2 CH2 + N (CH3)3 Vesicular ACh transporter Acetate Acetylcholine Postsynaptic cell Acetylcholinesterase CH3 C OO− Choline + HO CH2 CH2 + N (CH3)3 Acetylcholine receptors transporter loads approximately 10,000 molecules of ACh into each cholinergic vesicle. In contrast to most other small-molecule neurotransmitters, the postsynaptic actions of ACh at many cholinergic synapses (the neuromuscular junction in particular) is not terminated by reuptake but by a powerful hydrolytic enzyme, acetylcholinesterase (AChE). This enzyme is concentrated in the synaptic cleft, ensuring a rapid decrease in ACh concentration after its release from the presynaptic terminal. AChE has a very high catalytic activity (about 5000 molecules of ACh per AChE molecule per second) and hydrolyzes ACh into acetate and choline. The choline produced by ACh hydrolysis is transported back into nerve terminals and used to resynthesize ACh. Among the many interesting drugs that interact with cholinergic enzymes are the organophosphates. This group includes some potent chemical warfare agents. One such compound is the nerve gas “Sarin,” which was made notorious after a group of terrorists released this gas in Tokyo’s underground rail system. Organophosphates can be lethal because they inhibit AChE, causing ACh to accumulate at cholinergic synapses. This build-up of ACh depolarizes the postsynaptic cell and renders it refractory to subsequent ACh release, causing neuromuscular paralysis and other effects. The high sensitivity of insects to these AChE inhibitors has made organophosphates popular insecticides. Many of the postsynaptic actions of ACh are mediated by the nicotinic ACh receptor (nAChR), so named because the CNS stimulant, nicotine, also Neurotransmitters and Their Receptors 133 (A) (B) (C) N ACh γ δ ACh β α α C δ γ α α 6.5 nm Outside cell 3 nm Inside cell 2 nm 3 nm Figure 6.3 The structure of the nACh receptor/channel. (A) Each receptor subunit crosses the membrane four times. The membrane-spanning domain that lines the pore is shown in blue. (B) Five such subunits come together to form a complex structure containing 20 transmembrane domains that surround a central pore. (C) The openings at either end of the channel are very large—approximately 3 nm in diameter; even the narrowest region of the pore is approximately 0.6 nm in diameter. By comparison, the diameter of Na+ or K+ is less than 0.3 nm. (D) An electron micrograph of the nACh receptor, showing the actual position and size of the protein with respect to the membrane. (D from Toyoshima and Unwin, 1990.) (D) Receptor Membrane binds to these receptors. Nicotine consumption produces some degree of euphoria, relaxation, and eventually addiction (Box A), effects believed to be mediated in this case by nAChRs. Nicotinic receptors are the beststudied type of ionotropic neurotransmitter receptor. As described in Chapter 5, nAChRs are nonselective cation channels that generate excitatory postsynaptic responses. A number of biological toxins specifically bind to and block nicotinic receptors (Box B). The availability of these highly specific ligands—particularly a component of snake venom called α-bungarotoxin—has provided a valuable way to isolate and purify nAChRs. This pioneering work paved the way to cloning and sequencing the genes encoding the various subunits of the nAChR. Based on these molecular studies, the nAChR is now known to be a large protein complex consisting of five subunits arranged around a central membrane-spanning pore (Figure 6.3). In the case of skeletal muscle AChRs, the receptor pentamer contains two α subunits, each of which binds one molecule of ACh. Because both ACh binding sites must be occupied for the channel to open, only relatively high concentrations of this neurotransmitter lead to channel activation. These subunits also bind other ligands, such as nicotine and α-bungarotoxin. At the neuromuscular junction, the two α subunits are combined with up to four other types of subunit—β, γ, δ, ε—in the ratio 2α:β:ε:δ. Neuronal nAChRs typically differ from those of muscle in that they lack sensitivity to α-bungaro- 134 Chapter Six Box A Addiction Drug addiction is a chronic, relapsing disease with obvious medical, social, and political consequences. Addiction (also called substance dependence) is a persistent disorder of brain function in which compulsive drug use occurs despite serious negative consequences for the afflicted individual. The diagnostic manual of the American Psychiatric Association defines addiction in terms of both physical dependence and psychological dependence (in which an individual continues the drug-taking behavior despite obviously maladaptive consequences). The range of substances that can generate this sort of dependence is wide; the primary agents of abuse at present are opioids, cocaine, amphetamines, marijuana, alcohol, and nicotine. Addiction to more “socially acceptable” agents such as alcohol and nicotine are sometimes regarded as less problematic, but in fact involve medical and behavioral consequences that are at least as great as for drugs of abuse that are considered more dangerous. Importantly, the phenomenon of addiction is not limited to human behavior, but is demonstrable in laboratory animals. Most of these same agents are selfadministered if primates, rodents, or other species are provided with the opportunity to do so. In addition to a compulsion to take the agent of abuse, a major feature of addiction for many drugs is a constellation of negative physiological and emotional features, loosely referred to as “withdrawal syndrome,” that occur when the drug is not taken. The signs and symptoms of withdrawal are different for each agent of abuse, but in general are characterized by effects opposite those of the positive experience induced by the drug itself. Consider, as an example, cocaine, a drug that was estimated to be in regular use by 5 to 6 million Americans during the decade of the 1990s, with about 600,000 regular users either addicted or at high risk for addiction. The positive effects of the drug smoked or inhaled as a powder in the form of the alkaloidal free base is a “high” that is nearly immediate but generally lasts only a few minutes, typi- cally leading to a desire for additional drug in as little as 10 minutes to half an hour. The “high” is described as a feeling of well-being, self-confidence, and satisfaction. Conversely, when the drug is not available, frequent users experience depression, sleepiness, fatigue, drug-craving, and a general sense of malaise. Another aspect of addiction to cocaine or other agents is tolerance, defined as a reduction in the response to the drug upon repeated administration. Tolerance occurs as a consequence of persistent use of a number of drugs but is particularly significant in drug addiction, since it progressively increases the dose needed to experience the desired effects. Although it is fair to say that the neurobiology of addiction is incompletely understood, for cocaine and many other agents of abuse the addictive effects involve activation of dopamine receptors in critical brain regions involved in motivation and emotional reinforcement (see Chapter 28). The most important of these areas is the midbrain dopamine system, toxin, and comprise only two receptor subunit types (α and β), which are present in a ratio of 3α:2β. In all cases, however, five individual subunits assemble to form a functional, cation-selective nACh receptor. Each subunit of the nAChR molecule contains four transmembrane domains that make up the ion channel portion of the receptor, and a long extracellular region that makes up the ACh-binding domain (Figure 6.3A). Unraveling the molecular structure of this region of the nACh receptor has provided insight into the mechanisms that allow ligand-gated ion channels to respond rapidly to neurotransmitters: The intimate association of the ACh binding sites with the pore of the channel presumably accounts for the rapid response to ACh (Figure 6.3B–D). Indeed, this general arrangement is characteristic of all of the ligand-gated ion channels at fast-acting synapses, as summarized in Figure 6.4. Thus, the nicotinic receptor has served as a paradigm for studies of other ligand-gated ion channels, at the same time leading to a much deeper appreciation of several neuromuscular diseases (Box C). Neurotransmitters and Their Receptors 135 especially its projections from the ventral-tegmental area to the nucleus acumbens. Agents such as cocaine appear to act by raising dopamine levels in these areas, making this transmitter more available to receptors by interfering with re-uptake of synaptically released dopamine by the dopamine transporter. The reinforcement and motivation of drug-taking behaviors is thought to be related to the projections to the nucleus acumbens. The most common opioid drug of abuse is heroin. Heroin is a derivative of the opium poppy and is not legally available for clinical purposes in the United States. The number of heroin addicts in the United States is estimated to be between 750,000 and a million individuals. The positive feelings produced by heroin, generally described as the “rush,” are often compared to the feeling of sexual orgasm and begin in less than a minute after intravenous injection. There is then a feeling of general well-being (referred to as “on the nod”) that lasts about an hour. The symptoms of withdrawal can be intense; these are restlessness, irritability, nausea, muscle pain, depression, sleeplessness, and a sense of anxiety and malaise. The reinforcing aspects of the drug entail the same dopaminergic circuitry in the ventral tegmental area and nucleus acumbens as does cocaine, although additional areas are certainly involved, particularly the sites of opioid receptors described in Chapter 9. Interestingly, addiction to heroin or any other agent is not an inevitable consequence of drug use, but depends critically on the environment. For instance, returning veterans who were heroin addicts in Vietnam typically lost their addiction upon returning to the United States. Likewise, patients given other opioids (e.g., morphine) for painful conditions rarely become addicts. The treatment of any form of addiction is difficult and must be tailored to the circumstances of the individual. In addition to treating acute problems of withdrawal and “detoxification,” patterns of behavior must be changed that may take months or years. Addiction is thus a chronic disease state that requires A second class of ACh receptors is activated by muscarine, a poisonous alkaloid found in some mushrooms (see Box B), and thus they are referred to as muscarinic ACh receptors (mAChRs). mAChRs are metabotropic and mediate most of the effects of ACh in brain. Several subtypes of mAChR are known (Figure 6.5). Muscarinic ACh receptors are highly expressed in the striatum and various other forebrain regions, where they exert an inhibitory influence on dopamine-mediated motor effects. These receptors are also found in the ganglia of the peripheral nervous system. Finally, they mediate peripheral cholinergic responses of autonomic effector organs—such as heart, smooth muscle, and exocrine glands—and are responsible for the inhibition of heart rate by the vagus nerve. Numerous drugs act as mACh receptor agonists or antagonists, but most of these do not discriminate between different types of muscarinic receptors and often produce side effects. Nevertheless, mACh blockers that are therapeutically useful include atropine (used to dilate the pupil), scopolamine (effective in preventing motion sickness), and ipratropium (useful in the treatment of asthma). continual monitoring during the lifetime of susceptible individuals. References AMERICAN PSYCHIATRIC ASSOCIATION (1994) Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM IV). Washington, D.C. HYMAN, S. E. AND R. C. MALENKA (2001) Addiction and the brain: The neurobiology of compulsion and its persistence. Nature Rev. Neurosci. 2: 695–703. LAAKSO, A., A. R. MOHN, R. R. GAINETDINOV AND M. G. CARON (2002) Experimental genetic approaches to addiction. Neuron 36: 213–228. O’BRIEN, C. P. (2001) Goodman and Gilman’s The Pharmaceutical Basis of Therapeutics, 10th Edition. New York: McGraw-Hill, Chapter 24, pp. 621–642.. 136 Chapter Six Box B Neurotoxins that Act on Postsynaptic Receptors Poisonous plants and venomous animals are widespread in nature. The toxins they produce have been used for a variety of purposes, including hunting, healing, mind-altering, and, more recently, research. Many of these toxins have potent actions on the nervous system, often interfering with synaptic transmission by targeting neurotransmitter receptors. The poisons found in some organisms contain a single type of toxin, whereas others contain a mixture of tens or even hundreds of toxins. Given the central role of ACh receptors in mediating muscle contraction at neuromuscular junctions in numerous species, it is not surprising that a large number of natural toxins interfere with transmission at this synapse. In fact, the classification of nicotinic and muscarinic ACh receptors is based on the sensitivity of these receptors to the toxic plant alkaloids nicotine and muscarine, which activate nicotinic and muscarinic ACh receptors, respectively. Nicotine is derived from the dried leaves of the tobacco plant Nicotinia tabacum, and muscarine is from the poisonous red mushroom Amanita muscaria. Both toxins are stimulants that produce nausea, vomiting, mental confusion, and convulsions. Muscarine poisoning can also lead to circulatory collapse, coma, and death. The poison α-bungarotoxin, one of many peptides that together make up the venom of the banded krait, Bungarus multicinctus (Figure A), blocks transmission at neuromuscular junctions and is used by the snake to paralyze its prey. This 74-amino-acid toxin blocks neuromuscular transmission by irreversibly binding to nicotinic ACh receptors, thus preventing ACh from opening postsynaptic ion channels. Paralysis ensues because skeletal muscles can no longer be activated by motor neurons. As a result of its specificity and its high affinity for nicotinic ACh receptors, α-bungarotoxin has contributed greatly to understanding the ACh receptor mole- cule. Other snake toxins that block nicotinic ACh receptors are cobra α-neurotoxin and the sea snake peptide erabutoxin. The same strategy used by these snakes to paralyze prey was adopted by South American Indians who used curare, a mixture of plant toxins from Chondodendron tomentosum, as an arrowhead poison to immobilize their quarry. Curare also blocks nicotinic ACh receptors; the active agent is the alkaloid δtubocurarine. Another interesting class of animal toxins that selectively block nicotinic ACh and other receptors includes the peptides produced by fish-hunting marine cone snails (Figure B). These colorful snails kill small fish by “shooting” venomous darts into them. The venom contains hundreds of peptides, known as the conotoxins, many of which target proteins that are important in synaptic transmission. There are conotoxin peptides that block Ca2+ channels, Na+ channels, glutamate receptors, and ACh (A) (B) (C) (A) The banded krait Bungarus multicinctus. (B) A marine cone snail (Conus sp.) uses venomous darts to kill a small fish. (C) Betel nuts, Areca catechu, growing in Malaysia. (A, Robert Zappalorti/Photo Researchers, Inc.; B, Zoya Maslak and Baldomera Olivera, University of Utah; C, Fletcher Baylis/Photo Researchers, Inc.) Neurotransmitters and Their Receptors 137 receptors. The array of physiological responses produced by these peptides all serve to immobilize any prey unfortunate enough to encounter the cone snail. Many other organisms, including other mollusks, corals, worms and frogs, also utilize toxins containing specific blockers of ACh receptors. Other natural toxins have mind- or behavior-altering effects and in some cases have been used for thousands of years by shamans and, more recently, physicians. Two examples are plant alkaloid toxins that block muscarinic ACh receptors: atropine from deadly nightshade (belladonna), and scopolamine from henbane. Because these plants grow wild in many parts of the world, exposure is not unusual, and poisoning by either toxin can also be fatal. Another postsynaptic neurotoxin that, like nicotine, is used as a social drug is found in the seeds from the betel nut, Areca catechu (Figure C). Betel nut chewing, although unknown in the United States, is practiced by up to 25% of the population in India, Bangladesh, Ceylon, Malaysia, and the Philippines. Chewing these nuts produces a euphoria caused by arecoline, an alkaloid agonist of nicotinic ACh receptors. Like nicotine, arecoline is an addictive central nervous system stimulant. Many other neurotoxins alter transmission at noncholinergic synapses. For example, amino acids found in certain mushrooms, algae, and seeds are potent glutamate receptor agonists. The excitotoxic amino acids kainate, from the red alga Digenea simplex, and quisqualate, from the seed of Quisqualis indica, are used to distinguish two families of nonNMDA glutamate receptors (see text). Other neurotoxic amino acid activators of glutamate receptors include ibotenic acid and acromelic acid, both found in mushrooms, and domoate, which occurs in algae, seaweed, and mussels. Another large group of peptide neurotoxins blocks glutamate receptors. These include the α-agatoxins from the funnel web spider, NSTX-3 from the orb weaver spider, jorotoxin from the Joro spider, and β-philanthotoxin from wasp venom, as well as many cone snail toxins. All the toxins discussed so far target excitatory synapses. The inhibitory GABA and glycine receptors, however, have not been overlooked by the exigencies of survival. Strychnine, an alkaloid extracted from the seeds of Strychnos nux-vomica, is the only drug known to have specific actions on transmission at glycinergic synapses. Because the toxin blocks glycine receptors, strychnine poisoning causes overactivity in the spinal cord and brainstem, leading to seizures. Strychnine has long been used commercially as a poison for rodents, although alternatives such as the anticoagulant coumadin are now more popular because they are safer for humans. Neu- Glutamate Glutamate is the most important transmitter in normal brain function. Nearly all excitatory neurons in the central nervous system are glutamatergic, and it is estimated that over half of all brain synapses release this agent. Glutamate plays an especially important role in clinical neurology because elevated concentrations of extracellular glutamate, released as a result of neural injury, are toxic to neurons (Box D). Glutamate is a nonessential amino acid that does not cross the blood-brain barrier and therefore must be synthesized in neurons from local precursors. The most prevalent precursor for glutamate synthesis is glutamine, which is released by glial cells. Once released, glutamine is taken up into presynaptic rotoxins that block GABAA receptors include plant alkaloids such as bicuculline from Dutchman’s breeches and picrotoxin from Anamerta cocculus. Dieldrin, a commercial insecticide, also blocks these receptors. These agents are, like strychnine, powerful central nervous system stimulants. Muscimol, a mushroom toxin that is a powerful depressant as well as a hallucinogen, activates GABAA receptors. A synthetic analogue of GABA, baclofen, is a GABAB agonist that reduces EPSPs in some brainstem neurons and is used clinically to reduce the frequency and severity of muscle spasms. Chemical warfare between species has thus given rise to a staggering array of molecules that target synapses throughout the nervous system. Although these toxins are designed to defeat normal synaptic transmission, they have also provided a set of powerful tools to understand postsynaptic mechanisms. References ADAMS, M. E. AND B. M. OLIVERA (1994) Neurotoxins: Overview of an emerging research technology. TINS 17: 151–155. HUCHO, F. AND Y. OVCHINNIKOV (1990) Toxins as Tools in Neurochemistry. Berlin: Walter de Gruyer. MYERS, R. A., L. J. CRUZ, J. E. RIVIER AND B. M. OLIVERA (1993) Conus peptides as chemical probes for receptors and ion channels. Chem. Rev. 93: 1923–1926. 138 Chapter Six (A) (B) Assembled subunits Three transmembrane helices plus pore loop Four transmembrane helices N Transmitter N C Transmitter binding site Pore loop Outside Inside C (C) Receptor AMPA NMDA Kainate GABA Glycine nACh Serotonin Subunits (combination of 4 or 5 required for each receptor type) Glu R1 NR1 Glu R5 α1−7 α1 α2−9 5-HT3 P2X2 Purines P2X1 Glu R2 NR2A Glu R6 β1−4 α2 β1−4 Glu R3 NR2B Glu R7 γ1−4 α3 γ P2X3 Glu R4 NR2C KA1 δ α4 δ P2X4 NR2D KA2 ε β ρ1−3 P2X5 P2X6 P2X7 Figure 6.4 The general architecture of ligand-gated receptors. (A) One of the subunits of a complete receptor. The long N-terminal region forms the ligand-binding site, while the remainder of the protein spans the membrane either four times (left) or three times (right). (B) Assembly of either four or five subunits into a complete receptor. (C) A diversity of subunits come together to form functional ionotropic neurotransmitter receptors. terminals and metabolized to glutamate by the mitochondrial enzyme glutaminase (Figure 6.6). Glutamate can also be synthesized by transamination of 2-oxoglutarate, an intermediate of the tricarboxylic acid cycle. Hence, some of the glucose metabolized by neurons can also be used for glutamate synthesis. The glutamate synthesized in the presynaptic cytoplasm is packaged into synaptic vesicles by transporters, termed VGLUT. At least three different VGLUT genes have been identified. Once released, glutamate is removed from the synaptic cleft by the excitatory amino acid transporters (EAATs). There are five different types of high-affinity glutamate transporters exist, some of which are present in glial cells and others in presynaptic terminals. Glutamate taken up by glial cells is converted into glutamine by the enzyme Neurotransmitters and Their Receptors 139 (A) N Neurotransmitter binding site VII I VI II III V IV Figure 6.5 Structure and function of metabotropic receptors. (A) The transmembrane architecture of metabotropic receptors. These monomeric proteins contain seven transmembrane domains. Portions of domains II, III, VI, and VII make up the neurotransmitter-binding region. G-proteins bind to both the loop between domains V and VI and to portions of the C-terminal region. (B) Varieties of metabotropic neurotransmitter receptors. G-protein binding site C (B) Receptor class Glutamate GABAB Dopamine NE, Epi Histamine Serotonin Purines Muscarinic Receptor subtype Class I mGlu R1 GABAB R1 GABAB R2 D1A D1B α1 α2 H1 H2 5-HT 1 5-HT 2 M1 M2 D2 D3 D4 β1 β2 β3 H3 5-HT 3 5-HT 4 5-HT 5 A type A1 A2a A2b A3 mGlu R5 Class II mGlu R2 mGlu R3 Class III mGlu R4 5-HT 6 5-HT 7 mGlu R6 mGlu R7 mGlu R8 glutamine synthetase; glutamine is then transported out of the glial cells and into nerve terminals. In this way, synaptic terminals cooperate with glial cells to maintain an adequate supply of the neurotransmitter. This overall sequence of events is referred to as the glutamate-glutamine cycle (see Figure 6.6). Several types of glutamate receptors have been identified. Three of these are ionotropic receptors called, respectively, NMDA receptors, AMPA receptors, and kainate receptors (Figure 6.4C). These glutamate receptors are named after the agonists that activate them: NMDA (N-methyl-D-aspartate), AMPA (α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate), and kainic acid. All of the ionotropic glutamate receptors are nonselective cation channels similar to the nAChR, allowing the passage of Na+ and K+, and in some cases small amounts of Ca2+. Hence AMPA, kainate, and NMDA receptor activation always produces excitatory postsynaptic responses. Like other ionotropic receptors, AMPA/kainate and NMDA receptors are also formed P type P2x P2y P2z P2t P2u M3 M4 M5 140 Chapter Six Box C Myasthenia Gravis: An Autoimmune Disease of Neuromuscular Synapses EMG recording of muscle action potentials (mV) (A) 8 Normal Myasthenia gravis Myasthenia gravis after neostigmine treatment 0 0 4 0 –4 0 20 40 60 80 20 (B) Number of MEPPs Myasthenia gravis is a disease that interferes with transmission between motor neurons and skeletal muscle fibers and afflicts approximately 1 of every 200,000 people. Originally described by the British physician Thomas Willis in 1685, the hallmark of the disorder is muscle weakness, particularly during sustained activity. Although the course is variable, myasthenia commonly affects muscles controlling the eyelids (resulting in drooping of the eyelids, or ptosis) and eye movements (resulting in double vision, or diplopia). Muscles controlling facial expression, chewing, swallowing, and speaking are other common targets. An important indication of the cause of myasthenia gravis came from the clinical observation that the muscle weakness improves following treatment with inhibitors of acetylcholinesterase, the enzyme that normally degrades acetylcholine at the neuromuscular junction. Studies of muscle obtained by biopsy from myasthenic patients showed that both end plate potentials (EPPs) and miniature end plate potentials (MEPPs) are much smaller than normal (see figure; also see Chapter 5). Because both the frequency of MEPPs and the quantal content of EPPs are normal, it seemed likely that myasthenia gravis entails a disorder of the postsynaptic muscle cells. Indeed, electron microscopy shows that the structure of neuromuscular junctions is altered, obvious changes being a widening of the synaptic cleft and an apparent reduction in the number of acetylcholine receptors in the postsynaptic membrane. A chance observation in the early 1970s led to the discovery of the underlying cause of these changes. Jim Patrick and Jon Lindstrom, then working at the Salk Institute, were attempting to raise antibodies to nicotinic acetylcholine receptors by immunizing rabbits with 40 60 80 Time (ms) 20 40 60 80 15 Myasthenia gravis 10 Normal 5 0.05 0.10 0.20 0.50 1 MEPP amplitude (mV) 2 3 (A) Myasthenia gravis reduces the efficiency of neuromuscular transmission. Electromyographs show muscle responses elicited by stimulating motor nerves. In normal individuals, each stimulus in a train evokes the same contractile response. In contrast, transmission rapidly fatigues in myasthenic patients, although it can be partially restored by administration of acetylcholinesterase inhibitors such as neostigmine. (B) Distribution of MEPP amplitudes in muscle fibers from myasthenic patients (solid line) and controls (dashed line). The smaller size of MEPPs in myasthenics is due to a diminished number of postsynaptic receptors. (A after Harvey et al., 1941; B after Elmqvist et al., 1964.) the receptors. Unexpectedly, the immunized rabbits developed muscle weakness that improved after treatment with acetylcholinesterase inhibitors. Subsequent work showed that the blood of myasthenic patients contains antibodies directed against the acetylcholine receptor, and that these antibodies are present at neuromuscular synapses. Removal of antibodies by plasma exchange improves the weakness. Finally, injecting the serum of myasthenic patients into mice produces myasthenic effects (because the serum carries circulating antibodies). These findings indicate that myasthenia gravis is an autoimmune disease that targets nicotinic acetylcholine receptors. The immune response reduces the number of functional receptors at the neuromuscular junction and can eventually destroys them altogether, diminishing the efficiency of synaptic transmission; muscle weakness thus occurs because motor neurons are less capable of exciting the postsynaptic muscle cells. This causal sequence also explains why cholinesterase inhibitors alleviate the signs and symptoms of myasthenia: The inhibitors increase the concentration of acetylcholine in the synaptic cleft, allowing more effective activation of those postsynaptic receptors not yet destroyed by the immune system. Despite all these insights, it is still not clear what triggers the immune system to produce an autoimmune Neurotransmitters and Their Receptors 141 response to acetylcholine receptors. Surgical removal of the thymus is beneficial in young patients with hyperplasia of the thymus, though precisely how the thymus contributes to myasthenia gravis is incompletely understood. Many patients are treated with combi- nations of immunosuppression and cholinesterase inhibitors. References ELMQVIST, D., W. W. HOFMANN, J. KUGELBERG AND D. M. J. QUASTEL (1964) An electrophysiological investigation of neuromuscular transmission in myasthenia gravis. J. Physiol. (Lond.) 174: 417–434. PATRICK, J. AND J. LINDSTROM (1973) Autoimmune response to acetylcholine receptor. Science 180: 871–872. VINCENT, A. (2002) Unravelling the pathogenesis of myasthenia gravis. Nature Rev. Immunol. 2: 797–804. from the association of several protein subunits that can combine in many ways to produce a large number of receptor isoforms (see Figure 6.4C). NMDA receptors have especially interesting properties (Figure 6.7A). Perhaps most significant is the fact that NMDA receptor ion channels allow the entry of Ca2+ in addition to monovalent cations such as Na+ and K+. As a result, EPSPs produced by NMDA receptors can increase the concentration of Ca2+ within the postsynaptic neuron; the Ca2+ concentration change can then act as a second messenger to activate intracellular signaling cascades (see Chapter 7). Another key property is that they bind extracellular Mg2+. At hyperpolarized membrane potentials, this ion blocks the pore of the NMDA receptor channel. Depolarization, however, pushes Mg2+ out of the pore, allowing other cations to flow. This property provides the basis for a voltage-dependence to current flow through the receptor (dashed line in Figure 6.7B) and means that NMDA receptors pass cations (most notably Ca2+) Glial cell EATT Presynaptic terminal Glutamine Glutamine COO− + H3N CH O CH2 CH2 C Glutamine synthetase NH2 Glutamate Glutaminase Glutamate EATT COO− + NH3 CH CH2 CH2 COO− VGLUT Glutamate Postsynaptic cell Glutamate receptors Figure 6.6 Glutamate synthesis and cycling between neurons and glia. The action of glutamate released into the synaptic cleft is terminated by uptake into neurons and surrounding glial cells via specific transporters. Within the nerve terminal, the glutamine released by glial cells and taken up by neurons is converted back to glutamate. Glutamate is transported into cells via excitatory amino acid transporters (EATTs) and loaded into synaptic vesicles via vesicular glutamate transporters (VGLUT). 142 Chapter Six Ca2+ (C) Na+ Glutamate EPSC (pA) (A) 50 25 0 AMPA only Glycine −25 0 EPSC (pA) Channel pore 25 50 75 Time (ms) 100 50 75 Time (ms) 100 50 25 0 NMDA only −25 Mg2+ Mg2+ binding site K+ EPSC (pA) 0 50 25 0 AMPA and NMDA −25 0 (B) Glycine, no Mg2+ 150 EPSC (pA) 100 50 0 −50 Mg2+, glycine No glycine, no Mg2+ −100 −150 100 −100 −50 0 50 Membrane potential (mV) 25 25 50 75 Time (ms) 100 Figure 6.7 NMDA and AMPA/kainate receptors. (A) NMDA receptors contain binding sites for glutamate and the co-activator glycine, as well as an Mg2+-binding site in the pore of the channel. At hyperpolarized potentials, the electrical driving force on Mg2+ drives this ion into the pore of the receptor and blocks it. (B) Current flow across NMDA receptors at a range of postsynaptic voltages, showing the requirement for glycine, and Mg2+ block at hyperpolarized potentials (dotted line). (C) The differential effects of glutamate receptor antagonists indicate that activation of AMPA or kainate receptors produces very fast EPSCs (top panel) and activation of NMDA receptors causes slower EPSCs (middle panel), so that EPSCs recorded in the absence of antagonists have two kinetic components due to the contribution of both types of response (bottom panel). only during depolarization of the postsynaptic cell, due to either activation of a large number of excitatory inputs and/or by repetitive firing of action potentials in the presynaptic cell. These properties are widely thought to be the basis for some forms of information storage at synapses, such as memory, as described in Chapter 24. Another unusual property of NMRA receptors is that opening the channel of this receptor requires the presence of a coagonist, the amino acid glycine (Figure 6.7A,B). There are at least five forms of NMDA receptor subunits (NMDA-R1, and NMDA-R2A through NMDAR2D); different synapses have distinct combinations of these subunits, producing a variety of NMDA receptor-mediated postsynaptic responses. Whereas some glutamatergic synapses have only AMPA or NMDA receptors, most possess both AMPA and NMDA receptors. An antagonist of Neurotransmitters and Their Receptors 143 NMDA receptors, APV (2-amino-5-phosphono-valerate), is often used to differentiate between the two receptor types. The use of this drug has also revealed differences between the EPSPs produced by NMDA and those produced by AMPA/kainate receptors, such as the fact that the synaptic currents produced by NMDA receptors are slower and longer-lasting than the those produced by AMPA/kainate receptors (see Figure 6.7C). In addition to these ionotropic glutamate receptors, there are three types of metabotropic glutamate receptor (mGluRs) (Figure 6.5). These receptors, which modulate postsynaptic ion channels indirectly, differ in their coupling to intracellular signal transduction pathways (see Chapter 7) and in their sensitivity to pharmacological agents. Activation of many of these receptors leads to inhibition of postsynaptic Ca2+ and Na+ channels. Unlike the excitatory ionotropic glutamate receptors, mGluRs cause slower postsynaptic responses that can either increase or decrease the excitability of postsynaptic cells. As a result the physiological roles of mGluRs are quite varied. GABA and Glycine Most inhibitory synapses in the brain and spinal cord use either γ-aminobutyric acid (GABA) or glycine as neurotransmitters. Like glutamate, GABA was identified in brain tissue during the 1950s. The details of its synthesis and degradation were worked out shortly thereafter by the work of Ernst Florey and Eugene Roberts. During this era, David Curtis and Jeffrey Watkins first showed that GABA can inhibit action potential firing in mammalian neurons. Subsequent studies by Edward Kravitz and colleagues established that GABA serves as an inhibitory transmitter at lobster neuromuscular synapses. It is now known that as many as a third of the synapses in the brain use GABA as their inhibitory neurotransmitter. GABA is most commonly found in local circuit interneurons, although cerebellar Purkinje cells provide an example of a GABAergic projection neuron (see Chapter 18). The predominant precursor for GABA synthesis is glucose, which is metabolized to glutamate by the tricarboxylic acid cycle enzymes (pyruvate and glutamine can also act as precursors). The enzyme glutamic acid decarboxylase (GAD), which is found almost exclusively in GABAergic neurons, catalyzes the conversion of glutamate to GABA (Figure 6.8A). GAD requires a cofactor, pyridoxal phosphate, for activity. Because pyridoxal phosphate is derived from vitamin B6, a B6 deficiency can lead to diminished GABA synthesis. The significance of this became clear after a disastrous series of infant deaths was linked to the omission of vitamin B6 from infant formula. This lack of B6 resulted in a large reduction in the GABA content of the brain, and the subsequent loss of synaptic inhibition caused seizures that in some cases were fatal. Once GABA is synthesized, it is transported into synaptic vesicles via a vesicular inhibitory amino acid transporter (VIATT). The mechanism of GABA removal is similar to that for glutamate: Both neurons and glia contain high-affinity transporters for GABA, termed GATs (several forms of GAT have been identified). Most GABA is eventually converted to succinate, which is metabolized further in the tricarboxylic acid cycle that mediates cellular ATP synthesis. The enzymes required for this degradation, GABA transaminase and succinic semialdehyde dehydrogenase, are mitochondrial enzymes. Inhibition of GABA breakdown causes a rise in tissue GABA content and an increase in the activity of inhibitory neurons. There are also other pathways for degradation of GABA. The most noteworthy of these results in the production of γ-hydroxybutyrate, a GABA derivitive that has been abused as a “date rape” drug. Oral adminis- 144 Chapter Six Figure 6.8 Synthesis, release, and reuptake of the inhibitory neurotransmitters GABA and glycine. (A) GABA is synthesized from glutamate by the enzyme glutamic acid decarboxylase, which requires pyridoxal phosphate. (B) Glycine can be synthesized by a number of metabolic pathways; in the brain, the major precursor is serine. High-affinity transporters terminate the actions of these transmitters and return GABA or glycine to the synaptic terminals for reuse, with both transmitters being loaded into synaptic vesicles via the vesicular inhibitory amino acid transporter (VIATT). (A) Glial cell GABA breakdown Presynaptic terminal Glucose GAT Glutamate COO + NH3 CH − CH2 CH2 COO − Glutamic acid decarboxylase + pyridoxal phosphate GABA + H3N CH2 CH2 COO− CH2 GABA VIATT GABA receptors Postsynaptic cell (B) Glial cell Presynaptic terminal Glycine transporter Glucose Glucose Serine Serine ++ HH – HH3NN— CC— COO COO− 3 COOH COOH Glycine Glycine Serine transhydroxymethylase ++ HH H3H N3N — CC— COO COO–− HH Glycine VIATT Glycine receptors Postsynaptic cell Neurotransmitters and Their Receptors 145 Box D Excitotoxicity Following Acute Brain Injury Excitotoxicity refers to the ability of glutamate and related compounds to destroy neurons by prolonged excitatory synaptic transmission. Normally, the concentration of glutamate released into the synaptic cleft rises to high levels (approximately 1 mM), but it remains at this concentration for only a few milliseconds. If abnormally high levels of glutamate accumulate in the cleft, the excessive activation of neuronal glutamate receptors can literally excite neurons to death. The phenomenon of excitotoxicity was discovered in 1957 when D. R. Lucas and J. P. Newhouse serendipitously found that feeding sodium glutamate to infant mice destroys neurons in the retina. Roughly a decade later, John Olney at Washington University extended this discovery by showing that regions of glutamate-induced neuronal loss can occur throughout the brain. The damage was evidently restricted to the postsynaptic cells—the dendrites of the target neurons were grossly swollen— while the presynaptic terminals were spared. Olney also examined the relative potency of glutamate analogs and found that their neurotoxic actions paralleled their ability to activate postsynaptic glutamate receptors. Furthermore, glutamate receptor antagonists were effective in blocking the neurotoxic effects of glutamate. In light of this evidence, Olney postulated that glutamate destroys neurons by a mechanism similar to transmission at excitatory glutamatergic syn- apses, and coined the term excitotoxic to refer to this pathological effect. Evidence that excitotoxicity is an important cause of neuronal damage after brain injury has come primarily from studying the consequences of reduced blood flow. The most common cause of reduced blood flow to the brain (ischemia) is the occlusion of a cerebral blood vessel (i.e., a stroke; see Appendix 3). The idea that excessive synaptic activity contributes to ischemic injury emerged from the observation that concentrations of glutamate and aspartate in the extracellular space around neurons increase during ischemia. Moreover, microinjection of glutamate receptor antagonists in experimental animals protects neurons from ischemia-induced damage. Together, these findings imply that extracellular accumulation of glutamate during ischemia activates glutamate receptors excessively, and that this somehow triggers a chain of events that leads to neuronal death. The reduced supply of oxygen and glucose presumably elevates extracellular glutamate levels by slowing the energy-dependent removal of glutamate at synapses. Excitotoxic mechanisms have now been shown to be involved in other acute forms of neuronal insult, including hypoglycemia, trauma, and repeated intense seizures (called status epilepticus). Understanding excitotoxicity therefore has important implications for treating a variety of neurological disorders. For instance, a blockade of glutamate tration of γ-hydroxybutyrate can cause euphoria, memory deficits, and unconsciousness. Presumably these effects arise from actions on GABAergic synapses in the CNS. Inhibitory synapses employing GABA as their transmitter can exhibit three types of postsynaptic receptors, called GABAA, GABAB, and GABAC. GABAA and GABAC receptors are ionotropic receptors, while GABAB receptors are metabotropic. The ionotropic GABA receptors are usually receptors could, in principle, protect neurons from injury due to stroke, trauma, or other causes. Unfortunately, clinical trials of glutamate receptor antagonists have not led to much improvement in the outcome of stroke. The ineffectiveness of this quite logical treatment is probably due to several factors, one of which is that substantial excitotoxic injury occurs quite soon after ischemia, prior to the typical initiation of treatment. It is also likely that excitotoxicity is only one of several mechanisms by which ischemia damages neurons, other candidates including damage secondary to inflammation. Pharmacological interventions that target all these mechanisms nonetheless hold considerable promise for minimizing brain injury after stroke and other causes. References LUCAS, D. R. AND J. P. NEWHOUSE (1957) The toxic effects of sodium L-glutamate on the inner layers of the retina. Arch. Opthalmol. 58: 193–201. OLNEY, J. W. (1969) Brain lesions, obesity and other disturbances in mice treated with monosodium glutamate. Science 164: 719–721. OLNEY, J. W. (1971) Glutamate-induced neuronal necrosis in the infant mouse hypothalamus: An electron microscopic study. J. Neuropathol. Exp. Neurol. 30: 75–90. ROTHMAN, S. M. (1983) Synaptic activity mediates death of hypoxic neurons. Science 220: 536–537. SYNTICHAKI, P. AND N. TAVERNARAKIS (2003) The biochemistry of neuronal necrosis: Rogue biology? Nature Neurosci. Rev. 4: 672–684. 146 Chapter Six inhibitory because their associated channels are permeable to Cl– (Figure 6.9A); the flow of the negatively charged chloride ions inhibits postsynaptic cells since the reversal potential for Cl– is more negative than the threshold for neuronal firing (see Figure 5.19B). Like other ionotropic receptors, GABA receptors are pentamers assembled from a combination of five types of subunits (α, β, γ, δ, and ρ; see Figure 6.4C). As a result of this subunit diversity, as well as variable stoichiometry of subunits, the function of GABAA receptors differs widely among neuronal types. Drugs that act as agonists or modulators of postsynaptic GABA receptors, such as benzodiazepines and barbiturates, are used clinically to treat epilepsy and are effective sedatives and anesthetics. Binding sites for GABA, barbiturates, steroids, and picrotoxin are all located within the pore domain of the channel (Figure 6.9B). Another site, called the benzodiazepine binding site, lies outside the pore and modulates channel activity. Benzodiazepines, such as diazepam (Valium®) and chlordiazepoxide (Librium®), are tranquilizing (anxiety reducing) drugs that enhance GABAergic transmission by binding to the α and δ subunits of GABAA receptors. Barbiturates, such as phenobarbital and pentobarbital, are hypnotics that bind to the α and β subunits of some GABA receptors and are used therapeutically for anesthesia and to control epilepsy. Another drug that can alter the activity of GABA-mediated inhibitory circuits is alcohol; at least some aspects of drunken behavior are caused by the alcohol-mediated alterations in ionotropic GABA receptors. Metabotropic GABA receptors (GABAB) are also widely distributed in brain. Like the ionotropic GABAA receptors, GABAB receptors are inhibitory. Rather than activating Cl− selective channels, however, GABAB-mediated inhibition is due to the activation of K+ channels. A second mechanism for (A) (B) Membrane potential (mV) Benzodiazepine GABA 0 b Stimulate presynaptic neuron a a Subunit Channel pore –50 g 0 50 100 150 200 250 Time (ms) 300 350 b 400 Barbiturates Steroids Figure 6.9 Ionotropic GABA receptors. (A) Stimulation of a presynaptic GABAergic interneuron, at the time indicated by the arrow, causes a transient inhibition of action potential firing in its postynaptic target. This inhibitory response is caused by activation of postsynaptic GABAA receptors. (B) Ionotropic GABA receptors contain two binding sites for GABA and numerous sites at which drugs bind to and modulate these receptors. (A after Chavas and Marty, 2003). Picrotoxin Chloride ions Neurotransmitters and Their Receptors 147 GABAB-mediated inhibition is by blocking Ca2+ channels, which tends to hyperpolarize postsynaptic cells. Unlike most metabotropic receptors, GABAB receptors appear to assemble as heterodimers of GABAB R1 and R2 subunits. The distribution of the neutral amino acid glycine in the central nervous system is more localized than that of GABA. About half of the inhibitory synapses in the spinal cord use glycine; most other inhibitory synapses use GABA. Glycine is synthesized from serine by the mitochondrial isoform of serine hydroxymethyltransferase (Figure 6.8B), and is transported into synaptic vesicles via the same vesicular inhibitory amino acid transporter that loads GABA into vesicles. Once released from the presynaptic cell, glycine is rapidly removed from the synaptic cleft by the plasma membrane glycine transporters. Mutations in the genes coding for some of these enzymes result in hyperglycinemia, a devastating neonatal disease characterized by lethargy, seizures, and mental retardation. The receptors for glycine are also ligand-gated Cl– channels, their general structure mirroring that of the GABAA receptors. Glycine receptors are pentamers consisting of mixtures of the 4 gene products encoding glycine-binding α subunits, along with the accessory β subunit. Glycine receptors are potently blocked by strychnine, which may account for the toxic properties of this plant alkaloid (see Box B). COO− Tyrosine CH2 + NH3 HO O2 Tyrosine hydroxylase Dihydroxyphenylalanine (DOPA) COO− CH2 CH + NH3 H CH + NH3 H CH + NH3 H CH + NH2 HO OH CO2 DOPA decarboxylase Dopamine CH2 The Biogenic Amines Biogenic amine transmitters regulate many brain functions and are also active in the peripheral nervous system. Because biogenic amines are implicated in such a wide range of behaviors (ranging from central homeostatic functions to cognitive phenomena such as attention), it is not surprising that defects in biogenic amines function are implicated in most psychiatric disorders. The pharmacology of amine synapses is critically important in psychotherapy, with drugs affecting the synthesis, receptor binding, or catabolism of these neurotransmitters being among the most important agents in the armamentarium of modern pharmacology (Box E). Many drugs of abuse also act on biogenic amine pathways. There are five well-established biogenic amine neurotransmitters: the three catecholamines—dopamine, norepinephrine (noradrenaline), and epinephrine (adrenaline)—and histamine and serotonin (see Figure 6.1). All the catecholamines (so named because they share the catechol moiety) are derived from a common precursor, the amino acid tyrosine (Figure 6.10). The first step in catecholamine synthesis is catalyzed by tyrosine hydroxylase in a reaction requiring oxygen as a co-substrate and tetrahydrobiopterin as a cofactor to synthesize dihydroxyphenylalanine (DOPA). Histamine and serotonin are synthesized via other routes, as described below. • Dopamine is present in several brain regions (Figure 6.11A), although the major dopamine-containing area of the brain is the corpus striatum, which receives major input from the substantia nigra and plays an essential role in the coordination of body movements. In Parkinson’s disease, for instance, the dopaminergic neurons of the substantia nigra degenerate, leading to a characteristic motor dysfunction (see Box B in Chapter 17). Dopamine is also believed to be involved in motivation, reward, and reinforcement, and many drugs of abuse work by affecting dopaminergic synapses in the CNS (see Box A). In addition to these roles in the CNS, dopamine also plays a poorly understood role in some sympathetic ganglia. CH HO OH O2 Dopamine-β hydroxylase OH Norepinephrine CH HO OH RCH3 Phenylethanolamine N-methyltransferase R OH Epinephrine CH CH3 HO OH Figure 6.10 The biosynthetic pathway for the catecholamine neurotransmitters. The amino acid tyrosine is the precursor for all three catecholamines. The first step in this reaction pathway, catalyzed by tyrosine hydroxylase, is rate-limiting. 148 Chapter Six Box E Biogenic Amine Neurotransmitters and Psychiatric Disorders The regulation of the biogenic amine neurotransmitters is altered in a variety of psychiatric disorders. Indeed, most psychotropic drugs (defined as drugs that alter behavior, mood, or perception) selectively affect one or more steps in the synthesis, packaging, or degradation of biogenic amines. Sorting out how these drugs work has been extremely useful in beginning to understand the molecular mechanisms underlying some of these diseases. Based on their effects on humans, psychotherapeutic drugs can be divided into several broad categories: antipsychotics, antianxiety drugs, antidepressants, and stimulants. The first antipsychotic drug used to ameliorate disorders such as schizophrenia was reserpine. Reserpine was developed in the 1950s and initially used as an antihypertensive agent; it blocks the uptake of norepinephrine into synaptic vesicles and therefore depletes the transmitter at aminergic terminals, diminishing the ability of the sympathetic division of the visceral motor system to cause vasoconstriction (see Chapter 20). A major side effect in hypertensive patients treated with reserpine—behavioral depression—suggested the possibility of using it as an antipsychotic agent in patients suffering from agitation and pathological anxiety. (Its ability to cause depression in mentally healthy individuals also suggested that aminergic transmitters are involved in mood disorders; see Box E in Chapter 28.) Although reserpine is no longer used as an antipsychotic agent, its initial success stimulated the development of antipsychotic drugs such as chlorpromazine, haloperidol, and benperidol, which over the last several decades have radically changed the approach to treating psychotic disorders. Prior to the discovery of these drugs, psychotic patients were typically hospitalized for long periods, sometimes indefinitely, and in the 1940s were subjected to desperate measures such as frontal lobotomy (see Box B in Chapter 25). Modern antipsychotic drugs now allow most patients to be treated on an outpatient basis after a brief hospital stay. Importantly, the clinical effectiveness of these drugs is correlated with their ability to block brain dopamine receptors, implying that activation of dopamine receptors contributes to some types of psychotic illness. A great deal of effort continues to be expended on developing more effective antipsychotic drugs with fewer side effects, and on discovering the mechanism and site of action of these medications. The second category of psychotherapeutic drugs is the antianxiety agents. Anxiety disorders are estimated to afflict 10–35% of the population, making them the most common psychiatric problem. The two major forms of pathological anxiety—panic attacks and generalized anxiety disorder—both respond to drugs that affect aminergic transmission. The agents used to treat panic disorders include inhibitors of the enzyme monoamine oxidase (MAO inhibitors) required for the catabolism of the amine neurotransmitters, and blockers of serotonin receptors. The most effective drugs in treating generalized anxiety disorder have been benzodiazepines, such as chlordiazepoxide (Librium®), and diazepam (Valium®). In contrast to most other psychotherapeutic drugs, these agents increase the efficacy of transmission at GABAA synapses rather than acting at aminergic synapses. Antidepressants and stimulants also affect aminergic transmission. A large number of drugs are used clinically to treat depressive disorders. The three major classes of antidepressants—MAO inhibitors, tricyclic antidepressants, and serotonin uptake blockers such as fluoxetine (Prozac®) and trazodone—all influence various aspects of aminergic transmission. MAO inhibitors such as phenelzine block the breakdown of amines, whereas the tricyclic antidepressants such as desipramine block the reuptake of norepinephrine and other amines. The extraordinarily popular antidepressant fluoxetine (Prozac®) selectively blocks the reuptake of serotonin without affecting the reuptake of catecholamines. Stimulants such as amphetamine are also used to treat some depressive disorders. Amphetamine stimulates the release of norepinephrine from nerve terminals; the transient “high” resulting from taking amphetamine may reflect the emotional opposite of the depression that sometimes follows reserpine-induced norepinephrine depletion. Despite the relatively small number of aminergic neurons in the brain, this litany of pharmacological actions emphasizes that these neurons are critically important in the maintenance of mental health. References FRANKLE, W. G., J. LERMA AND M. LARUELLE (2003) The synaptic hypothesis of schizophrenia. Neuron 39: 205–216. FREEDMAN, R. (2003) Schizophrenia. N. Engl. J. Med. 349: 1738–1749. LEWIS, D. A. AND P. LEVITT (2002) Schizophrenia as a disorder of neurodevelopment. Annu. Rev. Neurosci. 25: 409–432. NESTLER, E. J., M. BARROT, R. J. DILEONE, A. J. EISCH, S. J. GOLD AND L. M. MONTEGGIA (2002) Neurobiology of depression. Neuron 34: 13–25. Neurotransmitters and Their Receptors 149 (A) Dopamine (B) Norepinephrine Corpus callosum Cerebral cortex Corpus callosum Cerebral cortex To striatum Thalamus Thalamus Cerebellum Substantia nigra Pons and ventral Medulla tegmental area To spinal cord Figure 6.11 The distribution in the human brain of neurons and their projections (arrows) containing catecholamine neurotransmitters. Curved arrows along the perimeter of the cortex indicate the innervation of lateral cortical regions not shown in this midsagittal plane of section. Cerebellum Pons Locus coeruleus Medulla (C) Epinephrine To spinal cord Corpus callosum Cerebral cortex Thalamus Hypothalamus Cerebellum Pons Medullary epinephrine neurons Dopamine is produced by the action of DOPA decarboxylase on DOPA (see Figure 6.10). Following its synthesis in the cytoplasm of presynaptic terminals, dopamine is loaded into synaptic vesicles via a vesicular monoamine transporter (VMAT). Dopamine action in the synaptic cleft is terminated by reuptake of dopamine into nerve terminals or surrounding glial cells by a Na+-dependent dopamine transporter, termed DAT. Cocaine apparently produces its psychotropic effects by binding to and inhibiting DAT, yielding a net increase in dopamine release from specific brain areas. Amphetamine, another addictive drug, also inhibits DAT as well as the transporter for norepinepherine (see below). The two major enzymes involved in the catabolism of dopamine are monoamine oxidase (MAO) and catechol O-methyltransferase (COMT). Both neurons and glia contain mitochondrial MAO and cytoplasmic COMT. Inhibitors of these enzymes, such as phenelzine and tranylcypromine, are used clinically as antidepressants (see Box E). Once released, dopamine acts exclusively by activating G-protein-coupled receptors. These are mainly dopamine-specific receptors, although β-adrenergic receptors also serve as important targets of norepinepherine and epinepherine (see below). Most dopamine receptor subtypes (see Figure 6.5B) Medulla To spinal cord 150 Chapter Six act by either activating or inhibiting adenylyl cyclase (see Chapter 7). Activation of these receptors generally contribute to complex behaviors; for example, administration of dopamine receptor agonists elicits hyperactivity and repetitive, stereotyped behavior in laboratory animals. Activation of another type of dopamine receptor in the medulla inhibits vomiting. Thus, antagonists of these receptors are used as emetics to induce vomiting after poisoning or a drug overdose. Dopamine receptor antagonists can also elicit catalepsy, a state in which it is difficult to initiate voluntary motor movement, suggesting a basis for this aspect of some psychoses. • Norepinephrine (also called noradrenaline) is used as a neurotransmitter in the locus coeruleus, a brainstem nucleus that projects diffusely to a variety of forebrain targets (Figure 6.11B) and influences sleep and wakefulness, attention, and feeding behavior. Perhaps the most prominent noradrenergic neurons are sympathetic ganglion cells, which employ norepinephrine as the major peripheral transmitter in this division of the visceral motor system (see Chapter 20). Norepinephrine synthesis requires dopamine β-hydroxylase, which catalyzes the production of norepinephrine from dopamine (see Figure 6.10). Norepinephrine is then loaded into synaptic vesicles via the same VMAT involved in vesicular dopamine transport. Norepinepherine is cleared from the synaptic cleft by the norepinepherine transporter (NET), which also is capable of taking up dopamine. As mentioned, NET serves as a molecular target of amphetamine, which acts as a stimulant by producing a net increase in the release of norepinepherine and dopamine. A mutation in the NET gene is a cause of orthostatic intolerance, a disorder that produces lightheadedness while standing up. Like dopamine, norepinepherine is degraded by MAO and COMT. Norepinepherine, as well as epinephrine, acts on α- and β-adrenergic receptors (Figure 6.5B). Both types of receptor are G-protein-coupled; in fact, the β-adrenergic receptor was the first identified metabotropic neurotransmitter receptor. Two subclasses of α-adrenergic receptors are now known. Activation of α1 receptors usually results in a slow depolarization linked to the inhibition of K+ channels, while activation of α2 receptors produces a slow hyperpolarization due to the activation of a different type of K+ channel. There are three subtypes of β-adrenergic receptor, two of which are expressed in many types of neurons. Agonists and antagonists of adrenergic receptors, such as the β blocker propanolol (Inderol®), are used clinically for a variety of conditions ranging from cardiac arrhythmias to migraine headaches. However, most of the actions of these drugs are on smooth muscle receptors, particularly in the cardiovascular and respiratory systems (see Chapter 20). • Epinephrine (also called adrenaline) is found in the brain at lower levels than the other catecholamines and also is present in fewer brain neurons than other catecholamines. Epinephrine-containing neurons in the central nervous system are primarily in the lateral tegmental system and in the medulla and project to the hypothalamus and thalamus (Figure 6.11C). The function of these epinepherine-secreting neurons is not known. The enzyme that synthesizes epinephrine, phenylethanolamine-Nmethyltransferase (see Figure 6.10), is present only in epinephrine-secreting neurons. Otherwise, the metabolism of epinepherine is very similar to that of norepinepherine. Epinepherine is loaded into vesicles via the VMAT. No plasma membrane transporter specific for epinepherine has been identified, though the NET is capable of transporting epinepherine. As already noted, epinepherine acts on both α- and β-adrenergic receptors. Neurotransmitters and Their Receptors 151 (A) Histamine (B) Serotonin Corpus callosum Cerebral cortex Corpus callosum Cerebral cortex Thalamus Thalamus Cerebellum Tuberomammillary nucleus of hypothalamus Pons Medulla To spinal cord Cerebellum Raphe nuclei • Histamine is found in neurons in the hypothalamus that send sparse but widespread projections to almost all regions of the brain and spinal cord (Figure 6.12A). The central histamine projections mediate arousal and attention, similar to central ACh and norepinephrine projections. Histamine also controls the reactivity of the vestibular system. Allergic reactions or tissue damage cause release of histamine from mast cells in the bloodstream. The close proximity of mast cells to blood vessels, together with the potent actions of histamine on blood vessels, also raises the possibility that histamine may influence brain blood flow. Histamine is produced from the amino acid histidine by a histidine decarboxylase (Figure 6.13A) and is transported into vesicles via the same VMAT as the catecholamines. No plasma membrane histamine transporter has been identified yet. Histamine is degraded by the combined actions of histamine methyltransferase and MAO. There are three known types of histamine receptors, all of which are Gprotein-coupled receptors (Figure 6.5B). Because of the importance of histamine receptors in the mediation of allergic responses, many histamine receptor antagonists have been developed as antihistamine agents. Antihistamines that cross the blood-brain barrier, such as diphenhydramine (Benadryl®), act as sedatives by interfering with the roles of histamine in CNS arousal. Antagonists of the H1 receptor also are used to prevent motion sickness, perhaps because of the role of histamine in controling vestibular function. H2 receptors control the secretion of gastric acid in the digestive system, allowing H2 receptor antagonists to be used in the treatment of a variety of upper gastrointestinal disorders (e.g., peptic ulcers). • Serotonin, or 5-hydroxytryptamine (5-HT), was initially thought to increase vascular tone by virtue of its presence in serum (hence the name serotonin). Serotonin is found primarily in groups of neurons in the raphe region of the pons and upper brainstem, which have widespread projections to the forebrain (see Figure 6.12B) and regulate sleep and wakefulness (see Chapter 27). 5-HT occupies a place of prominence in neuropharmacology because a large number of antipsychotic drugs that are valuable in the treatment of depression and anxiety act on serotonergic pathways (see Box E). 5-HT is synthesized from the amino acid tryptophan, which is an essential dietary requirement. Tryptophan is taken up into neurons by a plasma mem- Pons Medulla To spinal cord Figure 6.12 The distribution in the human brain of neurons and their projections (arrows) containing histamine (A) or serotonin (B). Curved arrows along the perimeter of the cortex indicate the innervation of lateral cortical regions not shown in this midsagittal plane of section. 152 Chapter Six (A) COO− Histidine CH2 HN CH + NH3 H CH + NH3 N Histidine decarboxylase CO2 Histamine CH2 HN N (B) COO− Tryptophan CH2 CH + NH3 N O2 Tryptophan-5hydroxylase brane transporter and hydroxylated in a reaction catalyzed by the enzyme tryptophan-5-hydroxylase (Figure 6.13B), the rate-limiting step for 5-HT synthesis. Loading of 5-HT into synaptic vesicles is done by the VMAT that is also responsible for loading of other monoamines into synaptic vesicles. The synaptic effects of serotonin are terminated by transport back into nerve terminals via a specific serotonin transporter (SERT). Many antidepressant drugs are selective serotonin reuptake inhibitors (SSRIs) that inhibit transport of 5-HT by SERT. Perhaps the best known example of an SSRI is Prozac (see Box E). The primary catabolic pathway for 5-HT is mediated by MAO. A large number of 5-HT receptors have been identified. Most 5-HT receptors are metabotropic (see Figure 6.5B). These have been implicated in behaviors, including the emotions, circadian rhythms, motor behaviors, and state of mental arousal. Impairments in the function of these receptors have been implicated in numerous psychiatric disorders, such as depression, anxiety disorders, and schizophrenia (see Chapter 28), and drugs acting on serotonin receptors are effective treatments for a number of these conditions. Activation of 5-HT receptors also mediates satiety and decreased food consumption, which is why serotonergic drugs are sometimes useful in treating eating disorders. Only one group of serotonin receptors, called the 5-HT3 receptors, are ligand-gated ion channels (see Figure 6.4C). These are non-selective cation channels and therefore mediate excitatory postsynaptic responses. Their general structure, with functional channels formed by assembly of multiple subunits, is similar to the other ionotropic receptors described in the chapter. Two types of 5-HT3 subunit are known, and form functional channels by assembling as a heteromultimer. 5-HT receptors are targets for a wide variety of therapeutic drugs including ondansetron (Zofran®) and granisetron (Kytril®), which are used to prevent postoperative nausea and chemotherapy-induced emesis. ATP and Other Purines COO− 5-Hydroxytryptophan HO CH2 CH + NH3 N Aromatic L-amino acid decarboxylase CO2 Serotonin (5-HT) HO CH2 H CH + NH3 N Figure 6.13 Synthesis of histamine and serotonin. (A) Histamine is synthesized from the amino acid histidine. (B) Serotonin is derived from the amino acid tryptophan by a two-step process that requires the enzymes tryptophan-5hydroxylase and a decarboxylase. Interestingly, all synaptic vesicles contain ATP, which is co-released with one or more “classical” neurotransmitters. This observation raises the possibility that ATP acts as a co-transmitter. It has been known since the 1920s that the extracellular application of ATP (or its breakdown products AMP and adenosine) can elicit electrical responses in neurons. The idea that some purines (so named because all these compounds contain a purine ring; see Figure 6.1) are also neurotransmitters has now received considerable experimental support. ATP acts as an excitatory neurotransmitter in motor neurons of the spinal cord, as well as sensory and autonomic ganglia. Postsynaptic actions of ATP have also been demonstrated in the central nervous system, specifically for dorsal horn neurons and in a subset of hippocampal neurons. Adenosine, however, cannot be considered a classical neurotransmitter because it is not stored in synaptic vesicles or released in a Ca2+-dependent manner. Rather, it is generated from ATP by the action of extracellular enzymes. A number of enzymes, such as apyrase and ecto-5′ nucleotidase, as well as nucleoside transporters are involved in the rapid catabolism and removal of purines from extracellular locations. Despite the relative novelty of this evidence, it suggests that excitatory transmission via purinergic synapses is widespread in the mammalian brain. In accord with this evidence, receptors for both ATP and adenosine are widely distributed in the nervous system, as well as many other tissues. Neurotransmitters and Their Receptors 153 Three classes of these purinergic receptors are now known. One of these classes consists of ligand-gated ion channels (see Figure 6.4C); the other two are G-protein-coupled metabotropic receptors (see Figure 6.5B). Like many ionotropic transmitter receptors, the ligand-gated channels are nonselective cation channels that mediate excitatory postsynaptic responses. The genes encoding these channels, however, are unique in that they appear to have only two transmembrane domains. Ionotropic purinergic receptors are widely distributed in central and peripheral neurons. In sensory nerves they evidently play a role in mechanosensation and pain; their function in most other cells, however, is not known. The two types of metabotropic receptors activated by purines differ in their sensitivity to agonists: One type is preferentially stimulated by adenosine, whereas the other is preferentially activated by ATP. Both receptor types are found throughout the brain, as well as in peripheral tissues such as the heart, adipose tissue, and the kidney. Xanthines such as caffeine and theophylline block adenosine receptors, and this activity is thought to be responsible for the stimulant effects of these agents. Peptide Neurotransmitters Many peptides known to be hormones also act as neurotransmitters. Some peptide transmitters have been implicated in modulating emotions (see Chapter 28). Others, such as substance P and the opioid peptides, are involved in the perception of pain (see Chapter 9). Still other peptides, such as melanocyte-stimulating hormone, adrenocorticotropin, and β-endorphin, regulate complex responses to stress. The mechanisms responsible for the synthesis and packaging of peptide transmitters are fundamentally different from those used for the smallmolecule neurotransmitters and are much like the synthesis of proteins that are secreted from non-neuronal cells (pancreatic enzymes, for instance). Peptide-secreting neurons generally synthesize polypeptides in their cell bodies that are much larger than the final, “mature” peptide. Processing these polypeptides in their cell bodies, which are called pre-propeptides (or pre-proproteins), takes place by a sequence of reactions in several intracellular organelles. Pre-propeptides are synthesized in the rough endoplasmic reticulum, where the signal sequence of amino acids—that is, the sequence indicating that the peptide is to be secreted—is removed. The remaining polypeptide, called a propeptide (or proprotein), then traverses the Golgi apparatus and is packaged into vesicles in the trans-Golgi network. The final stages of peptide neurotransmitter processing occur after packaging into vesicles and involve proteolytic cleavage, modification of the ends of the peptide, glycosylation, phosphorylation, and disulfide bond formation. Propeptide precursors are typically larger than their active peptide products and can give rise to more than one species of neuropeptide (Figure 6.14). The means that multiple neuroactive peptides can be released from a single vesicle. In addition, neuropeptides often are co-released with small-molecule neurotransmitters. Thus, peptidergic synapses often elicit complex postsynaptic responses. Peptides are catabolized into inactive amino acid fragments by enzymes called peptidases, usually located on the extracellular surface of the plasma membrane. The biological activity of the peptide neurotransmitters depends on their amino acid sequence (Figure 6.15). Based on their amino acid sequences, neuropeptide transmitters have been loosely grouped into five categories: 154 Chapter Six Figure 6.14 Proteolytic processing of the pre-propeptides pre-proopiomelanocortin (A) and pre-proenkephalin A (B). For each pre-propeptide, the signal sequence is indicated in orange at the left; the locations of active peptide products are indicated by different colors. The maturation of the pre-propeptides involves cleaving the signal sequence and other proteolytic processing. Such processing can result in a number of different neuroactive peptides such as ACTH, γ-lipotropin, and β-endorphin (A), or multiple copies of the same peptide, such as met-enkephalin (B). (A) Signal peptide Pre-proopiomelanocortin Pre-propeptide Proopiomelanocortin Propeptide ACTH β-lipotropin Active peptide γ-lipotropin β-endorphin Active peptides (B) Signal peptide Pre-proenkephalin A Pre-propeptide Proenkephalin A Propeptide Active peptides Met-enkephalin Met-enkephalin Leu-enkephalin the brain/gut peptides, opioid peptides, pituitary peptides, hypothalamic releasing hormones, and a catch-all category containing other peptides that are not easily classified. Substance P is an example of the first of these categories (Figure 6.15A). The study of neuropeptides actually began more than 60 years ago with the accidental discovery of substance P, a powerful hypotensive agent. (The peculiar name derives from the fact that this molecule was an unidentified component of powder extracts from brain and intestine.) This 11-amino-acid peptide (see Figure 6.15) is present in high concentrations in the human hippocampus, neocortex, and also in the gastrointestinal tract; hence its classification as a brain/gut peptide. It is also released from C fibers, the smalldiameter afferents in peripheral nerves that convey information about pain and temperature (as well as postganglionic autonomic signals). Substance P is a sensory neurotransmitter in the spinal cord, where its release can be inhibited by opioid peptides released from spinal cord interneurons, resulting in the suppression of pain (see Chapter 9). The diversity of neuropeptides is highlighted by the finding that the gene coding for substance P encodes a number of other neuroactive peptides including neurokinin A, neuropeptide K, and neuropeptide γ. An especially important category of peptide neurotransmitters is the family of opioids (Figure 6.15B). These peptides are so named because they bind Neurotransmitters and Their Receptors 155 (A) Brain–gut peptides Substance P Arg Pro Lys Pro Gln Gln Phe Phe Gly Leu Met Cholecystokinin octapeptide (CCK-8) Asp Tyr Met Gly Trp Met Asp Phe Vasoactive intestinal His Asp Ala Val Phe Thr Asp Asn Tyr Thr Arg Leu Arg Lys Gln Met Ala Val Lys Lys Tyr Leu Asn Ser Ile Leu Asn peptide (VIP) (B) Opioid peptides Leucine enkephalin Tyr Gly Gly Phe Leu Amino acid properties α-Endorphin Tyr Gly Gly Phe Met Thr Ser Glu Lys Ser Gln Thr Pro Leu Val Thr Hydrophobic Dynorphin A Tyr Gly Gly Phe Leu Arg Arg Ile Arg Pro Lys Leu Lys Trp Asp Asn Gln Polar, uncharged Acidic (C) Pituitary peptides Basic Vasopressin Cys Tyr Phe Gln Arg Cys Pro Leu Gly Oxytocin Cys Tyr Ile Gln Arg Cys Pro Arg/ Lys Gly Adrenocorticotropic Ser Tyr Ser Met Glu His Phe Arg Trp Gly Lys Pro Val Gly Lys Lys Arg Arg Pro Val Lys Val Tyr Pro hormone (ACTH) (D) Hypothalamic–releasing peptides Thyrotropin releasing hormone (TRH) Glu His Pro Leutinizing hormoneGlu His Trp Ser Tyr Gly Leu Arg Pro Gly releasing hormone (LHRH) Somatostatin-14 Ala Gly Cys Lys Asn Phe Phe Trp Lys Thr Phe Thr Ser Cys (E) Miscellaneous peptides Angiotensin-II Asp Arg Val Tyr Ile His Pro Phe Neuropeptide-γ Tyr Pro Ser Lys Pro Asp Asn Pro Gly Glu Asp Ala Pro Ala Glu Asp Leu Ala Arg Tyr Tyr Ser Ala Leu Arg His Tyr Ile Asn Leu Ile Thr Arg Gln Arg Tyr Neurotensin Glu Leu Tyr Glu Asn Lys Pro Arg Arg Pro Ile Leu to the same postsynaptic receptors activated by opium. The opium poppy has been cultivated for at least 5000 years, and its derivatives have been used as an analgesic since at least the Renaissance. The active ingredients in opium are a variety of plant alkaloids, predominantly morphine. Morphine, named for Morpheus, the Greek god of dreams, is still one of the most effective analgesics in use today, despite its addictive potential (see Box A). Synthetic opiates such as meperidine and methadone are also used as analgesics, and fentanyl, a drug with 80 times the analgesic potency of morphine, is widely used in clinical anesthesiology. The opioid peptides were discovered in the 1970s during a search for endorphins, endogenous compounds that mimicked the actions of morphine. It was hoped that such compounds would be analgesics, and that understanding them would shed light on drug addiction. The endogenous ligands of the opioid receptors have now been identified as a family of more than 20 opioid peptides that fall into three classes: the endorphins, the enkephalins, and the dynorphins (Table 6.2). Each of these classes are liberated from an inactive pre-propeptide (pre-proopiomelanocortin, preproenkephalin A, and pre-prodynorphin), derived from distinct genes (see Figure 6.14). Opioid precursor processing is carried out by tissue-specific processing enzymes that are packaged into vesicles, along with the precursor peptide, in the Golgi apparatus. Figure 6.15 Neuropeptides vary in length, but usually contain between 3 and 36 amino acids. The sequence of amino acids determines the biological activity of each peptide. 156 Chapter Six TABLE 6.2 Endogenous Opioid Peptides Name Endorphins α-Endorphin α-Neoendorphin β-Endorphin Amino acid sequencea γ-Endorphin Tyr-Gly-Gly-Phe-Met-Thr-Ser-Glu-Lys-Ser-Gln-Thr-Pro-Leu-Val-Thr Tyr-Gly-Gly-Phe-Leu-Arg-Lys-Tyr-Pro-Lys Tyr-Gly-Gly-Phe-Met-Thr-Ser-Glu-Lys-Ser-Gln-Thr-Pro-Leu-Val-Thr-Leu-Phe-Lys-Asn-Ala-IleVal-Lys-Asn-Ala-His-Lys-Gly-Gln Tyr-Gly-Gly-Phe-Met-Thr-Ser-Glu-Lys-Ser-Gln-Thr-Pro-Leu-Val-Thr-Leu Enkephalins Leu-enkephalin Met-enkephalin Tyr-Gly-Gly-Phe-Leu Tyr-Gly-Gly-Phe-Met Dynorphins Dynorphin A Dynorphin B Tyr-Gly-Gly-Phe-Leu-Arg-Arg-Ile-Arg-Pro-Lys-Leu-Lys-Trp-Asp-Asn-Gln Tyr-Gly-Gly-Phe-Leu-Arg-Arg-Gln-Phe-Lys-Val-Val-Thr a Note the initial homology, indicated by italics. Opioid peptides are widely distributed throughout the brain and are often co-localized with other small-molecule neurotransmitters, such as GABA and 5-HT. In general, these peptides tend to be depressants. When injected intracerebrally in experimental animals, they act as analgesics; on the basis of this and other evidence, opioids are likely to be involved in the mechanisms underlying acupuncture-induced analgesia. Opioids are also involved in complex behaviors such as sexual attraction and aggressive/submissive behaviors. They have also been implicated in psychiatric disorders such as schizophrenia and autism, although the evidence for this is debated. Unfortunately, the repeated administration of opioids leads to tolerance and addiction. Virtually all neuropeptides initiate their effects by activating G-proteincoupled receptors. The study of these metabotropic peptide receptors in the brain has been difficult because few specific agonists and antagonists are known. Peptides activate their receptors at low (nM to µM) concentrations compared to the concentrations required to activate receptors for small-molecule neurotransmitters. These properties allow the postsynaptic targets of peptides to be quite far removed from presynaptic terminals and to modulate the electrical properties of neurons that are simply in the vicinity of the site of peptide release. Neuropeptide receptor activation is especially important in regulating the postganglionic output from sympathetic ganglia and the activity of the gut (see Chapter 20). Peptide receptors, particularly the neuropeptide Y receptor, are also implicated in the initiation and maintenance of feeding behavior leading to satiety or obesity. Other behaviors ascribed to peptide receptor activation include anxiety and panic attacks, and antagonists of cholecystokinin receptors are clinically useful in the treatment of these afflictions. Other useful drugs have been developed by targeting the opiate receptors. Three well-defined opioid receptor subtypes (µ, δ, and κ) play a role in reward mechanisms as well as addiction. The µ-opiate receptor has been specifically identified as the primary site for drug reward mediated by opiate drugs Neurotransmitters and Their Receptors 157 Unconventional Neurotransmitters In addition to the conventional neurotransmitters already described, some unusual molecules are also used for signaling between neurons and their targets. These chemical signals can be considered as neurotransmitters because of their roles in interneuronal signaling and because their release from neurons is regulated by Ca2+. However, they are unconventional, in comparison to other neurotransmitters, because they are not stored in synaptic vesicles and are not released from presynaptic terminals via exocytotic mechanisms. In fact, these unconventional neurotransmitters need not be released from presynaptic terminals at all and are often associated with “retrograde” signaling from postsynaptic cells back to presynaptic terminals. • Endocannabinoids are a family of related endogenous signals that interact with cannabinoid receptors. These receptors are the molecular targets of ∆9-tetrahydrocannabinol, the psychoactive component of the marijuana plant, Cannabis (Box F). While some members of this emerging group of chemical signals remain to be determined, anandamide and 2-arachidonylglycerol (2-AG) have been established as endocannabinoids. These signals are unsaturated fatty acid with polar head groups and are produced by enzymatic degradation of membrane lipids (Figure 6.16A,B). Production of endocannabinoids is stimulated by a second messenger signal within postsynaptic neurons, typically a rise in postsynaptic Ca2+ concentration. Although the mechanism of endocannabinoid release is not entirely clear, it is likely that these hyrophobic signals diffuse through the postsynaptic membrane to reach cannabinoid receptors on other nearby cells. Endocannabinoid action is terminated by carrier-mediated transport of these signals back into the postsynaptic neuron. There they are hydrolyzed by the enzyme fatty acid hydrolase (FAAH). At least two types of cannabinoid receptor have been identified, with most actions of endocannabinoids in the CNS mediated by the type termed CB1. CB1 is a G-protein-coupled receptor that is related to the metabotropic receptors for ACh, glutamate, and the other conventional neurotransmitters. Several compounds that are structurally related to endocannabinoids and that bind to the CB1 receptor have been synthesized (see Figure 6.16C). These compounds act as agonists or antagonists of the CB1 receptor and serve as both tools for elucidating the physiological functions of endocannabinoids and as targets for developing therapeutically useful drugs. Endocannabinoids participate in several forms of synaptic regulation. The best-documented action of these agents is to inhibit communication between postsynaptic target cells and their presynaptic inputs. In both the hippocampus and the cerebellum, among other regions, endocannabinoids serve as retrograde signals to regulate GABA release at certain inhibitory terminals. At such synapses, depolarization of the postsynaptic neuron causes a transient reduction in inhibitory postsynaptic responses (Figure 6.17). Depolarization reduces synaptic transmission by elevating the concentration of Ca2+ within the postsynaptic neuron. This rise in Ca2+ triggers synthesis and release of endocannabinoids from the postsynaptic cells. The endocannabinoids then make their way to the presynaptic terminals and bind to CB1 receptors on these terminals. Activation of the CB1 receptors inhibits the amount of GABA released in response to presynaptic action potentials, thereby reducing inhibitory transmission. These mechanisms responsible for the reduction in GABA release are not entirely clear, but probably involve effects on voltage-gated Ca2+ channels and/or K+ channels in the presynaptic neurons. (A) (C) Alkyl O O Acyl C O C O O CH2 CH Phosphatidylethanolamine CH2 O –O N-Acyltransferase C O O C O O P O O O CH3 N WIN 55,212–2 N + NH3 O CH2 CH N-Archidonoyl phosphatidylethanolamine CH2 N O N O –O Phospholipase D P O O N O N NH Cl Cl HO O Rimonabant NH2 Cl Anandamide Figure 6.16 Endocannabinoid signals involved in synaptic transmission. Possible mechanism of production of the endocannabinoids (A) anandamide and (B) 2-AG. (C) Structures of the endocannabinoid receptor agonist WIN 55,212-2 and the antagonist rimonabant. (A,B after Freund et al., 2003; C after Iversen, 2003.) (B) Acyl Arachidonyl O C O O C O CH CH2 O Phosphatidylinositol Phospholipase C CH2 –O O O 1,2-Diacylglycerol (1,2-DAG) C O O C O Phospholipase A1 P O OH OH OH HO OH Inositol CH2 HO O CH C CH O CH2 CH2 O Lysophosphatidylinositol OH CH2 –O P O O 1,2-Diacylglycerol lipase HO O C CH2 CH O CH2 OH 2-Arachidonylglycerol (2-AG) OH OH OH HO OH Lysophospholipase C Neurotransmitters and Their Receptors 159 Inhibitory interneuron (B) (C) Depolarize Vpost Rimonabant–treated 100 Pyramidal neuron After depolarizing Vpost Record Stimulate IPSC (pA) 0 –100 IPSC amplitude (%) (A) Control –200 –300 0 50 100 150 Time (ms) 200 100 75 50 Control 25 0 –20 0 20 40 Time (s) 60 Figure 6.17 Endocannabinoid-mediated retrograde control of GABA release. (A) Experimental arrangement. Stimulation of a presynaptic interneuron causes release of GABA onto a postsynaptic pyramidal neuron. (B) IPSCs elicited by the inhibitory synapse (control) are reduced in amplitude following a brief depolarization of the postsynaptic neuron. This reduction in the IPSC is due to less GABA being released from the presynaptic interneuron. (C) The reduction in IPSC amplitude produced by postsynaptic depolarization lasts a few seconds and is mediated by endocannabinoids, because it is prevented by the endocannabinoid receptor antagonist rimonabant. (B,C after Ohno-Shosaku et al., 2001.) •Nitric oxide (NO) is an unusual but especially interesting chemical signal. NO is a gas that is produced by the action of nitric oxide synthase, an enzyme that converts the amino acid arginine into a metabolite (citrulline) and simultaneously generates NO (Figure 6.18). NO is produced by an enzyme, nitric oxide synthase. Neuronal NO synthase is regulated by Ca2+ binding to the Ca2+ sensor protein calmodulin (see Chapter 7). Once produced, NO can permeate the plasma membrane, meaning that NO generated inside one cell can travel through the extracellular medium and act within nearby cells. Thus, this gaseous signal has a range of influence that extends well beyond the cell of origin, diffusing a few tens of micrometers from its site of production before it is degraded. This property makes NO a Other target cells Arginine CH Guanylyl cyclase GTP COO− + NH3 Nitric oxide Figure 6.18 Synthesis, release, and termination of NO. NO synthase NH CH2 CH2 CH2 NH C NH2 cGMP Nitric oxide O2 Various nitrogen oxides + Citrulline COO− + NH3 CH O CH2 CH2 CH2 NH C NH2 Protein kinase G 160 Chapter Six Box F Marijuana and the Brain Medicinal use of the marijuana plant, Cannabis sativa (Figure A), dates back thousands of years. Ancient civilizations—including both Greek and Roman societies in Europe, as well as Indian and Chinese cultures in Asia—appreciated that this plant was capable of producing relaxation, euphoria, and a number of other psychopharmacological actions. In more recent times, medicinal use of marijuana has largely subsided (although it remains useful in relieving the symptoms of terminal cancer patients); the recreational use of marijuana (Figure B) has become so popular that some societies have decriminalized its use. Understanding the brain mechanisms underlying the actions of marijuana was advanced by the discovery that a cannabinoid, ∆9-tetrahydrocannabinol (THC; Figure C), is the active component of marijuana. This finding led to the development of synthetic derivatives, such as WIN 55,212-2 and rimonabant (see Figure 6.16), that have served as valuable tools for probing the brain actions of THC. Of particular interest is that receptors for these cannabinoids exist in the brain and exhibit marked regional variations in distribution, being especially enriched in the brain areas—such as substantia nigra and caudate putamen—that have been implicated in drug abuse (Figure D). The presence of these brain receptors for cannabinoids led in turn to a search for endogenous cannabinoid compounds in the brain, culiminating in the discovery of endocannabinoids such as 2-AG and anandamide (see Figure 6.16). This path of discovery closely parallels the identification of endogenous opioid peptides, which resulted from the search for endogenous morphine-like compounds in the brain (see text and Table 6.2). Given that THC interacts with brain endocannabinoid receptors, particularly (A) (B) Cannabis sativa (C) CH3 OH ∆9-Tetrahydrocannabinol (THC) H3C CH2 O CH2 CH2 CH2 CH3 CH3 (D) Caudate putamen Hippocampus Cerebellum Substantia nigra (A) Leaf of Cannabis sativa, the marijuana plant. (B) Smoking ground-up Cannabis leaves is a popular method of achieving the euphoric effects of marijuana. (C) Structure of THC (∆9tetrahydrocannabinol), the active ingredient of marijuana. (D) Distribution of brain CB1 receptors, visualized by examining the binding of CP-55,940, a CB1 receptor ligand. (B photo © Henry Diltz/Corbis; C after Iversen, 2003; D courtesy of M. Herkenham, NIMH.) Neurotransmitters and Their Receptors 161 the CB1 receptor, it is likely that such actions are responsible for the behavioral consequences of marijuana use. Indeed, many of the well-documented effects of marijuana are consistent with the distribution and actions of brain CB1 receptors. For example, marijuana effects on perception could be due to CB1 receptors in the neocortex, effects on psychomotor control due to endocannabinoid receptors in the basal ganglia and cerebellum, effects on shortterm memory due to cannabinoid receptors in the hippocampus, and the well-known effects of marijuana on stimulating appetite due to hypothalamic actions. While formal links between these behavioral consequences of marijuana and the underlying brain mechanisms are still being forged, studies of the actions of this drug have shed substantial light on basic synaptic mechanisms, which promise to further elucidate the mode of action of one of the world’s most popular drugs. potentially useful agent for coordinating the activities of multiple cells in a very localized region and may mediate certain forms of synaptic plasticity that spread within small networks of neurons. All of the known actions of NO are mediated within its cellular targets; for this reason, NO often is considered a second messenger rather than a neurotransmitter. Some of these actions of NO are due to the activation of the enzyme guanylyl cyclase, which then produces the second messenger cGMP within target cells (see Chapter 7). Other actions of NO are the result of covalent modification of target proteins via nitrosylation, the addition of a nitryl group to selected amino acids within the proteins. NO decays spontaneously by reacting with oxygen to produce inactive nitrogen oxides. As a result, NO signals last for only a short time, on the order of seconds or less. NO signaling evidently regulates a variety of synapses that also employ conventional neurotransmitters; so far, presynaptic terminals that release glutamate are the best-studied target of NO in the central nervous system. NO may also be involved in some neurological diseases. For example, it has been proposed that an imbalance between nitric oxide and superoxide generation underlies some neurodegenerative diseases. Summary The complex synaptic computations occurring at neural circuits throughout the brain arise from the actions of a large number of neurotransmitters, which act on an even larger number of postsynaptic neurotransmitter receptors. Glutamate is the major excitatory neurotransmitter in the brain, whereas GABA and glycine are the major inhibitory neurotransmitters. The actions of these small-molecule neurotransmitters are typically faster than those of the neuropeptides. Thus, most small-molecule transmitters mediate synaptic transmission when a rapid response is essential, whereas the neuropeptide transmitters, as well as the biogenic amines and some small-molecule neurotransmitters, tend to modulate ongoing activity in the brain or in peripheral target tissues in a more gradual and ongoing way. Two broadly different families of neurotransmitter receptors have evolved to carry out the postsynaptic signaling actions of neurotransmitters. Ionotropic or ligand- References ADAMS, A. R. (1941) Marihuana. Harvey Lect. 37: 168–168. FREUND, T. F., I. KATONA AND D. PIOMELLI (2003) Role of endogenous cannabinoids in synaptic signaling. Physiol. Rev. 83: 1017–1066. GERDEMAN, G. L., J. G. PARTRIDGE, C. R. LUPICA AND D. M. LOVINGER (2003) It could be habit forming: Drugs of abuse and striatal synaptic plasticity. Trends Neurosci. 26: 184–192. IVERSEN, L. (2003) Cannabis and the brain. Brain 126: 1252–1270. MECHOULAM, R. (1970) Marihuana chemistry. Science 168: 1159–1166. 162 Chapter Six gated ion channels combine the neurotransmitter receptor and ion channel in one molecular entity, and therefore give rise to rapid postsynaptic electrical responses. Metabotropic receptors regulate the activity of postsynaptic ion channels indirectly, usually via G-proteins, and induce slower and longer-lasting electrical responses. Metabotropic receptors are especially important in regulating behavior, and drugs targeting these receptors have been clinically valuable in treating a wide range of behavioral disorders. The postsynaptic response at a given synapse is determined by the combination of receptor subtypes, G-protein subtypes, and ion channels that are expressed in the postsynaptic cell. Because each of these features can vary both within and among neurons, a tremendous diversity of transmittermediated effects is possible. Drugs that influence transmitter actions have enormous importance in the treatment of neurological and psychiatric disorders, as well as in a broad spectrum of other medical problems. Additional Reading Reviews BARNES, N. M. AND T. SHARP (1999) A review of central 5-HT receptors and their function. Neuropharm. 38: 1083–1152. BOURIN, M., G. B. BAKER AND J. BRADWEJN (1998) Neurobiology of panic disorder. J. Psychosomatic Res. 44: 163–180. BURNSTOCK, G. (1999) Current status of purinergic signalling in the nervous system. Prog. Brain Res. 120: 3–10. CARLSSON, A. (1987) Perspectives on the discovery of central monoaminergic neurotransmission. Annu. Rev. Neurosci. 10: 19–40. CHANGEUX, J.-P. (1993) Chemical signaling in the brain. Sci. Amer. 269 (May): 58–62. CIVELLI, O. (1998) Functional genomics: The search for novel neurotransmitters and neuropeptides. FEBS Letters 430: 55–58. FREDHOLM, B. B. (1995) Adenosine, adenosine receptors and the actions of caffeine. Pharmacol. Toxicol. 76: 93–101. FREUND, T. F., I. KATONA AND D. PIOMELLI (2003) Role of endogenous cannabinoids in synaptic signaling. Physiol Rev. 83: 1017–1066. HÖKFELT, T. D. AND 10 OTHERS (1987) Coexistence of peptides with classical neurotransmitters. Experientia Suppl. 56: 154–179. HYLAND, K. (1999) Neurochemistry and defects of biogenic amine neurotransmitter metabolism. J. Inher. Metab. Dis. 22: 353–363. INESTROSSA, N. C. AND A. PERELMAN (1989) Distribution and anchoring of molecular forms of acetylcholinesterase. Trends Pharmacol. Sci. 10: 325–329. IVERSEN, L. (2003) Cannabis and the brain. Brain 126: 1252–1270. KOOB, G. F., P. P. SANNA AND F. E. BLOOM (1998) Neuroscience of addiction. Neuron 21: 467–476. KUPFERMANN, I. (1991) Functional studies of cotransmission. Physiol. Rev. 71: 683–732. LAUBE, B., G. MAKSAY, R. SCHEMM AND H. BETZ (2002) Modulation of glycine receptor function: A novel approach for therapeutic intervention at inhibitory synapses? Trends Pharmacol. Sci. 23: 519–527. LOVINGER, D. M. (1999) 5-HT3 receptors and the neural actions of alcohols: An increasingly exciting topic. Neurochem. Internat. 35: 125–30. MACKENZIE, A. B., A. SURPRENANT AND R. A. NORTH (1999) Functional and molecular diversity of purinergic ion channel receptors. Ann. NY Acad. Sci. 868: 716–729. MASSON, J., C. SAGN, M. HAMON AND S. E. MESTIKAWY (1999) Neurotransmitter transporters in the central nervous system. Pharmacol. Rev. 51: 439–464. MELDRUM, B. AND J. GARTHWAITE (1990) Glutamate neurotoxicity may underlie slowly progressive degenerative diseases such as Huntington’s Disease and Alzheimer’s Disease. Trends Pharmacol. Sci. 11: 379–387. NAKANISHI, S. (1992) Molecular diversity of glutamate receptors and implication for brain function. Science 258: 597–603. PERRY, E., M. WALKER, J. GRACE AND R. PERRY (1999) Acetylcholine in mind: A neurotransmitter correlate of consciousness? Trends Neurosci. 22: 273–280. PIERCE, K. L., R. T. PREMONT AND R. J. LEFKOWITZ (2002) Seven-transmembrane receptors. Nature Rev. Mol. Cell Biol. 3: 639–650. SCHWARTZ, J. C., J. M. ARRANG, M. GARBARG, H. POLLARD AND M. RUAT (1991) Histaminergic transmission in the mammalian brain. Physiol. Rev. 71: 1–51. SCHWARTZ, M. W., S. C. WOODS, D. PORTE JR., R. J. SEELEY AND D. G. BASKIN (2000) Central nervous system control of food intake. Nature 404: 661–671. STAMLER, J. S., E. J. TOONE, S. A. LIPTON AND N. J. SUCHER (1997) (S)NO Signals: Translocation, regulation, and a consensus motif. Neuron 18: 691–696. TUCEK, S., J. RICNY AND V. DOLEZAL (1990) Advances in the biology of cholinergic neurons. Adv. Neurol. 51: 109–115. WEBB, T. E. AND E. A. BARNARD (1999) Molecular biology of P2Y receptors expressed in the nervous system. Prog. Brain Res. 120: 23–31. WILSON, R. I. AND R. A. NICOLL (2002) Endocannabinoid signaling in the brain. Science 296: 678–682. Important Original Papers BRENOWITZ, S. D. AND W. G. REGEHR (2003) Calcium dependence of retrograde inhibition by endocannabinoids at synapses onto Purkinji cells. J. Neurosci. 23: 6373–6384. CHAVAS, J. AND A. MARTY (2003) Coexistence of excitatory and inhibitory GABA synapses in the cerebellar interneuron network. J. Neurosci. 23: 2019–2031. CHEN, Z. P., A. LEVY AND S. L. LIGHTMAN (1995) Nucleotides as extracellular signalling molecules. J. Neuroendocrinol. 7: 83–96. CURTIS, D. R., J. W. PHILLIS AND J. C. WATKINS (1959) Chemical excitation of spinal neurons. Nature 183: 611–612. DALE, H. H., W. FELDBERG AND M. VOGT (1936) Release of acetylcholine at voluntary motor nerve endings. J. Physiol. 86: 353–380. DAVIES, P. A. AND 6 OTHERS (1999) The 5-HT3B subunit is a major determinant of serotoninreceptor function. Nature 397: 359–363. GOMEZA, J. AND 6 OTHERS (2003) Inactivation of the glycine transporter 1 gene discloses vital role of glial glycine uptake in glycinergic inhibition. Neuron 40: 785–796. GU, J. G. AND A. B. MACDERMOTT (1997) Activation of ATP P2X receptors elicits glutamate release from sensory neuron synapses. Nature 389: 749–753. HÖKFELT, T., O. JOHANSSON, A. LJUNGDAHL, J. Neurotransmitters and Their Receptors 163 M. LUNDBERG AND M. SCHULTZBERG (1980) Peptidergic neurons. Nature 284: 515–521. HOLLMANN, M., C. MARON AND S. HEINEMANN (1994) N-glycosylation site tagging suggests a three transmembrane domain topology for the glutamate receptor GluR1. Neuron 13: 1331–1343. HUGHES, J., T. W. SMITH, H. W. KOSTERLITZ, L. A. FOTHERGILL, B. A. MORGAN AND H. R. MORRIS (1975) Identification of two related pentapeptides from the brain with potent opiate agonist activity. Nature 258: 577–580. KAUPMANN, K. AND 10 OTHERS (1997) Expression cloning of GABAβ receptors uncovers similarity to metabotropic glutamate receptors. Nature 386: 239–246. KREITZER, A. C. AND W. G. REGEHR (2001) Retrograde inhibition of presynaptic calcium influx by endogenous cannabinoids at excitatory synapses onto Purkinje cells. Neuron 29: 717–727. LEDEBT, C. AND 9 OTHERS (1997) Aggressiveness, hypoalgesia and high blood pressure in mice lacking the adenosine A2a receptor. Nature 388: 674–678. NAVEILHAN, P. AND 10 OTHERS (1999) Normal feeding behavior, body weight and leptin response require the neuropeptide Y Y2 receptor. Nature Med. 5: 1188–1193. OHNO-SHOSAKU, T., T. MAEJIMA, AND M. KANO (2001) Endogenous cannabinoids mediate retrograde signals from depolarized postsynaptic neurons to presynaptic terminals. Neuron 29: 729–738. ROSENMUND, C., Y. STERN-BACH AND C. F. STEVENS (1998) The tetrameric structure of a glutamate receptor channel. Science: 280: 1596–1599. THOMAS, S. A. AND R. D. PALMITER (1995) Targeted disruption of the tyrosine hydroxylase gene reveals that catecholamines are required for mouse fetal development. Nature 374: 640–643. UNWIN, N. (1995) Acetylcholine receptor channels imaged in the open state. Nature 373: 37–43. WANG, Y.M. AND 8 OTHERS (1997) Knockout of the vesicular monoamine transporter 2 gene results in neonatal death and supersensitivity to cocaine and amphetamine. Neuron 19: 1285–1296. Books BRADFORD, H. F. (1986) Chemical Neurobiology. New York: W. H. Freeman. COOPER, J. R., F. E. BLOOM AND R. H. ROTH (2003) The Biochemical Basis of Neuropharmacology. New York: Oxford University Press. FELDMAN, R. S., J. S. MEYER AND L. F. QUENZER (1997) Principles of Neuropharmacology, 2nd Edition. Sunderland, MA: Sinauer Associates. HALL, Z. (1992) An Introduction to Molecular Neurobiology. Sunderland, MA: Sinauer Associates. HILLE, B. (2002) Ion Channels of Excitable Membranes, 3rd Edition. Sunderland, MA: Sinauer Associates. MYCEK, M. J., R. A. HARVEY AND P. C. CHAMPE (2000) Pharmacology, 2nd Edition. Philadelphia, New York: Lippincott/Williams and Wilkins Publishers. NICHOLLS, D. G. (1994) Proteins, Transmitters, and Synapses. Boston: Blackwell Scientific. SIEGEL, G.J., B. W. AGRANOFF, R. W. ALBERS, S. K. FISHER AND M. D. UHLER (1999) Basic Neurochemistry. Philadelphia: Lippincott-Raven. Chapter 7 Molecular Signaling within Neurons Overview As is apparent in the preceding chapters, electrical and chemical signaling mechanisms allow one nerve cell to receive and transmit information to another. This chapter focuses on the related events within neurons and other cells that are triggered by the interaction of a chemical signal with its receptor. This intracellular processing typically begins when extracellular chemical signals, such as neurotransmitters, hormones, and trophic factors, bind to specific receptors located either on the surface or within the cytoplasm or nucleus of the target cells. Such binding activates the receptors and in so doing stimulates cascades of intracellular reactions involving GTP-binding proteins, second messenger molecules, protein kinases, ion channels, and many other effector proteins whose modulation temporarily changes the physiological state of the target cell. These same intracellular signal transduction pathways can also cause longer-lasting changes by altering the transcription of genes, thus affecting the protein composition of the target cells on a more permanent basis. The large number of components involved in intracellular signaling pathways allows precise temporal and spatial control over the function of individual neurons, thereby allowing the coordination of electrical and chemical activity in the related populations of neurons that comprise neural circuits and systems. Strategies of Molecular Signaling Chemical communication coordinates the behavior of individual nerve and glial cells in physiological processes that range from neural differentiation to learning and memory. Indeed, molecular signaling ultimately mediates and modulates all brain functions. To carry out such communication, a series of extraordinarily diverse and complex chemical signaling pathways has evolved. The preceding chapters have described in some detail the electrical signaling mechanisms that allow neurons to generate action potentials for conduction of information. These chapters also described synaptic transmission, a special form of chemical signaling that transfers information from one neuron to another. Chemical signaling is not, however, limited to synapses (Figure 7.1A). Other well-characterized forms of chemical communication include paracrine signaling, which acts over a longer range than synaptic transmission and involves the secretion of chemical signals onto a group of nearby target cells, and endocrine signaling, which refers to the secretion of hormones into the bloodstream where they can affect targets throughout the body. Chemical signaling of any sort requires three components: a molecular signal that transmits information from one cell to another, a receptor molecule 165 166 Chapter Seven (A) Synaptic Paracrine (B) Endocrine Signaling cell Capillary Signal Receptor Target molecules in distant cells Target molecule Activated receptors Target molecules Figure 7.1 Chemical signaling mechanisms. (A) Forms of chemical communication include synaptic transmission, paracrine signaling, and endocrine signaling. (B) The essential components of chemical signaling are: cells that initiate the process by releasing signaling molecules; specific receptors on target cells; second messenger target molecules; and subsequent cellular responses. Target molecules Blood flow Response that transduces the information provided by the signal, and a target molecule that mediates the cellular response (Figure 7.1B). The part of this process that take place within the confines of the target cell is called intracellular signal transduction. A good example of transduction in the context of intercellular communication is the sequence of events triggered by chemical synaptic transmission (see Chapter 5): Neurotransmitters serve as the signal, neurotransmitter receptors serve as the transducing receptor, and the target molecule is an ion channel that is altered to cause the electrical response of the postsynaptic cell. In many cases, however, synaptic transmission activates additional intracellular pathways that have a variety of functional consequences. For example, the binding of the neurotransmitter norepinephrine to its receptor activates GTP-binding proteins, which produces second messengers within the postsynaptic target, activates enzyme cascades, and eventually changes the chemical properties of numerous target molecules within the affected cell. A general advantage of chemical signaling in both intercellular and intracellular contexts is signal amplification. Amplification occurs because individual signaling reactions can generate a much larger number of molecular products than the number of molecules that initiate the reaction. In the case of norepinephrine signaling, for example, a single norepinephrine molecule binding to its receptor can generate many thousands of second messenger molecules (such as cyclic AMP), yielding an amplification of tens of thousands of phosphates transferred to target proteins (Figure 7.2). Similar amplification occurs in all signal transduction pathways. Because the transduction processes often are mediated by a sequential set of enzymatic reactions, each with its own amplification factor, a small number of signal molecules ultimately can activate a very large number of target molecules. Such amplification guarantees that a physiological response is evoked in the face of other, potentially countervailing, influences. Another rationale for these complex signal transduction schemes is to permit precise control of cell behavior over a wide range of times. Some molecular interactions allow information to be transferred rapidly, while others are slower and longer lasting. For example, the signaling cascades associated with synaptic transmission at neuromuscular junctions allow a person to respond to rapidly changing cues, such as the trajectory of a pitched ball, while the slower responses triggered by adrenal medullary hormones (epinephrine and norepinephrine) secreted during a challenging game produce slower (and longer lasting) effects on muscle metabolism (see Chapter 20) and emotional state (see Chapter 29). To encode information that varies so Molecular Signaling within Neurons 167 Receptor G-proteins Amplification Adenylyl cyclase No amplification Amplification Cyclic AMP Protein kinases No amplification widely over time, the concentration of the relevant signaling molecules must be carefully controlled. On one hand, the concentration of every signaling molecule within the signaling cascade must return to subthreshold values before the arrival of another stimulus. On the other hand, keeping the intermediates in a signaling pathway activated is critical for a sustained response. Having multiple levels of molecular interactions facilitates the intricate timing of these events. The Activation of Signaling Pathways The molecular components of these signal transduction pathways are always activated by a chemical signaling molecule. Such signaling molecules can be grouped into three classes: cell-impermeant, cell-permeant, and cellassociated signaling molecules (Figure 7.3). The first two classes are secreted molecules and thus can act on target cells removed from the site of signal synthesis or release. Cell-impermeant signaling molecules typically bind to receptors associated with cell membranes. Hundreds of secreted molecules have now been identified, including the neurotransmitters discussed in Chapter 6, as well as proteins such as neurotrophic factors (see Chapter 22), and peptide hormones such as glucagon, insulin, and various reproductive hormones. These signaling molecules are typically short-lived, either because they are rapidly metabolized or because they are internalized by endocytosis once bound to their receptors. Phosphates transferred to target proteins Amplification Figure 7.2 Amplification in signal transduction pathways. The activation of a single receptor by a signaling molecule, such as the neurotransmitter norepinephrine, can lead to the activation of numerous G-proteins inside cells. These activated proteins can bind to other signaling molecules, such as the enzyme adenylyl cyclase. Each activated enzyme molecule generates a large number of cAMP molecules. cAMP binds to and activates another family of enzymes, protein kinases. These enzymes can then phosphorylate many target proteins. While not every step in this signaling pathway involves amplification, overall the cascade results in a tremendous increase in the potency of the initial signal. 168 Chapter Seven Figure 7.3 Three classes of cell signaling molecules. (A) Cell-impermeant molecules, such as neurotransmitters, cannot readily traverse the plasma membrane of the target cell and must bind to the extracellular portion of transmembrane receptor proteins. (B) Cell-permeant molecules are able to cross the plasma membrane and bind to receptors in the cytoplasm or nucleus of target cells. (C) Cell-associated molecules are presented on the extracellular surface of the plasma membrane. These signals activate receptors on target cells only if they are directly adjacent to the signaling cell. (A) Cell-impermeant molecules Signaling molecules Transmembrane receptors (B) Cell-permeant molecules (C) Cell-associated molecules Signaling molecules Signaling molecules Receptor Intracellular receptor Nucleus Cell-permeant signaling molecules can cross the plasma membrane to act directly on receptors that are inside the cell. Examples include numerous steroid (glucocorticoids, estradiol, and testosterone) and thyroid (thyroxin) hormones, and retinoids. These signaling molecules are relatively insoluble in aqueous solutions and are often transported in blood and other extracellular fluids by binding to specific carrier proteins. In this form, they may persist in the bloodstream for hours or even days. The third group of chemical signaling molecules, cell-associated signaling molecules, are arrayed on the extracellular surface of the plasma membrane. As a result, these molecules act only on other cells that are physically in contact with the cell that carries such signals. Examples include proteins such as the integrins and neural cell adhesion molecules (NCAMs) that influence axonal growth (see Chapter 22). Membrane-bound signaling molecules are more difficult to study, but are clearly important in neuronal development and other circumstances where physical contact between cells provides information about cellular identities. Receptor Types Regardless of the nature of the initiating signal, cellular responses are determined by the presence of receptors that specifically bind the signaling molecules. Binding of signal molecules causes a conformational change in the receptor, which then triggers the subsequent signaling cascade within the affected cell. Given that chemical signals can act either at the plasma membrane or within the cytoplasm (or nucleus) of the target cell, it is not surprising that receptors are actually found on both sides of the plasma membrane. The receptors for impermeant signal molecules are membranespanning proteins. The extracellular domain of such receptors includes the binding site for the signal, while the intracellular domain activates intracellular signaling cascades after the signal binds. A large number of these receptors have been identified and are grouped into families defined by the mechanism used to transduce signal binding into a cellular response (Figure 7.4). Molecular Signaling within Neurons 169 1 Signal binds Ions Figure 7.4 Categories of cellular receptors. Membrane-impermeant signaling molecules can bind to and activate either channel-linked receptors (A), enzyme-linked receptors (B), or G-protein-coupled receptors (C). Membrane permeant signaling molecules activate intracellular receptors (D). (B) Enzyme-linked receptors (A) Channel-linked receptors 1 Signal binds 2 Enzyme activated 2 Channel opens Channel closed Enzyme inactive 3 Ions flow across membrane Substrate Product 3 Enzyme generates product (C) G-protein-coupled receptors (D) Intracellular receptors Signaling molecule 1 Signal binds 2 Receptor β γ Activated receptor regulates transcription 1 Signal binds α β α 2 G-protein binds γ G-protein 3 G-protein activated Receptor Channel-linked receptors (also called ligand-gated ion channels) have the receptor and transducing functions as part of the same protein molecule. Interaction of the chemical signal with the binding site of the receptor causes the opening or closing of an ion channel pore in another part of the same molecule. The resulting ion flux changes the membrane potential of the target cell and, in some cases, can also lead to entry of Ca2+ ions that serve as a second messenger signal within the cell. Good examples of such receptors are the ionotropic neurotransmitter receptors described in Chapters 5 and 6. Enzyme-linked receptors also have an extracellular binding site for chemical signals. The intracellular domain of such receptors is an enzyme whose catalytic activity is regulated by the binding of an extracellular signal. The great majority of these receptors are protein kinases, often tyrosine kinases, that phosphorylate intracellular target proteins, thereby changing the physiological function of the target cells. Noteworthy members of this 170 Chapter Seven group of receptors are the Trk family of neurotrophin receptors (see Chapter 22) and other receptors for growth factors. G-protein-coupled receptors regulate intracellular reactions by an indirect mechanism involving an intermediate transducing molecule, called the GTP-binding proteins (or G-proteins). Because these receptors all share the structural feature of crossing the plasma membrane seven times, they are also referred to as 7-transmembrane receptors (or metabotropic receptors; see Chapter 5). Hundreds of different G-protein-linked receptors have been identified. Well-known examples include the β-adrenergic receptor, the muscarinic type of acetylcholine receptor, metabotropic glutamate receptors, receptors for odorants in the olfactory system, and many types of receptors for peptide hormones. Rhodopsin, a light-sensitive, 7-transmembrane protein in retinal photoreceptors, is another form of G-protein-linked receptor (see Chapter 10). Intracellular receptors are activated by cell-permeant or lipophilic signaling molecules (Figure 7.4D). Many of these receptors lead to the activation of signaling cascades that produce new mRNA and protein within the target cell. Often such receptors comprise a receptor protein bound to an inhibitory protein complex. When the signaling molecule binds to the receptor, the inhibitory complex dissociates to expose a DNA-binding domain on the receptor. This activated form of the receptor can then move into the nucleus and directly interact with nuclear DNA, resulting in altered transcription. Some intracellular receptors are located primarily in the cytoplasm, while others are in the nucleus. In either case, once these receptors are activated they can affect gene expression by altering DNA transcription. G-Proteins and Their Molecular Targets Both G-protein-linked receptors and enzyme-linked receptors can activate biochemical reaction cascades that ultimately modify the function of target proteins. For both these receptor types, the coupling between receptor activation and their subsequent effects are the GTP-binding proteins. There are two general classes of GTP-binding protein (Figure 7.5). Heterotrimeric Gproteins are composed of three distinct subunits (α, β, and γ). There are many different α, β, and γ subunits, allowing a bewildering number of Gprotein permutations. Regardless of the specific composition of the heterotrimeric G-protein, its α subunit binds to guanine nucleotides, either GTP or GDP. Binding of GDP then allows the α subunit to bind to the β and γ subunits to form an inactive trimer. Binding of an extracellular signal to a Gprotein-coupled receptor in turn allows the G-protein to bind to the receptor and causes GDP to be replaced with GTP (Figure 7.5A). When GTP is bound to the G-protein, the α subunit dissociates from the βγ complex and activates the G-protein. Following activation, both the GTP-bound α subunit and the free βγ complex can bind to downstream effector molecules that mediate a variety of responses in the target cell. The second class of GTP-binding proteins are monomeric G-proteins (also called small G-proteins). These monomeric GTPases also relay signals from activated cell surface receptors to intracellular targets such as the cytoskeleton and the vesicle trafficking apparatus of the cell. The first small G-protein was discovered in a virus that causes rat sarcoma tumors and was therefore called ras. Ras is a molecule that helps regulate cell differentiation and proliferation by relaying signals from receptor kinases to the nucleus; the viral form of ras is defective, which accounts for the ability of the virus to cause the uncontrolled cell proliferation that leads to tumors. Since then, a Molecular Signaling within Neurons 171 (A) Heterotrimeric G-proteins (B) Monomeric G-proteins Chemical signaling molecule Chemical signaling molecule Receptor Receptor GDP α β γ GTP GDP GTP GDP α GTP G-protein β α GDP γ Effector protein Active Ras Ras GDP GAP GTP Inactive Pi Pi large number of small GTPases have been identified and can be sorted into five different subfamilies with different functions. For instance, some are involved in vesicle trafficking in the presynaptic terminal or elsewhere in the neuron, while others play a central role in protein and RNA trafficking in and out of the nucleus. Termination of signaling by both heterotrimeric and monomeric G-proteins is determined by hydrolysis of GTP to GDP. The rate of GTP hydrolysis is an important property of a particular G-protein that can be regulated by other proteins, termed GTPase-activating proteins (GAPs). By replacing GTP with GDP, GAPs return G-proteins to their inactive form. GAPs were first recognized as regulators of small G-proteins, but recently similar proteins have been found to regulate the α subunits of heterotrimeric G-proteins. Hence, monomeric and trimeric G-proteins function as molecular timers that are active in their GTP-bound state, and become inactive when they have hydrolized the bound GTP to GDP (Figure 7.5B). Activated G-proteins alter the function of many downstream effectors. Most of these effectors are enzymes that produce intracellular second messengers. Effector enzymes include adenylyl cyclase, guanylyl cyclase, phospholipase C, and others (Figure 7.6). The second messengers produced by these enzymes trigger the complex biochemical signaling cascades discussed in the next section. Because each of these cascades is activated by specific Gprotein subunits, the pathways activated by a particular receptor are determined by the specific identity of the G-protein subunits associated with it. As well as activating effector molecules, G-proteins can also directly bind to and activate ion channels. For example, some neurons, as well as heart muscle cells, have G-protein-coupled receptors that bind acetylcholine. Because these receptors are also activated by the agonist muscarine, they are usually called muscarinic receptors (see Chapters 6 and 20). Activation of muscarinic receptors can open K+ channels, thereby inhibiting the rate at which the neuron fires action potentials, or slowing the heartbeat of muscle GAP Figure 7.5 Types of GTP-binding protein. (A) Heterotrimeric G-proteins are composed of three distinct subunits (α, β, and γ). Receptor activation causes the binding of the G-protein and the α subunit to exchange GDP for GTP, leading to a dissociation of the α and βγ subunits. The biological actions of these Gproteins are terminated by hydrolysis of GTP, which is enhanced by GTPase-activating (GAP) proteins. (B) Monomeric G-proteins use similar mechanisms to relay signals from activated cell surface receptors to intracellular targets. Binding of GTP stimulates the biological actions of these G-proteins, and their activity is terminated by hydrolysis of GTP, which is also regulated by GAP proteins. 172 Chapter Seven Neurotransmitter Receptor Norepinephrine Glutamate Dopamine badrenergic mGluR Dopamine D2 G-protein Effector protein Second messengers Later effectors Target action Gs Adenylyl cyclase Gq Phospholipase C cAMP Diacylglycerol IP3 Protein kinase A Protein kinase C Ca2+ release Increase protein phosphorylation Figure 7.6 Effector pathways associated with G-protein-coupled receptors. In all three examples shown here, binding of a neurotransmitter to such a receptor leads to activation of a G-protein and subsequent recruitment of second messenger pathways. Gs , Gq, and Gi refer to three different types of heterotrimeric G-protein. Increase protein phosphorylation and activate calcium-binding proteins Gi Adenylyl cyclase cAMP Protein kinase A Decrease protein phosphorylation cells. These inhibitory responses are believed to be the result of βγ subunits of G-proteins binding to the K+ channels. The activation of α subunits can also lead to the rapid closing of voltage-gated Ca2+ and Na+ channels. Because these channels carry inward currents involved in generating action potentials, closing them makes it more difficult for target cells to fire (see Chapters 3 and 4). In summary, the binding of chemical signals to their receptors activates cascades of signal transduction events in the cytosol of target cells. Within such cascades, G-proteins serve a pivotal function as the molecular transducing elements that couple membrane receptors to their molecular effectors within the cell. The diversity of G-proteins and their downstream targets leads to many types of physiological responses. By directly regulating the gating of ion channels, G-proteins can influence the membrane potential of target cells. Second Messengers Neurons use many different second messengers as intracellular signals. These messengers differ in the mechanism by which they are produced and removed, as well as their downstream targets and effects (Figure 7.7A). This section summarizes the attributes of some of the principal second messengers. • Calcium. The calcium ion (Ca2+) is perhaps the most common intracellular messenger in neurons. Indeed, few neuronal functions are immune to the influence—direct or indirect—of Ca2+. In all cases, information is transmitted by a transient rise in the cytoplasmic calcium concentration, which Molecular Signaling within Neurons 173 (A) Second messenger Ca2+ Intracellular targets Plasma membrane: Voltage-gated Ca2+ channels Various ligandgated channels Calmodulin Plasma membrane: Na+/Ca2+ exchanger Protein kinases 2+ pump Ca Protein phosphatases Ion channels Endoplasmic reticulum: Synaptotagmin Ca2+ pump Many other Ca2+binding proteins Mitochondria Endoplasmic reticulum: IP3 receptors Figure 7.7 Neuronal second messengers. (A) Mechanisms responsible for producing and removing second messengers, as well as the downstream targets of these messengers. (B) Proteins involved in delivering calcium to the cytoplasm and in removing calcium from the cytoplasm. (C) Mechanisms of production and degradation of cyclic nucleotides. (D) Pathways involved in production and removal of diacylglycerol (DAG) and IP3. Removal mechanisms Sources Ryanodine receptors Cyclic AMP Adenylyl cyclase acts on ATP Protein kinase A Cyclic nucleotidegated channels cAMP phosphodiesterase Cyclic GMP Guanylyl cyclase acts on GTP Protein kinase G Cyclic nucleotidegated channels cGMP phosphodiesterase IP3 Phospholipase C acts on PIP2 IP3 receptors on endoplasmic reticulum Phosphatases Diacylglycerol Phospholipase C acts on PIP2 Protein kinase C Various enzymes (B) Voltage-gated Ca2+ channel Ligand-gated Ca 2+ channel (C) Na+/Ca2+ exchanger Adenylyl cyclase Ca2+ pump Na+ Cyclic nucleotidegated channel Guanylyl cyclase Cyclic nucleotidegated channel H+ ATP ATP GTP cAMP cGMP ADP cAMP phosphodiesterase Endoplasmic reticulum Endoplasmic reticulum PKA cGMP phosphodiesterase AMP [Ca2+]i [Ca2+]i (D) ATP PKG GMP Phosphatidylinositol biphosphate (PIP2) Diacylglycerol Phospholipase C IP3 receptor Ca2+-binding effector proteins Ryanodine receptor Mitochondrion ADP Ca2+ pump Ca2+binding buffer proteins PKC Phosphatases Inositol IP3 Ca2+ IP3 receptors 174 Chapter Seven allows Ca2+ to bind to a large number of Ca2+-binding proteins that serve as molecular targets. One of the most thoroughly studied targets of Ca2+ is calmodulin, a Ca2+-binding protein abundant in the cytosol of all cells. Binding of Ca2+ to calmodulin activates this protein, which then initiates its effects by binding to still other downstream targets, such as protein kinases. Ordinarily the concentration of Ca2+ ions in the cytosol is extremely low, typically 50–100 nanomolar (10–9 M). The concentration of Ca2+ ions outside neurons—in the bloodstream or cerebrospinal fluid, for instance—is several orders of magnitude higher, typically several millimolar (10–3 M). This steep Ca2+ gradient is maintained by a number of mechanisms (Figure 7.7B). Most important in this maintenance are two proteins that translocate Ca2+ from the cytosol to the extracellular medium: an ATPase called the calcium pump, and an Na+/Ca2+ exchanger, which is a protein that replaces intracellular Ca2+ with extracellular sodium ions (see Chapter 4). In addition to these plasma membrane mechanisms, Ca2+ is also pumped into the endoplasmic reticulum and mitochondria. These organelles can thus serve as storage depots of Ca2+ ions that are later released to participate in signaling events. Finally, nerve cells contain other Ca2+-binding proteins—such as calbindin—that serve as Ca2+ buffers. Such buffers reversibly bind Ca2+ and thus blunt the magnitude and kinetics of Ca2+ signals within neurons. The Ca2+ ions that act as intracellular signals enter cytosol by means of one or more types of Ca2+-permeable ion channels (see Chapter 4). These can be voltage-gated Ca2+ channels or ligand-gated channels in the plasma membrane, both of which allow Ca2+ to flow down the Ca2+ gradient and into the cell from the extracellular medium. In addition, other channels allow Ca2+ to be released from the interior of the endoplasmic reticulum into the cytosol. These intracellular Ca2+-releasing channels are gated, so they can be opened or closed in response to various intracellular signals. One such channel is the inositol trisphosphate (IP3) receptor. As the name implies, these channels are regulated by IP3, a second messenger described in more detail below. A second type of intracellular Ca2+-releasing channel is the ryanodine receptor, named after a drug that binds to and partially opens these receptors. Among the biological signals that activate ryanodine receptors are cytoplasmic Ca2+ and, at least in muscle cells, depolarization of the plasma membrane. These various mechanisms for elevating and removing Ca2+ ions allow precise control of both the timing and location of Ca2+ signaling within neurons, which in turn permit Ca2+ to control many different signaling events. For example, voltage-gated Ca2+ channels allow Ca2+ concentrations to rise very rapidly and locally within presynaptic terminals to trigger neurotransmitter release, as already described in Chapter 5. Slower and more widespread rises in Ca2+ concentration regulate a wide variety of other responses, including gene expression in the cell nucleus. • Cyclic nucleotides. Another important group of second messengers are the cyclic nucleotides, specifically cyclic adenosine monophosphate (cAMP) and cyclic guanosine monophosphate (cGMP) (Figure 7.7C). Cyclic AMP is a derivative of the common cellular energy storage molecule, ATP. Cyclic AMP is produced when G-proteins activate adenylyl cyclase in the plasma membrane. This enzyme converts ATP into cAMP by removing two phosphate groups from the ATP. Cyclic GMP is similarly produced from GTP by the action of guanylyl cyclase. Once the intracellular concentration of cAMP or cGMP is elevated, these nucleotides can bind to two different classes of targets. The most common targets of cyclic nucleotide action are protein kinases, either the cAMP-dependent protein kinase (PKA) or the cGMP-dependent Molecular Signaling within Neurons 175 protein kinase (PKG). These enzymes mediate many physiological responses by phosphorylating target proteins, as described in the following section. In addition, cAMP and cGMP can bind to certain ligand-gated ion channels, thereby influencing neuronal signaling. These cyclic nucleotide-gated channels are particularly important in phototransduction and other sensory transduction processes, such as olfaction. Cyclic nucleotide signals are degraded by phosphodiesterases, enzymes that cleave phosphodiester bonds and convert cAMP into AMP or cGMP into GMP. • Diacylglycerol and IP3. Remarkably, membrane lipids can also be converted into intracellular second messengers (Figure 7.7D). The two most important messengers of this type are produced from phosphatidylinositol bisphosphate (PIP2). This lipid component is cleaved by phospholipase C, an enzyme activated by certain G-proteins and by calcium ions. Phospholipase C splits the PIP2 into two smaller molecules that each act as second messengers. One of these messengers is diacylglycerol (DAG), a molecule that remains within the membrane and activates protein kinase C, which phosphorylates substrate proteins in both the plasma membrane and elsewhere. The other messenger is inositol trisphosphate (IP3), a molecule that leaves the cell membrane and diffuses within the cytosol. IP3 binds to IP3 receptors, channels that release calcium from the endoplasmic reticulum. Thus, the action of IP3 is to produce yet another second messenger (perhaps a third messenger, in this case!) that triggers a whole spectrum of reactions in the cytosol. The actions of DAG and IP3 are terminated by enzymes that convert these two molecules into inert forms that can be recycled to produce new molecules of PIP2. Second Messenger Targets: Protein Kinases and Phosphatases As already mentioned, second messengers typically regulate neuronal functions by modulating the phosphorylation state of intracellular proteins (Figure 7.8). Phosphorylation (the addition of phosphate groups) rapidly and reversibly changes protein function. Proteins are phosphorylated by a wide variety of protein kinases; phosphate groups are removed by other enzymes called protein phosphatases. The degree of phosphorylation of a target protein thus reflects a balance between the competing actions of protein kinases and phosphatases, thus integrating a host of cellular signaling pathways. The substrates of protein kinases and phosphatases include enzymes, neurotransmitter receptors, ion channels, and structural proteins. ATP Protein Pi Second messengers Second messengers ADP Protein kinase Phosphoprotein Pi Protein phosphatase Figure 7.8 Regulation of cellular proteins by phosphorylation. Protein kinases transfer phosphate groups (Pi) from ATP to serine, threonine, or tyrosine residues on substrate proteins. This phosphorylation reversibly alters the structure and function of cellular proteins. Removal of the phosphate groups is catalyzed by protein phosphatases. Both kinases and phosphatases are regulated by a variety of intracellular second messengers. 176 Chapter Seven Protein kinases and phosphatases typically act either on the serine and threonine residues (Ser/Thr kinases or phosphatases) or the tyrosine residues (Tyr kinases or phosphatases) of their substrates. Some of these enzymes act specifically on only one or a handful of protein targets, while others are multifunctional and have a broad range of substrate proteins. The activity of protein kinases and phosphatases can be regulated either by second messengers, such as cAMP or Ca2+, or by extracellular chemical signals, such as growth factors (see Chapter 22). Typically, second messengers activate Ser/Thr kinases, whereas extracellular signals activate Tyr kinases. Although thousands of protein kinases are expressed in the brain, a relatively small number function as regulators of neuronal signaling. • cAMP-dependent protein kinase (PKA). The primary effector of cAMP is the cAMP-dependent protein kinase (PKA). PKA is a tetrameric complex of two catalytic subunits and two inhibitory (regulatory) subunits. cAMP activates PKA by binding to the regulatory subunits and causing them to release active catalytic subunits. Such displacement of inhibitory domains is a general mechanism for activation of several protein kinases by second messengers (Figure 7.9A). The catalytic subunit of PKA phosphorylates serine and threonine residues of many different target proteins. Although this subunit is similar to the catalytic domains of other protein kinases, distinct amino acids allow the PKA to bind to specific target proteins, thus allowing only those targets to be phosphorylated in response to intracellular cAMP signals. • Ca2+/calmodulin-dependent protein kinase type II (CaMKII). Ca2+ ions binding to calmodulin can regulate protein phosphorylation/dephosphorylation. In neurons, the most abundant Ca2+/calmodulin-dependent protein kinase is CaMKII, a multifunctional Ser/Thr protein kinase. CaMKII is composed of approximately 14 subunits, which in the brain are the α and β types. Each subunit contains a catalytic domain and a regulatory domain, as well as other domains that allow the enzyme to oligomerize and target to the proper region within the cell. Ca2+/calmodulin activates CaMKII by displacing the inhibitory domain from the catalytic site (Figure 7.9B). CaMKII phosphorylates a large number of substrates, including ion channels and other proteins involved in intracellular signal transduction. • Protein kinase C (PKC). Another important group of Ser/Thr protein kinases is protein kinase C (PKC). PKCs are diverse monomeric kinases activated by the second messengers DAG and Ca2+. DAG causes PKC to move from the cytosol to the plasma membrane, where it also binds Ca2+ and phosphatidylserine, a membrane phospholipid (Figure 7.9C). These events relieve autoinhibition and cause PKC to phosphorylate various protein substrates. PKC also diffuses to sites other than the plasma membrane—such as the cytoskeleton, perinuclear sites, and the nucleus—where it phosphorylates still other substrate proteins. Prolonged activation of PKC can be accomplished with phorbol esters, tumor-promoting compounds that activate PKC by mimicking DAG. • Protein tyrosine kinases. Two classes of protein kinases transfer phosphate groups to tyrosine residues on substrate proteins. Receptor tyrosine kinases are transmembrane proteins with an extracellular domain that binds to protein ligands (growth factors, neurotrophic factors, or cytokines) and an intracellular catalytic domain that phosphorylates the relevant substrate proteins. Non-receptor tyrosine kinases are cytoplasmic or membrane-associated enzymes that are indirectly activated by extracellular signals. Tyrosine phosphorylation is less common than Ser/Thr phosphorylation, and it often serves to recruit signaling molecules to the phosphorylated protein. Tyrosine Molecular Signaling within Neurons 177 (A) PKA Inactive Catalytic domains Regulatory domain Phosphorylates substrates Active cAMP (B) CaMKII Inactive Phosphorylates substrates Active Ca2+/CaM Catalytic domain Regulatory domain (C) PKC Inactive Active DAG Ca2+ Regulatory domain Catalytic domain Phosphorylates substrates PS kinases are particularly important for cell growth and differentiation (see Chapters 21 and 22). • Mitogen-activated protein kinase (MAPK). In addition to protein kinases that are directly activated by second messengers, some of these molecules can be activated by other signals, such as phosphorylation by another protein kinase. Important examples of such protein kinases are the mitogen-activated protein kinases (MAPKs), also called extracellular signal-regulated kinases (ERKs). MAPKs were first identified as participants in the control of cell growth and are now known to have many other signaling functions. Figure 7.9 Mechanism of activation of protein kinases. Protein kinases contain several specialized domains with specific functions. Each of the kinases has homologous catalytic domains responsible for transferring phosphate groups to substrate proteins. These catalytic domains are kept inactive by the presence of an autoinhibitory domain that occupies the catalytic site. Binding of second messengers, such as cAMP, DAG, and Ca2+, to the appropriate regulatory domain of the kinase removes the autoinhibitory domain and allows the catalytic domain to be activated. For some kinases, such as PKC and CaMKII, the autoinhibitory and catalytic domains are part of the same molecule. For other kinases, such as PKA, the autoinhibitory domain is a separate subunit. 178 Chapter Seven MAPKs are normally inactive in neurons but become activated when they are phosphorylated by other kinases. In fact, MAPKs are part of a kinase cascade in which one protein kinase phosphorylates and activates the next protein kinase in the cascade. The extracellular signals that trigger these kinase cascades are often extracellular growth factors that bind to receptor tyrosine kinases that, in turn, activate monomeric G-proteins such as ras. Once activated, MAPKs can phosphorylate transcription factors, proteins that regulate gene expression. Among the wide variety of other MAPK substrates are various enzymes, including other protein kinases, and cytoskeletal proteins. The best-characterized protein phosphatases are the Ser/Thr phosphatases PP1, PP2A, and PP2B (also called calcineurin). In general, protein phosphatases display less substrate specificity than protein kinases. Their limited specificity may arise from the fact that the catalytic subunits of the three major protein phosphatases are highly homologous, though each still associates with specific targeting or regulatory subunits. PP1 dephosphorylates a wide array of substrate proteins and is probably the most prevalent Ser/Thr protein phosphatase in mammalian cells. PP1 activity is regulated by several inhibitory proteins expressed in neurons. PP2A is a multisubunit enzyme with a broad range of substrates that overlap with PP1. PP2B, or calcineurin, is present at high levels in neurons. A distinctive feature of this phosphatase is its activation by Ca2+/calmodulin. PP2B is composed of a catalytic and a regulatory subunit. Ca2+/calmodulin activates PP2B primarily by binding to the catalytic subunit and displacing the inhibitory regulatory domain. PP2B generally does not have the same molecular targets as CaMKII, even though both enzymes are activated by Ca2+/calmodulin. In summary, activation of membrane receptors can elicit complex cascades of enzyme activation, resulting in second messenger production and protein phosphorylation or dephosphorylation. These cytoplasmic signals produce a variety of rapid physiological responses by transiently regulating enzyme activity, ion channels, cytoskeletal proteins, and many other cellular processes. In addition, such signals can propagate to the nucleus to cause long-lasting changes in gene expression. Nuclear Signaling Second messengers elicit prolonged changes in neuronal function by promoting the synthesis of new RNA and protein. The resulting accumulation of new proteins requires at least 30–60 minutes, a time frame that is orders of magnitude slower than the responses mediated by ion fluxes or phosphorylation. Likewise, the reversal of such events requires hours to days. In some cases, genetic “switches” can be thrown to permanently alter a neuron, as in neuronal differentiation (see Chapter 21). The amount of protein present in cells is determined primarily by the rate of transcription of DNA into RNA (Figure 7.10). The first step in RNA synthesis is the decondensation of the structure of chromatin to provide binding sites for the RNA polymerase complex and for transcriptional activator proteins, also called transcription factors. Transcriptional activator proteins attach to binding sites that are present on the DNA molecule near the start of the target gene sequence; they also bind to other proteins that promote unwrapping of DNA. The net result of these actions is to allow RNA polymerase, an enzyme complex, to assemble on the promoter region of the DNA and begin transcription. In addition to clearing the promoter for RNA polymerase, activator proteins can stimulate transcription by interacting Molecular Signaling within Neurons 179 (A) (B) Chromosome Beads-on-a-string chromatin UAS Binding of transcriptional activator protein Transcriptional activator protein UAS Co-activator complex UAS Binding of co-activator complex Binding of RNA polymerase RNA polymerase and associated factors UAS Condensed chromatin Start site of RNA Transcription begins with the RNA polymerase complex or by interacting with other activator proteins that influence the polymerase. Intracellular signal transduction cascades regulate gene expression by converting transcriptional activator proteins from an inactive state to an active state in which they are able to bind to DNA. This conversion comes about in several ways. The key activator proteins and the mechanisms that allow them to regulate gene expression in response to signaling events are briefly summarized in the following sections. • CREB. The cAMP response element binding protein, usually abbreviated CREB, is a ubiquitous transcriptional activator (Figure 7.11). CREB is normally bound to its binding site on DNA (called the cAMP response element, or CRE), either as a homodimer or bound to another, closely related transcription factor. In unstimulated cells, CREB is not phosphorylated and has little or no transcriptional activity. However, phosphorylation of CREB greatly potentiates transcription. Several signaling pathways are capable of causing CREB to be phosphorylated. Both PKA and the ras pathway, for example, can phosphorylate CREB. CREB can also be phosphorylated in response to increased intracellular calcium, in which case the CRE site is also called the CaRE (calcium response element) site. The calcium-dependent phosphorylation of CREB is primarily caused by Ca2+/calmodulin kinase IV (a relative of CaMKII) and by MAP kinase, which leads to prolonged CREB phosphorylation. CREB phosphorylation must be maintained long enough for transcription to ensue, even though neuronal electrical activity only tran- Figure 7.10 Steps involved in transcription of DNA into RNA. Condensed chromatin (A) is decondensed into a beads-on-a-DNA-string array (B) in which an upstream activator site (UAS) is free of proteins and is bound by a sequence-specific transcriptional activator protein (transcription factor). The transcriptional activator protein then binds co-activator complexes that enable the RNA polymerase with its associated factors to bind at the start site of transcription and initiate RNA synthesis. 180 Chapter Seven Ca2+ channel G-protein-coupled receptor Receptor tyrosine kinase Ca2+ Electrical signal Outside cell γ α β Heterotrimeric G-protein Inside cell Adenylate cyclase Ca2+ cAMP ras Protein kinase A MAP kinase Newly synthesized protein, e.g., enzyme, structural protein, channels Ca2+/calmodulin kinase IV Translation mRNA mRNA CREB Pi Pi Transcription DNA CRE/CaRE Figure 7.11 Transcriptional regulation by CREB. Multiple signaling pathways converge by activating kinases that phosphorylate CREB. These include PKA, Ca2+/calmodulin kinase IV, and MAP kinase. Phosphorylation of CREB allows it to bind co-activators (not shown in the figure), which then stimulate RNA polymerase to begin synthesis of RNA. RNA is then processed and exported to the cytoplasm, where it serves as mRNA for translation into protein. RNA polymerase Target genes Nucleus siently raises intracellular calcium concentration. Such signaling cascades can potentiate CREB-mediated transcription by inhibiting a protein phosphatase that dephosphorylates CREB. CREB is thus an example of the convergence of multiple signaling pathways onto a single transcriptional activator. Many genes whose transcription is regulated by CREB have been identified. CREB-sensitive genes include the immediate early gene, c-fos (see below), the neurotrophin BDNF (see Chapter 22), the enzyme tyrosine hydroxylase (which is important for synthesis of catecholamine neurotransmitters; see Chapter 6), and many neuropeptides (including somatostatin, enkephalin, and corticotropin releasing hormone). CREB also is thought to mediate long-lasting changes in brain function. For example, CREB has been implicated in spatial learning, behavioral sensitization, long-term memory of odorant-conditioned behavior, and long-term synaptic plasticity (see Chapters 23 and 24). Molecular Signaling within Neurons 181 • Nuclear receptors. Nuclear receptors for membrane-permeant ligands also are transcriptional activators. The receptor for glucocorticoid hormones illustrates one mode of action of such receptors. In the absence of glucocorticoid hormones, the receptors are located in the cytoplasm. Binding of glucocorticoids causes the receptor to unfold and move to the nucleus, where it binds a specific recognition site on the DNA. This DNA binding activates the relevant RNA polymerase complex to initiate transcription and subsequent gene expression. Thus, a critical regulatory event for steroid receptors is their translocation to the nucleus to allow DNA binding. The receptors for thyroid hormone (TH) and other non-steroid nuclear receptors illustrate a second mode of regulation. In the absence of TH, the receptor is bound to DNA and serves as a potent repressor of transcription. Upon binding TH, the receptor undergoes a conformational change that ultimately opens the promoter for polymerase binding. Hence, TH binding switches the receptor from being a repressor to being an activator of transcription. • c-fos. A different strategy of gene regulation is apparent in the function of the transcriptional activator protein, c-fos. In resting cells, c-fos is present at a very low concentration. However, stimulation of the target cell causes cfos to be synthesized, and the amount of this protein rises dramatically over 30–60 minutes. Therefore, c-fos is considered to be an immediate early gene because its synthesis is directly triggered by the stimulus. Once synthesized, c-fos protein can act as a transcriptional activator to induce synthesis of second-order genes. These are termed delayed response genes because their activity is delayed by the fact that an immediate early gene—c-fos in this case—needs to be activated first. Multiple signals converge on c-fos, activating different transcription factors that bind to at least three distinct sites in the promoter region of the gene. The regulatory region of the c-fos gene contains a binding site that mediates transcriptional induction by cytokines and ciliary neurotropic factor. Another site is targeted by growth factors such as neurotrophins through ras and protein kinase C, and a CRE/CaRE that can bind to CREB and thereby respond to cAMP or calcium entry resulting from electrical activity. In addition to synergistic interactions among these c-fos sites, transcriptional signals can be integrated by converging on the same activator, such as CREB. Nuclear signaling events typically result in the generation of a large and relatively stable complex composed of a functional transcriptional activator protein, additional proteins that bind to the activator protein, and the RNA polymerase and associated proteins bound at the start site of transcription. Most of the relevant signaling events act to “seed” this complex by generating an active transcriptional activator protein by phosphorylation, by inducing a conformational change in the activator upon ligand binding, by fostering nuclear localization, by removing an inhibitor, or simply by making more activator protein. Examples of Neuronal Signal Transduction Understanding the general properties of signal transduction processes at the plasma membrane, in the cytosol, and within the nucleus make it possible to consider how these processes work in concert to mediate specific functions in the brain. Three important signal transduction pathways illustrate some of the roles of intracellular signal transduction processes in the nervous system. 182 Chapter Seven • NGF/TrkA. The first of these is signaling by the nerve growth factor (NGF). This protein is a member of the neurotrophin growth factor family and is required for the differentiation, survival, and synaptic connectivity of sympathetic and sensory neurons (see Chapter 22). NGF works by binding to a high-affinity tyrosine kinase receptor, TrkA, found on the plasma membrane of these target cells (Figure 7.12). NGF binding causes TrkA receptors to dimerize, and the intrinsic tyrosine kinase activity of each receptor then phosphorylates its partner receptor. Phosphorylated TrkA receptors trigger the ras cascade, resulting in the activation of multiple protein kinases. Some of these kinases translocate to the nucleus to activate transcriptional activators, such as CREB. This ras-based component of the NGF pathway is primarily responsible for inducing and maintaining differentiation of NGF-sensitive neurons. Phosphorylation of TrkA also causes this receptor to stimulate the activity of phospholipase C, which increases production of IP3 and DAG. IP3 induces release of Ca2+ from the endoplasmic reticulum, and diacylglycerol activates PKC. These two second messengers appear to target many of the same downstream effectors as ras. Finally, activation of TrkA receptors also causes activation of other protein kinases (such as Akt kinase) that inhibit cell death. This pathway, therefore, primarily mediates the NGFdependent survival of sympathetic and sensory neurons described in Chapter 22. • Long-term depression (LTD). The interplay between several intracellular signals can be observed at the excitatory synapses that innervate Purkinje NGF dimer TrkA Figure 7.12 Mechanism of action of NGF. NGF binds to a high-affinity tyrosine kinase receptor, TrkA, on the plasma membrane to induce phosphorylation of TrkA at two different tyrosine residues. These phosphorylated tyrosines serve to tether various adapter proteins or phospholipase C (PLC), which, in turn, activate three major signaling pathways: the PI 3 kinase pathway leading to activation of Akt kinase, the ras pathway leading to MAP kinases, and the PLC pathway leading to release of intracellular Ca2+ and activation of PKC. The ras and PLC pathways primarily stimulate processes responsible for neuronal differentiation, while the PI 3 kinase pathway is primarily involved in cell survival. Pi Pi Pi Pi Outside cell Inside cell PI 3 kinase pathway ras pathway PLC pathway Adapter proteins GEF PLC PI 3 kinase Akt kinase ras Kinases MAP Kinase Cell survival IP3 DAG Ca2+ release PKC from ER Neurite outgrowth and neuronal differentiation Molecular Signaling within Neurons 183 cells in the cerebellum. These synapses are central to information flow through the cerebellar cortex, which in turn helps coordinate motor movements (see Chapter 18). One of the synapses is between the parallel fibers (PFs) and their Purkinje cell targets. LTD is a form of synaptic plasticity that causes the PF synapses to become less effective (see Chapter 24). When PFs are active, they release the neurotransmitter glutamate onto the dendrites of Purkinje cells. This activates AMPA-type receptors, which are ligand-gated ion channels (see Chapter 6), and causes a small EPSP that briefly depolarizes the Purkinje cell. In addition to this electrical signal, PF synaptic transmission also generates two second messengers within the Purkinje cell (Figure 7.13). The glutamate released by PFs activates metabotropic glutamate receptors, which stimulates phospholipase C to produce IP3 and DAG. When the PF synapses alone are active, these intracellular signals are insufficient to open IP3 receptors or to stimulate PKC. LTD is induced when PF synapses are activated at the same time as the glutamatergic climbing fiber synapses that also innervate Purkinje cells. The climbing fiber synapses produce large EPSPs that strongly depolarize the membrane potential of the Purkinje cell. This depolarization allows Ca2+ to Presynaptic terminal of parallel fiber Na+ mGluR Glutamate AMPA receptor Na+ Phospholipase C Long-term depression DAG PKC PIP2 Dendritic spine Ca2+ Climbing fiber depolarizes VM IP3 Release Ca2+ Ca2+ Endoplasmic reticulum Figure 7.13 Signaling at cerebellar parallel fiber synapses. Glutamate released by parallel fibers activates both AMPAtype and metabotropic receptors. The latter produces IP3 and DAG within the Purkinje cell. When paired with a rise in Ca2+ associated with activity of climbing fiber synapses, the IP3 causes Ca2+ to be released from the endoplasmic reticulum, while Ca2+ and DAG together activate protein kinase C. These signals together change the properties of AMPA receptors to produce LTD. 184 Chapter Seven enter the Purkinje cell via voltage-gated Ca2+ channels. When both synapses are simultaneously activated, the rise in intracellular Ca2+ concentration caused by the climbing fiber synapse enhances the sensitivity of IP3 receptors to the IP3 produced by PF synapses and allows the IP3 receptors within the Purkinje cell to open. This releases Ca2+ from the endoplasmic reticulum and further elevates Ca2+ concentration locally near the PF synapses. This larger rise in Ca2+, in conjunction with the DAG produced by the PF synapses, activates PKC. PKC in turn phosphorylates a number of substrate proteins. Ultimately, these signaling processes change AMPA-type receptors at the PF synapse, so that these receptors produce smaller electrical signals in response to the glutamate released from the PFs. This weakening of the PF synapse is the final cause of LTD. In short, transmission at Purkinje cell synapses produces brief electrical signals and chemical signals that last much longer. The temporal interplay between these signals allows LTD to occur only when both PF and climbing fiber synapses are active. The actions of IP3, DAG and Ca2+ also are restricted to small parts of the Purkinje cell dendrite, which is a more limited spatial range than the EPSPs, which spread throughout the entire dendrite and cell body of the Purkinje cell. Thus, in contrast to the electrical signals, the second messenger signals can impart precise information about the location of active synapses and allow LTD to occur only in the vicinity of active PFs. • Phosphorylation of tyrosine hydroxylase. A third example of intracellular signaling in the nervous system is the regulation of the enzyme tyrosine hydroxylase. Tyrosine hydroxylase governs the synthesis of the catecholamine neurotransmitters: dopamine, norepinephrine, and epinephrine (see Chapter 6). A number of signals, including electrical activity, other neurotransmitters, and NGF, increase the rate of catecholamine synthesis by increasing the catalytic activity of tyrosine hydroxylase (Figure 7.14). The rapid increase of tyrosine hydroxylase activity is largely due to phosphorylation of this enzyme. Tyrosine hydroxylase is a substrate for several protein kinases, including PKA, CaMKII, MAP kinase, and PKC. Phosphorylation causes conformational changes that increase the catalytic activity of tyrosine hydroxylase. Stimuli that elevate cAMP, Ca2+, or DAG can all increase tyrosine hydroxylase activity and thus increase the rate of catecholamine biosynthesis. This regulation by several different signals allows for close control of tyrosine hydroxylase activity, and illustrates how several different pathways can converge to influence a key enzyme involved in synaptic transmission. Summary A diversity of signal transduction pathways exist within all neurons. Activation of these pathways typically is initiated by chemical signals such as neurotransmitters and hormones. These molecules bind to receptors that include ligand-gated ion channels, G-protein-coupled receptors and tyrosine kinase receptors. Many of these receptors activate either heterotrimeric or monomeric G-proteins that regulate intracellular enzyme cascades and/or ion channels. A common outcome of the activation of these receptors is the production of second messengers, such as cAMP, Ca2+, and IP3, that bind to effector enzymes. Particularly important effectors are protein kinases and phosphatases that regulate the phosphorylation state of their substrates, and thus their function. These substrates can be metabolic enzymes or other signal transduction molecules, such as ion channels, protein kinases, or transcription factors that regulate gene expression. Examples of transcription Molecular Signaling within Neurons 185 1 Action potential Tyrosine hydroxylase 4 Activation of protein kinases 5 Tyrosine hydroxlyase phosphorylated 3 Activation of second messengers Protein kinase 6 Increase in catecholamine synthesis Ca2+ Pi Ca2+ Calcium channel Figure 7.14 Regulation of tyrosine hydroxylase by protein phosphorylation. This enzyme governs the synthesis of the catecholamine neurotransmitters and is stimulated by a number of intracellular signals. In the example shown here, neuronal electrical activity (1) causes influx of Ca2+ (2). The resultant rise in intracellular Ca2+ concentration (3) activates protein kinases (4), which phosphorylates tyrosine hydroxylase (5) to stimulate catecholamine synthesis (6). This, in turn, increases release of catecholamines (7) and enhances the postsynaptic response produced by the synapse (8). 2 Calcium influx 7 Increase in transmitter release 8 Increase in postsynaptic response factors include CREB, steroid hormone receptors, and c-fos. This plethora of molecular components allows intracellular signal transduction pathways to generate responses over a wide range of times and distances, greatly augmenting and refining the information-processing ability of neuronal circuits and ultimately systems. Additional Reading Reviews AUGUSTINE, G. J., F. SANTAMARIA AND K. TANAKA (2003) Local calcium signaling in neurons. Neuron 40: 331–346. DEISSEROTH, K., P. G. MERMELSTEIN, H. XIA AND R. W. TSIEN (2003) Signaling from synapse to nucleus: The logic behind the mechanisms. Curr. Opin. Neurobiol. 13: 354–365. EXTON, J. H. (1998) Small GTPases. J. Biol. Chem. 273: 19923. FISCHER, E. H. (1999) Cell signaling by protein tyrosine phosphorylation. Adv. Enzyme Regul. Review 39: 359–369. FRIEDMAN, W. J. AND L. A. GREENE (1999) Neurotrophin signaling via Trks and p75. Exp. Cell Res. 253: 131–142. GILMAN, A. G. (1984) G proteins and dual control of adenylate cyclase. Cell 36: 577–579. GRAVES J. D. AND E. G. KREBS (1999) Protein phosphorylation and signal transduction. Pharmacol. Ther. 82: 111–121. KENNEDY, M. B. (2000) Signal-processing machines at the postsynaptic density. Science 290: 750–754. KUMER, S. AND K. VRANA (1996) Intricate regulation of tyrosine hydroxylase activity and gene expression. J. Neurochem. 67: 443–462. LEVITAN, I. B. (1999) Modulation of ion channels by protein phosphorylation. How the brain works. Adv. Second Mess. Phosphoprotein Res. 33: 3–22. NEER, E. J. (1995) Heterotrimeric G proteins: Organizers of transmembrane signals. Cell 80: 249–257. RODBELL, M. (1995) Nobel Lecture. Signal transduction: Evolution of an idea. Bioscience Reports 15: 117–133. SHENG, M. AND M. J. KIM (2002) Postsynaptic signaling and plasticity mechanisms. Science 298: 776–780. WEST, A. E. AND 8 OTHERS (2001) Calcium regulation of neuronal gene expression. Proc. Natl. Acad. Sci. USA 98: 11024–11031. 186 Chapter Seven Important Original Papers BACSKAI, B. J. AND 6 OTHERS (1993) Spatially resolved dynamics of cAMP and protein kinase A subunits in Aplysia sensory neurons. Science 260: 222–226. BURGESS, G. M., P. P. GODFREY, J. S. MCKINNEY, M. J. BERRIDGE, R. F. IRVINE AND J.W. PUTNEY JR. (1984) The second messenger linking receptor activation to internal Ca release in liver. Nature 309: 63–66. CONNOR, J. A. (1986) Digital imaging of free calcium changes and of spatial gradients in growing processes in single, mammalian central nervous system cells. Proc. Natl. Acad. Sci. USA 83: 6179–6183. DE KONINCK, P. AND H. SCHULMAN (1998) Sensitivity of CaM kinase II to the frequency of Ca2+ oscillations. Science 279: 227–230. FINCH, E. A. AND G. J. AUGUSTINE (1998) Local calcium signaling by IP3 in Purkinje cell dendrites. Nature 396: 753–756. HARRIS, B. A., J. D. ROBISHAW, S. M. MUMBY AND A. G. GILMAN (1985) Molecular cloning of complementary DNA for the alpha subunit of the G protein that stimulates adenylate cyclase. Science 229: 1274–1277. KAMMERMEIER, P. J. AND S. R. IKEDA (1999) Expression of RGS2 alters the coupling of metabotropic glutamate receptor 1a to M-type K+ and N-type Ca2+ channels. Neuron 22: 819–829. KRAFT, A. S. AND W. B. ANDERSON (1983) Phorbol esters increase the amount of Ca2+, phospholipid-dependent protein kinase associated with plasma membrane. Nature 301: 621–623. LINDGREN, N. AND 8 OTHERS (2000) Regulation of tyrosine hydroxylase activity and phosphorylation at ser(19) and ser(40) via activation of glutamate NMDA receptors in rat striatum. J. Neurochem. 74: 2470–2477. MILLER, S. G. AND M. B. KENNEDY (1986) Regulation of brain type II Ca2+/calmodulindependent protein kinase by autophosphorylation: A Ca2+-triggered molecular switch. Cell 44: 861–870. NORTHUP, J. K., P. C. STERNWEIS, M. D. SMIGEL, L. S. SCHLEIFER, E. M. ROSS AND A. G. GILMAN (1980) Purification of the regulatory component of adenylate cyclase. Proc. Natl. Acad. Sci. USA 77: 6516–6520. SAITOH, T. AND J. H. SCHWARTZ (1985) Phosphorylation-dependent subcellular translocation of a Ca2+/calmodulin-dependent protein kinase produces an autonomous enzyme in Aplysia neurons. J. Cell Biol. 100: 835–842. SHEN, K. AND T. MEYER (1999) Dynamic control of CaMKII translocation and localization in hippocampal neurons by NMDA receptor stimulation. Science 284: 162–166. SU, Y. AND 7 OTHERS (1995) Regulatory subunit of protein kinase A: Structure of deletion mutant with cAMP binding domains. Science 269: 807–813. TAO, X., S. FINKBEINER, D. B. ARNOLD, A. J. SHAYWITZ AND M. E. GREENBERG (1998) Ca2+ influx regulates BDNF transcription by a CREB family transcription factor-dependent mechanism. Neuron 20: 709–726. TESMER, J. J., R. K. SUNAHARA, A. G. GILMAN AND S. R. SPRANG (1997) Crystal structure of the catalytic domains of adenylyl cyclase in a complex with Gsα-GTPγS. Science 278: 1907–1916. ZHANG, G., M. G. KAZANIETZ, P. M. BLUMBERG AND J. H. HURLEY (1995) Crystal structure of the cys2 activator-binding domain of protein kinase C delta in complex with phorbol ester. Cell 81: 917–924. Books ALBERTS, B., A. JOHNSON, J. LEWIS, M. RAFF, K. ROBERTS AND P. WALTER (2002) Molecular Biology of the Cell, 4th Ed. New York: Garland Science. CARAFOLI, E. AND C. KLEE (1999) Calcium as a Cellular Regulator. New York: Oxford University Press. Sensation and Sensory Processing II UNIT II SENSATION AND SENSORY PROCESSING Surface view of the primary visual cortex illustrating patterns of neural activity visualized with intrinsic signal optical imaging techniques (see Box C in Chapter 11). Each panel illustrates the activity evoked by viewing a single thin vertical line. The smooth progression of the activated region from the upper left to the lower right panel illustrates the orderly mapping of visual space. The patchy appearance of the activated region in each panel reflects the columnar mapping of orientation preference. Red regions are the most active, black the least. (Courtesy of Bill Bosking, Justin Crowley, Tom Tucker, and David Fitzpatrick.) 8 9 10 11 12 13 14 The Somatic Sensory System Pain Vision: The Eye Central Visual Pathways The Auditory System The Vestibular System The Chemical Senses Sensation entails the ability to transduce, encode, and ultimately perceive information generated by stimuli arising from both the external and internal environments. Much of the brain is devoted to these tasks. Although the basic senses—somatic sensation, vision, audition, vestibular sensation, and the chemical senses—are very different from one another, a few fundamental rules govern the way the nervous system deals with each of these diverse modalities. Highly specialized nerve cells called receptors convert the energy associated with mechanical forces, light, sound waves, odorant molecules, or ingested chemicals into neural signals—afferent sensory signals—that convey information about the stimulus to the brain. Afferent sensory signals activate central neurons capable of representing both the qualitative and quantitative aspects of the stimulus (what it is and how strong it is) and, in some modalities (somatic sensation, vision, and audition) the location of the stimulus in space (where it is). The clinical evaluation of patients routinely requires an assessment of the sensory systems to infer the nature and location of potential neurological problems. Knowledge of where and how the different sensory modalities are transduced, relayed, represented, and further processed to generate appropriate behavioral responses is therefore essential to understanding and treating a wide variety of diseases. Accordingly, these chapters on the neurobiology of sensation also introduce some of the major structure/function relationships in the sensory components of the nervous system. Chapter 8 The Somatic Sensory System Overview The somatic sensory system has two major components: a subsystem for the detection of mechanical stimuli (e.g., light touch, vibration, pressure, and cutaneous tension), and a subsystem for the detection of painful stimuli and temperature. Together, these two subsystems give humans and other animals the ability to identify the shapes and textures of objects, to monitor the internal and external forces acting on the body at any moment, and to detect potentially harmful circumstances. This chapter focuses on the mechanosensory subsystem; the pain and temperature subsystem is taken up in the following chapter. Mechanosensory processing of external stimuli is initiated by the activation of a diverse population of cutaneous and subcutaneous mechanoreceptors at the body surface that relays information to the central nervous system for interpretation and ultimately action. Additional receptors located in muscles, joints, and other deep structures monitor mechanical forces generated by the musculoskeletal system and are called proprioceptors. Mechanosensory information is carried to the brain by several ascending pathways that run in parallel through the spinal cord, brainstem, and thalamus to reach the primary somatic sensory cortex in the postcentral gyrus of the parietal lobe. The primary somatic sensory cortex projects in turn to higher-order association cortices in the parietal lobe, and back to the subcortical structures involved in mechanosensory information processing. Cutaneous and Subcutaneous Somatic Sensory Receptors The specialized sensory receptors in the cutaneous and subcutaneous tissues are dauntingly diverse (Table 8.1). They include free nerve endings in the skin, nerve endings associated with specializations that act as amplifiers or filters, and sensory terminals associated with specialized transducing cells that influence the ending by virtue of synapse-like contacts. Based on function, this variety of receptors can be divided into three groups: mechanoreceptors, nociceptors, and thermoceptors. On the basis of their morphology, the receptors near the body surface can also be divided into free and encapsulated types. Nociceptor and thermoceptor specializations are referred to as free nerve endings because the unmyelinated terminal branches of these neurons ramify widely in the upper regions of the dermis and epidermis (as well as in some deeper tissues); their role in pain and temperature sensation is discussed in Chapter 9. Most other cutaneous receptors show some degree of encapsulation, which helps determine the nature of the stimuli to which they respond. Despite their variety, all somatic sensory receptors work in fundamentally the same way: Stimuli applied to the skin deform or otherwise change the 189 190 Chapter Eight TABLE 8.1 The Major Classes of Somatic Sensory Receptors Receptor type Anatomical characteristics Free nerve endings Minimally specialized nerve endings Encapsulated; between dermal papillae Encapsulated; onionlike covering Meissner’s corpuscles Pacinian corpuscles Merkel’s disks Ruffini’s corpuscles Muscle spindles Golgi tendon organs Joint receptors Encapsulated; associated with peptidereleasing cells Encapsulated; oriented along stretch lines Highly specialized (see Figure 8.5 and Chapter 15) Highly specialized (see Chapter 15) Minimally specialized Associated axonsa (and diameters) Axonal conduction velocities Location Function C, Aδ 2–20 m/s All skin Pain, Slow temperature, crude touch Touch, Rapid pressure (dynamic) Deep pressure, Rapid vibration (dynamic) High Touch, pressure (static) Slow Low Aβ 6–12 µm Aβ 6–12 µm Aβ Principally glabrous skin Subcutaneous tissue, interosseous membranes, viscera All skin, hair follicles Rate of adaptation Threshold of activation Low Low Aβ 6–12 µm All skin Stretching of skin Slow Low Ia and II Muscles Muscle length Both slow and rapid Low Ib Tendons Muscle tension Slow Low — Joints Joint position Rapid Low a In the 1920s and 1930s, there was a virtual cottage industry classifying axons according to their conduction velocity. Three main categories were discerned, called A, B, and C. A comprises the largest and fastest axons, C the smallest and slowest. Mechanoreceptor axons generally fall into category A. The A group is further broken down into subgroups designated a (the fastest), b, and d (the slowest). To make matters even more confusing, muscle afferent axons are usually classified into four additional groups—I (the fastest), II, III, and IV (the slowest)—with subgroups designated by lowercase roman letters! nerve endings, which in turn affects the ionic permeability of the receptor cell membrane. Changes in permeability generate a depolarizing current in the nerve ending, thus producing a receptor (or generator) potential that triggers action potentials, as described in Chapters 2 and 3. This overall process, in which the energy of a stimulus is converted into an electrical signal in the sensory neuron, is called sensory transduction and is the critical first step in all sensory processing. The quality of a mechanosensory (or any other) stimulus (i.e., what it represents and where it is) is determined by the properties of the relevant receptors and the location of their central targets (Figure 8.1). The quantity or strength of the stimulus is conveyed by the rate of action potential discharge triggered by the receptor potential (although this relationship is nonlinear and often quite complex). Some receptors fire rapidly when a stimulus is first presented and then fall silent in the presence of continued stimulation (which is to say they “adapt” to the stimulus), whereas others generate a sustained discharge in the presence of an ongoing stimulus (Figure 8.2). The usefulness of having some receptors that adapt quickly and others that do not is to provide information about both the dynamic and static qualities of a stimulus. Receptors that initially fire in the presence of a stimulus and then The Somatic Sensor y System 191 (A) Cerebrum Somatic sensory cortex Ventral posterior nuclear complex of thalamus Midbrain Gracile nucleus Cuneate nucleus Medial leminiscus Medulla Dorsal root ganglion cells Mechanosensory afferent fiber Spinal cord Receptor endings (B) Pain and temperature afferent fiber Central sulcus Primary somatic sensory cortex Figure 8.1 General organization of the somatic sensory system. (A) Mechanosensory information about the body reaches the brain by way of a three-neuron relay (shown in red). The first synapse is made by the terminals of the centrally projecting axons of dorsal root ganglion cells onto neurons in the brainstem nuclei (the local branches involved in segmental spinal reflexes are not shown here). The axons of these secondorder neurons synapse on third-order neurons of the ventral posterior nuclear complex of the thalamus, which in turn send their axons to the primary somatic sensory cortex (red). Information about pain and temperature takes a different course (shown in blue; the anterolateral system), and is discussed in the following chapter. (B) Lateral and midsagittal views of the human brain, illustrating the approximate location of the primary somatic sensory cortex in the anterior parietal lobe, just posterior to the central sulcus. 192 Chapter Eight become quiescent are particularly effective in conveying information about changes in the information the receptor reports; conversely, receptors that continue to fire convey information about the persistence of a stimulus. Accordingly, somatic sensory receptors and the neurons that give rise to them are usually classified into rapidly or slowly adapting types (see Table 8.1). Rapidly adapting, or phasic, receptors respond maximally but briefly to stimuli; their response decreases if the stimulus is maintained. Conversely, slowly adapting, or tonic, receptors keep firing as long as the stimulus is present. Stimulus Slowly adapting Rapidly adapting Mechanoreceptors Specialized to Receive Tactile Information 0 1 2 Time (s) 3 4 Figure 8.2 Slowly adapting mechanoreceptors continue responding to a stimulus, whereas rapidly adapting receptors respond only at the onset (and often the offset) of stimulation. These functional differences allow the mechanoreceptors to provide information about both the static (via slowly adapting receptors) and dynamic (via rapidly adapting receptors) qualities of a stimulus. Four major types of encapsulated mechanoreceptors are specialized to provide information to the central nervous system about touch, pressure, vibration, and cutaneous tension: Meissner’s corpuscles, Pacinian corpuscles, Merkel’s disks, and Ruffini’s corpuscles (Figure 8.3 and Table 8.1). These receptors are referred to collectively as low-threshold (or high-sensitivity) mechanoreceptors because even weak mechanical stimulation of the skin induces them to produce action potentials. All low-threshold mechanoreceptors are innervated by relatively large myelinated axons (type Aβ; see Table 8.1), ensuring the rapid central transmission of tactile information. Meissner’s corpuscles, which lie between the dermal papillae just beneath the epidermis of the fingers, palms, and soles, are elongated receptors formed by a connective tissue capsule that comprises several lamellae of Schwann cells. The center of the capsule contains one or more afferent nerve fibers that generate rapidly adapting action potentials following minimal skin depression. Meissner’s corpuscles are the most common mechanoreceptors of “glabrous” (smooth, hairless) skin (the fingertips, for instance), and their afferent fibers account for about 40% of the sensory innervation of the human hand. These corpuscles are particularly efficient in transducing information about the relatively low-frequency vibrations (30–50 Hz) that occur when textured objects are moved across the skin. Pacinian corpuscles are large encapsulated endings located in the subcutaneous tissue (and more deeply in interosseous membranes and mesenteries of the gut). These receptors differ from Meissner’s corpuscles in their morphology, distribution, and response threshold. The Pacinian corpuscle has an onion-like capsule in which the inner core of membrane lamellae is separated from an outer lamella by a fluid-filled space. One or more rapidly adapting afferent axons lie at the center of this structure. The capsule again acts as a filter, in this case allowing only transient disturbances at high frequencies (250–350 Hz) to activate the nerve endings. Pacinian corpuscles adapt more rapidly than Meissner’s corpuscles and have a lower response threshold. These attributes suggest that Pacinian corpuscles are involved in the discrimination of fine surface textures or other moving stimuli that produce high-frequency vibration of the skin. In corroboration of this supposition, stimulation of Pacinian corpuscle afferent fibers in humans induces a sensation of vibration or tickle. They make up 10–15% of the cutaneous receptors in the hand. Pacinian corpuscles located in interosseous membranes probably detect vibrations transmitted to the skeleton. Structurally similar endings found in the bills of ducks and geese and in the legs of cranes and herons detect vibrations in water; such endings in the wings of soaring birds detect vibrations produced by air currents. Because they are rapidly adapting, Pacinian corpuscles, like Meissner’s corpuscles, provide information primarily about the dynamic qualities of mechanical stimuli. The Somatic Sensor y System 193 Epidermis Dermis Sweat gland Meissner corpuscle Pacinian corpuscle Ruffini's corpuscles Merkel's disks Slowly adapting cutaneous mechanoreceptors include Merkel’s disks and Ruffini’s corpuscles (see Figure 8.3 and Table 8.1). Merkel’s disks are located in the epidermis, where they are precisely aligned with the papillae that lie beneath the dermal ridges. They account for about 25% of the mechanoreceptors of the hand and are particularly dense in the fingertips, lips, and external genitalia. The slowly adapting nerve fiber associated with each Merkel’s disk enlarges into a saucer-shaped ending that is closely applied to another specialized cell containing vesicles that apparently release peptides that modulate the nerve terminal. Selective stimulation of these receptors in humans produces a sensation of light pressure. These several properties have led to the supposition that Merkel’s disks play a major role in the static discrimination of shapes, edges, and rough textures. Ruffini’s corpuscles, although structurally similar to other tactile receptors, are not well understood. These elongated, spindle-shaped capsular specializations are located deep in the skin, as well as in ligaments and tendons. The long axis of the corpuscle is usually oriented parallel to the stretch lines in skin; thus, Ruffini’s corpuscles are particularly sensitive to the cutaneous stretching produced by digit or limb movements. They account for about 20% of the receptors in the human hand and do not elicit any particular tactile sensation when stimulated electrically. Although there is still some question as to their function, they probably respond primarily to internally generated stimuli (see the section on proprioception, below). Differences in Mechanosensory Discrimination across the Body Surface The accuracy with which tactile stimuli can be sensed varies from one region of the body to another, a phenomenon that illustrates some further principles Free nerve endings Figure 8.3 The skin harbors a variety of morphologically distinct mechanoreceptors. This diagram represents the smooth, hairless (also called glabrous) skin of the fingertip. The major characteristics of the various receptor types are summarized in Table 8.1. (After Darian-Smith, 1984.) 194 Chapter Eight Figure 8.4 Variation in the sensitivity of tactile discrimination as a function of location on the body surface, measured here by two-point discrimination. (After Weinstein, 1968.) 4 3 Fingers 2 1 Thumb Palm Forearm Cheek Nose Upper lip Forehead Upper arm Shoulder Breast Back Belly Thigh Calf 0 Sole Toe 5 10 15 20 25 30 35 40 45 50 Two-point discrimination threshold (mm) of somatic sensation. Figure 8.4 shows the results of an experiment in which variation in tactile ability across the body surface was measured by two-point discrimination. This technique measures the minimal interstimulus distance required to perceive two simultaneously applied stimuli as distinct (the indentations of the points of a pair of calipers, for example). When applied to the skin, such stimuli of the fingertips are discretely perceived if they are only 2 mm apart. In contrast, the same stimuli applied to the forearm are not perceived as distinct until they are at least 40 mm apart! This marked regional difference in tactile ability is explained by the fact that the encapsulated mechanoreceptors that respond to the stimuli are three to four times more numerous in the fingertips than in other areas of the hand, and many times more dense than in the forearm. Equally important in this regional difference are the sizes of the neuronal receptive fields. The receptive field of a somatic sensory neuron is the region of the skin within which a tactile stimulus evokes a sensory response in the cell or its axon (Boxes A and B). Analysis of the human hand shows that the receptive fields of mechanosensory neurons are 1–2 mm in diameter on the fingertips but 5–10 mm on the palms. The receptive fields on the arm are larger still. The importance of receptive field size is easy to envision. If, for instance, the receptive fields of all cutaneous receptor neurons covered the entire digital pad, it would be impossible to discriminate two spatially separate stimuli applied to the fingertip (since all the receptive fields would be returning the same spatial information). The Somatic Sensor y System 195 Box A Receptive Fields and Sensory Maps in the Cricket Two principles of somatiosensory organization have emerged from studies of the mammalian brain: (1) individual neurons are tuned to particular aspects of complex stimuli; and (2) these stimulus qualities are represented in an orderly fashion in relevant regions of the nervous system. These principles apply equally well to invertebrates, including the equivalent of the somatic sensory system in insects such as crickets, grasshoppers, and cockroaches. In the cricket, the salient tactile stimulation for the animal comes from air currents that displace sensory hairs of bilaterally symmetric sensory structures called cerci (sing. cercus). The location and structure of specific cercal hairs allow them to be displaced by air currents having different directions and speeds (Figure A). Accordingly, the peripheral sensory neurons associated with the hairs represent the full range of air current directions and velocities impinging on the animal. This information is carried centrally and is systematically represented in a region of the cricket central nervous system called the terminal ganglion. Individual neurons in this ganglion correspond to the cercal hairs, and have receptive fields and response properties that represent a full range of directions and speeds for extrinsic mechanical forces, including air currents (Figure B). For the cricket, the significance of this (A) (B) 0° 0° 90° 90° 180° % Maximal response Membrane potential (mV) 270° 135° % Maximal i response Speaker velocity 315° 0 100 200 Time (ms) References 100 50 0 0° 90° 180° 270° 360/0° 100 50 0 information is, among other things, detecting the direction and speed of oncoming objects to then execute motor programs for escape. (This is also the likely significance of this representation for cockroaches, which can therefore escape the consequences of a descending human foot.) Much like the somatic sensory system in mammals, the primary sensory afferents project to the terminal ganglion in an orderly fashion, such that there is a somatotopic map of air current directions. And, like mammals, individual neurons within this representation are tuned to specific aspects of the mechanical forces acting on the cricket. These facts about insects’ mechanosensory system emphasize that somatic sensory functions are basically similar across a wide range of animals. Indeed, regardless of sensory modality, nervous system organization, or the identity of the organism, it is likely that stimulus specificity will be reflected in receptive fields of individual neurons and there will be orderly mapping of those receptive fields into either a topographic or computational map in the animal’s brain. Front Right Rear Left Front Air current stimulus orientation (A) Intracellular recording of action potential activity of an individual sensory neuron’s responses to different directions of wind current. (B) The plots indicate this neuron’s receptive field for wind direction (top) and the tuning curve for the neuron’s selective firing to its preferred direction. (After Miller et al., 1991.) JACOBS, G. A. AND F. E. THEUNISSEN (1996) Functional organization of a neural map in the cricket cercal sensory system. J. Neurosci. 16: 769–784. MILLER, J. P., G. A. JACOBS AND F. E. THEUNISSEN (1991) Representation of sensory information in the cricket cercal sensory system. I. Response properties of the primary interneurons. J. Neurophys. 66: 1680–1688. MURPHEY, R. K. (1981) The structure and development of somatotopic map in crickets: The cercal afferent projection. Dev. Biol. 88: 236–246. MURPHEY, R. K. AND H. V. B. HIRSCH (1982) From cat to cricket: The genesis of response selectivity of interneurons. Curr. Topics Dev. Biol. 17: 241–256. 196 Chapter Eight Box B Dynamic Aspects of Somatic Sensory Receptive Fields When humans explore objects with their hands, multiple contacts between the skin and the object surface generate extraordinarily complex patterns of tactile stimuli. As a consequence, the somatic sensory system must process signals that change continuously in time. Nonetheless, we routinely discriminate the size, texture and shape of objects with great accuracy. Until recently, the temporal structure of such stimuli was not considered a major variable in characterizing the physiological properties of somatic sensory neurons. For instance, the classical definition of the receptive field of a somatic sensory neuron takes into account only the overall area of the body surface that elicits significant variation in the neuron’s firing rate. By the same token, the topographic maps in the somatic sensory system have been interpreted as evidence that tactile information processing involves primarily spatial criteria. The advent of multiple electrode recording to simultaneously monitor the activity of large populations of single neurons has begun to change this “static” view of the somatic sensory system. In both primates and rodents, this approach has shown that the receptive fields of cortical and subcortical neurons (A) SI VPM SpV PrV (A) Simultaneous electrode recordings in behaving rats allow monitoring of the spatiotemporal spread of neuronal activation across several levels of the somatic sensory system following stimulation (of a single facial whisker, in this example). These 3-D graphs represent patterns of neuronal ensemble activity at each level of the pathway. The x axis represents the poststimulus time in ms, the y axis the number of neurons recorded at each level; the color-coded gradient in the z axis shows the response of the neurons, with red the highest firing and green the lowest. SI, somatic sensory cortex; VPM, ventral posterior medial nucleus of the thalamus; SpV, spinal nucleus of the trigeminal brainstem complex; PrV, principal nucleus of the brainstem trigeminal complex. (From Nicholelis et al., 1997.) Receptor density and receptive field sizes in different regions are not the only factors determining somatic sensation. Psychophysical analysis of tactile performance suggests that something more than the cutaneous periphery is needed to explain variations in tactile perception. For instance, sensory thresholds in two-point discrimination tests vary with practice, fatigue, and stress. The contextual significance of stimuli is also important in determining what we actually feel; even though we spend most of the day wearing clothes, we usually ignore the tactile stimulation that they produce. Some aspect of the mechanosensory system allows us to filter out this information and pay attention to it only when necessary. The fascinating phenomenon of “phantom limb” sensations after amputation (see Box C in Chapter 9) provides further evidence that tactile perception is not fully explained by the The Somatic Sensor y System 197 (B) Whisker rows A 12−16 ms 16−20 ms 20−24 ms 24−28 ms B C D E A Whisker rows 8−12 ms 0 1 2 3 4 8−12 ms 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 Whisker columns 12−16 ms 16−20 ms 20−24 ms 0 1 2 3 4 24−28 ms B C area of the skin tends to excite more and more neurons as time goes by. Thus, many more neurons than those located in the area of the map directly representing the stimulated skin actually respond to the stimulus, albeit at longer latencies. The end result of these more complex neuronal responses is the emergence of spatiotemporal representations at all levels of the somatic sensory system. Thus, contrary to the classical notion of receptive fields, the somatic sensory system processes information in a dynamic way. Such processing is not only relevant for the normal operation of the system, but may also account for some aspects of adult plasticity (see Chapter 24). D References E 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 Whisker columns 0 1 2 3 4 (B) Receptive fields of two cortical neurons from two different animals. Each panel represents the matrix of whiskers on the animals’ snout (whisker columns are on the x axis and whisker rows on the y axis) for a 4-ms epoch of poststimulus time. Within a particular time period, the center of the receptive field is defined as the whisker eliciting the greatest response magnitude (yellow). Note that the receptive field centers shift as a function of time. (From Ghazanfar and Nicholelis, 1998.) vary as a function of time: The neuron responds differently to a spatially defined stimulus as the period of stimulation proceeds (see Figures A and B). This coupling of space and time can also be demonstrated at level of somatotopic maps. By recording the activity of single neurons located in different regions of the map simultaneously, it is apparent that the stimulation of a small peripheral information that travels centrally. The central nervous system clearly plays an active role in determining the perception of the mechanical forces that act on us. Mechanoreceptors Specialized for Proprioception Whereas cutaneous mechanoreceptors provide information derived from external stimuli, another major class of receptors provides information about mechanical forces arising from the body itself, the musculoskeletal system in particular. These are called proprioceptors, roughly meaning “receptors for self.” The purpose of proprioceptors is primarily to give detailed and continuous information about the position of the limbs and other body parts in GHAZANFAR, A. A. AND M. A. L. NICOLELIS (1999) Spatiotemporal properties of layer V neurons of the rat primary somatosensory cortex. Cereb. Cortex 4: 348–361. NICOLELIS, M. A. L., A. A. GHAZANFAR, B. FAGGIN, S. VOTAW AND L. M. O. OLIVEIRA (1997) Reconstructing the engram: Simultaneous, multiple site, many single neuron recordings. Neuron 18: 529–537. NICOLELIS, M. A. L. AND 7 OTHERS (1998) Simultaneous encoding of tactile information by three primate cortical areas. Nature Neurosci. 1: 621–630. 198 Chapter Eight space (specialized mechanoreceptors also exist in the heart and major vessels to provide information about blood pressure, but these neurons are considered to be part of the visceral motor system; see Chapter 20). Low-threshold mechanoreceptors, including muscle spindles, Golgi tendon organs, and joint receptors, provide this kind of sensory information, which is essential to the accurate performance of complex movements. Information about the position and motion of the head is particularly important; in this case, proprioceptors are integrated with the highly specialized vestibular system, which is considered separately in Chapter 13. The most detailed knowledge about proprioception derives from studies of muscle spindles, which are found in all but a few striated (skeletal) muscles. Muscle spindles consist of four to eight specialized intrafusal muscle fibers surrounded by a capsule of connective tissue. The intrafusal fibers are distributed among the ordinary (extrafusal) fibers of skeletal muscle in a parallel arrangement (Figure 8.5). In the largest of the several intrafusal fibers, the nuclei are collected in an expanded region in the center of the fiber called a bag; hence the name nuclear bag fibers. The nuclei in the remaining two to six smaller intrafusal fibers are lined up single file, with the result that these fibers are called nuclear chain fibers. Myelinated sensory axons belonging to group Ia innervate muscle spindles by encircling the middle portion of both types of intrafusal fibers (see Figure 8.5 and Table 8.1). The Ia axon terminal is known as the primary sensory ending of the spindle. Secondary innervation is provided by group II axons that innervate the nuclear chain fibers and give off a minor branch to the nuclear bag fibers. The intrafusal muscle fibers contract when commanded to do so by motor axons derived from a pool of specialized motor neurons in the spinal cord (called g motor neurons). The major function of muscle spindles is to provide information about muscle length (that is, the degree to which they are being stretched). A detailed account of how these important receptors function during movement is given in Chapters 15 and 16. The density of spindles in human muscles varies. Large muscles that generate coarse movements have relatively few spindles; in contrast, extraocular muscles and the intrinsic muscles of the hand and neck are richly supplied with spindles, reflecting the importance of accurate eye movements, the need to manipulate objects with great finesse, and the continuous demand for precise positioning of the head. This relationship between receptor den- Axon of a motor neuron Figure 8.5 A muscle spindle and several extrafusal muscle fibers. See text for description. (After Matthews, 1964.) Extrafusal muscle fibers Axons of g motor neurons Intrafusal Nuclear Subcapsular muscle fibers chain fiber space Group I and II afferent axons Nuclear bag fiber Capsule surrounding spindle The Somatic Sensor y System 199 sity and muscle size is consistent with the generalization that the sensory motor apparatus at all levels of the nervous system is much richer for the hands, head, speech organs, and other parts of the body that are used to perform especially important and demanding tasks. Spindles are lacking altogether in a few muscles, such as those of the middle ear, which do not require the kind of feedback that these receptors provide. Whereas muscle spindles are specialized to signal changes in muscle length, low-threshold mechanoreceptors in tendons inform the central nervous system about changes in muscle tension. These mechanoreceptors, called Golgi tendon organs, are innervated by branches of group Ib afferents and are distributed among the collagen fibers that form the tendons (see Chapter 15). Finally, rapidly adapting mechanoreceptors in and around joints gather dynamic information about limb position and joint movement. The function of these joint receptors is not well understood. Active Tactile Exploration Tactile discrimination—that is, perceiving the detailed shape or texture of an object—normally entails active exploration. In humans, this is typically accomplished by using the hands to grasp and manipulate objects, or by moving the fingers across a surface so that a sequence of contacts between the skin and the object of interest is established. Psychophysical evidence indicates that relative movement between the skin and a surface is the single most important requirement for accurate discrimination of texture. Animal experiments confirm the dependence of tactile discrimination on active exploration. Rats, for instance, discriminate the details of texture by rhythmically brushing their facial whiskers across surfaces. Active touching, which is called haptics, involves the interpretation of complex spatiotemporal patterns of stimuli that are likely to activate many classes of mechanoreceptors. Haptics also requires dynamic interactions between motor and sensory signals, which presumably induce sensory responses in central neurons that differ from the responses of the same cells during passive stimulation of the skin (see Box B). The Major Afferent Pathway for Mechanosensory Information: The Dorsal Column–Medial Lemniscus System The action potentials generated by tactile and other mechanosensory stimuli are transmitted to the spinal cord by afferent sensory axons traveling in the peripheral nerves. The neuronal cell bodies that give rise to these first-order axons are located in the dorsal root (or sensory) ganglia associated with each segmental spinal nerve (see Figure 8.1 and Box C). Dorsal root ganglion cells are also known as first-order neurons because they initiate the sensory process. The ganglion cells thus give rise to long peripheral axons that end in the somatic receptor specializations already described, and shorter central axons that reach the dorsolateral region of the spinal cord via the dorsal (sensory) roots of each spinal cord segment. The large myelinated fibers that innervate low-threshold mechanoreceptors are derived from the largest neurons in these ganglia, whereas the smaller ganglion cells give rise to smaller afferent nerve fibers that end in the high-threshold nociceptors and thermoceptors (see Table 8.1). Depending on whether they belong to the mechanosensory system or to the pain and temperature system, the first-order axons carrying information 200 Chapter Eight from somatic receptors have different patterns of termination in the spinal cord and define distinct somatic sensory pathways within the central nervous system (see Figure 8.1). The dorsal column–medial lemniscus pathway carries the majority of information from the mechanoreceptors that mediate tactile discrimination and proprioception (Figure 8.6); the spinothalamic (anterolateral) pathway mediates pain and temperature sensation and is described in Chapter 9. This difference in the afferent pathways of these modalities is one of the reasons that pain and temperature sensation is treated separately here. Upon entering the spinal cord, the first-order axons carrying information from peripheral mechanoreceptors bifurcate into ascending and descending branches, which in turn send collateral branches to several spinal segments. Some collateral branches penetrate the dorsal horn of the cord and synapse on neurons located mainly in a region called Rexed’s laminae III–V. These synapses mediate, among other things, segmental reflexes such as the “kneejerk” or myotatic reflex described in Chapter 1, and are further considered in Chapters 15 and 16. The major branch of the incoming axons, however, ascends ipsilaterally through the dorsal columns (also called the posterior funiculi) of the cord, all the way to the lower medulla, where it terminates by contacting second-order neurons in the gracile and cuneate nuclei (together referred to as the dorsal column nuclei; see Figures 8.1 and 8.6A). Axons in the dorsal columns are topographically organized such that the fibers that convey information from lower limbs are in the medial subdivision of the dorsal columns, called the gracile tract, a fact of some significance in the clinical localization of neural injury. The lateral subdivision, called the cuneate tract, contains axons conveying information from the upper limbs, trunk, and neck. At the level of the upper thorax, the dorsal columns account for more than a third of the cross-sectional area of the human spinal cord. Despite their size, lesions limited to the dorsal columns of the spinal cord in both humans and monkeys have only a modest effect on the performance of simple tactile tasks. Such lesions, however, do impede the ability to detect the direction and speed of tactile stimuli, as well as degrading the ability to sense the position of the limbs in space. Dorsal column lesions may also reduce a patient’s ability to initiate active movements related to tactile exploration. For instance, such individuals have difficulty recognizing numbers and letters drawn on their skin. The relatively mild deficit that follows dorsal column lesions is presumably explained by the fact that some axons responsible for cutaneous mechanoreception also run in the spinothalamic (pain and temperature) pathway, as described in Chapter 9. The second-order relay neurons in the dorsal column nuclei send their axons to the somatic sensory portion of the thalamus (see Figure 8.6A). The axons from dorsal column nuclei project in the dorsal portion of each side of the lower brainstem, where they form the internal arcuate tract. The internal arcuate axons subsequently cross the midline to form another named tract that is elongated dorsoventrally, the medial lemniscus. (The crossing of these fibers is called the decussation of the medial lemniscus, from the roman numeral “X,” or decem; the word lemniscus means “ribbon.”) In a cross-section through the medulla, such as the one shown in Figure 8.6A, the medial lemniscal axons carrying information from the lower limbs are located ventrally, whereas the axons related to the upper limbs are located dorsally (again, a fact of some clinical importance). As the medial lemniscus ascends through the pons and midbrain, it rotates 90° laterally, so that the upper body is eventually represented in the medial portion of the tract, and the lower body in the lateral portion. The axons of the medial lem- The Somatic Sensor y System 201 (A) (B) Cerebrum Primary somatic sensory cortex Ventral posterior lateral nucleus of the thalamus Ventral posterior medial nucleus of thalamus Midbrain Trigeminothalamic tract (trigeminal lemniscus) Medial leminiscus Trigeminal ganglion Mid-pons Medial lemniscus Rostral medulla Internal arcuate fibers Gracile nucleus (pathways from lower body) Cuneate nucleus (pathways from upper body) Caudal medulla Gracile tract Cuneate tract Cervical spinal cord Mechanosensory receptors from upper body Lumbar spinal cord Mechanosensory receptors from lower body Medial lemniscus Principal nucleus of trigeminal complex Mechanosensory receptors from face Figure 8.6 Schematic representation of the main mechanosensory pathways. (A) The dorsal column–medial lemniscus pathway carries mechanosensory information from the posterior third of the head and the rest of the body. (B) The trigeminal portion of the mechanosensory system carries similar information from the face. 202 Chapter Eight Box C C2 C3 C4 C5 Dermatomes Cervical The innervation arising from a single dorsal root ganglion and its spinal nerve is called a dermatome. The full set of sensory dermatomes is shown here for a typical adult. Knowledge of this arrangement is particularly important in defining the location of suspected spinal (and other) lesions. The numbers refer to the spinal segments by which each nerve is named. (After Rosenzweig et al., 2002.) Thoracic Trigeminal nerve branches T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 Sacral Lumbar Thoracic Cervical C5 C6 C7 C8 L1 L2 L3 L4 L5 S2−S4 Lumbar Sacral Each dorsal root (or sensory) ganglion and associated spinal nerve arises from an iterated series of embryonic tissue masses called somites. This fact of development explains the overall segmental arrangement of somatic nerves (and the targets they innervate) in the adult (see figure). The territory innervated by each spinal nerve is called a dermatome. In humans, the cutaneous area of each dermatome has been defined in patients in whom specific dorsal roots were affected S1 (as in herpes zoster, or “shingles”) or after surgical interruption (for relief of pain or other reasons). Such studies show that dermatomal maps vary among individuals. Moreover, dermatomes overlap substantially, so that injury to an individual dorsal root does not lead to complete loss of sensation in the relevant skin region, the overlap being more extensive for touch, pressure, and vibration than for pain and temperature. Thus, testing for pain sensation provides a more precise assessment of a segmental nerve injury than does testing for responses to touch, pressure, or vibration. Finally, the segmental distribution of proprioceptors does not follow the dermatomal map but is more closely allied with the pattern of muscle innervation. Despite these limitations, knowledge of dermatomes is essential in the clinical evaluation of neurological patients, particularly in determining the level of a spinal lesion. niscus thus reach the ventral posterior lateral (VPL) nucleus of the thalamus, whose cells are the third-order neurons of the dorsal column–medial lemniscus system (see Figure 8.7). The Trigeminal Portion of the Mechanosensory System As noted, the dorsal column–medial lemniscus pathway described in the preceding section carries somatic information from only the upper and lower body and from the posterior third of the head. Tactile and propriocep- The Somatic Sensor y System 203 tive information from the face is conveyed from the periphery to the thalamus by a different route. Information derived from the face is transmitted to the central nervous system via the trigeminal somatic sensory system (Figure 8.6B). Low-threshold mechanoreception in the face is mediated by firstorder neurons in the trigeminal (cranial nerve V) ganglion. The peripheral processes of these neurons form the three main subdivisions of the trigeminal nerve (the ophthalmic, maxillary, and mandibular branches), each of which innervates a well-defined territory on the face and head, including the teeth and the mucosa of the oral and nasal cavities. The central processes of trigeminal ganglion cells form the sensory roots of the trigeminal nerve; they enter the brainstem at the level of the pons to terminate on neurons in the subdivisions of the trigeminal brainstem complex. The trigeminal complex has two major components: the principal nucleus (responsible for processing mechanosensory stimuli), and the spinal nucleus (responsible for processing painful and thermal stimuli). Thus, most of the axons carrying information from low-threshold cutaneous mechanoreceptors in the face terminate in the principal nucleus. In effect, this nucleus corresponds to the dorsal column nuclei that relay mechanosensory information from the rest of the body. The spinal nucleus corresponds to a portion of the spinal cord that contains the second-order neurons in the pain and temperature system for the rest of the body (see Chapter 9). The secondorder neurons of the trigeminal brainstem nuclei give off axons that cross the midline and ascend to the ventral posterior medial (VPM) nucleus of the thalamus by way of the trigeminothalamic tract (also called the trigeminal lemniscus). The Somatic Sensory Components of the Thalamus Each of the several ascending somatic sensory pathways originating in the spinal cord and brainstem converge on the thalamus (Figure 8.7). The ventral posterior complex of the thalamus, which comprises a lateral and a medial nucleus, is the main target of these ascending pathways. As already mentioned, the more laterally located ventral posterior lateral (VPL) nucleus receives projections from the medial lemniscus carrying all somatosensory information from the body and posterior head, whereas the more medially located ventral posterior medial (VPM) nucleus receives axons from the trigeminal lemniscus (that is, mechanosensory and nociceptive information from the face). Accordingly, the ventral posterior complex of the thalamus contains a complete representation of the somatic sensory periphery. The Somatic Sensory Cortex The axons arising from neurons in the ventral posterior complex of the thalamus project to cortical neurons located primarily in layer IV of the somatic sensory cortex (see Figure 8.7; also see Box A in Chapter 25 for a more detailed description of cortical lamination). The primary somatic sensory cortex in humans (also called SI), which is located in the postcentral gyrus of the parietal lobe, comprises four distinct regions, or fields, known as Brodmann’s areas 3a, 3b, 1, and 2. Experiments carried out in nonhuman primates indicate that neurons in areas 3b and 1 respond primarily to cutaneous stimuli, whereas neurons in 3a respond mainly to stimulation of proprioceptors; area 2 neurons process both tactile and proprioceptive stimuli. Mapping studies in humans and other primates show further that each 204 Chapter Eight Somatic sensory cortex Primary somatic sensory cortex (SI) Posterior parietal cortex 1 4 7 3b 2 5 3a Central sulcus Postcentral gyrus Ventral posterior medial nucleus (VPM) Thalamus Secondary somatic sensory cortex (SII) Ventral posterior lateral nucleus (VPL) VP complex Figure 8.7 Diagram of the somatic sensory portions of the thalamus and their cortical targets in the postcentral gyrus. The ventral posterior nuclear complex comprises the VPM, which relays somatic sensory information carried by the trigeminal system from the face, and the VPL, which relays somatic sensory information from the rest of the body. Inset above shows organization of the primary somatosensory cortex in the postcentral gyrus, shown here in a section cutting across the gyrus from anterior to posterior. (After Brodal, 1992, and Jones et al., 1982.) of these four cortical areas contains a separate and complete representation of the body. In these somatotopic maps, the foot, leg, trunk, forelimbs, and face are represented in a medial to lateral arrangement, as shown in Figures 8.8A,B and 8.9. Although the topographic organization of the several somatic sensory areas is similar, the functional properties of the neurons in each region and their organization are distinct (Box D). For instance, the neuronal receptive fields are relatively simple in area 3b; the responses elicited in this region are generally to stimulation of a single finger. In areas 1 and 2, however, the majority of the receptive fields respond to stimulation of multiple fingers. Furthermore, neurons in area 1 respond preferentially to particular directions of skin stimulation, whereas many area 2 neurons require complex stimuli to activate them (such as a particular shape). Lesions restricted to area 3b produce a severe deficit in both texture and shape discrimination. In contrast, damage confined to area 1 affects the ability of monkeys to perform accurate texture discrimination. Area 2 lesions tend to produce deficits in finger coordination, and in shape and size discrimination. A salient feature of cortical maps, recognized soon after their discovery, is their failure to represent the body in actual proportion. When neurosurgeons determined the representation of the human body in the primary sensory (and motor) cortex, the homunculus (literally, “little man”) defined by such mapping procedures had a grossly enlarged face and hands compared to the torso and proximal limbs (Figure 8.8C). These anomalies arise because The Somatic Sensor y System 205 (A) Central sulcus Somatic sensory cortex (B) (C) Shoulder Neck Trunk Arm Head Leg Hand Feet Digits Thumb Toes Neck Eyes Nose Face Genitalia Lips Jaw Tongue Throat Lateral Medial Figure 8.8 Somatotopic order in the human primary somatic sensory cortex. (A) Diagram showing the region of the human cortex from which electrical activity is recorded following mechanosensory stimulation of different parts of the body. The patients in the study were undergoing neurosurgical procedures for which such mapping was required. Although modern imaging methods are now refining these classical data, the human somatotopic map first defined in the 1930s has remained generally valid. (B) Diagram along the plane in (A) showing the somatotopic representation of body parts from medial to lateral. (C) Cartoon of the homunculus constructed on the basis of such mapping. Note that the amount of somatic sensory cortex devoted to the hands and face is much larger than the relative amount of body surface in these regions. A similar disproportion is apparent in the primary motor cortex, for much the same reasons (see Chapter 17). (After Penfield and Rasmussen, 1950, and Corsi, 1991.) manipulation, facial expression, and speaking are extraordinarily important for humans, requiring more central (and peripheral) circuitry to govern them. Thus, in humans, the cervical spinal cord is enlarged to accommodate the extra circuitry related to the hand and upper limb, and as stated earlier, the density of receptors is greater in regions such as the hands and lips. Such distortions are also apparent when topographical maps are compared across species. In the rat brain, for example, an inordinate amount of the somatic sensory cortex is devoted to representing the large facial whiskers that pro- 206 Chapter Eight Figure 8.9 The primary somatic sensory map in the owl monkey based, as in Figure 8.8, on the electrical responsiveness of the cortex to peripheral stimulation. Much more detailed mapping is possible in experimental animals than in neurosurgical patients. The enlargement on the right shows areas 3b and 1, which process most cutaneous mechanosensory information. The arrangement is generally similar to that determined in humans. (After Kaas, 1983.) 3b Foot DV IV III II F 3b 1 1 P. leg Foot pads V DI I Foot A. leg Trunk Arm Hand DV DIV DIII DII Chin U. lip Hand pads IV III DV II I DI H L. lip Oral Arm U. lip Chin L. lip vide a key component of the somatic sensory input for rats and mice (see Boxes B and D), while raccoons overrepresent their paws and the platypus its bill. In short, the sensory input (or motor output) that is particularly significant to a given species gets relatively more cortical representation. Higher-Order Cortical Representations Somatic sensory information is distributed from the primary somatic sensory cortex to “higher-order” cortical fields (as well as to subcortical structures). One of these higher-order cortical centers, the secondary somatosensory cortex (sometimes called SII and adjacent to the primary cortex; see Figure 8.7), receives convergent projections from the primary somatic sensory cortex and sends projections in turn to limbic structures such as the amygdala and hippocampus (see Chapters 28 and 30). This latter pathway is believed to play an important role in tactile learning and memory. Neurons in motor cortical areas in the frontal lobe also receive tactile information from the anterior parietal cortex and, in turn, provide feedback projections to several cortical somatic sensory regions. Such integration of sensory and motor information is considered in Chapters 19 and 25, where the role of these “association” regions of the cerebral cortex are discussed in more detail. Finally, a fundamental but often neglected feature of the somatic sensory system is the presence of massive descending projections. These pathways originate in sensory cortical fields and run to the thalamus, brainstem, and spinal cord. Indeed, descending projections from the somatic sensory cortex outnumber ascending somatic sensory pathways! Although their physiological role is not well understood, it is generally assumed (with some experimental support) that descending projections modulate the ascending flow of sensory information at the level of the thalamus and brainstem. The Somatic Sensor y System 207 Box D Patterns of Organization within the Sensory Cortices: Brain Modules Observations over the last 40 years have made it clear that there is an iterated substructure within the somatic sensory (and many other) cortical maps. This substructure takes the form of units called modules, each involving hundreds or thousands of nerve cells in repeating patterns. The advantages of these iterated patterns for brain function remain largely mysterious; for the neurobiologist, however, such iterated arrangements have provided important clues about cortical connectivity and the mechanisms by which neural activity influences brain development (see Chapters 22 and 23). The observation that the somatic sensory cortex comprises elementary units of vertically linked cells was first noted in the 1920s by the Spanish neuroanatomist Rafael Lorente de Nó, based on his studies in the rat. The potential importance of cortical modularity remained largely unexplored until the 1950s, however, when electrophysiological experiments indicated an arrangement of repeating units in the brains of cats and, later, monkeys. Vernon Mountcastle, a neurophysiologist at Johns Hopkins, found that vertical microelectrode penetrations in the primary somatosensory cortex of these animals encountered cells that responded to the same sort of mechanical stimulus presented at the same location on the body surface. Soon after Mountcastle’s pioneering work, David Hubel and Torsten Wiesel discovered a similar arrangement in the cat primary visual cortex. These and other observations led Mountcastle to the general view that “the elementary pattern of organization of the cerebral cortex is a vertically oriented column or cylinder of cells capable of input-output functions of considerable complexity.” Since these discoveries in the late 1950s and early 1960s, the view that modular circuits represent a fundamental feature of the mammalian cerebral cortex has gained wide acceptance, and many such entities have now been described in various cortical regions (see figure). This wealth of evidence for such patterned circuits has led many neuroscientists to conclude, like Mountcastle, that modules are a fundamental feature of the cerebral cortex, essential for perception, cognition, and perhaps even consciousness. Despite the prevalence of iterated modules, there are some problems with the view that modular units are universally important in cortical function. First, although modular circuits of a given class are readily seen in the brains of some species, they have not been found in the same brain regions of other, sometimes closely related, animals. Second, not all regions of the mammalian cortex are organized in a modular fashion. And third, no clear function of such modules has been discerned, much effort and speculation notwithstanding. This salient feature of the organization of the somatic sensory cortex and other cortical (and some subcortical) regions therefore remains a tantalizing puzzle. (A) (B) (C) (D) (E) References HUBEL, D. H. (1988) Eye, Brain, and Vision. Scientific American Library. New York: W. H. Freeman. LORENTE DE NÓ, R. (1949) The structure of the cerebral cortex. Physiology of the Nervous System, 3rd Ed. New York: Oxford University Press. MOUNTCASTLE, V. B. (1957) Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J. Neurophysiol. 20: 408–434. MOUNTCASTLE, V. B. (1998) Perceptual Neuroscience: The Cerebral Cortex. Cambridge: Harvard University Press. PURVES, D., D. RIDDLE AND A. LAMANTIA (1992) Iterated patterns of brain circuitry (or how the cortex gets its spots). Trends Neurosci. 15: 362–368. WOOLSEY, T. A. AND H. VAN DER LOOS (1970) The structural organization of layer IV in the somatosensory region (SI) of mouse cerebral cortex. The description of a cortical field composed of discrete cytoarchitectonic units. Brain Res. 17: 205–242. (F) Examples of iterated, modular substructures in the mammalian brain. (A) Ocular dominance columns in layer IV in the primary visual cortex (V1) of a rhesus monkey. (B) Repeating units called “blobs” in layers II and III in V1 of a squirrel monkey. (C) Stripes in layers II and III in V2 of a squirrel monkey. (D) Barrels in layer IV in primary somatic sensory cortex of a rat. (E) Glomeruli in the olfactory bulb of a mouse. (F) Iterated units called “barreloids” in the thalamus of a rat. These and other examples indicate that modular organization is commonplace in the brain. These units are on the order of one hundred to several hundred microns across. (From Purves et al., 1992.) 208 Chapter Eight Summary The components of the somatic sensory system considered in this chapter process information conveyed by mechanical stimuli that impinge upon the body surface or that are generated within the body itself (proprioception). This processing is performed by neurons distributed across several brain structures that are connected by both ascending and descending pathways. Transmission of afferent mechanosensory information from the periphery to the brain begins with a variety of receptor types that initiate action potentials. This activity is conveyed centrally via a chain of neurons, referred to as the first-, second-, and third-order cells. First-order neurons are located in the dorsal root and cranial nerve ganglia. Second-order neurons are located in brainstem nuclei. Third-order neurons are found in the thalamus, from whence they project to the cerebral cortex. These pathways are topographically arranged throughout the system, the amount of cortical and subcortical space allocated to various body parts being proportional to the density of peripheral receptors. Studies of non-human primates show that specific cortical regions correspond to each functional submodality; area 3b, for example, processes information from low-threshold cutaneous receptors, and area 3a from proprioceptors. Thus, at least two broad criteria operate in the organization of the somatic sensory system: modality and somatotopy. The end result of this complex interaction is the unified perceptual representation of the body and its ongoing interaction with the environment. Additional Reading Reviews CHAPIN, J. K. (1987) Modulation of cutaneous sensory transmission during movement: Possible mechanisms and biological significance. In Higher Brain Function: Recent Explorations of the Brain’s Emergent Properties. S. P. Wise (ed.). New York: John Wiley and Sons, pp. 181–208. DARIAN-SMITH, I. (1982) Touch in primates. Annu. Rev. Psychol. 33: 155–194. JOHANSSON, R. S. AND A. B. VALLBO (1983) Tactile sensory coding in the glabrous skin of the human. Trends Neurosci. 6: 27–32. KAAS, J. H. (1990) Somatosensory system. In The Human Nervous System. G Paxinos (ed.). San Diego: Academic Press, pp. 813–844. KAAS, J. H. (1993) The functional organization of somatosensory cortex in primates. Ann. Anat. 175: 509–518. KAAS, J. H. AND C. E. COLLINS (2003) The organization of somatosensory cortex in anthropoid primates. Adv. Neurol. 2003: 93: 57–67. MOUNTCASTLE, V. B. (1975) The view from within: Pathways to the study of perception. Johns Hopkins Med. J. 136: 109–131. NICOLELIS, M. A. AND E. E. FANSELOW (2002) Thalamocortical optimization of tactile processing according to behavioral state. Nat. Neurosci. 5(6): 517–523. PETERSEN, R. S., S. PANZERI AND M. E. DIAMOND (2002) Population coding in somatosensory cortex. Curr. Opin. Neurobiol. 12(4): 441–447. WOOLSEY, C. (1958) Organization of somatic sensory and motor areas of the cerebral cortex. In Biological and Biochemical Bases of Behavior. H. F. Harlow and C. N. Woolsey (eds.). Madison, WI: University of Wisconsin Press, pp. 63–82. Important Original Papers ADRIAN, E. D. AND Y. ZOTTERMAN (1926) The impulses produced by sensory nerve endings. II. The response of a single end organ. J. Physiol. 61: 151–171. JOHANSSON, R. S. (1978) Tactile sensibility of the human hand: Receptive field characteristics of mechanoreceptive units in the glabrous skin. J. Physiol. (Lond.) 281: 101–123. JOHNSON, K. O. AND G. D. LAMB (1981) Neural mechanisms of spatial tactile discrimination: Neural patterns evoked by Braille-like dot patterns in the monkey. J. Physiol. (London) 310: 117–144. JONES, E. G. AND D. P. FRIEDMAN (1982) Projection pattern of functional components of thalamic ventrobasal complex on monkey somatosensory cortex. J. Neurophysiol. 48: 521–544. JONES, E. G. AND T. P. S. POWELL (1969) Connexions of the somatic sensory cortex of the rhesus monkey. I. Ipsilateral connexions. Brain 92: 477–502. LAMOTTE, R. H. AND M. A. SRINIVASAN (1987) Tactile discrimination of shape: Responses of rapidly adapting mechanoreceptive afferents to a step stroked across the monkey finger- pad. J. Neurosci. 7: 1672–1681. LAUBACH, M., J. WESSBER AND M. A. L. NICOLELIS (2000) Cortical ensemble activity increasingly predicts behavior outcomes during learning of a motor task. Nature 405: 567–571. MOORE, C. I. AND S. B. NELSON (1998) Spatiotemporal subthreshold receptive fields in the vibrissa representation of rat primary somatosensory cortex. J. Neurophysiol. 80: 2882– 2892. MOORE, C. I., S. B. NELSON AND M. SUR (1999) Dynamics of neuronal processing in rat somatosensory cortex. TINS 22: 513–520. NICOLELIS, M. A. L., L. A. BACCALA, R. C. S. LIN AND J. K. CHAPIN (1995) Sensorimotor encoding by synchronous neural ensemble activity at multiple levels of the somatosensory system. Science 268: 1353–1358. SUR, M. (1980) Receptive fields of neurons in areas 3b and 1 of somatosensory cortex in monkeys. Brain Res. 198: 465–471. WALL, P. D. AND W. NOORDENHOS (1977) Sensory functions which remain in man after complete transection of dorsal columns. Brain 100: 641–653. ZHU, J. J. AND B. CONNORS (1999) Intrinsic firing patterns and whisker-evoked synaptic responses of neurons in the rat barrel cortex. J. Neurophysiol. 81: 1171–1183. Chapter 9 Pain Overview A natural assumption is that the sensation of pain arises from excessive stimulation of the same receptors that generate other somatic sensations (i.e., those discussed in Chapter 8). This is not the case. Although similar in some ways to the sensory processing of ordinary mechanical stimulation, the perception of pain, called nociception, depends on specifically dedicated receptors and pathways. Since alerting the brain to the dangers implied by noxious stimuli differs substantially from informing it about innocuous somatic sensory stimuli, it makes good sense that a special subsystem be devoted to the perception of potentially threatening circumstances. The overriding importance of pain in clinical practice, as well as the many aspects of pain physiology and pharmacology that remain imperfectly understood, continue to make nociception an extremely active area of research. Nociceptors The relatively unspecialized nerve cell endings that initiate the sensation of pain are called nociceptors (noci is derived from the Latin nocere, “to hurt”). Like other cutaneous and subcutaneous receptors, they transduce a variety of stimuli into receptor potentials, which in turn trigger afferent action potentials (see Figure 8.2). Moreover, nociceptors, like other somatic sensory receptors, arise from cell bodies in dorsal root ganglia (or in the trigeminal ganglion) that send one axonal process to the periphery and the other into the spinal cord or brainstem (see Figure 8.1). Because peripheral nociceptive axons terminate in unspecialized “free endings,” it is conventional to categorize nociceptors according to the properties of the axons associated with them (see Table 8.1). As described in the previous chapter, the somatic sensory receptors responsible for the perception of innocuous mechanical stimuli are associated with myelinated axons that have relatively rapid conduction velocities. The axons associated with nociceptors, in contrast, conduct relatively slowly, being only lightly myelinated or, more commonly, unmyelinated. Accordingly, axons conveying information about pain fall into either the Aδ group of myelinated axons, which conduct at about 20 m/s, or into the C fiber group of unmyelinated axons, which conduct at velocities generally less than 2 m/s. Thus, even though the conduction of all nociceptive information is relatively slow, there are fast and slow pain pathways. In general, the faster-conducting Aδ nociceptors respond either to dangerously intense mechanical or to mechanothermal stimuli, and have receptive fields that consist of clusters of sensitive spots. Other unmyelinated nociceptors tend to respond to thermal, mechanical, and chemical stimuli, and are 209 210 Chapter Nine (A) (B) Record Heat stimulus Nociceptor 45° Stimulus Nonnociceptive thermoreceptor (C) Figure 9.1 Experimental demonstration that nociception involves specialized neurons, not simply greater discharge of the neurons that respond to normal stimulus intensities. (A) Arrangement for transcutaneous nerve recording. (B) In the painful stimulus range, the axons of thermoreceptors fire action potentials at the same rate as at lower temperatures; the number and frequency of action potential discharge in the nociceptive axon, however, continues to increase. (Note that 45°C is the approximate threshold for pain.) (C) Summary of results. (After Fields, 1987.) Thermoreceptor Magnitude of afferent response (action potentials per second) Nociceptor 0 40 45 Temperature (°C) 50 therefore said to be polymodal. In short, there are three major classes of nociceptors in the skin: Aδ mechanosensitive nociceptors; Aδ mechanothermal nociceptors; and polymodal nociceptors, the latter being specifically associated with C fibers. The receptive fields of all pain-sensitive neurons are relatively large, particularly at the level of the thalamus and cortex, presumably because the detection of pain is more important than its precise localization. Studies carried out in both humans and experimental animals demonstrated some time ago that the rapidly conducting axons that subserve somatic sensory sensation are not involved in the transmission of pain. A typical experiment of this sort is illustrated in Figure 9.1. The peripheral axons responsive to nonpainful mechanical or thermal stimuli do not discharge at a greater rate when painful stimuli are delivered to the same region of the skin surface. The nociceptive axons, on the other hand, begin to discharge only when the strength of the stimulus (a thermal stimulus in the example in Figure 9.1) reaches high levels; at this same stimulus intensity, other thermoreceptors discharge at a rate no different from the maximum rate already achieved within the nonpainful temperature range, indicating that there are both nociceptive and nonnociceptive thermoreceptors. Equally important, direct stimulation of the large-diameter somatic sensory afferents at any frequency in humans does not produce sensations that are described as painful. In contrast, the smaller-diameter, more slowly conducting Aδ and C fibers are active when painful stimuli are delivered; and when stimulated electrically in human subjects, they produce pain. How, then, do these different classes of nociceptors lead to the perception of pain? As mentioned, one way of determining the answer has been to stimulate different nociceptors in human volunteers while noting the sensations reported. In general, two categories of pain perception have been described: a sharp first pain and a more delayed, diffuse, and longer-lasting sensation that is generally called second pain (Figure 9.2A). Stimulation of the large, rapidly conducting Aα and Aβ axons in peripheral nerves does not elicit the sensation of pain. When the stimulus intensity is raised to a level that activates a subset of Aδ fibers, however, a tingling sensation or, if the stimulation is intense enough, a feeling of sharp pain is reported. If the stimulus intensity is increased still further, so that the small-diameter, slowly conducting C fiber axons are brought into play, then a duller, longer-lasting Pain 211 (A) (B) Subjective pain intensity Aδ fiber (C) C fiber X X First pain Second pain Time sensation of pain is experienced. It is also possible to selectively anesthetize C fibers and Aδ fibers; in general, these selective blocking experiments confirm that the Aδ fibers are responsible for first pain, and that C fibers are responsible for the duller, longer-lasting second pain (Figure 9.2B,C). Transduction of Nociceptive Signals Given the variety of stimuli (mechanical, thermal, and chemical) that can give rise to painful sensations, the transduction of nociceptive signals is a complex task. While many puzzles remain, some insights have come from the identification of specific receptors associated with nociceptive afferent endings. These receptors are sensitive to both heat and to capsaicin, the ingredient in chili peppers that is responsible for the familiar tingling or burning sensation produced by spicy foods (Box A). The so-called vanilloid receptor (VR-1 or TRPV1) is found in C and Aδ fibers and is activated by moderate heat (45°C—a temperature that is perceived as uncomfortable) as well as by capsaicin. Another type of receptor (vanilloid-like receptor, VRL-1 or TRPV2) has a higher threshold response to heat (52°C), is not sensitive to capsaicin, and is found in Aδ fibers. Both are members of the larger family of transient receptor potential (TRP) channels, first identified in studies of the phototransduction pathway in fruit flies and now known to comprise a large number of receptors sensitive to different ranges of heat and cold. Structurally, TRP channels resemble voltage-gated potassium or cyclic nucleotide-gated channels, having six transmembrane domains with a pore between domains 5 and 6. Under resting conditions the pore of the channel is closed. In the open, activated state, these receptors allow an influx of sodium and calcium that initiates the generation of action potentials in the nociceptive fibers. Since the same receptor is responsive to heat as well as capsaicin, it is not surprising that chili peppers seem “hot.” A puzzle, however, is why the nervous system has evolved receptors that are sensitive to a chemical in chili peppers. As with the case of other plant compounds that selectively activate neural receptors (see the discussion of opiates below), it seems likely that TRPV1 receptors detect endogenous substances whose chemical structure resembles that of capsaicin. In fact, there is now some evidence that ‘endovanilloids’ that are produced by peripheral tissues in response to injury, Figure 9.2 Pain can be separated into an early perception of sharp pain and a later sensation that is described as having a duller, burning quality. (A) First and second pain, as these sensations are called, are carried by different axons, as can be shown by (B) the selective blockade of the more rapidly conducting myelinated axons that carry the sensation of first pain, or (C) blockade of the more slowly conducting C fibers that carry the sensation of second pain. (After Fields, 1990.) 212 Chapter Nine Box A Capsaicin Capsaicin, the principle ingredient responsible for the pungency of hot peppers, is eaten daily by over a third of the world’s population. Capsaicin activates responses in a subset of nociceptive C fibers (polymodal nociceptors; see Chapter 9) by opening ligand-gated ion channels that permit the entry of Na+ and Ca2+. One of these channels (VR-1) has been cloned and has been found to be activated by capsaicin, acid, and anandamide (an endogeneous compound that also activates cannabanoid receptors), and by heating the tissue to about 43°C. It follows that anandamide and temperature are probably the endogenous activators of these channels. Mice whose VR-1 receptors have been knocked out drink capsaicin solutions as if they were water. Receptors for capsaicin have been found in polymodal nociceptors of all mammals, but are not present in birds (leading to the produc- tion of squirrel-proof birdseed laced with capsaicin!). When applied to the mucus membranes of the oral cavity, capsaicin acts as an irritant, producing protective reactions. When injected into skin, it produces a burning pain and elicits hyperalgesia to thermal and mechanical stimuli. Repeated applications of capsaicin also desensitize pain fibers and prevent neuromodulators such as substance P, VIP, and somatostatin from being released by peripheral and central nerve terminals. Consequently, capsaicin is used clinically as an analgesic and anti-inflammatory agent; it is usually applied topically in a cream (0.075%) to relieve the pain associated with arthritis, postherpetic neuralgia, mastectomy, and trigeminal neuralgia. Thus, this remarkable chemical irritant not only gives gustatory pleasure on an enormous scale, but is also a useful pain reliever! (A) References CATERINA, M. J., M. A. SCHUMACHER, M. TOMINAGA, T. A ROSEN, J. D. LEVINE AND D. JULIUS (1997) The capsaicin receptor: A heat-activated ion channel in the pain pathway. Nature 389: 816–766. CATERINA, M. J. AND 8 OTHERS (2000) Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science 288: 306–313. SZALLASI, A. AND P. M. BLUMBERG (1999) Vanilloid (capsaicin) receptors and mechanisms. Pharm. Reviews 51: 159–212. TOMINAGA, M. AND 8 OTHERS (1998) The cloned capsaicin receptor integrates multiple pain-producing stimuli. Neuron 21: 531–543. ZYGMUNT, P. M. AND 7 OTHERS (1999) Vanilloid receptors on sensory nerves mediate the vasodilator action of anandamide. Nature 400: 452–457. (B) Capsaicin Red chile CH3O Habañero O N H HO Jalapeño (D) Capsaicin Ca2+ Na+ Heat H+ (C) Outside Inside (A) Some popular peppers that contain capsaicin. (B) The chemical structure of capsaicin. (C) The capsaicin molecule. (D) Schematic of the VR-1/capsaicin receptor channel. This channel can be activated by capsaicin intercellularly, or by heat or protons (H+) at the cell surface. VR-1 receptor Pain 213 and that these substances, along with other factors, contribute to the nocieceptive response to injury. Central Pain Pathways The pathways that carry information about noxious stimuli to the brain, as might be expected for such an important and multifaceted system, are also complex (see Boxes B and C). It helps in understanding this complexity to distinguish two components of pain: the sensory discriminative component, which signals the location, intensity, and quality of the noxious stimululation, and the affective-motivational component of pain—which signals the unpleasant quality of the experience, and enables the autonomic activation that follows a noxious stimulus (the classic fight-or-flight reaction; see Chapter 20). The discriminative component is thought to depend on pathways that target the traditional somatosensory areas of cortex, while the affective- motivational component is thought to depend on additional cortical and brainstem pathways. The major pathways are summarized in Figure 9.3. Pathways responsible for the discriminative component of pain originate with other sensory neurons, in dorsal root ganglia and, like other sensory nerve cells the central axons of nociceptive nerve cells enter the spinal cord via the dorsal roots (Figure 9.3A). When these centrally projecting axons reach the dorsal horn of the spinal cord, they branch into ascending and descending collaterals, forming the dorsolateral tract of Lissauer (named after the German neurologist who first described this pathway in the late nineteenth century). Axons in Lissauer’s tract typically run up and down for one or two spinal cord segments before they penetrate the gray matter of the dorsal horn. Once within the dorsal horn, the axons give off branches that contact neurons located in several of Rexed’s laminae (these laminae are the descriptive divisions of the spinal gray matter in cross section, again named after the neuroanatomist who described these details in the 1950s). The axons of these second-order neurons in the dorsal horn of the spinal cord cross the midline and ascend all the way to the brainstem and thalamus in the anterolateral (also called ventrolateral) quadrant of the contralateral half of the spinal cord. These fibers form the spinothalamic tract, the major ascending pathway for information about pain and temperature. This overall pathway is also referred to as the anterolateral system, much as the mechanosensory pathway is referred to as the dorsal column–medial lemniscus system. The location of the spinothalamic tract is particularly important clinically because of the characteristic sensory deficits that follow certain spinal cord injuries. Since the mechanosensory pathway ascends ipsilaterally in the cord, a unilateral spinal lesion will produce sensory loss of touch, pressure, vibration, and proprioception below the lesion on the same side. The pathways for pain and temperature, however, cross the midline to ascend on the opposite side of the cord. Therefore, diminished sensation of pain below the lesion will be observed on the side opposite the mechanosensory loss (and the lesion). This pattern is referred to as a dissociated sensory loss and (together with local dermatomal signs; see Box C in Chapter 8) helps define the level of the lesion (Figure 9.4). As is the case of the mechanosensory pathway, information about noxious and thermal stimulation of the face follows a separate route to the thalamus (see Figure 9.3B). First-order axons originating from the trigeminal ganglion cells and from ganglia associated with nerves VII, IX, and X carry information from facial nociceptors and thermoreceptors into the brainstem. After 214 Chapter Nine (A) (B) Cerebrum Cerebrum Primary somatic sensory cortex Ventral posterior medial nucleus of thalamus Midbrain Ventral posterior lateral nucleus of the thalamus Midbrain Trigeminothalamic tract Spinothalamic tract Mid-pons Mid-pons Pain and temperature information from face Middle medulla Middle medulla Spinal trigeminal tract (afferent axons) Caudal medulla Anterolateral system Cervical spinal cord Lumbar spinal cord Caudal medulla Pain and temperature information from upper body (excluding the face) Pain and temperature information from lower body Spinal nucleus of the trigeminal complex Figure 9.3 Major pathways for the discriminative aspects of pain and temperature sensation. (A) The spinothalamic system. (B) The trigeminal pain and temperature system, which carries information about these sensations from the face. Pain 215 Box B Referred Pain Surprisingly, there are few, if any, neurons in the dorsal horn of the spinal cord that are specialized solely for the transmission of visceral pain. Obviously, we recognize such pain, but it is conveyed centrally via dorsal horn neurons that are also concerned with cutaneous pain. As a result of this economical arrangement, the disorder of an internal organ is sometimes perceived as cutaneous pain. A patient may therefore present to the physician with the complaint of pain at a site other than its actual source, a potentially confusing phenomenon called referred pain. The most common clinical example is anginal pain (pain arising from heart muscle that is not being adequately perfused with blood) referred to the upper chest wall, with radiation into the left arm and hand. Other important examples are gallbladder pain referred to the scapular region, esophogeal pain referred to the chest wall, ureteral pain (e.g., from passing a kidney stone) referred to the lower abdominal wall, bladder pain referred to the perineum, and the pain from an inflamed appendix referred to the anterior abdominal wall around the umbilicus. Understanding referred pain can lead to an astute diagnosis that might otherwise be missed. References CAPPS, J. A. AND G. H. COLEMAN (1932) An Experimental and Clinical Study of Pain in the Pleura, Pericardium, and Peritoneum. New York: Macmillan. HEAD, H. (1893) On disturbances of sensation with special reference to the pain of visceral disease. Brain 16: 1–32. KELLGREW, J. H. (1939–1942) On the distribution of pain arising from deep somatic structures with charts of segmental pain areas. Clin. Sci. 4: 35–46. Examples of pain arising from a visceral disorder referred to a cutaneous region (color). Esophagus Heart Left ureter Urinary/bladder Right prostate 216 Chapter Nine Normal sensation Zone of complete loss of sensation Reduced sensation of temperature and pain Reduced sensation of two-point discrimination, vibration, and proprioception Figure 9.4 Pattern of “dissociated” sensory loss following a spinal cord hemisection at the 10th thoracic level on the left side. This pattern, together with motor weakness on the same side as the lesion, is sometimes referred to as Brown-Séquard syndrome. entering the pons, these small myelinated and unmyelinated trigeminal fibers descend to the medulla, forming the spinal trigeminal tract (or spinal tract of cranial nerve V), and terminate in two subdivisions of the spinal trigeminal complex: the pars interpolaris and pars caudalis. Axons from the second-order neurons in these two trigeminal nuclei, like their counterparts in the spinal cord, cross the midline and ascend to the contralateral thalamus in the trigeminothalamic tract. The principal target of the spinothalamic and trigeminothalamic pathway is the ventral posterior nucleus of the thalamus. Similar to the organization of the mechanosensory pathways, information from the body terminates in the VPL, while information from the face terminate in the VPM. These nuclei send their axons to primary and secondary somatosensory cortex. The nociceptive information transmitted to these cortical areas is thought to be responsible for the discriminative component of pain: identifying the location, the intensity, and quality of the stimulation. Consistent with this interpretation, electrophysiological recordings from nociceptive neurons in S1, show that these neurons have small localized receptive fields, properties commensurate with behavioral measures of pain localization. The affective–motivational aspect of pain is evidently mediated by separate projections of the anterolateral system to the reticular formation of the midbrain (in particular the parabrachial nucleus), and to thalamic nuclei that lie medial to the ventral posterior nucleus (including the so-called intralaminar nuclei; see Figure 9.5). Studies in rodents show that neurons in the parabrachial nucleus respond to most types of noxious stimuli, and have large receptive fields that can include the whole surface of the body. Neurons in the parabrachial nucleus project in turn to the hypothalamus and the amygdala, thus providing nociceptive information to circuits known to be concerned with motivation and affect (see Chapter 28). These parabrachial targets are also the source of projections to the periaqueductal grey of the midbrain, a structure that plays an important role in the descending control of activity in the pain pathway. Nociceptive inputs to the parabrachial nucleus and to the ventral posterior nucleus arise from separate populations of neurons in the dorsal horn of the spinal cord. Parabrachial inputs arise from neurons in the most superficial part of the dorsal horn (lamina I), while ventral posterior inputs arise from deeper parts of the dorsal horn (e.g., lamina V). By taking advantage of the unique molecular signature of these two sets of neurons, it has been possible to selectively eliminate the nociceptive inputs to the parabrachial nucleus in rodents. In these animals, the behavioral responses to the presentation of noxious stimulation (capsaicin, for example) are substantially attenuated. Projections from the anterolateral system to the medial thalamic nuclei provide nociceptive signals to areas in the frontal lobe, the insula and the cingulate cortex (Figure 9.5). In accord with this anatomy, functional imaging studies in humans have shown a strong correlation between activity in the anterior cingulate cortex and the experience of a painful stimulus. Moreover, experiments using hypnosis have been able to tease apart the neural response to changes in the intensity of a painful stimulus from changes in its unpleasantness. Changes in intensity are accompanied by changes in the activity of neurons in somatosensory cortex, with little change in the activity of cingulate cortex, whereas changes in unpleasantness are correlated with changes in the activity of neurons in cingulate cortex. From this description, it should be evident that the full experience of pain involves the cooperative action of an extensive network of brain regions whose properties are only beginning to be understood (Box C). The cortical Pain 217 Cerebrum Insula Cingulate cortex Projections to the amygdala and hypothalamus Mid-pons Parabrachial nucleus Intralaminar nuclei of the thalamus Superior cerebellar peduncle Reticular formation Middle medulla Caudal medulla Anterolateral system Cervical spinal cord Lumbar spinal cord Information from upper body (excluding the face) Information from lower body Figure 9.5 Affective–motivational pain pathways. Nociceptive information critical for signaling the unpleasant quality of pain is mediated by projections to the reticular formation (including the parabrachial nucleus) and to the intralaminar nuclei of the thalamus. 218 Chapter Nine Box C A Dorsal Column Pathway for Visceral Pain Chapters 8 and 9 present a framework for considering the central neural pathways that convey innocuous mechanosensory signals and painful signals from cutaneous and deep somatic sources. Considering just the signals derived from the body below the head, discriminitive mechanosensory and proprioceptive information travels to the ventral posterior thalamus via the dorsal-column medial lemniscal system (see Figure 8.6A), while nociceptive information travels to the same (and additional) thalamic relays via the anterolateral systems (see Figure 9.3A). But how do painful signals that arise in the visceral organs of the pelvis, abdomen, and thorax enter the central nervous system and ultimately reach consciousness? The answer is via a newly discovered component of the dorsal column medial lemniscal pathway that conveys visceral nociception. Although Chapter 20 will present more information on the systems that receive and process visceral sensory information, at this juncture it is worth considering this component of the pain pathways and how this particular pathway has begun to impact clinical medicine. Primary visceral afferents from the pelvic and abdominal viscera enter the spinal cord and synapse on second-order neurons in the dorsal horn of the lumbar-sacral spinal cord. As discussed in Box A and Chapter 20, some of these second-order neurons are cells that give rise to the anterolateral systems and contribute to referred visceral pain patterns. However, other neurons—perhaps primarily those that give rise to nociceptive signals—synapse upon neurons in the intermediate gray region of the spinal cord near the central canal. These neurons, in turn, send their axons not through the anterolateral white matter of the spinal cord (as might be expected for a pain pathway) but through the dorsal columns in a position very near the midline (see Figure A). Similarly, secondorder neurons in the thoracic spinal cord that convey nociceptive signals from thoracic viscera send their axons rostrally through the dorsal columns along the dorsal intermediate septum, near the division of the gracile and cuneate fasciculi. These second order axons then synapse in the dorsal column nuclei of the caudal medulla, where neurons give rise to arcuate fibers that form the contralateral medial lemniscus and eventually synapse on thalamocortical projection neurons in the ventral-posterior thalamus. This dorsal column visceral sensory projection now appears to be the principal pathway by which painful sensations arising in the viscera are detected and discriminated. Several observations support this conclusion: (1) neurons in the ventral posterior lateral nucleus, nucleus gracilis and near the central canal of the spinal cord all respond to noxious visceral stimulation; (2) responses of neurons in the ventral posterior lateral nucleus and nucleus gracilis to such stimulation are greatly reduced by spinal lesions of the dorsal columns (see Figure B), but not lesions of the anterolateral white matter; and (3) infusion of drugs that block nociceptive synaptic transmission into the intermediate gray region of the sacral spinal cord blocks the responses of neurons in the nucleus gracilis to noxious visceral stimulation, but not to innocuous cutaneous stimulation. The discovery of this visceral sensory component in the dorsal-column medial lemniscal system has helped to explain why surgical transection of the axons that run in the medial part of the dorsal columns (a procedure termed midline myelotomy) generates significant relief from the debilitating pain that can result from visceral cancers in the abdomen and pelvis. Although the initial develop- ment of this surgical procedure preceded the elucidation of this visceral pain pathway, these new discoveries have renewed interest in midline myelotomy as a palliative neurosurgical intervention for cancer patients whose pain is otherwise unmanageable. Indeed, precise knowledge of the visceral sensory pathway in the dorsal columns has led to further refinements that permit a minimally invasive (“punctate”) surgical procedure that attempts to interupt the secondorder axons of this pathway within just a single spinal segment (typically, a midor lower-thoracic level; see Figure C). In so doing, this procedure offers some hope to patients who struggle to maintain a reasonable quality of life in extraordinarily difficult circumstances. References AL-CHAER, E. D., N. B. LAWAND, K. N. WESTW. D. WILLIS (1996) Visceral nociceptive input into the ventral posterolateral nucleus of the thalamus: a new function for the dorsal column pathway. J. Neurophys. 76: 2661–2674. AL-CHAER, E. D., N. B. LAWAND, K. N. WESTLUND AND W. D. WILLIS (1996) Pelvic visceral input into the nucleus gracilis is largely mediated by the postsynaptic dorsal column pathway. J. Neurophys. 76: 2675–2690. BECKER, R., S. GATSCHER, U. SURE AND H. BERTALANFFY (2001) The punctate midline myelotomy concept for visceral cancer pain control – case reort and review of the literature. Acta Neurochir. [Suppl.] 79: 77–78. HITCHCOCK, E. R. (1970) Stereotactic cervical myelotomy. J. Neurol. Neurosurg. Psychiatry 33: 224–230. KIM, Y. S. AND S. J. KWON (2000) High thoracic midline dorsal column myelotomy for severe visceral pain due to advanced stomach cancer. Neurosurg. 46:85-90. NAUTA, H. AND 8 OTHERS (2000) Punctate midline myelotomy for the relief of visceral cancer pain. J. Neurosurg. (Spine 2) 92: 125–130. WILLIS, W. D., E. D. AL-CHAER, M. J. QUAST AND K. N. WESTLUND (1999) A visceral pain pathway in the dorsal column of the spinal cord. Proc. Natl. Acad. Sci. USA 96: 7675–7679. LUND AND Pain 219 (A) Cerebrum (B) Sham lesion Dorsal column lesion Before surgery 4 months after surgery (C) Ventral posterior nuclear complex of thalamus Needle Insular cortex Dorsal columns Dorsal horn Midbrain Gastrointestinal tract Gracile nucleus Cuneate nucleus Medial leminiscus Medulla Dorsal root ganglion cells Spinal cord (A) A visceral pain pathway in the dorsal-column medial lemniscal system. For simplicity, only the pathways that mediate visceral pain from the pelvis and lower abdomen are illustrated. The mechanosensory component of this system for the discrimination of tactile stimuli and the anterolateral system for the detection of painful and thermal cutaneous stimuli are also shown for comparison (see also Figures 8.6A and 9.3A). (B) Empirical evidence supporting the existence of the visceral pain pathway shown in (A). Increased neural activity was observed with functional MRI techniques in the thalamus of monkeys that were subjected to noxious distention of the colon and rectum, indicating the processing of visceral pain. This activity was abolished by lesion of the dorsal columns at T10, but not by “sham” surgery. (From Willis et al., 1999.) (C) Top, one method of punctate midline myelotomy for the relief of severe visceral pain. Bottom, myelinstained section of the thoracic spinal cord (T10) from a patient who underwent midline myelotomy for the treatment of colon cancer pain that was not controlled by analgesics. After surgery, the patient experienced relief from pain during the remaining three months of his life. (From Hirshberg et al., 1996; drawing after Nauta et al., 1997.) 220 Chapter Nine representation of pain is the least well documented aspect of the central pathways for nociception, and further studies will be needed to elucidate the contribution of regions outside the somatosensory areas of the parietal lobe. Nevertheless, a prominent role for these areas in the perception of pain is suggested by the fact that ablations of the relevant regions of the parietal cortex do not generally alleviate chronic pain (although they impair contralateral mechanosensory perception, as expected). Sensitization Following a painful stimulus associated with tissue damage (e.g., cuts, scrapes, and bruises), , stimuli in the area of the injury and the surrounding region that would ordinarily be perceived as slightly painful are perceived as significantly more so, a phenomenon referred to as hyperalgesia. A good example of hyperalgesia is the increased sensitivity to temperature that occurs after a sunburn. This effect is due to changes in neuronal sensitivity that occur at the level of peripheral receptors as well as their central targets. Peripheral sensitization results from the interaction of nociceptors with the “inflammatory soup” (Figure 9.6) of substances released when tissue is damaged. These products of tissue damage include extracellular protons, arachidonic acid and other lipid metabolites, bradykinin, histamine, serotonin, prostaglandins, nucleotides, and nerve growth factor (NGF), all of which can interact with receptors or ion channels of nociceptive fibers, augmenting their response. For example, the responses of the TRPV1 receptor to heat can be potentiated by direct interaction of the channel with extracellular protons or lipid metabolites. NGF and bradykinin also potentiate the Tissue injury Bradykinin ATP Prostaglandin 5–HT H+ Blood vessel Figure 9.6 Inflammatory response to tissue damage. Substances released by damaged tissues augment the response of nociceptive fibers. In addition, electrical activation of nociceptors causes the release of peptides and neurotransmitters that further contribute to the inflammatory response. Mast cell or neutrophil Histamine Substance P Anterolateral system CGRP Substance P Dorsal root ganglion cell body Spinal cord Pain 221 activity of the TRPV1 receptors, but do so indirectly through the actions of separate cell-surface receptors (TrkA and bradykinin receptors respectively) and their associated intracellular signalling pathways. The prostaglandins are thought to contribute to peripheral sensitization by binding to G-proteincoupled receptors that increase levels of cyclic AMP within nociceptors. Prostaglandins also reduce the threshold depolarization required for generating action potentials via phosphorylation of a specific class of TTX-resistant Na channels that are expressed in nociceptors. In addition, electrical activity in the nociceptors causes them to release peptides and neurotransmitters such as substance P, calcitonin-gene–related peptide (CGRP) and ATP which further contribute to the inflammatory response, causing vasodilation, swelling, and the release of histamine from mast cells. The presumed purpose of the complex chemical signaling arising from local damage is not only to protect the injured area (as a result of the painful perceptions produced by ordinary stimuli close to the site of damage), but also to promote healing and guard against infection by means of local effects such as increased blood flow and the migration of white blood cells to the site. Obviously the identification of the components of the inflammatory soup and their mechanisms of action is a fertile area to explore for potential analgesics (i.e., compounds that reduce pain intensity). For example, so-called nonsteroidal anti-inflammatory drugs (NSAIDs), which include aspirin and ibuprofen, act by inhibiting cyclooxygenase (COX), an enzyme important in the biosynthesis of prostaglandins. Central sensitization refers to an immediate onset, activity dependent increase in the excitability of neurons in the dorsal horn of the spinal cord following high levels of activity in the nociceptive afferents. As a result, activity levels in nociceptive afferents that were subthreshold prior to the sensitizing event, become sufficient to generate action potentials in dorsal horn neurons, contributing to an increase in pain sensitivity. Although central sensitization is triggered in dorsal horn neurons by activity in nociceptors, the effects generalize to other inputs that arise from low threshold mechanoreceptors. Thus, stimuli that under normal conditions would be innocuous (such as brushing the surface of the skin) activate second-order neurons in the dorsal horn that receive nociceptive inputs, and give rise to a sensation of pain. The induction of pain by what is normally an innocuous stimulus is referred to as allodynia. This phenomenon typically occurs immediately after the painful event and can outlast the stimulus by several hours. Like its peripheral counterpart, a number of different mechanisms contribute to central sensitization, and these can be divided broadly into transcription independent and dependent processes. One form of transcription independent central sensitization called “windup” involves a progressive increase in the discharge rate of dorsal horn neurons in response to repeated low frequency activation of nociceptive afferents. A behavioral correlate of the windup phenomenon has been studied by examining the perceived intensity of pain in response to multiple presentations of a noxious stimulus. Although the intensity of the stimulation is constant, the perceived intensity increases with each stimulus presentation. Windup lasts only during the period of stimulation and arises from the summation of the slow synaptic potentials that are evoked in dorsal horn neurons by nociceptive inputs. The sustained depolarization of the dorsal horn neurons results in part from the activation of voltage dependent L-type calcium channels, and from the removal of the Mg block of NMDA receptors, increasing the sensitivity of the 222 Chapter Nine Box D Phantom Limbs and Phantom Pain Following the amputation of an extremity, nearly all patients have an illusion that the missing limb is still present. Although this illusion usually diminishes over time, it persists in some degree throughout the amputee’s life and can often be reactivated by injury to the stump or other perturbations. Such phantom sensations are not limited to amputated limbs; phantom breasts following mastectomy, phantom genitalia following castration, and phantoms of the entire lower body following spinal cord transection have all been reported. Phantoms are also common after local nerve block for surgery. During recovery from brachial plexus anesthesia, for example, it is not unusual for the patient to experience a phantom arm, perceived as whole and intact, but displaced from the real arm. When the real arm is viewed, the phantom appears to “jump into” the arm and may emerge and reenter intermittently as the anesthesia wears off. These sensory phantoms demonstrate that the central machinery for processing somatic sensory information is not idle in the absence of peripheral stimuli; apparently, the central sensory processing apparatus continues to operate independently of the periphery, giving rise to these bizarre sensations. Interestingly, considerable functional reorganization of somatotopic maps in the primary somatosensory cortex occurs in amputees (see Chapter 24). This reorganization starts immediately after the amputation and tends to evolve for several years. One of the effects of this process is that neurons that have lost their original inputs (i.e., from the removed limb) respond to tactile stimulation of other body parts. A surprising consequence is that stimulation of the face, for example, can be experienced as if the missing limb had been touched. Further evidence that the phenomenon of phantom limb is the result of a central representation is the experience of children born without limbs. Such individuals have rich phantom sensations, despite the fact that a limb never developed. This observation suggests that a full represenation of the body exists independently of the peripheral elements that are mapped. Based on these results, Ronald Melzack proposed that the loss of a limb generates an internal mismatch between the brain’s representation of the body and the pattern of peripheral tactile input that reaches the neocortex. The consequence would be an illusory sensation that the missing body part is still present and functional. With time, the brain may adapt to this loss and alter its intrinsic somatic representation to better accord with the new configuration of the body. This change could explain why the phantom sensation appears almost immediately after limb loss, but usually decreases in intensity over time. Phantoms might simply be a curiosity—or a provocative clue about higherorder somatic sensory processing—were it not for the fact that a substantial number of amputees also develop phantom pain. This common problem is usually described as a tingling or burning sensation in the missing part. Sometimes, however, the sensation becomes a more seri- the dorsal horn neuron to glutamate, the transmitter in nociceptive afferents. Other forms of central sensitization that last longer than the period of sensory stimulation (such as allodynia) are thought to involve an LTP-like enhancement of postsynaptic potentials (see Chapter 24). The longest lasting forms, resulting from transcription dependent processes, can be elicited by changes in neuronal activity or by humoral signals. Those elicited by neuronal activity are localized to the site of the injury, while humoral activation can lead to more widespread changes. For example, cytokines released from microglia and from other sources promote the widespread transcription of COX-2 and the production of prostaglandins in dorsal horn neurons. As described for nociceptive afferents, increased levels of prostaglandins in CNS neurons augments neuronal excitability. Thus the analgesic effects of drugs that inhibit COX are due to actions in both the periphery and within the dorsal horn. As injured tissue heals, the sensitization induced by peripheral and central mechanisms typically declines and the theshold for pain returns to Pain 223 Drawings of phantom arms and legs, based on patients’ reports. The phantom is indicated by a dashed line, with the colored regions showing the most vividly experienced parts. Note that some phantoms are telescoped into the stump. (After Solonen, 1962.) ous pain that patients find increasingly debilitating. Phantom pain is, in fact, one of the more common causes of chronic pain syndromes and is extraordinarily difficult to treat. Because of the widespread nature of central pain processing, ablation of the spinothalamic tract, portions of the thalamus, or even primary sensory cortex does not generally relieve the discomfort felt by these patients. References MELZACK, R. (1989) Phantom limbs, the self, and the brain. The D.O. Hebb Memorial Lecture. Canad. Psychol. 30: 1–14. MELZACK, R. (1990) Phantom limbs and the concept of a neuromatrix. TINS 13: 88–92. preinjury levels. However, when the afferent fibers or central pathways themselves are damaged—a frequent complication in pathological conditions that include diabetes, shingles, AIDs, multiple sclerosis, and stroke— these processes can persist. The resulting condition is refered to as neuropathic pain, a chronic, intensely painful experience that is difficult to treat with conventional analgesic medications. (See Box D for a description of neuropathic pain associated with amputation of an extremity.) The pain can arise spontaneously (without a stimulus) or can be produced by mild forms of stimulation that are common to everyday experience, such as the gentle touch and pressure of clothing, or warm and cool temperatures. Patients often describe their experience as a constant burning sensation interrupted by episodes of shooting, stabbing, or electric shocklike jolts. Because the disability and psychological stress associated with chronic neuropathic pain can be severe, much present research is being devoted to better understanding of the mechanisms of peripheral and central sensitization with the hope of more effective therapies for this debilitating syndrome. NASHOLD, B. S., JR. (1991) Paraplegia and pain. In Deafferentation Pain Syndromes: Pathophysiology and Treatment, B. S. Nashold, Jr. and J. Ovelmen-Levitt (eds.). New York: Raven Press, pp. 301–319. RAMACHANDRAN, V. S. AND S. BLAKESLEE (1998) Phantoms in the Brain. New York: William Morrow & Co. SOLONEN, K. A. (1962) The phantom phenomenon in amputated Finnish war veterans. Acta. Orthop. Scand. Suppl. 54: 1–37. 224 Chapter Nine Descending Control of Pain Perception With respect to the interpretation of pain, observers have long commented on the difference between the objective reality of a painful stimulus and the subjective response to it. Modern studies of this discrepancy have provided considerable insight into how circumstances affect pain perception and, ultimately, into the anatomy and pharmacology of the pain system. During World War II, Henry Beecher and his colleagues at Harvard Medical School made a fundamental observation. In the first systematic study of its kind, they found that soldiers suffering from severe battle wounds often experienced little or no pain. Indeed, many of the wounded expressed surprise at this odd dissociation. Beecher, an anesthesiologist, concluded that the perception of pain depends on its context. For instance, the pain of an injured soldier on the battlefield would presumably be mitigated by the imagined benefits of being removed from danger, whereas a similar injury in a domestic setting would present quite a different set of circumstances that could exacerbate the pain (loss of work, financial liability, and so on). Such observations, together with the well-known placebo effect (discussed in the next section), make clear that the perception of pain is subject to central modulation (all sensations are subject to at least some degree of this kind of modification). This statement, incidentally, should not be taken as a vague acknowledgment about the importance of psychological or “topdown” influences on sensory experience. On the contrary, there has been a gradual realization among neuroscientists and neurologists that such “psychological” effects are as real and important as any other neural phenomenon. This appreciation has provided a much more rational view of psychosomatic problems in general, and pain in particular. The Placebo Effect The placebo effect is defined as a physiological response following the administration of a pharmacologically inert “remedy.” The word placebo means “I will please,” and the placebo effect has a long history of use (and abuse) in medicine. The reality of the effect is undisputed. In one classic study, medical students were given one of two different pills, one said to be a sedative and the other a stimulant. In fact, both pills contained only inert ingredients. Of the students who received the “sedative,” more than twothirds reported that they felt drowsy, and students who took two such pills felt sleepier than those who had taken only one. Conversely, a large fraction of the students who took the “stimulant” reported that they felt less tired. Moreover, about a third of the entire group reported side effects ranging from headaches and dizziness to tingling extremities and a staggering gait! Only 3 of the 56 students studied reported that the pills they took had no appreciable effect. In another study of this general sort, 75% of patients suffering from postoperative wound pain reported satisfactory relief after an injection of sterile saline. The researchers who carried out this work noted that the responders were indistinguishable from the nonresponders, both in the apparent severity of their pain and psychological makeup. Most tellingly, this placebo effect in postoperative patients could be blocked by naloxone, a competitive antagonist of opiate receptors, indicating a substantial pharmacological basis for the pain relief experienced (see the next section). A common misunderstanding about the placebo effect is the view that patients who Pain 225 respond to a therapeutically meaningless reagent are not suffering real pain, but only “imagining” it; this is certainly not the case. Among other things, the placebo effect probably explains the efficacy of acupuncture anesthesia and the analgesia that can sometimes be achieved by hypnosis. In China, surgery has often been carried out under the effect of a needle (often carrying a small electrical current) inserted at locations dictated by ancient acupuncture charts. Before the advent of modern anesthetic techniques, operations such as thyroidectomies for goiter were commonly done without extraordinary discomfort, particularly among populations where stoicism was the cultural norm. The mechanisms of pain amelioration on the battlefield, in acupuncture anesthesia, and with hypnosis are presumably related. Although the mechanisms by which the brain affects the perception of pain are only beginning to be understood, the effect is neither magical nor a sign of a suggestible intellect. In short, the placebo effect is quite real. The Physiological Basis of Pain Modulation Understanding the central modulation of pain perception (on which the placebo effect is presumably based) was greatly advanced by the finding that electrical or pharmacological stimulation of certain regions of the midbrain produces relief of pain. This analgesic effect arises from activation of descending pain-modulating pathways that project to the dorsal horn of the spinal cord (as well as to the spinal trigeminal nucleus) and regulate the transmission of information to higher centers (Figure 9.7A). One of the major brainstem regions that produce this effect is located in the periaqueductal gray of the midbrain. Electrical stimulation at this site in experimental animals not only produces analgesia by behavioral criteria, but also demonstrably inhibits the activity of nociceptive projection neurons in the dorsal horn of the spinal cord. Further studies of descending pathways to the spinal cord that regulate the transmission of nociceptive information have shown that they arise from a number of brainstem sites, including the parabrachial nucleus, the dorsal raphe, and locus coeruleus and the medullary reticular formation (see Figure 9.7A). The analgesic effects of stimulating the periaqueductal gray are mediated through these brainstem sites. These centers employ a wealth of different neurotransmitters (noradrenaline, serotonin, dopamine, histamine, acetylcholine) and can exert both facilitatory and inhibitory effects on the the activity of neurons in the dorsal horn. The complexity of these interactions is made even greater by the fact that descending projections can exert their effects on a variety of sites within the dorsal horn including the synaptic terminals of nociceptive afferents, excitatory and inhibitory interneurons, the synaptic terminals of the other descending pathways, as well as by contacting the projection neurons themselves. Although these descending projections were originally viewed as a mechanism that served primarily to inhibit the transmission of nociceptive signals, it is now evident that these projections provide a balance of facilitatory and inhibitory influences that ultimately determines the efficacy of nociceptive transmission. In addition to descending projections, local interactions between mechanoreceptive afferents and neural circuits within the dorsal horn can modulate the transmission of nociceptive information to higher centers (Figure 9.7B). These interactions are thought to explain the ability to reduce the sensation of sharp pain by activating low-threshold mechanoreceptors: If you crack your shin or stub a toe, a natural (and effective) reaction is to vigor- 226 Chapter Nine (B) (A) Somatic sensory cortex Aβ fiber (mechanoreceptor) Hypothalamus Amygdala Inhibitory local circuit neuron Anterolateral system Midbrain periaqueductal gray To dorsal columns + C fiber (nociceptor) − + Dorsal horn projection neuron − Parabrachial nucleus Medullary reticular formation Locus coeruleus + Raphe nuclei To anterolateral system Dorsal horn of spinal cord (C) Figure 9.7 The descending systems that modulate the transmission of ascending pain signals. (A) These modulatory systems originate in the somatic sensory cortex, the hypothalamus, the periaqueductal gray matter of the midbrain, the raphe nuclei, and other nuclei of the rostral ventral medulla. Complex modulatory effects occur at each of these sites, as well as in the dorsal horn. (B) Gate theory of pain. Activation of mechanoreceptors modulates the transmission of nociceptive information to higher centers. (C) The role of enkephalin-containing local circuit neurons in the descending control of nociceptive signal transmission. Descending inputs, e.g. raphe nuclei + Axon terminal of enkephalin-containing local circuit neuron Dorsal horn projection neuron C fiber (nociceptor) − + ously rub the site of injury for a minute or two. Such observations, buttressed by experiments in animals, led Ronald Melzack and Patrick Wall to propose that the flow of nociceptive information through the spinal cord is modulated by concomitant activation of the large myelinated fibers associated with low-threshold mechanoreceptors. Even though further investigation led to modification of some of the original propositions in Melzack and Wall’s gate theory of pain, the idea stimulated a great deal of work on pain modulation and has emphasized the importance of synaptic interactions within the dorsal horn for modulating the perception of pain intensity. The most exciting advance in this long-standing effort to understand central mechanisms of pain regulation has been the discovery of endogenous opioids. For centuries it had been apparent that opium derivatives such as morphine are powerful analgesics—indeed, they remain a mainstay of analgesic therapy today. Modern animal studies have shown that a variety of brain regions are susceptible to the action of opiate drugs, particularly—and significantly—the periaqueductal gray matter and other sources of descend- Pain 227 ing projections. There are, in addition, opiate-sensitive neurons within the dorsal horn of the spinal cord. In other words, the areas that produce analgesia when stimulated are also responsive to exogenously administered opiates. It seems likely, then, that opiate drugs act at most or all of the sites shown in Figure 9.7 in producing their dramatic pain-relieving effects. The analgesic action of opiates implied the existence of specific brain and spinal cord receptors for these drugs long before the receptors were actually found during the 1960s and 1970s. Since such receptors are unlikely to exist for the purpose of responding to the administration of opium and its derivatives, the conviction grew that there must be endogenous compounds for which these receptors had evolved (see Chapter 6). Several categories of endogenous opioids have now been isolated from the brain and intensively studied. These agents are found in the same regions that are involved in the modulation of nociceptive afferents, although each of the families of endogenous opioid peptides has a somewhat different distribution. All three of the major groups (enkephalins, endorphins, and dynorphins; see Table 6.2) are present in the periaqueductal gray matter. The enkephalins and dynorphins have also been found in the rostral ventral medulla and in the spinal cord regions involved in the modulation of pain. One of the most compelling examples of the mechanism by which endogenous opiates modulate transmission of nociceptive information occurs at the first synapse in the pain pathway between nociceptive afferents and projection neurons in the dorsal horn of the spinal cord (see Figure 9.7B). A class of enkephalin-containing local circuit neurons within the dorsal horn synapses with the axon terminals of nociceptive afferents, which synapse in turn with dorsal horn projection neurons. The release of enkephalin onto the nociceptive terminals inhibits their release of neurotransmitter onto the projection neuron, reducing the level of activity that is passed on to higher centers. Enkephalin-containing local circuit neurons are themselves the targets of descending projections, thus providing a powerful mechanism by which higher centers can decrease the activity relayed by nociceptive afferents. A particularly impressive aspect of this story is the wedding of physiology, pharmacology, and clinical research to yield a much richer understanding of the intrinsic modulation of pain. This information has finally begun to explain the subjective variability of painful stimuli and the striking dependence of pain perception on the context of the experience. Precisely how pain is modulated is being explored in many laboratories at present, motivated by the tremendous clinical (and economic) benefits that would accrue from still deeper knowledge of the pain system and its molecular underpinnings. Summary Whether from a structural or functional perspective, pain is an extraordinarily complex sensory modality. Because of the importance of warning an animal about dangerous circumstances, the mechanisms and pathways that subserve nociception are widespread and redundant. A distinct set of pain afferents with membrane receptors known as nociceptors transduces noxious stimulation and conveys this information to neurons in the dorsal horn of the spinal cord. The major central pathway responsible for transmitting the discriminative aspects of pain (location, intensity and quality) differs from the mechanosensory pathway primarily in that the central axons of dorsal root ganglion cells synapse on second-order neurons in the dorsal horn; the axons of the second-order neurons then cross the midline in the spinal cord and ascend to 228 Chapter Nine thalamic nuclei that relay information to the somatic sensory cortex of the postcentral gyrus. Additional pathways involving a number of centers in the brainstem, thalamus, and cortex mediate the affective and motivational responses to painful stimuli. Descending pathways interact with local circuits in the spinal cord to regulate the transmission of nociceptive signals to higher centers. Tremendous progress in understanding pain has been made in the last 25 years, and much more seems likely, given the importance of the problem. No patients are more distressed—or more difficult to treat—than those with chronic pain. Indeed, some aspects of pain seem much more destructive to the sufferer than required by any physiological purposes. Perhaps such seemingly excessive effects are a necessary but unfortunate by-product of the protective benefits of this vital sensory modality. Additional Reading Reviews CATERINA, M. J. AND D. JULIUS (1999) Sense and specificity: A molecular identity for nociceptors. Curr. Opin. Neurobiol. 9: 525–530. DI MARZO, V., P. M. BLUMBERG AND A. SZALLASI (2002) Endovanilloid signaling in pain. Curr. Opin. Neurobiol. 12: 372–379. DUBNER, R. AND M. S. GOLD (1999) The neurobiology of pain. Proc. Natl. Acad. Sci. USA 96: 7627–7630. FIELDS, H. L. AND A. I. BASBAUM (1978) Brainstem control of spinal pain transmission neurons. Annu. Rev. Physiol. 40: 217–248. HUNT, S. P. AND P. W. MANTYH (2001) The molecular dynamics of pain control. Nat. Rev. Neurosci. 2: 83–91. JI, R. R., T. KOHNO, K. A. MOORE AND C. J. WOOLF (2003) Central sensitization and LTP: Do pain and memory share similar mechanisms? TINS 26: 696–705. JULIUS, D. AND A. I. BASBAUM (2001) Molecular mechanisms of nociception. Nature 413: 203–209. MILLAN, M. J. (2002) Descending control of pain. Prog. Neurobiol. 66: 355–474. PATAPOUTIAN, A., A. M. PEIER, G. M. STORY AND V. VISWANATH (2003) ThermoTRP channels and beyond: Mechanisms of temperature sensation. Nat. Rev. Neurosci. 4: 529–539. RAINVILLE, P. (2002) Brain mechanisms of pain affect and pain modulation. Curr. Opin. Neurobiol. 12: 195–204. SCHOLZ, J. AND C. J. WOOLF (2002) Can we conquer pain? Nat. Rev. Neurosci. 5 (Suppl): 1062–1067. TREEDE, R. D., D. R. KENSHALO, R. H. GRACELY AND A. K. JONES (1999) The cortical representation of pain. Pain 79: 105–111. CRAIG, A. D., M. C. BUSHNELL, E.-T. ZHANG AND A. BLOMQVIST (1994) A thalamic nucleus specific for pain and temperature sensation. Nature 372: 770–773. CRAIG, A. D., E. M. REIMAN, A. EVANS AND M. C. BUSHNELL (1996) Functional imaging of an illusion of pain. Nature 384: 258–260. LEVINE, J. D., H. L. FIELDS AND A. I. BASBAUM (1993) Peptides and the primary afferent nociceptor. J. Neurosci. 13: 2273–2286. MOGIL, J. S. AND J. E. GRISEL (1998) Transgenic studies of pain. Pain 77: 107–128. Important Original Papers FIELDS, H. L. (1987) Pain. New York: McGrawHill. FIELDS, H. L. (ed.) (1990) Pain Syndromes in Neurology. London: Butterworths. KOLB, L. C. (1954) The Painful Phantom. Springfield, IL: Charles C. Thomas. SKRABANEK, P. AND J. MCCORMICK (1990) Follies and Fallacies in Medicine. New York: Prometheus Books. WALL, P. D. AND R. MELZACK (1989) Textbook of Pain. New York: Churchill Livingstone. BASBAUM, A. I. AND H. L. FIELDS (1979) The origin of descending pathways in the dorsolateral funiculus of the spinal cord of the cat and rat: Further studies on the anatomy of pain modulation. J. Comp. Neurol. 187: 513–522. BEECHER, H. K. (1946) Pain in men wounded in battle. Ann. Surg. 123: 96. BLACKWELL, B., S. S. BLOOMFIELD AND C. R. BUNCHER (1972) Demonstration to medical students of placebo response and non-drug factors. Lancet 1: 1279–1282. CATERINA, M. J. AND 8 OTHERS (2000) Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science 288: 306–313. Books Chapter 10 Vision: The Eye Overview The human visual system is extraordinary in the quantity and quality of information it supplies about the world. A glance is sufficient to describe the location, size, shape, color, and texture of objects and, if the objects are moving, their direction and speed. Equally remarkable is the fact that visual information can be discerned over a wide range of stimulus intensities, from the faint light of stars at night to bright sunlight. The next two chapters describe the molecular, cellular, and higher-order mechanisms that allow us to see. The first steps in the process of seeing involve transmission and refraction of light by the optics of the eye, the transduction of light energy into electrical signals by photoreceptors, and the refinement of these signals by synaptic interactions within the neural circuits of the retina. Anatomy of the Eye The eye is a fluid-filled sphere enclosed by three layers of tissue (Figure 10.1). Only the innermost layer of the eye, the retina, contains neurons that are sensitive to light and are capable of transmitting visual signals to central targets. The immediately adjacent layer of tissue includes three distinct but continuous structures collectively referred to as the uveal tract. The largest component of the uveal tract is the choroid, which is composed of a rich capillary bed (important for nourishing the photoreceptors of the retina) as well as a high concentration of the light absorbing pigment melanin. Extending from the choroid near the front of the eye is the ciliary body, a ring of tissue that encircles the lens and consists of a muscular component that is important for adjusting the refractive power of the lens, and a vascular component (the so-called ciliary processes) that produces the fluid that fills the front of the eye. The most anterior component of the uveal tract is the iris, the colored portion of the eye that can be seen through the cornea. It contains two sets of muscles with opposing actions, which allow the size of the pupil (the opening in its center) to be adjusted under neural control. The sclera forms the outermost tissue layer of the eye and is composed of a tough white fibrous tissue. At the front of the eye, however, this opaque outer layer is transformed into the cornea, a specialized transparent tissue that permits light rays to enter the eye. Beyond the cornea, light rays pass through two distinct fluid environments before striking the retina. In the anterior chamber, just behind the cornea and in front of the lens, lies aqueous humor, a clear, watery liquid that supplies nutrients to both of these structures. Aqueous humor is produced by the ciliary processes in the posterior chamber (the region between 229 230 Chapter Ten Pupil Iris Cornea Aqueous humor in anterior chamber Zonule fibers Posterior chamber Ciliary muscle Lens Choroid Sclera Vitreous humor Retina Fovea Optic disk Optic nerve and retinal vessels Figure 10.1 Anatomy of the human eye. the lens and the iris) and flows into the anterior chamber through the pupil. The amount of fluid produced by the ciliary processes is substantial: it is estimated that the entire volume of fluid in the anterior chamber is replaced 12 times a day. Thus the rates of a aqueous humor production must be balanced by comparable rates of drainage from the anterior chamber in order to ensure a constant intraocular pressure. A specialized meshwork of cells that lies at the junction of the iris and the cornea (a region called the limbus) is responsible for aqueous drainage. Failure of adequate drainage results in a disorder known as glaucoma, in which high levels of intraocular pressure can reduce the blood supply to the eye and eventually damage retinal neurons. The space between the back of the lens and the surface of the retina is filled with a thick, gelatinous substance called the vitreous humor, which accounts for about 80% of the volume of the eye. In addition to maintaining the shape of the eye, the vitreous humor contains phagocytic cells that remove blood and other debris that might otherwise interfere with light transmission. The housekeeping abilities of the vitreous humor are limited, however, as a large number of middle-aged and elderly individuals with vitreal “floaters” will attest. Floaters are collections of debris too large for phagocytic consumption that therefore remain to cast annoying shadows on the retina; they typically arise when the aging vitreous membrane pulls away from the overly long eyeball of myopic individuals (Box A). Vision: The Eye 231 The Formation of Images on the Retina Normal vision requires that the optical media of the eye be transparent, and both the cornea and the lens are remarkable examples of tissue specializations that achieve a level of transparency that rivals that found in inorganic materials such as glass. Not surprisingly, alterations in the composition of the cornea or the lens can significantly reduce their transparency and have serious consequences for visual perception. Indeed, cataracts (opacities in the lens) account for roughly half the cases of blindness in the world, and almost everyone over the age of 70 will experience some loss of transparency in the lens that ultimately degrades the quality of visual experience. Fortunately, there are successful surgical treatments for cataracts that can restore vision in most cases. Furthermore, the recognition that a major factor in the production of cataracts is exposure to ultraviolet (UV) solar radiation has heightened public awareness of the need to protect the lens (and the retina) by reducing UV exposure through the use of sunglasses. Beyond efficiently transmitting light energy, the primary function of the optical components of the eye is to achieve a focused image on the surface of the retina. The cornea and the lens are primarily responsible for the refraction (bending) of light that is necessary for formation of focused images on the photoreceptors of the retina (Figure 10.2). The cornea contributes most of the necessary refraction, as can be appreciated by considering the hazy, outof-focus images experienced when swimming underwater. Water, unlike air, has a refractive index close to that of the cornea; as a result, immersion in water virtually eliminates the refraction that normally occurs at the air/cornea interface; thus the image is no longer focused on the retina. The lens has considerably less refractive power than the cornea; however, the refraction supplied by the lens is adjustable, allowing objects at various distances from the observer to be brought into sharp focus. Dynamic changes in the refractive power of the lens are referred to as accommodation. When viewing distant objects, the lens is made relatively thin and flat and has the least refractive power. For near vision, the lens becomes thicker and rounder and has the most refractive power (see Figure 10.2). These changes result from the activity of the ciliary muscle that surrounds the lens. The lens is held in place by radially arranged connective tissue bands (called zonule fibers) that are attached to the ciliary muscle. The shape of the lens is thus determined by two opposing forces: the elasticity of the lens, which tends to keep it rounded up (removed from the eye, the lens Unaccommodated Cornea Accommodated Iris Aqueous humor Ciliary muscle Lens Zonule fibers Vitreous humor Figure 10.2 Diagram showing the anterior part of the human eye in the unaccommodated (left) and accommodated (right) state. Accommodation for focusing on near objects involves the contraction of the ciliary muscle, which reduces the tension in the zonule fibers and allows the elasticity of the lens to increase its curvature. 232 Chapter Ten Box A Myopia and Other Refractive Errors Optical discrepancies among the various components of the eye cause a majority of the human population to have some form of refractive error, called ametropia. People who are unable to bring distant objects into clear focus are said to be nearsighted, or myopic (Figures A and B). Myopia can be caused by the corneal surface being too curved, or by the eyeball being too long. In either case, with the lens as flat as it can be, the image of distant objects focuses in front of, rather than on, the retina. People who are unable to focus on near objects are said to be farsighted, or hyperopic. Hyperopia can be caused by the eyeball being too short or the refracting system too weak (Figure C). Even with the lens in its most rounded-up state, the image is out of focus on the retinal surface (focusing at some point behind it). Both myopia and hyperopia are correctable by appropriate lenses—concave (minus) and convex (plus), respectively—or by the increasingly popular technique of corneal surgery. Myopia, or nearsightedness, is by far the most common ametropia; an estimated 50% of the population in the United States is affected. Given the large number of people who need glasses, contact lenses, or surgery to correct this refractive error, one naturally wonders how nearsighted people coped before spectacles were invented only a few centuries ago. From what is now known about myopia, most people’s vision may have been considerably better in ancient times. The basis for this assertion is the surprising finding that the growth of the eyeball is strongly influenced by focused light falling on the retina. This phenomenon was first described in 1977 by Torsten Wiesel and Elio Raviola at Harvard Medical School, who studied monkeys reared with their lids sutured (the same approach used to demonstrate the effects of visual deprivation on cortical connections in the visual system; see Chapter 23), a procedure that deprives the eye of focused retinal images. They found that animals growing to maturity under these conditions show an elongation of the eyeball. The effect of focused light deprivation appears to be a local one, since the abnormal growth of the eye occurs in experimental animals even if the optic nerve is cut. Indeed, if only a portion of the retinal surface is deprived of focused light, then only that region of the eyeball grows abnormally. Although the mechanism of lightmediated control of eye growth is not fully understood, many experts now believe that the prevalence of myopia is due to some aspect of modern civilization—perhaps learning to read and write at an early age—that interferes with the normal feedback control of vision on eye development, leading to abnormal elongation of the eyeball. A corollary of this hypothesis is that if children (or, more (A) Emmetropia (normal) (B) Myopia (nearsighted) (C) Hyperopia (farsighted) Refractive errors. (A) In the normal eye, with ciliary muscles relaxed, an image of a distant object is focused on the retina. (B) In myopia, light rays are focused in front of the retina. (C) In hyperopia, images are focused at a point beyond the retina. likely, their parents) wanted to improve their vision, they might be able to do so by practicing far vision to counterbalance the near work “overload.” Practically, of becomes spheroidal), and the tension exerted by the zonule fibers, which tends to flatten it. When viewing distant objects, the force from the zonule fibers is greater than the elasticity of the lens, and the lens assumes the flatter shape appropriate for distance viewing. Focusing on closer objects requires relaxing the tension in the zonule fibers, allowing the inherent elasticity of the lens to increase its curvature. This relaxation is accomplished by the sphincter-like contraction of the ciliary muscle. Because the ciliary muscle forms a ring around the lens, when the muscle contracts, the attachment points of the zonule fibers move toward the central axis of the eye, thus Vision: The Eye 233 Amplitude of accommodation (diopters) course, most people would probably choose wearing glasses or contacts or having corneal surgery rather than indulging in the onerous daily practice that would presumably be required. Not everyone agrees, however, that such a remedy would be effective, and a number of investigators (and drug companies) are exploring the possibility of pharmacological intervention during the period of childhood when abnormal eye growth is presumed to occur. In any event, it is a remarkable fact that deprivation of focused light on the retina causes a compensatory growth of the eye and that this feedback loop is so easily perturbed. Even people with normal (emmetropic) vision as young adults eventually experience difficulty focusing on near objects. One of the many consequences of aging is that the lens loses its elasticity; as a result, the maximum curvature the lens can achieve when the ciliary muscle contracts is gradually reduced. The near point (the closest point that can be brought into clear focus) thus recedes, and objects (such as this book) must be farther and farther away from the eye in order to focus them on the retina. At some point, usually during (D) 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 0.07 0.1 Accommodation range 0.2 Near point (m) (D) Changes in the ability of the lens to round up (accommodate) with age. The graph also shows how the near point (the closest point to the eye that can be brought into focus) changes. Accommodation, which is an optical measurement of the refractive power of the lens, is given in diopters. (After Westheimer, 1974.) 0.5 1.0 15 20 25 30 35 40 45 Age (years) early middle age, the accommodative ability of the eye is so reduced that near vision tasks like reading become difficult or impossible (Figure D). This condition is referred to as presbyopia, and can be corrected by convex lenses for near-vision tasks, or by bifocal lenses if myopia is also present (which requires a negative correction). Bifocal correction presents a particular problem for those who prefer contact lenses. Because contact lenses float on the surface of the cornea, having the distance correction above and the near correction below (as in conventional bifocal glasses) doesn’t work (although “omnifocal” contact lenses have recently been used with some success). A surprisingly effective solution to this problem for some contact lens wearers has been to put a near correcting lens in one eye and a distance correcting lens in the other! The success of this approach is another reducing the tension on the lens. Unfortunately, changes in the shape of the lens are not always able to produce a focused image on the retina, in which case a sharp image can be focused only with the help of additional corrective lenses (see Box A). Adjustments in the size of the pupil also contribute to the clarity of images formed on the retina. Like the images formed by other optical instruments, those generated by the eye are affected by spherical and chromatic aberrations, which tend to blur the retinal image. Since these aberrations are greatest for light rays that pass farthest from the center of the lens, narrow- 50 55 60 65 70 testament to the remarkable ability of the visual system to adjust to a wide variety of unusual demands. References BOCK, G. AND K. WIDDOWS (1990) Myopia and the Control of Eye Growth. Ciba Foundation Symposium 155. Chichester: Wiley. COSTER, D. J. (1994) Physics for Ophthalmologists. Edinburgh: Churchill Livingston. KAUFMAN, P. L. AND A. ALM (EDS.) (2002) Adler’s Physiology of the Eye: Clinical Application, 10th Ed. St. Louis, MO: Mosby Year Book. SHERMAN, S. M., T. T. NORTON AND V. A. CASAGRANDE (1977) Myopia in the lidsutured tree shrew. Brain Res. 124: 154–157. WALLMAN, J., J. TURKEL AND J. TRACTMAN (1978) Extreme myopia produced by modest changes in early visual experience. Science 201: 1249–1251. WIESEL, T. N. AND E. RAVIOLA (1977) Myopia and eye enlargement after neonatal lid fusion in monkeys. Nature 266: 66–68. 234 Chapter Ten ing the pupil reduces both spherical and chromatic aberration, just as closing the iris diaphragm on a camera lens improves the sharpness of a photographic image. Reducing the size of the pupil also increases the depth of field—that is, the distance within which objects are seen without blurring. However, a small pupil also limits the amount of light that reaches the retina, and, under conditions of dim illumination, visual acuity becomes limited by the number of available photons rather than by optical aberrations. An adjustable pupil thus provides an effective means of reducing optical aberrations, while maximizing depth of field to the extent that different levels of illumination permit. The size of the pupil is controlled by innervation from both sympathetic and parasympathetic divisions of the visceral motor system, which are in turn modulated by several brainstem centers (see Chapters 19 and 20). The Retina Figure 10.3 Development of the human eye. (A) The retina develops as an outpocketing from the neural tube, called the optic vesicle. (B) The optic vesicle invaginates to form the optic cup. (C, D) The inner wall of the optic cup becomes the neural retina, while the outer wall becomes the pigment epithelium. (A–C from Hilfer and Yang, 1980; D courtesy of K. Tosney.) (A) 4-mm embryo Despite its peripheral location, the retina or neural portion of the eye, is actually part of the central nervous system. During development, the retina forms as an outpocketing of the diencephalon, called the optic vesicle, which undergoes invagination to form the optic cup (Figure 10.3; see also Chapter 21). The inner wall of the optic cup gives rise to the retina, while the outer wall gives rise to the retinal pigment epithelium. This epithelium is a thin melanin-containing structure that reduces backscattering of light that enters the eye; it also plays a critical role in the maintenance of photoreceptors, renewing photopigments and phagocytosing the photoreceptor disks, whose turnover at a high rate is essential to vision. Consistent with its status as a full-fledged part of the central nervous system, the retina comprises complex neural circuitry that converts the graded electrical activity of photoreceptors into action potentials that travel to the brain via axons in the optic nerve. Although it has the same types of functional elements and neurotransmitters found in other parts of the central nervous system, the retina comprises fewer classes of neurons, and these are arranged in a manner that has been less difficult to unravel than the circuits in other areas of the brain. There are five types of neurons in the retina: photoreceptors, bipolar cells, ganglion cells, horizontal cells, and amacrine cells. The cell bodies and processes of these neurons are stacked in alternating layers, with the cell bodies located in the inner nuclear, outer nuclear, and ganglion cell layers, and the processes and synaptic contacts located in the inner plexiform and outer plexiform layers (Figure 10.4). A direct three- (B) 4.5-mm embryo (C) 5-mm embryo (D) 7-mm embryo Lens forming Lens Retina Optic cup Ventricle Optic vesicle Pigment epithelium Vision: The Eye 235 (A) Section of retina (B) Pigment epithelium Cone Cone Rod Cone Rod Rod Photoreceptor outer segments Light Distal Figure 10.4 Structure of the retina. (A) Section of the retina showing overall arrangement of retinal layers. (B) Diagram of the basic circuitry of the retina. A three-neuron chain—photoreceptor, bipolar cell, and ganglion cell—provides the most direct route for transmitting visual information to the brain. Horizontal cells and amacrine cells mediate lateral interactions in the outer and inner plexiform layers, respectively. The terms inner and outer designate relative distances from the center of the eye (inner, near the center of the eye; outer, away from the center, or toward the pigment epithelium). Horizontal cell Lateral information flow Bipolar cell Amacrine cell Virtical information Vertical information flow flow Outer nuclear layer Outer plexiform layer Inner nuclear layer Inner plexiform layer Proximal Ganglion cell Ganglion cell layer To optic nerve Nerve fiber layer Light neuron chain—photoreceptor cell to bipolar cell to ganglion cell—is the major route of information flow from photoreceptors to the optic nerve. There are two types of photoreceptors in the retina: rods and cones. Both types have an outer segment composed of membranous disks that contain light-sensitive photopigment and lies adjacent to the pigment epithelium, and an inner segment that contains the cell nucleus and gives rise to synaptic terminals that contact bipolar or horizontal cells (see also Figure 10.8). Absorption of light by the photopigment in the outer segment of the photoreceptors initiates a cascade of events that changes the membrane potential of the receptor, and therefore the amount of neurotransmitter released by the photoreceptor synapses onto the cells they contact. The synapses between photoreceptor terminals and bipolar cells (and horizontal cells) occur in the outer plexiform layer; more specifically, the cell bodies of photoreceptors make up the outer nuclear layer, whereas the cell bodies of bipolar cells lie in the inner nuclear layer. The short axonal processes of bipolar cells make synaptic contacts in turn on the dendritic processes of ganglion cells in the inner plexiform layer. The much larger axons of the ganglion cells form the optic 236 Chapter Ten nerve and carry information about retinal stimulation to the rest of the central nervous system. The two other types of neurons in the retina, horizontal cells and amacrine cells, have their cell bodies in the inner nuclear layer and have processes that are limited to the outer and inner plexiform layers respectively (see Figure 10.4). The processes of horizontal cells enable lateral interactions between photoreceptors and bipolar cells that maintain the visual system’s sensitivity to luminance contrast over a wide range of light intensities. The processes of amacrine cells are postsynaptic to bipolar cell terminals and presynaptic to the dendrites of ganglion cells. Different subclasses of amacrine cells are thought to make distinct contributions to visual function. One class of amacrine cells, for example, plays an important role in transforming the sustained responses of bipolar cells to step changes in light intensity into transient onset or offset responses exhibited by some types of ganglion cells. Another type serves as an obligatory step in the pathway that transmits information from rod photoreceptors to retinal ganglion cells. The variety of amacrine cell subtypes illustrates the more general rule that although there are only five basic retinal cell types, there can be considerable diversity within a given cell type. This diversity is also a hallmark of retinal ganglion cells and the basis for pathways that convey different sorts of information to central targets in a parallel manner (see Chapter 11). At first glance, the spatial arrangement of retinal layers seems counterintuitive, since light rays must pass through various non-light-sensitive elements of the retina as well as the retinal vasculature (which branches extensively on the inner surface of the retina—see Figure 11.1) before reaching the outer segments of the photoreceptors, where photons are absorbed (Figure 10.4). The reason for this curious feature of retinal organization lies in the special relationship that exists among the outer segments of the photoreceptors, the pigment epithelium, and the underlying choroid. Recall that the outer segments contain membranous disks that house the light-sensitive photopigment and other proteins involved in the transduction process. These disks are formed near the inner segment of the photoreceptor and move toward the tip of the outer segment, where they are shed. The pigment epithelium plays an essential role in removing the expended receptor disks; this is no small task, since all the disks in the outer segments are replaced every 12 days. In addition, the pigment epithelium contains the biochemical machinery that is required to regenerate photopigment molecules after they have been exposed to light. Finally, the capillaries in the choroid underlying the pigment epithelium are the primary source of nourishment for retinal photoreceptors. These functional considerations presumably explain why rods and cones are found in the outermost rather than the innermost layer of the retina. They also explain why disruptions in the normal relationships between the pigment epithelium and retinal photoreceptors such as those that occur in retinitis pigmentosa have severe consequences for vision (Box B). Phototransduction In most sensory systems, activation of a receptor by the appropriate stimulus causes the cell membrane to depolarize, ultimately stimulating an action potential and transmitter release onto the neurons it contacts. In the retina, however, photoreceptors do not exhibit action potentials; rather, light activation causes a graded change in membrane potential and a corresponding change in the rate of transmitter release onto postsynaptic neurons. Indeed, Vision: The Eye 237 Membrane potential (mV) Light flash Least intense flash response −40 −45 −50 −55 −60 Most intense flash response −65 0 100 200 300 Time (ms) 400 500 600 Figure 10.5 An intracellular recording from a single cone stimulated with different amounts of light (the cone has been taken from the turtle retina, which accounts for the relatively long time course of the response). Each trace represents the response to a brief flash that was varied in intensity. At the highest light levels, the response amplitude saturates (at about –65 mV). The hyperpolarizing response is characteristic of vertebrate photoreceptors; interestingly, some invertebrate photoreceptors depolarize in response to light. (After Schnapf and Baylor, 1987.) much of the processing within the retina is mediated by graded potentials, largely because action potentials are not required to transmit information over the relatively short distances involved. Perhaps even more surprising is that shining light on a photoreceptor, either a rod or a cone, leads to membrane hyperpolarization rather than depolarization (Figure 10.5). In the dark, the receptor is in a depolarized state, with a membrane potential of roughly –40 mV (including those portions of the cell that release transmitters). Progressive increases in the intensity of illumination cause the potential across the receptor membrane to become more negative, a response that saturates when the membrane potential reaches about –65 mV. Although the sign of the potential change may seem odd, the only logical requirement for subsequent visual processing is a consistent relationship between luminance changes and the rate of transmitter release from the photoreceptor terminals. As in other nerve cells, transmitter release from the synaptic terminals of the photoreceptor is dependent on voltage-sensitive Ca2+ channels in the terminal membrane. Thus, in the dark, when photoreceptors are relatively depolarized, the number of open Ca2+ channels in the synaptic terminal is high, and the rate of transmitter release is correspondingly great; in the light, when receptors are hyperpolarized, the number of open Ca2+ channels is reduced, and the rate of transmitter release is also reduced. The reason for this unusual arrangement compared to other sensory receptor cells is not known. The relatively depolarized state of photoreceptors in the dark depends on the presence of ion channels in the outer segment membrane that permit Na+ and Ca2+ ions to flow into the cell, thus reducing the degree of inside negativity (Figure 10.6). The probability of these channels in the outer segment being open or closed is regulated in turn by the levels of the nucleotide cyclic guanosine monophosphate (cGMP) (as in many other second messenger systems; see Chapter 7). In darkness, high levels of cGMP in the outer segment keep the channels open. In the light, however, cGMP levels drop and some of the channels close, leading to hyperpolarization of the outer segment membrane, and ultimately the reduction of transmitter release at the photoreceptor synapse. The series of biochemical changes that ultimately leads to a reduction in cGMP levels begins when a photon is absorbed by the photopigment in the receptor disks. The photopigment contains a light-absorbing chromophore (retinal, an aldehyde of vitamin A) coupled to one of several possible proteins called opsins that tune the molecule’s absorption of light to a particular region of the spectrum. Indeed, it is the different protein component of 238 Chapter Ten Figure 10.6 Cyclic GMP-gated channels in the outer segment membrane are responsible for the light-induced changes in the electrical activity of photoreceptors (a rod is shown here, but the same scheme applies to cones). In the dark, cGMP levels in the outer segment are high; this molecule binds to the Na+permeable channels in the membrane, keeping them open and allowing sodium (and other cations) to enter, thus depolarizing the cell. Exposure to light leads to a decrease in cGMP levels, a closing of the channels, and receptor hyperpolarization. Dark Light Rod outer segment Rod Rod outer segment Na+ Ca2+ Na+ cGMP cGMP cGMP cGMP Inside Outside 0 0 − − + (A) + Rod inner segment Rod inner segment Figure 10.7 Details of phototransduction in rod photoreceptors. (A) The molecular structure of rhodopsin, the pigment in rods. (B) The second messenger cascade of phototransduction. Light stimulation of rhodopsin in the receptor disks leads to the activation of a G-protein (transducin), which in turn activates a phosphodiesterase (PDE). The phosphodiesterase hydrolyzes cGMP, reducing its concentration in the outer segment and leading to the closure of sodium channels in the outer segment membrane. Ca2+ Na+ the photopigment in rods and cones that contributes to the functional specialization of these two receptor types. Most of what is known about the molecular events of phototransduction has been gleaned from experiments in rods, in which the photopigment is rhodopsin (Figure 10.7A). When the retinal moiety in the rhodopsin molecule absorbs a photon, its configuration changes from the 11-cis isomer to all-trans retinal; this change then triggers a series of alterations in the protein component of the molecule (Figure 10.7B). The changes lead, in turn, to the activation of an intracellular messenger called transducin, which activates a phosphodiesterase that hydrolyzes (B) Outer segment membrane C 3 PDE hydrolyzes cGMP, reducing its concentration Opsin GMP GMP GMP 1 Light stimulation of rhodopsin leads to activation of a G-protein, transducin 2 Activated G-protein activates cGMP phosphodiesterase (PDE) Light α 11-cis retinal GTP Transducin βγ GDP Na+ Na+ Closed Na+ channel Disk Rhodopsin N cGMP cGMP cGMP Open Na+ channel GTP 4 This leads to closure of Na+ channels Disk membrane α PDE Inside of cell Outside of cell Vision: The Eye 239 Box B Retinitis Pigmentosa Retinitis pigmentosa (RP) refers to a heterogeneous group of hereditary eye disorders characterized by progressive vision loss due to a gradual degeneration of photoreceptors. An estimated 100,000 people in the United States have RP. In spite of the name, inflammation is not a prominent part of the disease process; instead the photoreceptor cells appear to die by apoptosis (determined by the presence of DNA fragmentation). Classification of this group of disorders under one rubric is based on the clinical features commonly observed in these patients. The hallmarks of RP are night blindness, a reduction of peripheral vision, narrowing of the retinal vessels, and the migration of pigment from disrupted retinal pigment epithelium into the retina, forming clumps of various sizes, often next to retinal blood vessels (see figure). Typically, patients first notice difficulty seeing at night due to the loss of rod photoreceptors; the remaining cone Characteristic appearance of the retina in patients with retinitis pigmentosa. Note the dark clumps of pigment that are the hallmark of this disorder. photoreceptors then become the mainstay of visual function. Over many years, the cones also degenerate, leading to a progressive loss of vision. In most RP patients, visual field defects begin in the midperiphery, between 30° and 50° from the point of foveal fixation. The defective regions gradually enlarge, leaving islands of vision in the periphery and a constricted central field—a condition known as tunnel vision. When the visual field contracts to 20° or less and/or central vision is 20/200 or worse, the patient is categorized as legally blind. Inheritance patterns indicate that RP can be transmitted in an X-linked (XLRP), autosomal dominant (ADRP), or recessive (ARRP) manner. In the United States, the percentage of these genetic types is estimated to be 9%, 16%, and 41%, respectively. When only one member of a pedigree has RP, the case is classified as “simplex,” which accounts for about a third of all cases. Among the three genetic types of RP, ADRP is the mildest. These patients often retain good central vision until 60 years of age or older. In contrast, patients with the XLRP form of the disease are usually legally blind by 30 to 40 years of age. However, the severity and age of onset of the symptoms varies greatly among patients with the same type of RP, and even within the same family (when, presumably, all the affected members have the same genetic mutation). To date, RP-inducing mutations of 30 genes have been identified. Many of these genes encode photoreceptor-specific proteins, several being associated with phototransduction in the rods. Among the latter are genes for rhodopsin, subunits of the cGMP phosphodiesterase, and the cGMP-gated channel. Multiple mutations have been found in each of these cloned genes. For example, in the case of the rhodopsin gene, 90 different mutations have been identified among ADRP patients. The heterogeneity of RP at all levels, from genetic mutations to clinical symptoms, has important implications for understanding the pathogenesis of the disease and designing therapies. Given the complex molecular etiology of RP, it is unlikely that a single cellular mechanism will explain the disease in all cases. Regardless of the specific mutation or causal sequence, the vision loss that is most critical to RP patients is due to the gradual degeneration of cones. In many cases, the protein that the RP-causing mutation affects is not even expressed in the cones; the prime example is rhodopsin—the rod-specific visual pigment. Therefore, the loss of cones may be an indirect result of a rod-specific mutation. In consequence, understanding and treating this disease presents a particularly difficult challenge. References WELEBER, R. G. AND K. GREGORY-EVANS (2001) Retinitis pigmentosa and allied disorders. In Retina, 3rd Ed., Vol. 1: Basic Science and Inherited Retinal Diseases. S. J. Ryan (ed. in chief). St. Louis, MO: Mosby Year Book, pp. 362–460. RATTNER, A., A. SUN AND J. NATHANS (1999) Molecular genetics of human retinal disease. Annu. Rev. Genet. 33: 89–131. THE FOUNDATION FIGHTING BLINDNESS of Hunt Valley, MD, maintains a web site that provides updated information about many forms of retinal degeneration: www.blindness.org RETNET provides updated information, including references to original articles, on genes and mutations associated with retinal diseases: www.sph.uth.tmc.edu/RetNet 240 Chapter Ten cGMP. All of these events take place within the disk membrane. The hydrolysis by phosphodiesterase at the disk membrane lowers the concentration of cGMP throughout the outer segment, and thus reduces the number of cGMP molecules that are available for binding to the channels in the surface of the outer segment membrane, leading to channel closure. One of the important features of this complex biochemical cascade initiated by photon capture is that it provides enormous signal amplification. It has been estimated that a single light-activated rhodopsin molecule can activate 800 transducin molecules, roughly eight percent of the transducin molecules on the disk surface. Although each transducin molecule activates only one phosphodiesterase molecule, each of these is in turn capable of catalyzing the breakdown of as many as six cGMP molecules. As a result, the absorption of a single photon by a rhodopsin molecule results in the closure of approximately 200 ion channels, or about 2% of the number of channels in each rod that are open in the dark. This number of channel closures causes a net change in the membrane potential of about 1 mV. Equally important is the fact that the magnitude of this amplification varies with the prevailing levels of illumination, a phenomenon known as light adaptation. At low levels of illumination, photoreceptors are the most sensitive to light. As levels of illumination increase, sensitivity decreases, preventing the receptors from saturating and thereby greatly extending the range of light intensities over which they operate. The concentration of Ca2+ in the outer segment appears to play a key role in the light-induced modulation of photoreceptor sensitivity. The cGMP-gated channels in the outer segment are permeable to both Na+ and Ca2+; thus, light-induced closure of these channels leads to a net decrease in the internal Ca2+ concentration. This decrease triggers a number of changes in the phototransduction cascade, all of which tend to reduce the sensitivity of the receptor to light. For example, the decrease in Ca2+ increases the activity of quanylate cyclase, the cGMP synthesizing enzyme, leading to an increase in cGMP levels. Likewise, the decrease in Ca2+ increases the affinity of the cGMP-gated channels for cGMP, reducing the impact of the light-induced reduction of cGMP levels. The regulatory effects of Ca2+ on the phototransduction cascade are only one part of the mechanism that adapts retinal sensitivity to background levels of illumination; another important contribution comes from neural interactions between horizontal cells and photoreceptor terminals (see below). Once initiated, additional mechanisms limit the duration of this amplifying cascade and restore the various molecules to their inactivated states. The protein arrestin, for instance, blocks the ability of activated rhodopsin to activate transducin, and facilitates the breakdown of activated rhodopsin. The all-trans retinal then dissociates from the opsin, diffuses into the cytosol of the outer segment, is converted to all-trans retinol and is transported out of the outer segment and into the pigment epithelium, where appropriate enzymes ultimately convert it to 11-cis retinal. After it is transported back into the outer segment, the 11-cis retinal recombines with opsin in the receptor disks. The recycling of rhodopsin is critically important for maintaining the light sensitivity of photoreceptors. Even under intense levels of illumination, the rate of regeneration is sufficient to maintain a significant number of active photopigment molecules. Functional Specialization of the Rod and Cone Systems The two types of photoreceptors, rods and cones, are distinguished by shape (from which they derive their names), the type of photopigment they con- Vision: The Eye 241 (A) Rod Figure 10.8 Structural differences between rods and cones. Although generally similar in structure, rods (A) and cones (B) differ in their size and shape, as well as in the arrangement of the membranous disks in their outer segments. (B) Cone Disks Outer segment Cytoplasmic space Plasma membrane Outer segment Cilium Mitochondria Inner segment Inner segment Nucleus Synaptic terminal Synaptic vesicles Synaptic terminal tain, distribution across the retina, and pattern of synaptic connections (Figure 10.8). These properties reflect the fact that the rod and cone systems (the receptors and their connections within the retina) are specialized for different aspects of vision. The rod system has very low spatial resolution but is extremely sensitive to light; it is therefore specialized for sensitivity at the expense of resolution. Conversely, the cone system has very high spatial resolution but is relatively insensitive to light; it is therefore specialized for acuity at the expense of sensitivity. The properties of the cone system also allow humans and many other animals to see color. The range of illumination over which the rods and cones operate is shown in Figure 10.9. At the lowest levels of light, only the rods are activated. Such rod-mediated perception is called scotopic vision. The difficulty of making fine visual discriminations under very low light conditions where only the rod system is active is a common experience. The problem is primarily the poor resolution of the rod system (and, to a lesser degree, the fact that there is no perception of color in dim light because the cones are not involved to a significant degree). Although cones begin to contribute to visual perception at about the level of starlight, spatial discrimination at this light level is still very poor. As illumination increases, cones become more and more dominant in determining what is seen, and they are the major determinant of perception under relatively bright conditions such as normal indoor lighting or sunlight. The contributions of rods to vision drops out nearly entirely in socalled photopic vision because their response to light saturates—that is, the membrane potential of individual rods no longer varies as a function of illumination because all of the membrane channels are closed (see Figure 10.5). Mesopic vision occurs in levels of light at which both rods and cones contribute—at twilight, for example. From these considerations it should be clear that most of what we think of as normal “seeing” is mediated by the cone system, and that loss of cone function is devastating, as occurs in 242 Chapter Ten −6 −4 −2 Starlight Luminance (log cd/m−2) 0 2 Moonlight 4 Indoor lighting 6 8 Sunlight Luminance of white paper in: Good color vision Best acuity No color vision Poor acuity Visual function Scotopic Absolute threshold Mesopic Cone threshold Figure 10.9 The range of luminance values over which the visual system operates. At the lowest levels of illumination, only rods are activated. Cones begin to contribute to perception at about the level of starlight and are the only receptors that function under relatively bright conditions. Photopic Rod saturation begins 50% bleach Best acuity Indirect ophthalmoscope Damage possible elderly individuals suffering from macular degeneration (Box C). People who have lost cone function are legally blind, whereas those who have lost rod function only experience difficulty seeing at low levels of illumination (night blindness; see Box B). Differences in the transduction mechanisms utilized by the two receptor types is a major factor in the ability of rods and cones to respond to different ranges of light intensity. For example, rods produce a reliable response to a single photon of light, whereas more than 100 photons are required to produce a comparable response in a cone. It is not, however, that cones fail to effectively capture photons. Rather, the change in current produced by single photon capture in cones is comparatively small and difficult to distinguish from noise. Another difference is that the response of an individual cone does not saturate at high levels of steady illumination, as does the rod response. Although both rods and cones adapt to operate over a range of luminance values, the adaptation mechanisms of the cones are more effective. This difference in adaptation is apparent in the time course of the response of rods and cones to light flashes. The response of a cone, even to a bright light flash that produces the maximum change in photoreceptor current, recovers in about 200 milliseconds, more than four times faster than rod recovery. The arrangement of the circuits that transmit rod and cone information to retinal ganglion cells also contributes to the different characteristics of scotopic and photopic vision. In most parts of the retina, rod and cone signals converge on the same ganglion cells; i.e., individual ganglion cells respond to both rod and cone inputs, depending on the level of illumination. The early stages of the pathways that link rods and cones to ganglion cells, however, are largely independent. For example, the pathway from rods to ganglion cells involves a distinct class of bipolar cell (called rod bipolar) that, unlike cone bipolar cells, does not contact retinal ganglion cells. Instead, rod bipolar cells synapse with the dendritic processes of a specific class of amacrine cell that makes gap junctions and chemical synapses with the terminals of cone bipolars; these processes, in turn, make synaptic contacts on the dendrites of ganglion cells in the inner plexiform layer. As a consequence, the circuits linking the rods and cones to retinal ganglion cells differ dramatically in their degree of convergence. Each rod bipolar cell is contacted by a number of rods, and many rod bipolar cells contact a given amacrine cell. In contrast, the cone system is much less convergent. Thus, each retinal ganglion cell that dominates central vision (called midget gan- Vision: The Eye 243 Box C Macular Degeneration An estimated six million people in the United States suffer from a condition known as age-related macular degeneration (AMD), which causes a progressive loss of central vision. Since central vision is critical for sight, diseases that affect the macula (see Figure 11.1) severely limit the ability to perform visual tasks. Indeed, AMD is the most common cause of vision loss in people over age 55, and its incidence is rising with the increasing percentage of elderly individuals in the population. The underlying problem, which remains poorly understood, is degeneration of the photoreceptors. Usually, patients first notice a blurring of central vision when performing tasks such as reading. Images may also appear distorted. A graph paper-like chart known as the Amsler grid is used as a simple test for early signs of AMD. By focusing on a marked spot in the middle of the grid, the patient can assess whether the parallel and perpendicular lines on the grid appear blurred or distorted. Blurred central vision often progresses to having blind spots within central vision, and in most cases both eyes are eventually involved. Although the risk of developing AMD clearly increases with age, the causes of the disease are not known. Various studies have implicated hereditary factors, cardiovascular disease, environmental factors such as smoking and light exposure, and nutritional causes. Indeed, it may be that all these contribute to the risk of developing AMD. In descriptive terms, macular degeneration is broadly divided into two types. In the exudative-neovascular form, or “wet” AMD, which accounts for 10% of all cases, abnormal blood vessel growth occurs under the macula. These blood vessels leak fluid and blood into the retina and cause damage to the photore- ceptors. Wet AMD tends to progress rapidly and can cause severe damage; rapid loss of central vision may occur over just a few months. The treatment for this form of the disease is laser therapy. By transferring thermal energy, the laser beam destroys the leaky blood vessels under the macula, thus slowing the rate of vision loss. A disadvantage of this approach is that the high thermal energy delivered by the beam also destroys nearby healthy tissue. An improvement in the laser treatment of AMD involves a light-activated drug to target abnormal blood vessels. Once the drug is administered, relatively low energy laser pulses aimed at the abnormal blood vessels are delivered to stimulate the drug, which in turn destroys the abnormal blood vessels with minimal damage to the surrounding tissue. The remaining 90% of AMD cases are the nonexudative, or “dry” form. In these patients there is a gradual disappearance of the retinal pigment epithelium, resulting in circumscribed areas of atrophy. Since photoreceptor loss follows the disappearance of the pigment epithelium, the affected retinal areas have little or no visual function. Vision loss from dry AMD occurs more gradually, typically over the course of many years. These patients usually retain some central vision, although the loss can be severe enough to compromise performance of tasks that require seeing details. Unfortunately, at the present time there is no treatment for dry AMD. A radical and quite fascinating new approach that offers some promise entails surgically repositioning the retina away from the abnormal area. Occasionally, macular degeneration occurs in much younger individuals. Many of these cases are caused by various mutations, each with its own clinical manifestations and genetic cause. The most common form of juvenile macular degeneration is known as Stargardt disease, which is inherited as an autosomal recessive. Patients are usually diagnosed before they reach the age of 20. Although the progression of vision loss is variable, most of these patients are legally blind by age 50. Mutations that cause Stargardt disease have been identified in the ABCR gene, which codes for a protein that transports retinoids across the photoreceptor membrane. Thus, the visual cycle of photopigment regeneration may be disrupted in this form of macular degeneration, presumably by dysfunctional proteins encoded by the abnormal gene. Interestingly, the ABCR gene is expressed only in rods, suggesting that the cones may have their own visual cycle enzymes. References FINE, S. L., J. W. BERGER, M. G. MAGUIRE AND A. C. HO (2000) Drug therapy: Age-related macular degeneration. NEJM 342: 483–492. SARKS, S. H. AND J. P. SARKS (2001) Age-related macular degeneration—atrophic form. In Retina, 3rd Ed., Vol. 2: Medical Retina. S. J. Ryan (ed.-in-chief). St. Louis, MO: Mosby Year Book, pp. 1071–1102. ELMAN, M. J. AND S. L. FINE (2001) Exudative age-related macular degeneration. In Retina, 3rd Ed., Vol. 2: Medical Retina. S. J. Ryan (ed.in-chief). St. Louis, MO: Mosby Year Book, pp. 1103–114. DEUTMAN, A. F. (2001) Macular dystrophies. In Retina, 3rd Ed., Vol. 2: Medical Retina. S. J. Ryan (ed.-in-chief). St. Louis, MO: Mosby Year Book, pp. 1186–1240. THE FOUNDATION FIGHTING BLINDNESS of Hunt Valley, MD, maintains a web site that provides updated information about many forms of retinal degeneration: www.blindness.org RETNET provides updated information, including references to original articles, on genes and mutations associated with retinal diseases: www.sph.uth.tmc.edu/RetNet 244 Chapter Ten glion cells) receives input from only one cone bipolar cell, which, in turn, is contacted by a single cone. Convergence makes the rod system a better detector of light, because small signals from many rods are pooled to generate a large response in the bipolar cell. At the same time, convergence reduces the spatial resolution of the rod system, since the source of a signal in a rod bipolar cell or retinal ganglion cell could have come from anywhere within a relatively large area of the retinal surface. The one-to-one relationship of cones to bipolar and ganglion cells is, of course, just what is required to maximize acuity. Anatomical Distribution of Rods and Cones 160 140 120 100 Rods 80 60 Optic disk Figure 10.10 Distribution of rods and cones in the human retina. Graph illustrates that cones are present at a low density throughout the retina, with a sharp peak in the center of the fovea. Conversely, rods are present at high density throughout most of the retina, with a sharp decline in the fovea. Boxes at top illustrate the appearance of face on sections through the outer segments of the photoreceptors at different eccentricities. The increased density of cones in the fovea is accompanied by a striking reduction in the diameter of their outer segments. Receptor density (mm−2 × 103) The distribution of rods and cones across the surface of the retina also has important consequences for vision (Figure 10.10). Despite the fact that perception in typical daytime light levels is dominated by cone-mediated vision, the total number of rods in the human retina (about 90 million) far exceeds the number of cones (roughly 4.5 million). As a result, the density of rods is much greater than cones throughout most of the retina. However, this relationship changes dramatically in the fovea, a highly specialized region of the central retina that measures about 1.2 millimeters in diameter (Figure 10.11). In the fovea, cone density increases almost 200-fold, reaching, at its center, the highest receptor packing density anywhere in the retina. This high density is achieved by decreasing the diameter of the cone outer segments such that foveal cones resemble rods in their appearance. The increased density of cones in the fovea is accompanied by a sharp decline in the density of rods. In fact, the central 300 µm of the fovea, called the foveola, is totally rod-free. The extremely high density of cone receptors in the fovea, and the one-toone relationship with bipolar cells and retinal ganglion cells (see earlier), endows this component of the cone system with the capacity to mediate high visual acuity. As cone density declines with eccentricity and the degree of convergence onto retinal ganglion cells increases, acuity is markedly reduced. Just 6° eccentric to the line of sight, acuity is reduced by 75%, a fact Rods 40 20 Cones 0 80 60 Temporal 40 Cones 20 0 20 40 Eccentricity (degrees) 60 80 Nasal Vision: The Eye 245 Figure 10.11 Diagrammatic cross section through the human fovea. The overlying cellular layers and blood vessels are displaced so that light is subjected to a minimum of scattering before photons strike the outer segments of the cones in the center of the fovea, called the foveola. Cones Capillaries Rods Bipolar cells Pigment epithelium Choroid Outer nuclear layer Inner nuclear layer Avascular zone Ganglion cell layer Foveola Fovea that can be readily appreciated by trying to read the words on any line of this page beyond the word being fixated on. The restriction of highest acuity vision to such a small region of the retina is the main reason humans spend so much time moving their eyes (and heads) around—in effect directing the foveas of the two eyes to objects of interest (see Chapter 19). It is also the reason why disorders that affect the functioning of the fovea have such devastating effects on sight (see Box C). Conversely, the exclusion of rods from the fovea, and their presence in high density away from the fovea, explain why the threshold for detecting a light stimulus is lower outside the region of central vision. It is easier to see a dim object (such as a faint star) by looking slightly away from it, so that the stimulus falls on the region of the retina that is richest in rods (see Figure 10.10). Another anatomical feature of the fovea (which literally means “pit”) that contributes to the superior acuity of the cone system is that the layers of cell bodies and processes that overlie the photoreceptors in other areas of the retina are displaced around the fovea, and especially the foveola (see Figure 10.11). As a result, photons are subjected to a minimum of scattering before they strike the photoreceptors. Finally, another potential source of optical distortion that lies in the light path to the receptors—the retinal blood vessels—are diverted away from the foveola. This central region of the fovea is therefore dependent on the underlying choroid and pigment epithelium for oxygenation and metabolic sustenance. Cones and Color Vision A special property of the cone system is color vision. Perceiving color allows humans (and many other animals) to discriminate objects on the basis of the distribution of the wavelengths of light that they reflect to the eye. While differences in luminance (i.e., overall light intensity) are often sufficient to distinguish objects, color adds another perceptual dimension that is especially useful when differences in luminance are subtle or nonexistent. Color obviously gives us a quite different way of perceiving and describing the world we live in. Ganglion cells 246 Chapter Ten Unlike rods, which contain a single photopigment, there are three types of cones that differ in the photopigment they contain. Each of these photopigments has a different sensitivity to light of different wavelengths, and for this reason are referred to as “blue,” “green,” and “red” or, more appropriately, short (S), medium (M), and long (L) wavelength cones—terms that more or less describe their spectral sensitivities (Figure 10.12). This nomenclature implies that individual cones provide color information for the wavelength of light that excites them best. In fact, individual cones, like rods, are entirely color blind in that their response is simply a reflection of the number of photons they capture, regardless of the wavelength of the photon (or, more properly, its vibrational energy). It is impossible, therefore, to determine whether the change in the membrane potential of a particular cone has arisen from exposure to many photons at wavelengths to which the receptor is relatively insensitive, or fewer photons at wavelengths to which it is most sensitive. This ambiguity can only be resolved by comparing the activity in different classes of cones. Based on the responses of individual ganglion cells, and cells at higher levels in the visual pathway (see Chapter 11), comparisons of this type are clearly involved in how the visual system extracts color information from spectral stimuli. Despite these insights, a full understanding of the neural mechanisms that underlie color perception has been elusive (Box D). Much additional information about color vision has come from studies of individuals with abnormal color detecting abilities. Color vision deficiencies result either from the inherited failure to make one or more of the cone pigments or from an alteration in the absorption spectra of cone pigments (or, rarely, from lesions in the central stations that process color information; see Chapter 11). Under normal conditions, most people can match any color in a test stimulus by adjusting the intensity of three superimposed light sources generating long, medium, and short wavelengths. The fact that only three such sources are needed to match (nearly) all the perceived colors is strong Short Rods Medium Long Figure 10.12 Color vision. The light absorption spectra of the four photopigments in the normal human retina. (Recall that light is defined as electromagnetic radiation having wavelengths between ~400 and 700 nm.) The solid curves indicate the three kinds of cone opsins; the dashed curve shows rod rhodopsin for comparison. Absorbance is defined as the log value of the intensity of incident light divided by intensity of transmitted light. Relative spectral absorbance 100 50 0 400 450 500 Wavelength (nm) 550 600 650 Vision: The Eye 247 Box D The Importance of Context in Color Perception Seeing color logically demands that retinal responses to different wavelengths in some way be compared. The discovery of the three human cone types and their different absorption spectra is correctly regarded, therefore, as the basis for human color vision. Nevertheless, how these human cone types and the higherorder neurons they contact (see Chapter 11) produce the sensations of color is still unclear. Indeed, this issue has been debated by some of the greatest minds in science (Hering, Helmholtz, Maxwell, Schroedinger, and Mach, to name only a few) since Thomas Young first proposed that humans must have three different receptive “particles”—i.e., the three cone types. A fundamental problem has been that, although the relative activities of three cone types can more or less explain the colors perceived in color-matching experiments performed in the laboratory, the perception of color is strongly influenced by context. For example, a patch returning the exact same spectrum of wavelengths to the eye can appear quite different depending on its surround, a phenomenon called color contrast (Figure A). Moreover, test patches returning different spectra to the eye can appear to be the same color, an effect called color constancy (Figure B). Although these phenomena were well known in the nineteenth century, they were not accorded a central place in color vision theory until Edwin Land’s work in the 1950s. In his most famous demonstration, Land (who among other achievements founded the Polaroid company and became a billionaire) used a collage of colored papers that have been referred to as “the Land Mondrians” because of their similarity to the work of the Dutch artist Piet Mondrian. Using a telemetric photometer and three adjustable illuminators generating short, middle, and long wavelength light, Land showed that two patches that in white light appeared quite different in color (e. g., green and brown) continued to look their respective colors even when the three illuminators were adjusted so that the light being returned from the “green” surfaces produced exactly the same readings on the three telephotometers as had previously come from the “brown” surface—a striking demonstration of color constancy. The phenomena of color contrast and color constancy have led to a heated modern debate about how color percepts are generated that now spans several decades. For Land, the answer lay in a series of ratiometric equations that could integrate the spectral returns of different regions over the entire scene. It was rec(A) ognized even before Land’s death in 1991, however, that his so-called retinex theory did not work in all circumstances and was in any event a description rather than an explanation. An alternative explanation of these contextual aspects of color vision is that color, like brightness, is generated empirically according to what spectral stimuli have typically signified in past experience (see Box E). References LAND, E. (1986) Recent advances in Retinex theory. Vis. Res. 26: 7–21. PURVES, D. AND R. B. LOTTO (2003) Why We See What We Do: An Empirical Theory of Vision, Chapters 5 and 6. Sunderland MA: Sinauer Associates, pp. 89–138. (B) The genesis of contrast and constancy effects by exactly the same context. The two panels demonstrate the effects on apparent color when two similarly reflective target surfaces (A) or two differently reflective target surfaces (B) are presented in the same context in which all the information provided is consistent with illumination that differs only in intensity. The appearances of the relevant target surfaces in a neutral context are shown in the insets below. (From Purves and Lotto, 2003) 248 Chapter Ten confirmation of the fact that color sensation is based on the relative levels of activity in three sets of cones with different absorption spectra. That color vision is trichromatic was first recognized by Thomas Young at the beginning of the nineteenth century (thus, people with normal color vision are called trichromats). For about 5–6% of the male population in the United States and a much smaller percentage of the female population, however, color vision is more limited. Only two bandwidths of light are needed to match all the colors that these individuals can perceive; the third color category is simply not seen. Such dichromacy, or “color blindness” as it is commonly called, is inherited as a recessive, sex-linked characteristic and exists in two forms: protanopia, in which all color matches can be achieved by using only green and blue light, and deuteranopia, in which all matches can be achieved by using only blue and red light. In another major class of color deficiencies, all three light sources (i.e., short, medium, and long wavelengths) are needed to make all possible color matches, but the matches are made using values that are significantly different from those used by most individuals. Some of these anomalous trichromats require more red than normal to match other colors (protanomalous trichromats); others require more green than normal (deuteranomalous trichromats). Jeremy Nathans and his colleagues at Johns Hopkins University have provided a deeper understanding of these color vision deficiencies by identifying and sequencing the genes that encode the three human cone pigments (Figure 10.13). The genes that encode the red and green pigments show a high degree of sequence homology and lie adjacent to each other on the X chromosome, thus explaining the prevalence of color blindness in males. In contrast, the blue-sensitive pigment gene is found on chromosome 7 and is quite different in its amino acid sequence. These facts suggest that the red and green pigment genes evolved relatively recently, perhaps as a result of the duplication of a single ancestral gene; they also explain why most color vision abnormalities involve the red and green cone pigments. Human dichromats lack one of the three cone pigments, either because the corresponding gene is missing or because it exists as a hybrid of the red and green pigment genes (see Figure 10.13). For example, some dichromats lack the green pigment gene altogether, while others have a hybrid gene that is thought to produce a red-like pigment in the “green” cones. Anomalous trichromats also possess hybrid genes, but these genes elaborate pigments Red pigment gene Green pigment gene Crossover event Figure 10.13 Many deficiencies of color vision are the result of genetic alterations in the red or green cone pigments due to the crossing over of chromosomes during meiosis. This recombination can lead to the loss of a gene, the duplication of a gene, or the formation of a hybrid with characteristics distinct from those of normal genes. Different crossover events can lead to: (1) Hybrid gene (2) Loss of gene (3) Duplication of gene (does not affect color vision) Patterns in color-blind men Vision: The Eye 249 whose spectral properties lie between those of the normal red and green pigments. Thus, although most anomalous trichromats have distinct sets of medium and long-wavelength cones, there is more overlap in their absorption spectra than in normal trichromats, and thus less difference in how the two sets of cones respond to a given wavelength (with resulting anomalies in color perception). Retinal Circuits for Detecting Luminance Change Despite the esthetically pleasing nature of color vision, most of the information in visual scenes consists of spatial variations in light intensity (a black and white movie, for example, has most of the information a color version has, although it is deficient in some respects and usually is less fun to watch). How the spatial patterns of light and dark that fall on the photoreceptors are deciphered by central targets has been a vexing problem (Box E). To understand what is accomplished by the complex neural circuits within the retina during this process, it is useful to start by considering the responses of individual retinal ganglion cells to small spots of light. Stephen Kuffler, working at Johns Hopkins University in the 1950s, pioneered this approach by characterizing the responses of single ganglion cells in the cat retina. He found that each ganglion cell responds to stimulation of a small circular patch of the retina, which defines the cell’s receptive field (see Chapter 8 for discussion of receptive fields). Based on these responses, Kuffler distinguished two classes of ganglion cells, “on”-center and “off”-center (Figure 10.14). Turning on a spot of light in the receptive field center of an on-center ganglion cell produces a burst of action potentials. The same stimulus applied to the receptive field center of an off-center ganglion cell reduces the rate of (A) (B) (C) Dark spot in center Light spot in center Figure 10.14 The responses of on-center and off-center retinal ganglion cells to stimulation of different regions of their receptive fields. Upper panels indicate the time sequence of stimulus changes. (A) Effects of light spot in the receptive field center. (B) Effects of dark spot in the receptive field center. (C) Effects of light spot in the center followed by the addition of light in the surround. Center plus surround Center only t3 t2 t2 t2 t1 t1 t1 t0 t0 t0 ++ +++ ++ + + +++ ++ On-center ganglion cell +++++ ++++++++ +++++++++ ++++ ++++ +++ +++ +++ +++ +++ +++ ++++ ++++ + + + + + + + ++ ++++++ +++ Off-center ganglion cell t0 t1 t2 Time t0 t1 t2 Time t0 t1 t2 t3 Time 250 Chapter Ten Box E The Perception of Light Intensity Understanding the link between retinal stimulation and what we see (perception) is arguably the central problem in vision, and the relation of luminance (a physical measurement of light intensity) and brightness (the sensation elicited by light intensity) is probably the simplest place to consider this challenge. As indicated in the text, how we see the brightness differences (i.e., contrast) between adjacent territories with distinct luminances depends in the first instance on the relative firing rate of retinal ganglion cells, modified by lateral interactions. However, there is a problem with the assumption that the central nervous system simply “reads out” these relative rates of ganglion cell activity to sense brightness. The difficulty, as in perceiving color, is that the brightness of a given target is markedly affected by its context in ways that are difficult or impossible to explain in terms of the retinal output as such. The accompanying figures, which illustrate two simultaneous brightness contrast illusions, help make this point. In Figure A, two photometrically identical (equiluminant) gray squares appear differently bright as a function of the background in which they are presented. A conventional interpretation of this phenomenon is that the receptive field properties illustrated in Figures 10.14 through 10.17 cause ganglion cells to fire differently depending on whether the surround of the equiluminant target is dark or light. The demonstration in Figure B, however, undermines this explanation, since in this case the target surrounded by more dark area actually looks darker than the same target surrounded by more light area. An alternative interpretation of luminance perception that can account for these puzzling phenomena is that brightness percepts are generated on a statistical basis as a means of contending with the inherent ambiguity of luminance (i.e., the fact that a given value of lumi- nance can be generated by many different combinations of illumination and surface reflectance properties). Since to be successful an observer has to respond to the real-world sources of luminance and not to light intensity as such, this ambiguity of the retinal stimulus presents a quandary. A plausible solution to (A) (B) (C) (A) Standard illusion of simultaneous brightness contrast. (B) Another illusion of simultaneous brightness contrast that is difficult to explain in conventional terms. (C) Cartoons of some possible sources of the standard simultaneous brightness contrast illusion in (A). (Courtesy of R. Beau Lotto and Dale Purves.) Vision: The Eye 251 the inherent uncertainty of the relationship between luminance values and their actual sources would be to generate the sensation of brightness elicited by a given luminance (e.g., in the brightness of the identical test patches in the figure) on the basis of what the luminance of the test patches had typically turned out to be in the past experience of human observers. To get the gist of this explanation consider Figure C, which illustrates the point that the two equiluminant target patches in Figure A could have been generated by two differently painted surfaces in different illuminants, as in a comparison of the target patches on the left and middle cubes, or two similarly reflecting surfaces in similar amounts of light, as in a comparison of the target patches on the middle and right cubes. An expedient—and perhaps the only— way the visual system can cope with this ambiguity is to generate the perception of the stimulus in Figure A (and in Figure B) empirically, i.e., based on what the target patches typically turned out to signify in the past. Since the equiluminant targets will have arisen from a variety of possible sources, it makes sense to have the brightness elicited by the patches determined statistically by the relative frequency of occurrence of that luminance in the particular context in which it is presented. The advantage of seeing discharge, and when the spot of light is turned off, the cell responds with a burst of action potentials (Figure 10.14A). Complementary patterns of activity are found for each cell type when a dark spot is placed in the receptive field center (Figure 10.14B). Thus, on-center cells increase their discharge rate to luminance increments in the receptive field center, whereas off-center cells increase their discharge rate to luminance decrements in the receptive field center. On- and off-center ganglion cells are present in roughly equal numbers. The receptive fields have overlapping distributions, so that every point on the retinal surface (that is, every part of visual space) is analyzed by several on-center and several off-center ganglion cells. A rationale for having these two distinct types of retinal ganglion cells was suggested by Peter Schiller and his colleagues at the Massachusetts Institute of Technology, who examined the effects of pharmacologically inactivating on-center ganglion cells on a monkey’s ability to detect a variety of visual stimuli. After silencing oncenter ganglion cells, the animals showed a deficit in their ability to detect stimuli that were brighter than the background; however, they could still see objects that were darker than the background. These observations imply that information about increases or decreases in luminance is carried separately to the brain by the axons of these two different types of retinal ganglion cells. Having separate luminance “channels” means that changes in light intensity, whether increases or decreases, are always conveyed to the brain by an increased number of action potentials. Because ganglion cells rapidly adapt to changes in luminance, their “resting” discharge rate in constant illumination is relatively low. Although an increase in discharge rate above resting level serves as a reliable signal, a decrease in firing rate from an initially low rate of discharge might not. Thus, having luminance changes signaled by two classes of adaptable cells provides unambiguous information about both luminance increments and decrements. The functional differences between these two ganglion cell types can be understood in terms of both their anatomy and their physiological proper- luminance according to the relative probabilities of the possible sources of the stimulus is that percepts generated in this way give the observer the best chance of making appropriate behavioral responses to profoundly ambiguous stimuli. References ADELSON, E. H. (1999) Light perception and lightness illusions. In The Cognitive Neurosciences, 2nd Ed. M. Gazzaniga (ed.). Cambridge, MA: MIT Press, pp. 339–351. PURVES, D. AND R. B. LOTTO (2003) Why We See What We Do: An Empirical Theory of Vision, Chapters 3 and 4. Sunderland MA: Sinauer Associates, pp. 41–87. 252 Chapter Ten Figure 10.15 Circuitry responsible for generating receptive field center responses of retinal ganglion cells. (A) Functional anatomy of cone inputs to the center of a ganglion cell receptive field. A plus indicates a sign-conserving synapse; a minus represents a signinverting synapse. (B) Responses of various cell types to the presentation of a light spot in the center of the ganglion cell receptive field. (C) Responses of various cell types to the presentation of a dark spot in the center of the ganglion cell receptive field. ties and relationships (Figure 10.15). On- and off-center ganglion cells have dendrites that arborize in separate strata of the inner plexiform layer, forming synapses selectively with the terminals of on- and off-center bipolar cells that respond to luminance increases and decreases, respectively (Figure 10.15A). As mentioned previously, the principal difference between ganglion cells and bipolar cells lies in the nature of their electrical response. Like most other cells in the retina, bipolar cells have graded potentials rather than action potentials. Graded depolarization of bipolar cells leads to an increase in transmitter release (glutamate) at their synapses and consequent depolarization of the on-center ganglion cells that they contact via AMPA, kainate, and NMDA receptors. The selective response of on- and off-center bipolar cells to light increments and decrements is explained by the fact that they express different types of glutamate receptors (Figure 10.15A). Off-center bipolar cells have ionotropic receptors (AMPA and kainate) that cause the cells to depolarize in response to glutamate released from photoreceptor terminals. In contrast, on-center bipolar cells express a G-protein-coupled metabotropic glutamate receptor (mGluR6). When bound to glutamate, these receptors activate an intracellular cascade that closes cGMP-gated Na+ channels, reducing inward (A) (B) Surround Center (C) Light spot in center Surround Dark spot in center t2 t2 t1 t1 t0 t0 Center cone Center cone Glutamate mGluR6 – AMPA kainate + On-center bipolar cell On-center ganglion cell t0 t1 Time t2 t0 t1 Time t2 On-center bipolar cell Off-center bipolar cell On-center bipolar cell Off-center bipolar cell t1 t1 t1 t1 Off-center bipolar cell Glutamate AMPA, kainate, NMDA Center cone + + t2 t2 t2 t2 On-center ganglion cell Off-center ganglion cell On-center ganglion cell Off-center ganglion cell t1 t1 t1 t1 Off-center ganglion cell t2 t2 t2 t2 Vision: The Eye 253 current and hyperpolarizing the cell. Thus, glutamate has opposite effects on these two classes of cells, depolarizing off-center bipolar cells and hyperpolarizing on-center cells. Photoreceptor synapses with off-center bipolar cells are called sign-conserving, since the sign of the change in membrane potential of the bipolar cell (depolarization or hyperpolarization) is the same as that in the photoreceptor (Figure 10.15B,C). Photoreceptor synapses with oncenter bipolar cells are called sign-inverting because the change in the membrane potential of the bipolar cell is the opposite of that in the photoreceptor. In order to understand the response of on- and off-center bipolar cells to changes in light intensity, recall that photoreceptors hyperpolarize in response to light increments, decreasing their release of neurotransmitter (Figure 10.15B). Under these conditions, on-center bipolar cells contacted by the photoreceptors are freed from the hyperpolarizing influence of the photoreceptor’s transmitter, and they depolarize. In contrast, for off-center cells, the reduction in glutamate represents the withdrawal of a depolarizing influence, and these cells hyperpolarize. Decrements in light intensity naturally have the opposite effect on these two classes of bipolar cells, hyperpolarizing on-center cells and depolarizing off-center ones (Figure 10.15C). Kuffler’s work also called attention to the fact that retinal ganglion cells do not act as simple photodetectors. Indeed, most ganglion cells are relatively poor at signaling differences in the level of diffuse illumination. Instead, they are sensitive to differences between the level of illumination that falls on the receptive field center and the level of illumination that falls on the surround—that is, to luminance contrast. The center of a ganglion cell receptive field is surrounded by a concentric region that, when stimulated, antagonizes the response to stimulation of the receptive field center (see Figure 10.14C). For example, as a spot of light is moved from the center of the receptive field of an on-center cell toward its periphery, the response of the cell to the spot of light decreases (Figure 10.16). When the spot falls completely outside the center (that is, in the surround), the response of the cell falls below its resting level; the cell is effectively inhibited until the distance from the center is so great that the spot no longer falls on the receptive field at all, in which case the cell returns to its resting level of firing. Off-center Light ++ +++ ++ + + +++ ++ ++ +++ ++ + + +++ ++ ++ +++ ++ + + +++ ++ ++ +++ ++ + + +++ ++ ++ +++ ++ + + +++ ++ Response rate (impulses/s) 100 80 60 40 20 Spontaneous level of activity 0 1 2 3 4 Distance (degrees) from center of receptive field 5 Figure 10.16 Rate of discharge of an on-center ganglion cell to a spot of light as a function of the distance of the spot from the receptive field center. Zero on the x axis corresponds to the center; at a distance of 5°, the spot falls outside the receptive field. 254 Chapter Ten Dark A B ++ +++ ++ + + +++ ++ Light Edge C ++ +++ ++ + + +++ ++ D ++ +++ ++ + + +++ ++ E ++ +++ ++ + + +++ ++ On-center ganglion cells ++ +++ ++ + + +++ ++ D Response rate Figure 10.17 Responses of a hypothetical population of on-center ganglion cells whose receptive fields (A–E) are distributed across a light-dark edge. Those cells whose activity is most affected have receptive fields that lie along the light-dark edge. E C A Spontaneous level of activity B Position cells exhibit a similar surround antagonism. Stimulation of the surround by light opposes the decrease in firing rate that occurs when the center is stimulated alone, and reduces the response to light decrements in the center (compare Figures 10.14A and 10.14C). Because of their antagonistic surrounds, ganglion cells respond much more vigorously to small spots of light confined to their receptive field centers than to large spots, or to uniform illumination of the visual field (see Figure 10.14C). To appreciate how center-surround antagonism makes the ganglion cell sensitive to luminance contrast, consider the activity levels in a hypothetical population of on-center ganglion cells whose receptive fields are distributed across a retinal image of a light-dark edge (Figure 10.17). The neurons whose firing rates are most affected by this stimulus—either increased (neuron D) or decreased (neuron B)—are those with receptive fields that lie along the light-dark border; those with receptive fields completely illuminated (or completely darkened) are less affected (neurons A and E). Thus, the information supplied by the retina to central visual stations for further processing does not give equal weight to all regions of the visual scene; rather, it emphasizes the regions where there are differences in luminance. Contribution of Retinal Circuits to Light Adaptation In addition to making ganglion cells especially sensitive to light-dark borders in the visual scene, center-surround mechanisms make a significant contribution to the process of light adaptation. As illustrated for an on-center cell in Figure 10.18, the response rate of a ganglion cell to a small spot of light turned on in its receptive field center varies as a function of the spot’s intensity. In fact, response rate is proportional to the spot’s intensity over a range of about one log unit. However, the intensity of spot illumination required to evoke a given discharge rate is dependent on the background level of illumination. Increases in background level of illumination are accompanied by adaptive shifts in the cell’s operating range such that Vision: The Eye 255 400 −5 Discharge rate (spikes/s) −4 −3 300 −1 −2 0 200 100 0 9 x 10−5 9 x 10−4 9 x 10−3 9 x 10−2 Test spot luminance (cd/m2) 9 x 10−1 9 greater stimulus intensities are required to achieve the same discharge rate. Thus, firing rate is not an absolute measure of light intensity, but rather signals the difference from background level of illumination. Because the range of light intensities over which we can see is enormous compared to the narrow range of ganglion cell discharge rates (see Figure 10.9), adaptational mechanisms are essential. By scaling the ganglion cell’s response to ambient levels of illumination, the entire dynamic range of a neuron’s firing rate is used to encode information about intensity differences over the range of luminance values that are relevant for a given visual scene. Due to the antagonistic center-surround organization of retinal ganglion cells, the signal sent to the brain from the retina downplays the background level of illumination (see Figure 10.14). This arrangement presumably explains why the relative brightness of objects remains much the same over a wide range of lighting conditions. In bright sunlight, for example, the print on this page reflects considerably more light to the eye than it does in room light. In fact, the print reflects more light in sunlight than the paper reflects in room light; yet it continues to look black and the page white, indoors or out. Like the mechanism responsible for generating the on- and off-center response, the antagonistic surround of ganglion cells is a product of interactions that occur at the early stages of retinal processing (Figure 10.19). Much of the antagonism is thought to arise via lateral connections established by horizontal cells and receptor terminals. Horizontal cells receive synaptic inputs from photoreceptor terminals and are linked via gap junctions with a vast network of other horizontal cells distributed over a wide area of the retinal surface. As a result, the activity in horizontal cells reflects levels of illumination over a broad area of the retina. Although the details of their actions are not entirely clear, horizontal cells are thought to exert their influence via the release of neurotransmitter directly onto photoreceptor terminals, regulating the amount of transmitter that the photoreceptors release onto bipolar cell dendrites. Glutamate release from photoreceptor terminals has a depolarizing effect on horizontal cells (sign-conserving synapse), while the transmitter released from horizontal cells (GABA) has a hyperpolarizing influence on photoreceptor terminals (sign-inverting synapse) (Figure 10.19A). As a result, the net effect of inputs from the horizontal cell network is to oppose changes in the Figure 10.18 A series of curves illustrating the discharge rate of a single oncenter ganglion cell to the onset of a small test spot of light in the center of its receptive field. Each curve represents the discharge rate evoked by spots of varying intensity at a constant background level of illumination, which is given by the red numbers at the top of each curve (the highest background level is 0, the lowest –5). The response rate is proportional to stimulus intensity over a range of 1 log unit, but the operating range shifts to the right as the background level of illumination increases. 256 Chapter Ten (A) (B) Center only Center plus surround t2 t1 t0 Center cone Surround Center Transductionmediated hyperpolarization Surround cone Surround cone Center cone – – + – + + Horizontal cell Horizontal cell mediated depolarization t0 t1 t2 Time Horizontal cell – + – t1 t2 On-center bipolar cell On-center bipolar cell t1 t2 Horizontal cell + t1 t2 On-center ganglion cell On-center ganglion cell t1 t2 Figure 10.19 Circuitry responsible for generating the receptive field surround of an on-center retinal ganglion cell. (A) Functional anatomy of horizontal cell inputs responsible for surround antagonism. A plus indicates a sign-conserving synapse; a minus represents a sign-inverting synapse. (B) Responses of various cell types to the presentation of a light spot in the center of the receptive field (t1) followed by the addition of light stimulation in the surround (t2). Light stimulation of the surround leads to hyperpolarization of the horizontal cells and a decrease in the release of inhibitory transmitter (GABA) onto the photoreceptor terminals. The net effect is to depolarize the center cone terminal, offsetting much of the hyperpolarization induced by the transduction cascade in the center cone’s outer segment. Vision: The Eye 257 membrane potential of the photoreceptor that are induced by phototransduction events in the outer segment. How these events lead to surround suppression in an on-center ganglion cell is illustrated in Figure 10.19. A small spot of light centered on a photoreceptor supplying input to the center of the ganglion cell’s receptive field produces a strong hyperpolarizing response in the photoreceptor. Under these conditions, changes in the membrane potential of the horizontal cells that synapse with the photoreceptor terminal are relatively small, and the response of the photoreceptor to light is largely determined by its phototransduction cascade (Figure10.19B). With the addition of light to the surround, however, the impact of the horizontal network becomes significantly greater; the light-induced reduction in the release of glutamate from the photoreceptors in the surround leads to a strong hyperpolarization of the horizontal cells whose processes converge on the terminal of the photoreceptor in the receptive field center. The reduction in GABA release from the horizontal cells has a depolarizing effect on the membrane potential of the central photoreceptor, reducing the lightevoked response and ultimately reducing the firing rate of the on-center ganglion cell. Thus, even at the earliest stages in visual processing, neural signals do not represent the absolute numbers of photons that are captured by receptors, but rather the relative intensity of stimulation—how much the current level of stimulation differs from ambient levels. While it may seem that the actions of horizontal cells decrease the sensitivity of the retina, they play a critical role in allowing the full range of the photoreceptor’s electrical response (about 30 mV) to be applied to the limited range of stimulus intensities that are present at any given moment. The network mechanisms of adaptation described here function in conjunction with cellular mechanisms in the receptor outer segments that regulate the sensitivity of the phototransduction cascade at different light levels. Together, they allow retinal circuits to convey the most salient aspects of luminance changes to the central stages of the visual system described in the following chapter. Summary The light that falls on photoreceptors is transformed by retinal circuitry into a pattern of action potentials that ganglion cell axons convey to the visual centers in the brain. This process begins with phototransduction, a biochemical cascade that ultimately regulates the opening and closing of ion channels in the membrane of the photoreceptor’s outer segment, and thereby the amount of neurotransmitter the photoreceptor releases. Two systems of photoreceptors—rods and cones—allow the visual system to meet the conflicting demands of sensitivity and acuity, respectively. Retinal ganglion cells operate quite differently from the photoreceptor cells. The centersurround arrangement of ganglion cell receptive fields makes these neurons particularly sensitive to luminance contrast and relatively insensitive to the overall level of illumination. It also allows the retina to adapt, such that it can respond effectively over the enormous range of illuminant intensities in the world. The underlying organization is generated by the synaptic interactions between photoreceptors, horizontal cells, and bipolar cells in the outer plexiform layer. As a result, the signal sent to the visual centers in the brain is already highly processed when it leaves the retina, emphasizing those aspects of the visual scene that convey the most information. Additional Reading Reviews ARSHAVSKY, V. Y., T. D. LAMB AND E. N. PUGH JR. (2002) G proteins and phototransduction. Annu. Rev. Physiol. 64: 153–187. BURNS, M. E. AND D. A. BAYLOR (2001) Activation, deactivation, and adaptation in vertebrate photoreceptor cells. Annu. Rev. Neurosci. 24: 779–805. NATHANS, J. (1987) Molecular biology of visual pigments. Annu. Rev. Neurosci. 10: 163–194. SCHNAPF, J. L. AND D. A. BAYLOR (1987) How photoreceptor cells respond to light. Sci. Amer. 256 (April): 40–47. STERLING, P. (1990) Retina. In The Synaptic Organization of the Brain, G. M. Shepherd (ed.). New York: Oxford University Press, pp. 170–213. STRYER, L. (1986) Cyclic GMP cascade of vision. Annu. Rev. Neurosci. 9: 87–119. Important Original Papers BAYLOR, D. A., M. G. F. FUORTES AND P. M. O’BRYAN (1971) Receptive fields of cones in the retina of the turtle. J. Physiol. (Lond.) 214: 265–294. DOWLING, J. E. AND F. S. WERBLIN (1969) Organization of the retina of the mud puppy, Necturus maculosus. I. Synaptic structure. J. Neurophysiol. 32: 315–338. ENROTH-CUGELL, C. AND R. M. SHAPLEY (1973) Adaptation and dynamics of cat retinal ganglion cells. J. Physiol. 233: 271–309. FASENKO, E. E., S. S. KOLESNIKOV AND A. L. LYUBARSKY (1985) Induction by cyclic GMP of cationic conductance in plasma membrane of retinal rod outer segment. Nature 313: 310–313. KUFFLER, S. W. (1953) Discharge patterns and functional organization of mammalian retina. J. Neurophysiol. 16: 37–68. NATHANS, J., D. THOMAS AND D. S. HOGNESS (1986) Molecular genetics of human color vision: The genes encoding blue, green and red pigments. Science 232: 193–202. NATHANS, J., T. P. PIANTANIDA, R. EDDY, T. B. SHOWS AND D. S. HOGNESS (1986) Molecular genetics of inherited variation in human color vision. Science 232: 203–210. SCHILLER, P. H., J. H. SANDELL AND J. H. R. MAUNSELL (1986) Functions of the “on” and “off” channels of the visual system. Nature 322: 824–825. WERBLIN, F. S. AND J. E. DOWLING (1969) Organization of the retina of the mud puppy, Necturus maculosus. II. Intracellular recording. J. Neurophysiol. 32: 339–354. Books BARLOW, H. B. AND J. D. MOLLON (1982) The Senses. London: Cambridge University Press. DOWLING, J. E. (1987) The Retina: An Approachable Part of the Brain. Cambridge, MA: Belknap Press. FAIN, G. L. (2003) Sensory Transduction. Sunderland, MA: Sinauer Associates. HART, W. M. J. (ed.) (1992) Adler’s Physiology of the Eye: Clinical Application, 9th Ed. St. Louis, MO: Mosby Year Book. HELMHOLTZ, H. L. F. VON (1924) Helmholtz’s Treatise on Physiological Optics, Vol. I–III. Transl. from the Third German Edition by J. P. C. Southall. Menasha, WI: George Banta Publishing Company. HOGAN, M. J., J. A. ALVARADO AND J. E. WEDDELL (1971) Histology of the Human Eye: An Atlas and Textbook. Philadelphia: Saunders. HUBEL, D. H. (1988) Eye, Brain, and Vision, Scientific American Library Series. New York: W. H. Freeman. HURVICH, L. (1981) Color Vision. Sunderland, MA: Sinauer Associates, pp. 180–194. OGLE, K. N. (1964) Researches in Binocular Vision. Hafner: New York. OYSTER, C. (1999) The Human Eye: Structure and Function. Sunderland, MA: Sinauer Associates. POLYAK, S. (1957) The Vertebrate Visual System. Chicago: The University of Chicago Press. RODIECK, R. W. (1973) The Vertebrate Retina. San Francisco: W. H. Freeman. RODIECK, R. W. (1998) First Steps in Seeing. Sunderland, MA: Sinauer Associates. WANDELL, B. A. (1995) Foundations of Vision. Sunderland, MA: Sinauer Associates. Chapter 11 Central Visual Pathways Overview Information supplied by the retina initiates interactions between multiple subdivisions of the brain that eventually lead to conscious perception of the visual scene, at the same time stimulating more conventional reflexes such as adjusting the size of the pupil, directing the eyes to targets of interest, and regulating homeostatic behaviors that are tied to the day/night cycle. The pathways and structures that mediate this broad range of functions are necessarily diverse. Of these, the primary visual pathway from the retina to the dorsal lateral geniculate nucleus in the thalamus and on to the primary visual cortex is the most important and certainly the most thoroughly studied component of the visual system. Different classes of neurons within this pathway encode the varieties of visual information—luminance, spectral differences, orientation, and motion—that we ultimately see. The parallel processing of different categories of visual information continues in cortical pathways that extend beyond primary visual cortex, supplying a variety of visual areas in the occipital, parietal, and temporal lobes. Visual areas in the temporal lobe are primarily involved in object recognition, whereas those in the parietal lobe are concerned with motion. Normal vision depends on the integration of information in all these cortical areas. The processes underlying visual perception are not understood and remain one of the central challenges of modern neuroscience. Central Projections of Retinal Ganglion Cells Ganglion cell axons exit the retina through a circular region in its nasal part called the optic disk (or optic papilla), where they bundle together to form the optic nerve. This region of the retina contains no photoreceptors and, because it is insensitive to light, produces the perceptual phenomenon known as the blind spot (Box A). The optic disk is easily identified as a whitish circular area when the retina is examined with an ophthalmoscope; it also is recognized as the site from which the ophthalmic artery and veins enter (or leave) the eye (Figure 11.1). In addition to being a conspicuous retinal landmark, the appearance of the optic disk is a useful gauge of intracranial pressure. The subarachnoid space surrounding the optic nerve is continuous with that of the brain; as a result, increases in intracranial pressure—a sign of serious neurological problems such as a space-occupying lesion—can be detected as papilledema, a swelling of the optic disk. Axons in the optic nerve run a straight course to the optic chiasm at the base of the diencephalon. In humans, about 60% of these fibers cross in the chiasm, while the other 40% continue toward the thalamus and midbrain targets on the same side. Once past the chiasm, the ganglion cell axons on each 259 260 Chapter Eleven Figure 11.1 The retinal surface of the left eye, viewed with an ophthalmoscope. The optic disk is the region where the ganglion cell axons leave the retina to form the optic nerve; it is also characterized by the entrance and exit, respectively, of the ophthalmic arteries and veins that supply the retina. The macula lutea can be seen as a distinct area at the center of the optical axis (the optic disk lies nasally); the macula is the region of the retina that has the highest visual acuity. The fovea is a depression or pit about 1.5 mm in diameter that lies at the center of the macula (see Chapter 10). Macula lutea Fovea Optic disk (papilla) Branch of ophthalmic vein Branch of ophthalmic artery side form the optic tract. Thus, the optic tract, unlike the optic nerve, contains fibers from both eyes. The partial crossing (or decussation) of ganglion cell axons at the optic chiasm allows information from corresponding points on the two retinas to be processed by approximately the same cortical site in each hemisphere, an important issue that is considered in the next section. The ganglion cell axons in the optic tract reach a number of structures in the diencephalon and midbrain (Figure 11.2). The major target in the diencephalon is the dorsal lateral geniculate nucleus of the thalamus. Neurons in the lateral geniculate nucleus, like their counterparts in the thalamic relays of other sensory systems, send their axons to the cerebral cortex via the internal capsule. These axons pass through a portion of the internal capsule called the optic radiation and terminate in the primary visual cortex, or striate cortex (also referred to as Brodmann’s area 17 or V1), which lies largely along and within the calcarine fissure in the occipital lobe. The retinogeniculostriate pathway, or primary visual pathway, conveys information that is essential for most of what is thought of as seeing. Thus, damage anywhere along this route results in serious visual impairment. A second major target of the ganglion cell axons is a collection of neurons that lies between the thalamus and the midbrain in a region known as the pretectum. Although small in size compared to the lateral geniculate nucleus, the pretectum is particularly important as the coordinating center for the pupillary light reflex (i.e., the reduction in the diameter of the pupil that occurs when sufficient light falls on the retina) (Figure 11.3). The initial component of the pupillary light reflex pathway is a bilateral projection from the retina to the pretectum. Pretectal neurons, in turn, project to the EdingerWestphal nucleus, a small group of nerve cells that lies close to the nucleus of the oculomotor nerve (cranial nerve III) in the midbrain. The Edinger-Westphal nucleus contains the preganglionic parasympathetic neurons that send their axons via the oculomotor nerve to terminate on neurons in the ciliary Central Visual Pathways 261 Optic tract Optic nerve Hypothalamus: regulation of circadian rhythms Optic chiasm Lateral geniculate nucleus Figure 11.2 Central projections of retinal ganglion cells. Ganglion cell axons terminate in the lateral geniculate nucleus of the thalamus, the superior colliculus, the pretectum, and the hypothalamus. For clarity, only the crossing axons of the right eye are shown (view is looking up at the inferior surface of the brain). Pretectum: reflex control of pupil and lens Optic radiation Superior colliculus: orienting the movements of head and eyes Striate cortex ganglion (see Chapter 19). Neurons in the ciliary ganglion innervate the constrictor muscle in the iris, which decreases the diameter of the pupil when activated. Shining light in the eye thus leads to an increase in the activity of pretectal neurons, which stimulates the Edinger-Westphal neurons and the ciliary ganglion neurons they innervate, thus constricting the pupil. In addition to its normal role in regulating the amount of light that enters the eye, the pupillary reflex provides an important diagnostic tool that allows the physician to test the integrity of the visual sensory apparatus, the motor outflow to the pupillary muscles, and the central pathways that medi- Figure 11.3 The circuitry responsible for the pupillary light reflex. This pathway includes bilateral projections from the retina to the pretectum and projections from the pretectum to the Edinger-Westphal nucleus. Neurons in the Edinger-Westphal nucleus terminate in the ciliary ganglion, and neurons in the ciliary ganglion innervate the pupillary constrictor muscles. Notice that the afferent axons activate both Edinger-Westphal nuclei via the neurons in the pretectum. Postganglionic parasympathetic fiber Pupillary constrictor muscle Iris Optic nerve Ciliary ganglion Preganglionic parasympathetic fiber in cranial nerve III EdingerWestphal nucleus Pretectum Superior colliculus Cornea Aqueous humor Lens Retina Vitreous humor 262 Chapter Eleven Box A The Blind Spot It is logical to suppose that a visual field defect (called a scotoma) arising from damage to the retina or central visual pathways would be obvious to the individual suffering from such pathology. When the deficit involves a peripheral region of the visual field, however, a scotoma often goes unnoticed until a car accident or some other mishap all too dramatically reveals the sensory loss. In fact, all of us have a physiological scotoma of which we are quite unaware, the so-called “blind spot.” The blind spot is the substantial gap in each monocular visual field that corresponds to the location of the optic disk, the receptor-free region of the retina where the optic nerve leaves the eye (see Figure 11.1). To find the “blind spot” of the right eye, close the left eye and fixate on the X shown in the figure here, holding the book about 30–40 centimeters away. Now take a pencil in your right hand and, without breaking fixation, move the tip slowly toward the X from the right side of the page. At some point, the tip of the pencil (indeed the whole end of the pencil) will disappear; mark this point and continue to move the pencil to the left until it reappears; then make another mark. The borders of the blind spot along the vertical axis can be determined in the same way by moving the pencil X up and down so that its path falls between the two horizontal marks. To prove that information from the region of visual space bounded by the marks is really not perceived, put a penny inside the demarcated area. When you fixate the X with both eyes and then close the left eye, the penny will disappear, a seemingly magical event that amazed the French royal court when it was first reported by the natural philosopher Edmé Mariotte in 1668. How can we be unaware of such a large defect in the visual field (typically about 5°–8°)? The optic disk is located in the nasal retina of each eye. With both eyes open, information about the corresponding region of visual space is, of course, available from the temporal retina of the other eye. But this fact does not explain why the blind spot remains undetected with one eye closed. When the world is viewed monocularly, the visual system appears to “fill-in” the missing part of the scene based on the information supplied by the regions surrounding the optic disk. To observe this phenomenon, notice what happens when a pencil or some other object lies across the optic disk representation. Remarkably, the pencil looks complete! Although electrophysiological recordings have shown that neurons in the visual cortex whose receptive fields lie in the optic disk representation can be activated by stimulating the regions that surround the optic disk of the contralateral eye, suggesting that “filling-in” the blind spot is based on cortical mechanisms that integrate information from different points in the visual field, the mechanism of this striking phenomenon is not clear. Herman von Helmholtz pointed out in the nineteenth century that it may just be that this part of the visual world is ignored, the pencil being completed across the blind spot because the rest of the scene simply “collapses” around it. References FIORANI, M., M. G. P. ROSA, R. GATTASS AND C. E. ROCHA-MIRANDA (1992) Dynamic surrounds of receptive fields in striate cortex: A physiological basis for perceptual completion? Proc. Natl. Acad. Sci. USA 89: 8547–8551. GILBERT, C. D. (1992) Horizontal integration and cortical dynamics. Neuron 9: 1–13. RAMACHANDRAN, V. S. AND T. L. GREGORY (1991) Perceptual filling in of artificially induced scotomas in human vision. Nature 350: 699–702. VON HELMHOLTZ, H. (1968). Helmholtz’s Treatise on Physiological Optics, Vols. I–III (Translated from the Third German Ed. published in 1910). J. P. C. Southall (ed.). New York: Dover Publications. See pp. 204ff in Vol. III. Central Visual Pathways 263 ate the reflex. Under normal conditions, the pupils of both eyes respond identically, regardless of which eye is stimulated; that is, light in one eye produces constriction of both the stimulated eye (the direct response) and the unstimulated eye (the consensual response; see Figure 11.3). Comparing the response in the two eyes is often helpful in localizing a lesion. For example, a direct response in the left eye without a consensual response in the right eye suggests a problem with the visceral motor outflow to the right eye, possibly as a result of damage to the oculomotor nerve or Edinger-Westphal nucleus in the brainstem. Failure to elicit a response (either direct or indirect) to stimulation of the left eye if both eyes respond normally to stimulation of the right eye suggests damage to the sensory input from the left eye, possibly to the left retina or optic nerve. There are several other important targets of retinal ganglion cell axons. One is the suprachiasmatic nucleus of the hypothalamus, a small group of neurons at the base of the diencephalon (see Box A in Chapter 20). The retinohypothalamic pathway is the route by which variation in light levels influences the broad spectrum of visceral functions that are entrained to the day/night cycle (see Chapters 20 and 27). Another target is the superior colliculus, a prominent structure visible on the dorsal surface of the midbrain (see Figure 1.14). The superior colliculus coordinates head and eye movements to visual (as well as other) targets; its functions are considered in Chapter 19. The type of visual information required to perform the functions of these different retinal targets is quite different. Reading the text on this page, for example, requires a high-resolution sampling of the retinal image, whereas regulating circadian rhythms and adjusting the pupil accordingly require only a measure of overall changes in light levels, and little or no information about the features of the image. It should come as no surprise, then, that there is a diversity of ganglion cell types that provide information appropriate to the functions of these different targets. Projections to the lateral geniculate nucleus (which are described in more detail later) arise from at least three broad classes of ganglion cells, whose tuning properties are appropriate for mediating the richness of visual perception (high acuity, color, motion). In contrast, projections to the hypothalamus and the pretectum arise from ganglion cells that lack these properties and are highly suited for detecting luminance flux. The retinal specializations responsible for constructing these distinct classes of retinal ganglion cells are only beginning to be identified; they include not only differences in ganglion cell synaptic connections, but in the locus of the phototransduction event itself. Unlike the majority of ganglion cells, which depend on rods and cones for their sensitivity to light, the ganglion cells that project to the hypothalamus and pretectum express their own light-sensitive photopigment (melanopsin) and are capable of modulating their response to changes in light levels in the absence of signals from rods and cones. The presence of light sensitivity within this class of ganglion cells presumably explains why normal circadian rhythms are maintained in animals that have completely lost form vision due to degeneration of rod and cone photoreceptors. The Retinotopic Representation of the Visual Field The spatial relationships among the ganglion cells in the retina are maintained in most of their central targets as orderly representations or “maps” of visual space. Most of these structures receive information from both eyes, requiring that these inputs be integrated to form a coherent map of individ- 264 Chapter Eleven Eye (A) Lens F Binocular visual field (B) F Monocular portion of visual field Monocular portion of visual field Left visual Right visual field field Left monocular visual field Right monocular visual field Superior (S) F Temporal (T) Inferior (I) T Nasal (N) Fixation point S F Left retina S I F T I S N Fovea Right retina F Figure 11.4 Projection of the visual fields onto the left and right retinas. (A) Projection of an image onto the surface of the retina. The passage of light rays through the pupil of the eye results in images that are inverted and left–right reversed on the retinal surface. (B) Retinal quadrants and their relation to the organization of monocular and binocular visual fields, as viewed from the back surface of the eyes. Vertical and horizontal lines drawn through the center of the fovea define retinal quadrants (bottom). Comparable lines drawn through the point of fixation define visual field quadrants (center). Color coding illustrates corresponding retinal and visual field quadrants. The overlap of the two monocular visual fields is shown at the top. T I ual points in space. As a general rule, information from the left half of the visual world, whether it originates from the left or right eye, is represented in the right half of the brain, and vice versa. Understanding the neural basis for the appropriate arrangement of inputs from the two eyes requires considering how images are projected onto the two retinas, and the central destination of the ganglion cells located in different parts of the retina. Each eye sees a part of visual space that defines its visual field (Figure 11.4A). For descriptive purposes, each retina and its corresponding visual field are divided into quadrants. In this scheme, the surface of the retina is subdivided by vertical and horizontal lines that intersect at the center of the fovea (Figure 11.4B). The vertical line divides the retina into nasal and temporal divisions and the horizontal line divides the retina Central Visual Pathways 265 into superior and inferior divisions. Corresponding vertical and horizontal lines in visual space (also called meridians) intersect at the point of fixation (the point in visual space that falls on the fovea) and define the quadrants of the visual field. The crossing of light rays diverging from different points on an object at the pupil causes the images of objects in the visual field to be inverted and left-right reversed on the retinal surface. As a result, objects in the temporal part of the visual field are seen by the nasal part of the retina, and objects in the superior part of the visual field are seen by the inferior part of the retina. (It may help in understanding Figure 11.4B to imagine that you are looking at the back surfaces of the retinas, with the corresponding visual fields projected onto them.) With both eyes open, the two foveas are normally aligned on a single target in visual space, causing the visual fields of both eyes to overlap extensively (see Figure 11.4B and Figure 11.5). This binocular field of view consists of two symmetrical visual hemifields (left and right). The left binocular hemifield includes the nasal visual field of the right eye and the temporal visual field of the left eye; the right hemifield includes the temporal visual field of Binocular visual field Left visual field Right visual field A B FP C D Fixation point Left visual field B Right visual field C D A Temporal retina Temporal retina Nasal retina Optic chiasm Left optic tract Right optic tract Figure 11.5 Projection of the binocular field of view onto the two retinas and its relation to the crossing of fibers in the optic chiasm. Points in the binocular portion of the left visual field (B) fall on the nasal retina of the left eye and the temporal retina of the right eye. Points in the binocular portion of the right visual field (C) fall on the nasal retina of the right eye and the temporal retina of the left eye. Points that lie in the monocular portions of the left and right visual fields (A and D) fall on the left and right nasal retinas, respectively. The axons of ganglion cells in the nasal retina cross in the optic chiasm, whereas those from the temporal retina do not. As a result, information from the left visual field is carried in the right optic tract, and information from the right visual field is carried in the left optic tract. 266 Chapter Eleven Figure 11.6 Visuotopic organization of the striate cortex in the right occipital lobe, as seen in mid-sagittal view. (A) The primary visual cortex occupies a large part of the occipital lobe. The area of central vision (the fovea) is represented over a disproportionately large part of the caudal portion of the lobe, whereas peripheral vision is represented more anteriorly. The upper visual field is represented below the calcarine sulcus, the lower field above the calcarine sulcus. (B) Photomicrograph of a coronal section of the human striate cortex, showing the characteristic myelinated band, or stria, that gives this region of the cortex its name. The calcarine sulcus on the medial surface of the occipital lobe is indicated. (B courtesy of T. Andrews and D. Purves.) the right eye and the nasal visual field of the left eye. The temporal visual fields are more extensive than the nasal visual fields, reflecting the size of the nasal and temporal retinas respectively. As a result, vision in the periphery of the field of view is strictly monocular, mediated by the most medial portion of the nasal retina. Most of the rest of the field of view can be seen by both eyes; i.e., individual points in visual space lie in the nasal visual field of one eye and the temporal visual field of the other. It is worth noting, however, that the shape of the face and nose impact the extent of this region of binocular vision. In particular, the inferior nasal visual fields are less extensive than the superior nasal fields, and consequently the binocular field of view is smaller in the lower visual field than in the upper (see Figure 11.4B). Ganglion cells that lie in the nasal division of each retina give rise to axons that cross in the chiasm, while those that lie in the temporal retina give rise to axons that remain on the same side (see Figure 11.5). The boundary (or line of decussation) between contralaterally and ipsilaterally projecting ganglion cells runs through the center of the fovea and defines the border between the nasal and temporal hemiretinas. Images of objects in the left visual hemifield (such as point B in Figure 11.5) fall on the nasal retina of the left eye and the temporal retina of the right eye, and the axons from ganglion cells in these regions of the two retinas project through the right optic tract. Objects in the right visual hemifield (such as point C in Figure 11.5) fall on the nasal retina of the right eye and the temporal retina of the left eye; the axons from ganglion cells in these regions project through the left optic tract. As mentioned previously, objects in the monocular portions of the visual hemifields (points A and D in Figure 11.5) are seen only by the most peripheral nasal retina of each eye; the axons of ganglion cells in these regions (like the rest of the nasal retina) run in the contralateral optic tract. Thus, unlike the optic nerve, the optic tract contains the axons of ganglion cells that originate in both eyes and represent the contralateral field of view. Optic tract axons terminate in an orderly fashion within their target structures thus generating well ordered maps of the contralateral hemifield. For the primary visual pathway, the map of the contralateral hemifield that is established in the lateral geniculate nucleus is maintained in the projections of the lateral geniculate nucleus to the striate cortex (Figure 11.6). Thus the (A) (B) Parieto-occipital sulcus Binocular portion Macula Calcarine sulcus Myelinated stria Monocular portion Medial surface Left visual field Right occipital lobe Central Visual Pathways 267 fovea is represented in the posterior part of the striate cortex, whereas the more peripheral regions of the retina are represented in progressively more anterior parts of the striate cortex. The upper visual field is mapped below the calcarine sulcus, and the lower visual field above it. As in the somatic sensory system, the amount of cortical area devoted to each unit area of the sensory surface is not uniform, but reflects the density of receptors and sensory axons that supply the peripheral region. Like the representation of the hand region in the somatic sensory cortex, the representation of the macula is therefore disproportionately large, occupying most of the caudal pole of the occipital lobe. Visual Field Deficits A variety of retinal or more central pathologies that involve the primary visual pathway can cause visual field deficits that are limited to particular regions of visual space. Because the spatial relationships in the retinas are maintained in central visual structures, a careful analysis of the visual fields can often indicate the site of neurological damage. Relatively large visual field deficits are called anopsias and smaller ones are called scotomas (see Box A). The former term is combined with various prefixes to indicate the specific region of the visual field from which sight has been lost (Figures 11.7 and 11.8). Damage to the retina or one of the optic nerves before it reaches the chiasm results in a loss of vision that is limited to the eye of origin. In contrast, damage in the region of the optic chiasm—or more centrally—results in specific types of deficits that involve the visual fields of both eyes (Figure 11.8). Damage to structures that are central to the optic chiasm, including the optic tract, lateral geniculate nucleus, optic radiation, and visual cortex, results in deficits that are limited to the contralateral visual hemifield. For example, interruption of the optic tract on the right results in a loss of sight in the left visual field (that is, blindness in the temporal visual field of the left eye and the nasal visual field of the right eye). Because such damage affects corresponding parts of the visual field in each eye, there is a complete loss of vision in the affected region of the binocular visual field, and the deficit is referred to as a homonymous hemianopsia (in this case, a left homonymous hemianopsia). Lateral ventricles Lateral geniculate nucleus Fibers representing superior retinal quadrants (inferior visual field) Meyer’s loop Fibers representing inferior retinal quadrants (superior visual field) Figure 11.7 Course of the optic radiation to the striate cortex. Axons carrying information about the superior portion of the visual field sweep around the lateral horn of the ventricle in the temporal lobe (Meyer’s loop) before reaching the occipital lobe. Those carrying information about the inferior portion of the visual field travel in the parietal lobe. 268 Chapter Eleven Left eye visual field Right eye visual field (A) Left Right (B) Optic nerve Optic chiasm Optic tract A (C) B C D Lateral geniculate nucleus (D) Optic radiation E (E) Striate cortex Figure 11.8 Visual field deficits resulting from damage at different points along the primary visual pathway. The diagram on the left illustrates the basic organization of the primary visual pathway and indicates the location of various lesions. The right panels illustrate the visual field deficits associated with each lesion. (A) Loss of vision in right eye. (B) Bitemporal (heteronomous) hemianopsia. (C) Left homonymous hemianopsia. (D) Left superior quadrantanopsia. (E) Left homonymous hemianopsia with macular sparing. In contrast, damage to the optic chiasm results in visual field deficits that involve noncorresponding parts of the visual field of each eye. For example, damage to the middle portion of the optic chiasm (which is often the result of pituitary tumors) can affect the fibers that are crossing from the nasal retina of each eye, leaving the uncrossed fibers from the temporal retinas intact. The resulting loss of vision is confined to the temporal visual field of each eye and is known as bitemporal hemianopsia. It is also called heteronomous hemianopsia to emphasize that the parts of the visual field that are lost in each eye do not overlap. Individuals with this condition are able to see in both left and right visual fields, provided both eyes are open. However, all information from the most peripheral parts of visual fields (which are seen only by the nasal retinas) is lost. Damage to central visual structures is rarely complete. As a result, the deficits associated with damage to the chiasm, optic tract, optic radiation, or visual cortex are typically more limited than those shown in Figure 11.8. This is especially true for damage along the optic radiation, which fans out under the temporal and parietal lobes in its course from the lateral geniculate nucleus to the striate cortex. Some of the optic radiation axons run out into the temporal lobe on their route to the striate cortex, a branch called Meyer’s loop (see Figure 11.7). Meyer’s loop carries information from the superior portion of the contralateral visual field. More medial parts of the optic radiation, which pass under the cortex of the parietal lobe, carry information from the inferior portion of the contralateral visual field. Damage to parts of the temporal lobe with involvement of Meyer’s loop can thus result in a superior Central Visual Pathways 269 homonymous quadrantanopsia; damage to the optic radiation underlying the parietal cortex results in an inferior homonymous quadrantanopsia. Injury to central visual structures can also lead to a phenomenon called macular sparing, i.e., the loss of vision throughout wide areas of the visual field, with the exception of foveal vision. Macular sparing is commonly found with damage to the cortex, but can be a feature of damage anywhere along the length of the visual pathway. Although several explanations for macular sparing have been offered, including overlap in the pattern of crossed and uncrossed ganglion cells supplying central vision, the basis for this selective preservation is not clear. The Functional Organization of the Striate Cortex Much in the same way that Stephen Kuffler explored the response properties of individual retinal ganglion cells (see Chapter 10), David Hubel and Torsten Wiesel used microelectrode recordings to examine the properties of neurons in more central visual structures. The responses of neurons in the lateral geniculate nucleus were found to be remarkably similar to those in the retina, with a center-surround receptive field organization and selectivity for luminance increases or decreases. However, the small spots of light that were so effective at stimulating neurons in the retina and lateral geniculate nucleus were largely ineffective in visual cortex. Instead, most cortical neurons in cats and monkeys responded vigorously to light–dark bars or edges, and only if the bars were presented at a particular range of orientations within the cell’s receptive field (Figure 11.9). The responses of cortical neurons are thus tuned to the orientation of edges, much like cone receptors are tuned to the wavelength of light; the peak in the tuning curve (the orientation to which a cell is most responsive) is referred to as the neuron’s preferred orientation. By sampling the responses of a large number of single cells, Hubel and Weisel demonstrated that all edge orientations were roughly equally represented in visual cortex. As a (A) Experimental setup (B) Stimulus orientation Stimulus presented Light bar stimulus projected on screen Recording from visual cortex Record 0 1 2 Time (s) 3 Figure 11.9 Neurons in the primary visual cortex respond selectively to oriented edges. (A) An anesthetized animal is fitted with contact lenses to focus the eyes on a screen, where images can be projected; an extracellular electrode records the neuronal responses. (B) Neurons in the primary visual cortex typically respond vigorously to a bar of light oriented at a particular angle and weakly—or not at all—to other orientations. 270 Chapter Eleven result, a given orientation in a visual scene appears to be “encoded” in the activity of a distinct population of orientation-selective neurons. Hubel and Wiesel also found that there are subtly different subtypes within a class of neurons that preferred the same orientation. For example, the receptive fields of some cortical cells, which they called simple cells, were composed of spatially separate “on” and “off” response zones, as if the “on” and “off” centers of lateral geniculate cells that supplied these neurons were arrayed in separate parallel bands. Other neurons, referred to as complex cells, exhibited mixed “on” and “off” responses throughout their receptive field, as if they received their inputs from a number of simple cells. Further analysis uncovered cortical neurons sensitive to the length of the bar of light that was moved across their receptive field, decreasing their rate of response when the bar exceeded a certain length. Still other cells responded selectively to the direction in which an edge moved across their receptive field. Although the mechanisms responsible for generating these selective responses are still not well understood, there is little doubt that the specificity of the receptive field properties of neurons in the striate cortex (and beyond) plays an important role in determining the basic attributes of visual scenes. Another feature that distinguishes the responses of neurons in the striate cortex from those at earlier stages in the primary visual pathway is binocularity. Although the lateral geniculate nucleus receives inputs from both eyes, the axons terminate in separate layers, so that individual geniculate Lateral geniculate nucleus Striate cortex (B) I II Layer (A) 1 2 3 4 5 6 III IV V VI Parietooccipital sulcus Calcarine sulcus Posterior pole of occipital lobe Figure 11.10 Mixing of the pathways from the two eyes first occurs in the striate cortex. (A) Although the lateral geniculate nucleus receives inputs from both eyes, these are segregated in separate layers (see also Figure 11.14). In many species, including most primates, the inputs from the two eyes remain segregated in the ocular dominance columns of layer IV, the primary cortical target of lateral geniculate axons. Layer IV neurons send their axons to other cortical layers; it is at this stage that the information from the two eyes converges onto individual neurons. (B) Pattern of ocular dominance columns in human striate cortex. The alternating left and right eye columns in layer IV have been reconstructed from tissue sections and projected onto a photograph of the medial wall of the occipital lobe. (B from Horton and Hedley-Whyte, 1984.) Central Visual Pathways 271 neurons are monocular, driven by either the left or right eye but not by both (Figure 11.10; see also Figure 11.14). In some species, including most (but not all) primates, inputs from the left and right eyes remain segregated to some degree even beyond the geniculate because the axons of geniculate neurons terminate in alternating eye-specific columns within cortical layer IV—the so-called ocular dominance columns (see the next section). Beyond this point, the signals from the two eyes are combined at the cellular level. Thus, most cortical neurons have binocular receptive fields, and these fields are almost identical, having the same size, shape, preferred orientation, and roughly the same position in the visual field of each eye. Bringing together the inputs from the two eyes at the level of the striate cortex provides a basis for stereopsis, the special sensation of depth that arises from viewing nearby objects with two eyes instead of one. Because the two eyes look at the world from slightly different angles, objects that lie in front of or behind the plane of fixation project to noncorresponding points on the two retinas. To convince yourself of this fact, hold your hand at arm’s length and fixate on the tip of one finger. Maintain fixation on the finger as you hold a pencil in your other hand about half as far away. At this distance, the image of the pencil falls on noncorresponding points on the two retinas and will therefore be perceived as two separate pencils (a phenomenon called double vision, or diplopia). If the pencil is now moved toward the finger (the point of fixation), the two images of the pencil fuse and a single pencil is seen in front of the finger. Thus, for a small distance on either side of the plane of fixation, where the disparity between the two views of the world remains modest, a single image is perceived; the disparity between the two eye views of objects nearer or farther than the point of fixation is interpreted as depth (Figure 11.11). Although the neurophysiological basis of stereopsis is not understood, some neurons in the striate cortex and in other visual cortical areas have receptive field properties that make them good candidates for extracting information about binocular disparity. Unlike many binocular cells whose monocular receptive fields sample the same region of visual space, these neurons have monocular fields that are slightly displaced (or perhaps differ in their internal organization) so that the cell is maximally activated by stimuli that fall on noncorresponding parts of the retinas. Some of these neurons (so-called far cells) discharge to disparities beyond the plane of fixation, while others (near cells) respond to disparities in front of the plane of fixation. The pattern of activity in these different classes of neurons seems likely to contribute to sensations of stereoscopic depth (Box B). Interestingly, the preservation of the binocular responses of cortical neurons is contingent on the normal activity from the two eyes during early postnatal life. Anything that creates an imbalance in the activity of the two eyes—for example, the clouding of one lens or the abnormal alignment of the eyes during infancy (strabismus)—can permanently reduce the effectiveness of one eye in driving cortical neurons, and thus impair the ability to use binocular information as a cue for depth. Early detection and correction of visual problems is therefore essential for normal visual function in maturity (see Chapter 23). The Columnar Organization of the Striate Cortex The variety of response properties exhibited by cortical neurons raises the question of how neurons with different receptive fields are arranged within striate cortex. For the most part, the responses of neurons are qualitatively c b Fixation point a Right Left Far disparities Images of fixation point Near disparities Figure 11.11 Binocular disparities are generally thought to be the basis of stereopsis. When the eyes are fixated on point b, points that lie beyond the plane of fixation (point c) or in front of the point of fixation (point a) project to noncorresponding points on the two retinas. When these disparities are small, the images are fused and the disparity is interpreted by the brain as small differences in depth. When the disparities are greater, double vision occurs (although this normal phenomenon is generally unnoticed). 272 Chapter Eleven Box B Random Dot Stereograms and Related Amusements An important advance in studies of stereopsis was made in 1959 when Bela Julesz, then working at the Bell Laboratories in Murray Hill, New Jersey, discovered an ingenious way of showing that stereoscopy depends on matching information seen by the two eyes without any prior recognition of what object(s) such matching might generate. Julesz, a Hungarian whose background was in engineering and physics, was working on the problem of how to “break” camouflage. He surmised that the brain’s ability to fuse the slightly different views of the two eyes to bring out new information would be an aid in overcoming military camouflage. Julesz also realized that, if his hypothesis was correct, a hidden figure in a random pattern presented to the two eyes should emerge when a portion of the otherwise identical pattern was shifted horizontally in the view of one eye or the other. A horizontal shift in one direction would cause the hidden object to appear in front of the plane of the background, whereas a shift in the other direction would cause the hidden object to appear in back of the plane. Such a figure, called a random dot stereogram, and the method of its creation are shown in Figures A and B. The two images can be easily fused in a stereoscope (like the familiar Viewmaster® toy) but can also be fused simply by allowing the eyes to diverge. Most people find it easiest to do this by imagining that they are looking “through” the figure; after some seconds, during which the brain tries to make sense of what it is presented with, the two images merge and the hidden figure appears (in this case, a square that occupies the middle portion of the figure). The random dot stereogram has been widely used in stereoscopic research for about 40 years, although how such stimuli elicit depth remains very much a matter of dispute. (A) (B) Random dot stereograms and autostereograms. (A) to construct a random dot stereogram, a random dot pattern is created to be observed by one eye. The stimulus for the other eye is created by copying the first image, displacing a particular region horizontally, and then filling in the gap with a random sample of dots. (B) When the right and left images are viewed simultaneously but independently by the two eyes (by using a stereoscope or fusing the images by converging or diverging the eyes), the shifted region (a square) appears to be in a different plane from the other dots. (A after Wandell, 1995.) An impressive—and extraordinarily popular—derivative of the random dot stereogram is the autostereogram (Figure C). The possibility of autostereograms was first discerned by the nineteenthcentury British physicist David Brewster. While staring at a Victorian wallpaper with an iterated but offset pattern, he noticed that when the patterns were fused, he perceived two different planes. The plethora of autostereograms that can be seen today in posters, books, and newspapers are close cousins of the random dot stereogram in that computers are used to shift patterns of iterated Binocular fusion produces sensation that the shifted square is in front of the background plane. Central Visual Pathways 273 information with respect to each other. The result is that different planes emerge from what appears to be a meaningless array of visual information (or, depending on the taste of the creator, an apparently “normal” scene in which the iterated and displaced information is hidden). Some autostereograms are designed to reveal the hidden figure when the eyes diverge, and others when they converge. (Looking at a plane more distant than the plane of the surface causes divergence; looking at a plane in front of the picture causes the eyes to converge; see Figure 11.11.) The elevation of the autostereogram to a popular art form should probably be attributed to Chris W. Tyler, a student of Julesz’s and a visual psychophysicist, who was among the first to create commercial autostereograms. Numerous graphic artists—preeminently in Japan, where the popularity of the autostereogram has been enormous—have gener- ated many of such images. As with the random dot stereogram, the task in viewing the autostereogram is not clear to the observer. Nonetheless, the hidden figure emerges, often after minutes of effort in which the brain automatically tries to make sense of the occult information. (C) (C) An autostereogram. The hidden figure (three geometrical forms) emerges by diverging the eyes in this case. (C courtesy of Jun Oi.) similar at any one point in primary visual cortex, but tend to shift smoothly across its surface. With respect to orientation, for example, all the neurons encountered in an electrode penetration perpendicular to the surface at a particular point will very likely have the same orientation preference, forming a “column” of cells with similar response properties. Adjacent columns, however, usually have slightly different orientation preferences; the sequence of orientation preferences encountered along a tangential electrode penetration gradually shifts as the electrode advances (Figure 11.12). Thus, orientation preference is mapped in the cortex, much like receptive field References JULESZ, B. (1971) Foundations of Cyclopean Perception. Chicago: The University of Chicago Press. JULESZ, B. (1995) Dialogues on Perception. Cambridge, MA: MIT Press. N. E. THING ENTERPRISES (1993) Magic Eye: A New Way of Looking at the World. Kansas City: Andrews and McMeel. 274 Chapter Eleven Figure 11.12 Columnar organization of orientation selectivity in the monkey striate cortex. Vertical electrode penetrations encounter neurons with the same preferred orientations, whereas oblique penetrations show a systematic change in orientation across the cortical surface. The circles denote the lack of orientation-selective cells in layer IV. Vertical electrode penetration I II−III IV V Oblique electrode penetration VI location (Box C). Unlike the map of visual space, however, the map of orientation preference is iterated many times, such that the same orientation preference is repeated at approximately 1-mm intervals across the striate cortex. This iteration presumably ensures that there are neurons for each region of visual space that represent the full range of orientation values. The orderly progression of orientation preference (as well as other properties that are mapped in this systematic way) is accommodated within the orderly map of visual space by the fact that the mapping is relatively coarse. Each small region of visual space is represented by a set of neurons whose receptive fields cover the full range of orientation preferences, the set being distributed over several millimeters of the cortical surface The columnar organization of the striate cortex is equally apparent in the binocular responses of cortical neurons. Although most neurons in the striate cortex respond to stimulation of both eyes, the relative strength of the inputs from the two eyes varies from neuron to neuron. At the extremes of this continuum are neurons that respond almost exclusively to the left or right eye; in the middle are those that respond equally well to both eyes. As in the case of orientation preference, vertical electrode penetrations tend to encounter neurons with similar ocular preference (or ocular dominance, as it is usually called), whereas tangential penetrations show gradual shifts in ocular dominance. And, like the arrangement of orientation preference, a movement of about a millimeter across the surface is required to sample the full complement of ocular dominance values (Figure 11.13). These shifts in ocular dominance result from the ocular segregation of the inputs from lateral geniculate nucleus within cortical layer IV (see Figure 11.10). Although the modular arrangement of the visual cortex was first recognized on the basis of these orientation and ocular dominance columns, further work has shown that other stimulus features such as color, direction of motion, and spatial frequency also tend to be distributed in iterated patterns that are systematically related to each other (for example, orientation columns tend to intersect ocular dominance columns at right angles). In short, the striate cortex is composed of repeating units, or modules, that contain all the neuronal machinery necessary to analyze a small region of visual space for a variety of different stimulus attributes. As described in Box D in Chapter 8, a number of other cortical regions show a similar columnar arrangement of their processing circuitry. Central Visual Pathways 275 (A) Left (B) Distance along electrode track Right 1 Ocular dominance groups 2 1 2 3 4 5 6 7 Recording site 3 Cortical cell 4 Electr Electrode ode tr ack 5 Layer IV 6 Left Right eye eye Left eye Right eye Left Right eye eye 7 Division of Labor within the Primary Visual Pathway In addition to being specific for input from one eye or the other, the layers in the lateral geniculate are also distinguished on the basis of cell size: Two ventral layers are composed of large neurons and are referred to as the magnocellular layers, while more dorsal layers are composed of small neurons and are referred to as the parvocellular layers. The magno- and parvocellular layers receive inputs from distinct populations of ganglion cells that exhibit corresponding differences in cell size. M ganglion cells that terminate in the magnocellular layers have larger cell bodies, more extensive dendritic fields, and larger-diameter axons than the P ganglion cells that terminate in the parvocellular layers (Figure 11.14A). Moreover, the axons of relay cells in the magno- and parvocellular layers of the lateral geniculate nucleus terminate on distinct populations of neurons located in separate strata within layer 4 of striate cortex. Thus the retinogeniculate pathway is composed of parallel magnocellular and parvocellular streams that convey distinct types of information to the initial stages of cortical processing. The response properties of the M and P ganglion cells provide important clues about the contributions of the magno- and parvocellular streams to visual perception. M ganglion cells have larger receptive fields than P cells, and their axons have faster conduction velocities. M and P ganglion cells also differ in ways that are not so obviously related to their morphology. M cells respond transiently to the presentation of visual stimuli, while P cells respond in a sustained fashion. Moreover, P ganglion cells can transmit information about color, whereas M cells cannot. P cells convey color information because their receptive field centers and surrounds are driven by different classes of cones (i.e., cones responding with greatest sensitivity to Figure 11.13 Columnar organization of ocular dominance. (A) Cortical neurons in all layers vary in the strength of their response to the inputs from the two eyes, from complete domination by one eye to equal influence of the two eyes. (B) Tangential electrode penetration across the superficial cortical layers reveals a gradual shift in the ocular dominance of the recorded neurons from one eye to the other. In contrast, all neurons encountered in a vertical electrode penetration (other than those neurons that lie in layer IV) tend to have the same ocular dominance. 276 Chapter Eleven Box C Optical Imaging of Functional Domains in the Visual Cortex The recent availability of optical imaging techniques has made it possible to visualize how response properties, such as the selectivity for edge orientation or ocular dominance, are mapped across the cortical surface. These methods generally rely on intrinsic signals (changes in the amount of light reflected from the cortical surface) that correlate with levels of neural activity. Such signals are thought to arise at least in part from local changes in the ratio of oxyhemoglobin and deoxyhemoglobin that accompany such activity, more active areas having a higher deoxyhemoglobin/oxyhemoglobin ratio (see also Box A in Chapter 1). This change can be detected when the cortical surface is illuminated with red light (605–700 nm). Under these conditions, active cortical regions absorb more light than less active ones. With the use of a sensitive video camera, and averaging over a number of trials (the changes are small, 1 or 2 parts per thousand), it is possible to visualize these differences and use them to map cortical patterns of activity (Figure A). This approach has now been successfully applied to both striate and extrastri- ate areas in both experimental animals and human patients undergoing neurosurgery. The results emphasize that maps of stimulus features are a general principle of cortical organization. For example, orientation preference is mapped in a continuous fashion such that adjacent positions on the cortical surface tend to have only slightly shifted orientation preferences. However, there are points where continuity breaks down. Around these points, orientation preference is represented in a radial pattern resembling a pinwheel, covering the whole 180° of possible orientation values (Figure B). This powerful technique can also be used to determine how maps for different stimulus properties are arranged relative to one another, and to detect additional maps such as that for direction of motion. A comparison of ocular dominance bands and orientation preference maps, for example, shows that pinwheel centers are generally located in the center of ocular dominance bands, and that the iso-orientation contours that emanate from the pinwheel centers run orthogonal to the borders of ocular dominance bands (Figure C). An orderly relation- (A) References BLASDEL, G. G. AND G. SALAMA (1986) Voltagesensitive dyes reveal a modular organization in monkey striate cortex. Nature 321: 579–585. BONHOEFFER, T. AND A. GRINVALD (1993) The layout of iso-orientation domains in area 18 of the cat visual cortex: Optical imaging reveals a pinwheel-like organization. J. Neurosci 13: 4157–4180. BONHOEFFER, T. AND A. GRINVALD (1996) Optical imaging based on intrinsic signals: The methodology. In Brain Mapping: The Methods, A. Toge (ed.). New York: Academic Press. OBERMAYER, K. AND G. G. BLASDEL (1993) Geometry of orientation and ocular dominance columns in monkey striate cortex. J. Neurosci. 13: 4114–4129. WELIKY, M., W. H. BOSKING AND D. FITZPATRICK (1996) A systematic map of direction preference in primary visual cortex. Nature 379: 725–728. (B) M ganglion cell 6 Koniocellular layers Parvocellular layers 5 4 3 2 Magnocellular layers P ganglion cell ship between maps of orientation selectivity and direction selectivity has also been demonstrated. These systematic relationships between the functional maps that coexist within primary visual cortex are thought to ensure that all combinations of stimulus features (orientation, direction, ocular dominance, and spatial frequency) are analyzed for all regions of visual space. K ganglion cell 1 1 mm Central Visual Pathways 277 (A) (B) Illuminator Video camera Macro lens Visual stimulation computer Optical chamber Imaging computer (C) Data display Monitor (A) The technique of optical imaging. A sensitive video camera is used to record changes in light absorption that occur as the animal views various stimuli presented on a video Purves Neuroscience 3E monitor. Images are digitized and stored in a computer in order to construct maps that Pyramis Studios compare patterns of activity associated with different stimuli. (B) Maps of orientation preference in the visual cortex visualized with optical imaging. Each color represents the P3_12BXC angle 121503 of an edge that was most effective in activating the neurons at a given site. Orientation preference changes in a continuous fashion, rotating around pinwheel centers. (Note that this image shows only a small region of the overall map of orientation) (C) Comparison of optical image maps of orientation preference and ocular dominance in monkey visual cortex. The thick black lines represent the borders between ocular dominance columns. The thin gray lines represent the iso-orientation contours, which converge at orientation pinwheel centers (arrow). Iso-orientation contour lines generally intersect the borders of ocular dominance bands at right angles. (B from Bonhoeffer and Grinvald, 1993; C from Obermeyer and Blasdel, 1993.) ▲ short-, medium-, or long-wavelength light). For example, some P ganglion cells have centers that receive inputs from long-wavelength (“red”) sensitive cones and surrounds that receive inputs from medium-wavelength (“green”) cones. Others have centers that receive inputs from “green cones” and surrounds from “red cones” (see Chapter 10). As a result, P cells are sensitive to differences in the wavelengths of light striking their receptive field center Figure 11.14 Magno- and parvocellular streams. (A) Tracings of M and P ganglion cells as seen in flat mounts of the retina after staining by the Golgi method. M cells have large-diameter cell bodies and large dendritic fields. They supply the magnocellular layers of the lateral geniculate nucleus. P cells have smaller cell bodies and dendritic fields. They supply the parvocellular layers of the lateral geniculate nucleus. (B) Photomicrograph of the human lateral geniculate nucleus showing the magnocellular and parvocellular layers. (A after Watanabe and Rodieck, 1989; B courtesy of T. Andrews and D. Purves.) 278 Chapter Eleven and surround. Although M ganglion cells also receive inputs from cones, there is no difference in the type of cone input to the receptive field center and surround; the center and surround of each M cell receptive field is driven by all cone types. The absence of cone specificity to center-surround antagonism makes M cells largely insensitive to differences in the wavelengths of light that strike their receptive field centers and surrounds, and they are thus unable to transmit color information to their central targets. The contribution of the magno- and parvocellular streams to visual perception has been tested experimentally by examining the visual capabilities of monkeys after selectively damaging either the magno- or parvocellular layers of the lateral geniculate nucleus. Damage to the magnocellular layers has little effect on visual acuity or color vision, but sharply reduces the ability to perceive rapidly changing stimuli. In contrast, damage to the parvocellular layers has no effect on motion perception but severely impairs visual acuity and color perception. These observations suggest that the visual information conveyed by the parvocellular stream is particularly important for high spatial resolution vision—the detailed analysis of the shape, size, and color of objects. The magnocellular stream, on the other hand, appears critical for tasks that require high temporal resolution, such as evaluating the location, speed and direction of a rapidly moving object. In addition to the magno- and parvocellular streams, a third distinct anatomical pathway—the koniocellular, or K-cell pathway—has been identified within the lateral geniculate nucleus. Neurons contributing to the Kcell pathway reside in the interlaminar zones that separate lateral geniculate layers; these neurons receive inputs from fine-caliber retinal axons and project in a patchy fashion to the superficial layers (layers II and III) of striate cortex. Although the contribution of the K-cell pathway to perception is not understood, it appears that some aspects of color vision, especially information derived from short-wavelength-sensitive cones, may be transmitted via the K-cell rather than the P-cell pathway. Why short-wavelength-sensitive cone signals should be processed differently from middle- and long-wavelength information is not clear, but the distinction may reflect the earlier evolutionary origin of the K-cell pathway (see Chapter 10). The Functional Organization of Extrastriate Visual Areas Anatomical and electrophysiological studies in monkeys have led to the discovery of a multitude of areas in the occipital, parietal, and temporal lobes that are involved in processing visual information (Figure 11.15). Each of these areas contains a map of visual space, and each is largely dependent on the primary visual cortex for its activation. The response properties of the neurons in some of these regions suggest that they are specialized for different aspects of the visual scene. For example, the middle temporal area (MT) contains neurons that respond selectively to the direction of a moving edge without regard to its color. In contrast, neurons in another cortical area called V4 respond selectively to the color of a visual stimulus without regard to its direction of movement. These physiological findings are supported by behavioral evidence; thus, damage to area MT leads to a specific impairment in a monkey’s ability to perceive the direction of motion in a stimulus pattern, while other aspects of visual perception remain intact. Recent functional imaging studies have indicated a similar arrangement of visual areas within human extrastriate cortex. Using retinotopically restricted stimuli, it has been possible to localize at least 10 separate repre- Central Visual Pathways 279 (A) (B) MT V4 Motor V2 V1 Visual areas V2 Somatic sensory MT V3 Auditory V4 V1 V3 V2 V2 V1 V2 V4 sentations of the visual field (Figure 11.16). One of these areas exhibits a large motion-selective signal, suggesting that it is the homologue of the motion-selective middle temporal area described in monkeys. Another area exhibits color-selective responses, suggesting that it may be similar to V4 in non-human primates. A role for these areas in the perception of motion and color, respectively, is further supported by evidence for increases in activity not only during the presentation of the relevant stimulus, but also during periods when subjects experience motion or color afterimages. The clinical description of selective visual deficits after localized damage to various regions of extrastriate cortex also supports functional specialization of extrastriate visual areas in humans. For example, a well-studied patient who suffered a stroke that damaged the extrastriate region thought to be comparable to area MT in the monkey was unable to appreciate the motion of objects. The neurologist who treated her noted that she had difficulty in pouring tea into a cup because the fluid seemed to be “frozen.” In addition, she could not stop pouring at the right time because she was unable to perceive when the fluid level had risen to the brim. The patient also had trouble following a dialogue because she could not follow the movements of the speaker’s mouth. Crossing the street was potentially terrifying because she couldn’t judge the movement of approaching cars. As the patient related, “When I’m looking at the car first, it seems far away. But Figure 11.15 Subdivisions of the extrastriate cortex in the macaque monkey. (A) Each of the subdivisions indicated in color contains neurons that respond to visual stimulation. Many are buried in sulci, and the overlying cortex must be removed in order to expose them. Some of the more extensively studied extrastriate areas are specifically identified (V2, V3, V4, and MT). V1 is the primary visual cortex; MT is the middle temporal area. (B) The arrangement of extrastriate and other areas of neocortex in a flattened view of the monkey neocortex. There are at least 25 areas that are predominantly or exclusively visual in function, plus 7 other areas suspected to play a role in visual processing. (A after Maunsell and Newsome, 1987; B after Felleman and Van Essen, 1991.) 280 Chapter Eleven (A) Lateral (C) Brain “inflated” to reveal buried cortex MT V3a VP Flattened occipital lobe (B) Medial V3 V1 V2 V3a V3 V3a Calcarine sulcus V2 V1 V2 MST VP V4 Figure 11.16 Localization of multiple visual areas in the human brain using fMRI. (A,B) Lateral and medial views (respectively) of the human brain, illustrating the location of primary visual cortex (V1) and additional visual areas V2, V3, VP (ventral posterior area), V4, MT (middle temporal area), and MST (medial superior temporal area). (C) Unfolded and flattened view of retinotopically defined visual areas in the occipital lobe. Dark grey areas correspond to cortical regions that were buried in sulci; light regions correspond to regions that were located on the surface of gyri. Visual areas in humans show a close resemblance to visual areas originally defined in monkeys (compare with Figure 11.15). (After Sereno et al., 1995.) Sulci MT V1 V4 VP V2 Gyri then, when I want to cross the road, suddenly the car is very near.” Her ability to perceive other features of the visual scene, such as color and form, was intact. Another example of a specific visual deficit as a result of damage to extrastriate cortex is cerebral achromatopsia. These patients lose the ability to see the world in color, although other aspects of vision remain in good working order. The normal colors of a visual scene are described as being replaced by “dirty” shades of gray, much like looking at a poor quality black-and-white movie. Achromatopsic individuals know the normal colors of objects—that a school bus is yellow, an apple red—but can no longer see them. Thus, when asked to draw objects from memory, they have no difficulty with shapes but are unable to appropriately color the objects they have represented. It is important to distinguish this condition from the color blindness that arises from the congenital absence of one or more cone pigments in the retina (see Chapter 10). In achromatopsia, the three types of cones are functioning normally; it is damage to specific extrastriate cortical areas that renders the patient unable to use the information supplied by the retina. Central Visual Pathways 281 Based on the anatomical connections between visual areas, differences in electrophysiological response properties, and the effects of cortical lesions, a consensus has emerged that extrastriate cortical areas are organized into two largely separate systems that eventually feed information into cortical association areas in the temporal and parietal lobes (see Chapter 25). One system, called the ventral stream, includes area V4 and leads from the striate cortex into the inferior part of the temporal lobe. This system is thought to be responsible for high-resolution form vision and object recognition. The dorsal stream, which includes the middle temporal area, leads from striate cortex into the parietal lobe. This system is thought to be responsible for spatial aspects of vision, such as the analysis of motion, and positional relationships between objects in the visual scene (Figure 11.17). The functional dichotomy between these two streams is supported by observations on the response properties of neurons and the effects of selective cortical lesions. Neurons in the ventral stream exhibit properties that are important for object recognition, such as selectivity for shape, color, and texture. At the highest levels in this pathway, neurons exhibit even greater selectivity, responding preferentially to faces and objects (see Chapter 25). In contrast, those in the dorsal stream are not tuned to these properties, but show selectivity for direction and speed of movement. Consistent with this interpretation, lesions of the parietal cortex severely impair an animal’s ability to distinguish objects on the basis of their position, while having little effect on its ability to perform object recognition tasks. In contrast, lesions of the inferotemporal cortex produce profound impairments in the ability to perform recognition tasks but no impairment in spatial tasks. These effects are remarkably similar to the syndromes associated with damage to the parietal and temporal lobe in humans (see Chapters 25 and 26). What, then, is the relationship between these “higher-order” extrastriate visual pathways and the magno- and parvocellular pathways that supply the primary visual cortex? Not long ago, it seemed that these intracortical pathways were simply a continuation of the geniculostriate pathways—that is, the magnocellular pathway provided input to the dorsal stream and the parvocellular pathway provided input to the ventral stream. However, more recent work has indicated that the situation is more complicated. The temporal pathway clearly has access to the information conveyed by both the magno- and parvocellular streams; and the parietal pathway, while dominated by inputs from the magnocellular stream, also receives inputs from the parvocellular stream. Thus, interaction and cooperation between the magno- and parvocellular streams appear to be the rule in complex visual perceptions. Summary Distinct populations of retinal ganglion cells send their axons to a number of central visual structures that serve different functions. The most important projections are to the pretectum for mediating the pupillary light reflex, to the hypothalamus for the regulation of circadian rhythms, to the superior colliculus for the regulation of eye and head movements, and—most important of all—to the lateral geniculate nucleus for mediating vision and visual perception. The retinogeniculostriate projection (the primary visual pathway) is arranged topographically such that central visual structures contain an organized map of the contralateral visual field. Damage anywhere along the primary visual pathway, which includes the optic nerve, optic tract, lateral geniculate nucleus, optic radiation, and striate cortex, results in a loss of vision confined to a predictable region of visual space. Compared to retinal Parietal lobe Dorsal (spatial vision) pathway MT V4 Temporal lobe V2 V2 V1 Ventral (object recognition) pathway Figure 11.17 The visual areas beyond the striate cortex are broadly organized into two pathways: a ventral pathway that leads to the temporal lobe, and a dorsal pathway that leads to the parietal lobe. The ventral pathway plays an important role in object recognition, the dorsal pathway in spatial vision. 282 Chapter Eleven ganglion cells, neurons at higher levels of the visual pathway become increasingly selective in their stimulus requirements. Thus, most neurons in the striate cortex respond to light–dark edges only if they are presented at a certain orientation; some are selective for the length of the edge, and others to movement of the edge in a specific direction. Indeed, a point in visual space is related to a set of cortical neurons, each of which is specialized for processing a limited set of the attributes in the visual stimulus. The neural circuitry in the striate cortex also brings together information from the two eyes; most cortical neurons (other than those in layer IV, which are segregated into eye-specific columns) have binocular responses. Binocular convergence is presumably essential for the detection of binocular disparity, an important component of depth perception. The primary visual pathway is composed of separate functional streams that convey information from different types of retinal ganglion cells to the initial stages of cortical processing. The magnocellular stream conveys information that is critical for the detection of rapidly changing stimuli, the parvocellular stream mediates high acuity vision and appears to share responsibility for color vision with the koniocellular stream. Finally, beyond striate cortex, parcellation of function continues in the ventral and dorsal streams that lead to the extrastriate and association areas in the temporal and parietal lobes, respectively. Areas in the inferotemporal cortex are especially important in object recognition, whereas areas in the parietal lobe are critical for understanding the spatial relations between objects in the visual field. Additional Reading Reviews BERSON, D. M. (2003) Strange vision: Ganglion cells as circadian photoreceptors. Trends Neurosci. 26: 314–320. COURTNEY, S. M. AND L. G. UNGERLEIDER (1997) What fMRI has taught us about human vision. Curr. Op. Neurobiol. 7: 554–561. FELLEMAN, D. J. AND D. C. VAN ESSEN (1991) Distributed hierarchical processing in primate cerebral cortex. Cerebral Cortex 1: 1–47. HORTON, J. C. (1992) The central visual pathways. In Alder’s Physiology of the Eye. W. M. Hart (ed.). St. Louis: Mosby Yearbook. HENDRY, S. H. AND R. C. REID (2000) The koniocellular pathway in primate vision. Annu. Rev. Neurosci. 23: 127–153. HUBEL, D. H. AND T. N. WIESEL (1977) Functional architecture of macaque monkey visual cortex. Proc. R. Soc. (Lond.) 198: 1–59. MAUNSELL, J. H. R. (1992) Functional visual streams. Curr. Opin. Neurobiol. 2: 506–510. SCHILLER, P. H. AND N. K. LOGOTHETIS (1990) The color-opponent and broad-band channels of the primate visual system. Trends Neurosci. 13: 392–398. TOOTELL, R.B., A. M. DALE, M. I. SERENO AND R. MALACH (1996) New images from human visual cortex. Trends Neurosci. 19: 481–489. UNGERLEIDER, J. G. AND M. MISHKIN (1982) Two cortical visual systems. In Analysis of Visual Behavior. D. J. Ingle, M. A. Goodale and R. J. W. Mansfield (eds.). Cambridge, MA: MIT Press, pp. 549–586. HUBEL, D. H. AND T. N. WIESEL (1968) Receptive fields and functional architecture of monkey striate cortex. J. Physiol. (Lond.) 195: 215–243. SERENO, M. I. AND 7 OTHERS (1995) Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging. Science 268: 889–893. ZIHL, J., D. VON CRAMON AND N. MAI (1983) Selective disturbance of movement vision after bilateral brain damage. Brain 106: 313–340. Important Original Papers Books HATTAR, S., H. W. LIAO, M. TAKAO, D. M. BERSON AND K. W. YAU (2002) Melanopsincontaining retinal ganglion cells: Architecture, projections, and intrinsic photosensitivity. Science 295: 1065–1070. HUBEL, D. H. AND T. N. WIESEL (1962) Receptive fields, binocular interaction and functional architecture in the cat’s visual cortex. J. Physiol. (Lond.) 160: 106–154. CHALUPA, L. M. AND J. S. WERNER (EDS.) (2004) The Visual Neurosciences. Cambridge, MA: MIT Press. HUBEL, D. H. (1988) Eye, Brain, and Vision. New York: Scientific American Library. RODIECK, R. W. (1998) The First Steps in Seeing. Sunderland, MA: Sinauer Associates. ZEKI, S. (1993) A Vision of the Brain. Oxford: Blackwell Scientific Publications. Chapter 12 The Auditory System Overview The auditory system is one of the engineering masterpieces of the human body. At the heart of the system is an array of miniature acoustical detectors packed into a space no larger than a pea. These detectors can faithfully transduce vibrations as small as the diameter of an atom, and they can respond a thousand times faster than visual photoreceptors. Such rapid auditory responses to acoustical cues facilitate the initial orientation of the head and body to novel stimuli, especially those that are not initially within the field of view. Although humans are highly visual creatures, much human communication is mediated by the auditory system; indeed, loss of hearing can be more socially debilitating than blindness. From a cultural perspective, the auditory system is essential not only to understanding speech, but also to music, one of the most aesthetically sophisticated forms of human expression. For these and other reasons, audition represents a fascinating and especially important mode of sensation. Sound In physical terms, sound refers to pressure waves generated by vibrating air molecules (somewhat confusingly, sound is used more casually to refer to an auditory percept). Sound waves are much like the ripples that radiate outward when a rock is thrown in a pool of water. However, instead of occurring across a two-dimensional surface, sound waves propagate in three dimensions, creating spherical shells of alternating compression and rarefaction. Like all wave phenomena, sound waves have four major features: waveform, phase, amplitude (usually expressed in log units known as decibels, abbreviated dB), and frequency (expressed in cycles per second or Hertz, abbreviated Hz). For human listeners, the amplitude and frequency of a sound pressure change at the ear roughly correspond to loudness and pitch, respectively. The waveform of a sound stimulus is its amplitude plotted against time. It helps to begin by visualizing an acoustical waveform as a sine wave. At the same time, it must be kept in mind that sounds composed of single sine waves (i.e., pure tones) are extremely rare in nature; most sounds in speech, for example, consist of acoustically complex waveforms. Interestingly, such complex waveforms can often be modeled as the sum of sinusoidal waves of varying amplitudes, frequencies, and phases. In engineering applications, an algorithm called the Fourier transform decomposes a complex signal into its sinusoidal components. In the auditory system, as will be apparent later in the chapter, the inner ear acts as a sort of acoustical prism, decomposing complex sounds into a myriad of constituent tones. 283 284 Chapter Twelve Tuning fork aves ric w nt once C Sinusoidal wave + Air pressure − Normal atmospheric pressure Distance Figure 12.1 Diagram of the periodic condensation and rarefaction of air molecules produced by the vibrating tines of a tuning fork. The molecular disturbance of the air is pictured as if frozen at the instant the constituent molecules responded to the resultant pressure wave. Shown below is a plot of the air pressure versus distance from the fork. Note its sinusoidal quality. Figure 12.1 diagrams the behavior of air molecules near a tuning fork that vibrates sinusoidally when struck. The vibrating tines of the tuning fork produce local displacements of the surrounding molecules, such that when the tine moves in one direction, there is molecular condensation; when it moves in the other direction, there is rarefaction. These changes in density of the air molecules are equivalent to local changes in air pressure. Such regular, sinusoidal cycles of compression and rarefaction can be thought of as a form of circular motion, with one complete cycle equivalent to one full revolution (360°). This point can be illustrated with two sinusoids of the same frequency projected onto a circle, a strategy that also makes it easier to understand the concept of phase (Figure 12.2). Imagine that two tuning forks, both of which resonate at the same frequency, are struck at slightly different times. At a given time t = 0, one wave is at position P and the other at position Q. By projecting P and Q onto the circle, their respective phase angles, θ1 and θ2, are apparent. The sine wave that starts at P reaches a particular point on the circle, say 180°, at time t1, whereas the wave that starts at Q reaches 180° at time t2. Thus, phase differences have corresponding time differences, a concept that is important in appreciating how the auditory system locates sounds in space. The human ear is extraordinarily sensitive to sound pressure. At the threshold of hearing, air molecules are displaced an average of only 10 picometers (10–11 m), and the intensity of such a sound is about one-trillionth of a watt per square meter! This means a listener on an otherwise noiseless planet could hear a 1-watt, 3-kHz sound source located over 450 km away (consider that even a very dim light bulb consumes more than 1 watt of power). Even dangerously high sound pressure levels (>100 dB) have power at the eardrum that is only in the milliwatt range (Box A). The Audible Spectrum P θ1 θ2 Q O t2 t1 Time Figure 12.2 A sine wave and its projection as circular motion. The two sinusoids shown are at different phases, such that point P corresponds to phase angle θ1 and point Q corresponds to phase angle θ2. Humans can detect sounds in a frequency range from about 20 Hz to 20 kHz. Human infants can actually hear frequencies slightly higher than 20 kHz, but lose some high-frequency sensitivity as they mature; the upper limit in average adults is closer to 15–17 kHz. Not all mammalian species are sensitive to the same range of frequencies. Most small mammals are sensitive to very high frequencies, but not to low frequencies. For instance, some species of bats are sensitive to tones as high as 200 kHz, but their lower limit is around 20 kHz—the upper limit for young people with normal hearing. One reason for these differences is that small objects, including the auditory structures of these small mammals, resonate at high frequencies, whereas large objects tend to resonate at low frequencies—which explains why the violin has a higher pitch than the cello. Different animal species tend to emphasize frequency bandwidths in both their vocalizations and their range of hearing. In general, vocalizations by virtue of their periodicity can be distinguished from the noise “barrier” created by environmental sounds, such as wind and rustling leaves. Animals that echolocate, such as bats and dolphins, rely on very high-frequency vocal sounds to maximally resolve spatial features of the target, while animals intent on avoiding predation have auditory systems “tuned” to the low frequency vibrations that approaching predators transmit through the substrate. These behavioral differences are mirrored by a wealth of anatomical and functional specializations throughout the auditory system. The Auditor y System 285 Box A Four Causes of Acquired Hearing Loss Acquired hearing loss is an increasingly common sensory deficit that can often lead to impaired oral communication and social isolation. Four major causes of acquired hearing loss are acoustical trauma, infection of the inner ear, ototoxic drugs, and presbyacusis (literally, the hearing of the old). The exquisite sensitivity of the auditory periphery, combined with the direct mechanical linkage between the acoustical stimulus and the receptor cells, make the ear especially susceptible to acute or chronic acoustical trauma. Extremely loud, percussive sounds, such as those generated by explosives or gunfire, can rupture the eardrum and so severely distort the inner ear that the organ of Corti is torn. The resultant loss of hearing is abrupt and often quite severe. Less well appreciated is the fact that repeated exposure to less dramatic but nonetheless loud sounds, including those produced by industrial or household machinery or by amplified musical instruments, can also damage the inner ear. Although these sounds leave the eardrum intact, specific damage is done to the hair bundle itself; the stereocilia of cochlear hair cells of animals exposed to loud sounds shear off at their pivot points with the hair cell body, or fuse together in a platelike fashion that impedes movement. In humans, the mechanical resonance of the ear to stimulus frequencies centered about 3 kHz means that exposure to loud, broadband noises (such as those generated by jet engines) results in especially pronounced deficits near this resonant frequency. Ototoxic drugs include aminoglycoside antibiotics (such as gentamycin and kanamycin), which directly affect hair cells, and ethacrynic acid, which poisons the potassium-extruding cells of the stria vascularis that generate the endocochlear potential. In the absence of these ion pumping cells, the endocochlear potential, which supplies the energy to drive the transduction process, is lost. Although still a matter of some debate, the relatively nonselective transduction channel apparently affords a means of entry for aminoglycoside antibiotics, A Synopsis of Auditory Function The auditory system transforms sound waves into distinct patterns of neural activity, which are then integrated with information from other sensory systems to guide behavior, including orienting movements to acoustical stimuli and intraspecies communication. The first stage of this transformation occurs at the external and middle ears, which collect sound waves and amplify their pressure, so that the sound energy in the air can be successfully transmitted to the fluid-filled cochlea of the inner ear. In the inner ear, a series of biomechanical processes occur that break up the signal into simpler, sinusoidal components, with the result that the frequency, amplitude, and phase of the original signal are all faithfully transduced by the sensory hair cells and encoded by the electrical activity of the auditory nerve fibers. One product of this process of acoustical decomposition is the systematic representation of sound frequency along the length of the cochlea, referred to as tonotopy, which is an important organizational feature preserved which then poison hair cells by disrupting phosphoinositide metabolism. In particular, outer hair cells and those inner hair cells that transduce high-frequency stimuli are more affected, simply because of their greater energy requirements. Finally, presbyacusis, the hearing loss associated with aging, may in part stem from atherosclerotic damage to the especially fine microvasculature of the inner ear, as well as from genetic predispositions to hair cell damage. Recent advances in understanding the genetic transmission of acquired hearing loss in both humans and mice point to mutations in myosin isoforms unique to hair cells as a likely culprit. References HOLT, J. R. AND D. P. COREY (1999) Ion channel defects in hereditary hearing loss. Neuron 22: 217–219. KEATS, B. J. AND D. P. COREY (1999) The usher syndromes. Amer. J. Med. Gen. 89: 158–166. PRIUSKA, E. M. AND J. SCHACT (1997) Mechanism and prevention of aminoglycoside ototoxicity: Outer hair cells as targets and tools. Ear, Nose, Throat J. 76: 164–171. 286 Chapter Twelve Box B Music Even though we all recognize it when we hear it, the concept of music is vague. The Oxford English Dictionary defines it as “The art or science of combining vocal or instrumental sounds with a view toward beauty or coherence of form and expression of emotion.” In terms of the present chapter, music chiefly concerns the aspect of human audition that is experienced as tones. The stimuli that give rise to tonal percepts are periodic, meaning that they repeat systematically over time, as in the sine wave in Figure 12.1. Periodic stimuli, which do not occur naturally as sine waves but rather as complex repetitions involving a number of different frequencies, give rise to a sense of harmony when sounded together in appropriate combinations, and a sense of melody when they occur sequentially. Although we usually take the way tone-evoking stimuli are heard for granted, this aspect of audition presents some profoundly puzzling qualities. The most obvious of these is that humans perceive periodic stimuli whose fundamental frequencies have a 2:1 ratio as highly similar, and, for the most part, musically interchangeable. Thus in West- ern musical terminology, any two tones related by an interval of one or more octaves are given the same name (i.e., A, B, C…G), and are distinguished only by a qualifier that denotes relative ordinal position (e.g., C1, C2, C3, etc.). As a result, music is framed in repeating intervals (called octaves) defined by these more or less interchangeable tones. A key question, then, is why periodic sound stimuli whose fundamentals have a 2:1 ratio are perceived as similar when there is no obvious physical or physiological basis for this phenomenon. A second puzzling feature is that most if not all musical traditions subdivide octaves into a relatively small set of intervals for composition and performance, each interval being defined by its relationship to the lowest tone of the set. Such sets are called musical scales. The scales predominantly employed in all cultures over the centuries have used some (or occasionally all) of the 12 tonal intervals that in Western musical terminology are referred to as the chromatic scale (see figure). Moreover, some intervals of the chromatic scale, such as the fifth, the fourth, the major third, and the major sixth, are more often used in com- Illustration of 10 of the 12 tones in the chromatic scale, related to a piano keyboard. The function above the keyboard indicates that these tones correspond statistically to peaks of power in normalized human speech. (After Schwartz et al., 2003.) position and performance than others. These form the majority of the intervals employed in the pentatonic and diatonic major scales, the two most frequently used scales in music world-wide. Again, throughout the central auditory pathways. The earliest stage of central processing occurs at the cochlear nucleus, where the peripheral auditory information diverges into a number of parallel central pathways. Accordingly, the output of the cochlear nucleus has several targets. One of these is the superior olivary complex, the first place that information from the two ears interacts and the site of the initial processing of the cues that allow listeners to localize sound in space. The cochlear nucleus also projects to the inferior colliculus of the midbrain, a major integrative center and the first place where auditory information can interact with the motor system. The inferior colliculus is an obligatory relay for information traveling to the thalamus and cortex, where additional integrative aspects (such as harmonic and temporal combinations) of sound especially germane to speech and music are processed (Box B). The large number of stations between the auditory periphery and the cortex far exceeds those in other sensory systems, providing a hint that the perception of communication and environmental sounds The Auditor y System 287 there is no principled explanation of these preferences among all the possible intervals within the octave. Perhaps the most fundamental question in music—and arguably the common denominator of all musical tonality—is why certain combinations of tones are perceived as relatively consonant or ‘harmonious’ and others relatively dissonant or ‘inharmonious’. These perceived differences among the possible combinations of tones making up the chromatic scale are the basis for polytonal music, in which the perception of relative harmoniousness guides the composition of chords and melodic lines. The more compatible of these combinations are typically used to convey ‘resolution’ at the end of a musical phrase or piece, whereas less compatible combinations are used to indicate a transition, a lack of resolution, or to introduce a sense of tension in a chord or melodic sequence. Like octaves and scales, the reason for this phenomenology remains a mystery. The classical approaches to rationalizing octaves, scales and consonance have been based on the fact that the musical intervals corresponding to octaves, fifths, and fourths (in modern musical terminology) are produced by physical sources whose relative proportions (e.g., the relative lengths of two plucked strings or their fundamental frequencies) have ratios of 2:1, 3:2, or 4:3, respectively (these relationships were first described by Pythagoras). This coincidence of numerical simplicity and perceptual effect has been so impressive over the centuries that attempts to rationalize phenomena such as consonance and scale structure in terms of mathematical relationships have tended to dominate the thinking about these issues. This conceptual framework, however, fails to account for many of the perceptual observations that have been made over the last century. Another way to consider the problem is in terms of the biological rationale for evolving a sense of tonality in the first place. A pertinent fact in this regard is that only a small minority of naturally occurring sound stimuli are periodic. Since the auditory system evolved in the world of natural sounds, this point is presumably critical for thinking about the biological purposes of tonality and music. Indeed, the majority of periodic sounds that humans would have been exposed to during evolution are those made by the human vocal tract in the process of communication, initially prelinguistic but more recently speech sounds (see Chapter 26). Thus developing a sense of tonality would enable listeners to respond not only the distinc- is an especially intensive neural process. Furthermore, both the peripheral and central auditory system are “tuned” to conspecific communication vocalizations, pointing to the interdependent evolution of neural systems used for generating and perceiving these signals. The External Ear The external ear, which consists of the pinna, concha, and auditory meatus, gathers sound energy and focuses it on the eardrum, or tympanic membrane (Figure 12.3). One consequence of the configuration of the human auditory meatus is that it selectively boosts the sound pressure 30- to 100fold for frequencies around 3 kHz via passive resonance effects. This amplification makes humans especially sensitive to frequencies in the range of 2–5 kHz—and also explains why they are particularly prone to hearing loss near this frequency following exposure to loud broadband noises, such as those tions among the different speech sounds that are important for understanding spoken language, but to information about the probable sex, age, and emotional state of the speaker. It may thus be that music reflects the advantage of facilitating a listener’s ability to glean the linguistic intent and biological state of fellow humans through vocal utterances. References BURNS, E. M. (1999) Intervals, scales, and tuning. In The Psychology of Music, D. Deutsch (ed.). New York: Academic Press, pp. 215–264. CARTERETTE, E. C. AND R. A. KENDALL (1999) Comparative music perception and cognition. In The Psychology of Music, D. Deutsch (ed.). New York: Academic Press. LEWICKI, M. S. (2002) Efficient coding of natural sounds. Nature Neurosci. 5: 356–363. PIERCE, J. R. (1983, 1992) The Science of Musical Sound. New York: W.H. Freeman and Co., Chapters 4–6. PLOMP, R. AND W. J. LEVELT (1965) Tonal consonance and critical bandwidth. J. Acoust. Soc. Amer. 28: 548–560. RASCH, R. AND R. PLOMP (1999) The perception of musical tones. In The Psychology of Music, D. Deutsch (ed.). New York: Academic Press, pp. 89–112. SCHWARTZ, D. A., C. Q. HOWE AND D. PURVES (2003) The statistical structure of human speech sounds predicts musical universals. J. Neurosci. 23: 7160–7168. TERHARDT, E. (1974) Pitch, consonance, and harmony. J. Acoust. Soc. Amer. 55: 1061–1069. 288 Chapter Twelve Malleus Bone Malleus Concha Stapes Incus Incus Stapes Semicircular canals Oval window Vestibular nerve Cochlear nerve Tympanic membrane Base of stapes in oval window Cochlea Vestibule Inner ear Round window Eustachian tube Pinna Tympanic membrane Outer ear Middle ear External auditory meatus Figure 12.3 The human ear. Note the large surface area of the tympanic membrane (eardrum) relative to the oval window, a feature that facilitates transmission of airborne sounds to the fluid-filled cochlea. generated by heavy machinery or high explosives (see Box A). The sensitivity to this frequency range in the human auditory system appears to be directly related to speech perception: although human speech is a broadband signal, the energy of the plosive consonants (e.g., ba and pa) that distinguish different phonemes (the elementary human speech sounds) is concentrated around 3 kHz (see Box A in Chapter 26). Therefore, selective hearing loss in the 2–5 kHz range disproportionately degrades speech recognition. Most vocal communication occurs in the low-kHz range to overcome environmental noise; as already noted, generation of higher frequencies is difficult for animals the size of humans. A second important function of the pinna and concha is to selectively filter different sound frequencies in order to provide cues about the elevation of the sound source. The vertically asymmetrical convolutions of the pinna are shaped so that the external ear transmits more high-frequency components from an elevated source than from the same source at ear level. This effect can be demonstrated by recording sounds from different elevations after they have passed through an “artificial” external ear; when the recorded sounds are played back via earphones, so that the whole series is at the same elevation relative to the listener, the recordings from higher elevations are perceived as coming from positions higher in space than the recordings from lower elevations. The Auditor y System 289 The Middle Ear Sounds impinging on the external ear are airborne; however, the environment within the inner ear, where the sound-induced vibrations are converted to neural impulses, is aqueous. The major function of the middle ear is to match relatively low-impedance airborne sounds to the higher-impedance fluid of the inner ear. The term “impedance” in this context describes a medium’s resistance to movement. Normally, when sound waves travel from a low-impedance medium like air to a much higher-impedance medium like water, almost all (more than 99.9%) of the acoustical energy is reflected. The middle ear (see Figure 12.3) overcomes this problem and ensures transmission of the sound energy across the air–fluid boundary by boosting the pressure measured at the tympanic membrane almost 200-fold by the time it reaches the inner ear. Two mechanical processes occur within the middle ear to achieve this large pressure gain. The first and major boost is achieved by focusing the force impinging on the relatively large-diameter tympanic membrane on to the much smaller-diameter oval window, the site where the bones of the middle ear contact the inner ear. A second and related process relies on the mechanical advantage gained by the lever action of the three small interconnected middle ear bones, or ossicles (i.e., the malleus, incus, and stapes; see Figure 12.3), which connect the tympanic membrane to the oval window. Conductive hearing losses, which involve damage to the external or middle ear, lower the efficiency at which sound energy is transferred to the inner ear and can be partially overcome by artificially boosting sound pressure levels with an external hearing aid (Box C). In normal hearing, the efficiency of sound transmission to the inner ear also is regulated by two small muscles in the middle ear, the tensor tympani, innervated by cranial nerve V, and the stapedius, innervated by cranial nerve VII (see Appendix A). Flexion of these muscles, which is triggered automatically by loud noises or during self-generated vocalization, stiffens the ossicles and reduces the amount of sound energy transmitted to the cochlea, serving to protect the inner ear. Conversely, conditions that lead to flaccid paralysis of either of these muscles, such as Bell’s palsy (nerve VII), can trigger a painful sensitivity to moderate or even low intensity sounds known as hyperacusis. Bony and soft tissues, including those surrounding the inner ear, have impedances close to that of water. Therefore, even without an intact tympanic membrane or middle ear ossicles, acoustical vibrations can still be transferred directly through the bones and tissues of the head to the inner ear. In the clinic, bone conduction can be exploited using a simple test involving a tuning fork to determine whether hearing loss is due to conductive problems or is due to damage to the hair cells of the inner ear or to the auditory nerve itself (sensorineural hearing loss; see Boxes A and C) The Inner Ear The cochlea of the inner ear is arguably the most critical structure in the auditory pathway, for it is there that the energy from sonically generated pressure waves is transformed into neural impulses. The cochlea not only amplifies sound waves and converts them into neural signals, but it also acts as a mechanical frequency analyzer, decomposing complex acoustical waveforms into simpler elements. Many features of auditory perception derive from aspects of the physical properties of the cochlea; hence, it is important to consider this structure in some detail. 290 Chapter Twelve Box C Sensorineural Hearing Loss and Cochlear Implants The same features that make the auditory periphery exquisitely sensitive to detecting airborne sounds also make it highly vulnerable to damage. By far the most common forms of hearing loss involve the peripheral auditory system, namely to those structures that transmit and transduce sounds into neural impulses. Monaural hearing deficits are the defining symptom of a peripheral hearing loss, because unilateral damage at or above the auditory brainstem results in a binaural deficit (due to the extensive bilateral organization of the central auditory system). Peripheral hearing insults can be further divided into conductive hearing losses, which involve damage to the outer or middle ear, and sensorineural hearing losses, which stem from damage to the inner ear, most typically the cochlear hair cells or the VIIIth nerve itself. Although both forms of peripheral hearing loss manifest themselves as a raised threshold for hearing on the affected side, their diagnoses and treatments differ. Conductive hearing loss can be due to occlusion of the ear canal by wax or foreign objects, rupture of the tympanic membrane itself, or arthritic ossification of the middle ear bones. In contrast, sensorineural hearing loss usually is due to congenital or environmental insults that lead to hair cell death (see Box A) or damage to the eighth nerve. As hair cells are relatively few in number and do not regenerate in humans, their depletion leads to a diminished ability to detect sounds. The Weber test, a simple test involving a tuning fork, can be used to distinguish between these two forms of hearing loss. If a resonating tuning fork (∼256 Hz) is placed on the vertex, a patient with conductive hearing loss will report that the sound is louder in the affected ear. In the “plugged” state, sounds propagating through the skull do not dissipate so freely back out through the auditory meatus, and thus a greater amount of sound energy is transmitted to the cochlea on the blocked side. In contrast, a patient with a monaural sensorineural hearing loss will report that a Weber test sounds louder on the intact side, because even though the inner ear may vibrate equally on the two sides, the damaged side cannot transduce this vibration into a neural signal. Treatment also differs for these two types of deafness. An external hearing aid is used to boost sounds to compensate for the reduced efficiency of the conductive apparatus in conductive hearing losses. These miniature devices are inserted in the ear canal, and contain a microphone and speaker, as well as an amplifier. One limitation of hearing aids is that they often provide rather flat amplification curves, which can interfere with listening in noisy environments; moreover, they do not achieve a high degree of directionality. The use of digi- tal signal processing strategies partly overcomes these problems, and hearing aids obviously provide significant benefits to many people. The treatment of sensorineural hearing loss is more complicated and invasive; conventional hearing aids are useless, because no amount of mechanical amplification can compensate for the inability to generate or convey a neural impulse from the cochlea. However, if the VIIIth nerve is intact, cochlear implants can be used to partially restore hearing. The cochlear implant consists of a peripherally mounted microphone and digital signal processor that transforms a sound into its spectral components, and additional electronics that use this information to activate different combinations of contacts on a threadlike multi-site stimulating electrode array. The electrode is inserted into the cochlea through the round window (see figure) and positioned along the length of the tonotopically organized basilar membrane and VIIIth nerve endings. This placement enables electrical stimulation of the nerve in a manner that mimics some aspects of the spectral decomposition naturally performed by the cochlea. Cochlear implants can be remarkably effective in restoring hearing to people with hair cell damage, permitting them to engage in spoken communication. Despite such success in treating those who have lost their hearing after having The cochlea (from the Latin for “snail”) is a small (about 10 mm wide) coiled structure, which, were it uncoiled, would form a tube about 35 mm long (Figures 12.4 and 12.5). Both the oval window and, the round window, another region where the bone is absent surrounding the cochlea, are at the basal end of this tube. The cochlea is bisected from its basal almost to its apical end by the cochlear partition, which is a flexible structure that supports the basilar membrane and the tectorial membrane. There are fluid-filled chambers on each side of the cochlear partition, named the scala vestibuli and the scala tympani; a distinct channel, the scala media, runs within the The Auditor y System 291 Cochlea Microphone Implantable cochlear stimulator Auditory nerve Headpiece Round window Electrode array learned to speak, whether cochlear implants can enable development of spoken language in the congenitally deaf is still a matter of debate. Although cochlear implants cannot help patients with VIIIth nerve damage, brainstem implants are being developed that use a conceptually similar approach to stimulate the cochlear nuclei directly, bypassing the auditory periphery altogether. References Cable to speech processor Acoronal section at the level of the auditory meatus shows the components of the cochlear implant, including the filamentous stimulatory electrode inserted into the cochlea through the round window. cochlear partition. The cochlear partition does not extend all the way to the apical end of the cochlea; instead there is an opening, known as the helicotrema, that joins the scala vestibuli to the scala tympani, allowing their fluid, known as perilymph, to mix. One consequence of this structural arrangement is that inward movement of the oval window displaces the fluid of the inner ear, causing the round window to bulge out slightly and deforming the cochlear partition. The manner in which the basilar membrane vibrates in response to sound is the key to understanding cochlear function. Measurements of the vibra- RAMSDEN, R. T. (2002) Cochlear implants and brain stem implants. Brit. Med. Bull. 63: 183–193. RAUSCHECKER, J. P. AND R. V. SHANNON (2002) Sending sound to the brain. Science. 295: 1025–1029. 292 Chapter Twelve Cross section of cochlea Cochlea Auditory nerve Vestibular nerve Scala media Tectorial membrane Auditory nerve Scala vestibuli Spiral ganglion Oval window Round window Scala tympani Cochlea Tectorial membrane Organ of Corti Stereocilia of inner hair cells Inner hair cells Stereocilia Outer hair cells Basilar membrane Stereocilia of outer hair cells Figure 12.4 The cochlea, viewed face-on (upper left) and in cross section (subsequent panels). The stapes transfers force from the tympanic membrane to the oval window. The cross section of the cochlea shows the scala media between the scalae vestibuli and tympani. Blowup of the organ of Corti shows that the hair cells are located between the basilar and tectorial membranes; the latter is rendered transparent in the line drawing and removed in the scanning electron micrograph. The hair cells are named for their tufts of stereocilia; inner hair cells receive afferent inputs from cranial nerve VIII, whereas outer hair cells receive mostly efferent input. (Micrograph from Kessel and Kardon, 1979.) Afferent axons Basilar membrane Inner hair cells Tunnel of Corti Efferent axons Outer hair cells tion of different parts of the basilar membrane, as well as the discharge rates of individual auditory nerve fibers that terminate along its length, show that both these features are highly tuned; that is, they respond most intensely to a sound of a specific frequency. Frequency tuning within the inner ear is attributable in part to the geometry of the basilar membrane, which is wider and more flexible at the apical end and narrower and stiffer at the basal end. One feature of such a system is that regardless of where energy is supplied to it, movement always begins at the stiff end (i.e., the base), and then propagates to the more flexible end (i.e., the apex). Georg von Békésy, working at Harvard University, showed that a membrane that varies systematically in its width and flexibility vibrates maximally at different positions as a function of the stimulus frequency (Figure 12.5). Using tubular models and human cochleas taken from cadavers, he found that an acoustical stimulus initiates a traveling wave of the same frequency in the cochlea, which propagates from the base toward the apex of the basilar membrane, growing in The Auditor y System 293 1 Cochlea 2 3 Apex is “tuned” for low frequencies “Uncoiled” cochlea Cochlear base Scala vestibuli Stapes on oval window Round window 1 2 3 Relative amplitude Base of basilar membrane is “tuned” for high frequencies Helicotrema 4 5 6 4 5 6 Scala tympani 800 Hz 400 Hz 200 Hz 100 Hz 50 Hz 7 7 Traveling wave 1600 Hz Basilar membrane amplitude and slowing in velocity until a point of maximum displacement is reached. This point of maximal displacement is determined by the sound frequency. The points responding to high frequencies are at the base of the basilar membrane where it is stiffer, and the points responding to low frequencies are at the apex, giving rise to a topographical mapping of frequency (that is, to tonotopy). An important feature is that complex sounds cause a pattern of vibration equivalent to the superposition of the vibrations generated by the individual tones making up that complex sound, thus accounting for the decompositional aspects of cochlear function mentioned earlier. This process of spectral decomposition appears to be an important strategy for detecting the various harmonic combinations that distinguish different natural sounds. Indeed, tonotopy is conserved throughout much of the auditory system, including the auditory cortex, suggesting that it is important to speech processing. Von Békésy’s model of cochlear mechanics was a passive one, resting on the premise that the basilar membrane acts like a series of linked resonators, much as a concatenated set of tuning forks. Each point on the basilar membrane was postulated to have a characteristic frequency at which it vibrated most efficiently; because it was physically linked to adjacent areas of the membrane, each point also vibrated (if somewhat less readily) at other frequencies, thus permitting propagation of the traveling wave. It is now clear, however, that the tuning of the auditory periphery, whether measured at the basilar membrane or recorded as the electrical activity of auditory nerve fibers, is too sharp to be explained by passive mechanics alone. At very low sound intensities, the basilar membrane vibrates one hundred-fold more than would be predicted by linear extrapolation from the motion measured at high intensities. Therefore, the ear’s sensitivity arises from an active biomechanical process, as well as from its passive resonant properties (Box D). The outer hair cells, which together with the inner hair cells comprise the Cochlear apex 0 25 Hz 10 20 Distance from stapes (mm) 30 Figure 12.5 Traveling waves along the cochlea. A traveling wave is shown at a given instant along the cochlea, which has been uncoiled for clarity. The graphs on the right profile the amplitude of the traveling wave along the basilar membrane for different frequencies and show that the position (i.e., 1–7) where the traveling wave reaches its maximum amplitude varies directly with the frequency of stimulation. (Drawing after Dallos, 1992; graphs after von Békésy, 1960.) 294 Chapter Twelve Box D The Sweet Sound of Distortion As early as the first half of the eighteenth century, musical composers such as G. Tartini and W. A. Sorge discovered that upon playing pairs of tones, other tones not present in the original stimulus are also heard. These combination tones, fc, are mathematically related to the played tones f1 and f2 (f2 > f1) by the formula fc = mf1 ± nf2 where m and n are positive integers. Combination tones have been used for a variety of compositional effects, as they can strengthen the harmonic texture of a chord. Furthermore, organ builders sometimes use the difference tone (f2 – f1) created by two smaller organ pipes to produce the extremely low tones that would otherwise require building one especially large pipe. Modern experiments suggest that this distortion product is due at least in part to the nonlinear properties of the inner ear. M. Ruggero and his colleagues placed small glass beads (10–30 mm in diameter) on the basilar membrane of an anesthetized animal and then determined the velocity of the basilar mem- brane in response to different combinations of tones by measuring the Doppler shift of laser light reflected from the beads. When two tones were played into the ear, the basilar membrane vibrated not only at those two frequencies, but also at other frequencies predicted by the above formula. Related experiments on hair cells studied in vitro suggest that these nonlinearities result from the properties of the mechanical linkage of the transduction apparatus. By moving the hair bundle sinusoidally with a metal-coated glass fiber, A. J. Hudspeth and his coworkers found that the hair bundle exerts a force at the same frequency. However, when two sinusoids were applied simultaneously, the forces exerted by the hair bundle occurred not only at the primary frequencies, but at several combination frequencies as well. These distortion products are due to the transduction apparatus, since blocking the transduction channels causes the forces exerted at the combination frequencies to disappear, even though the forces at the primary frequencies remain unaffected. It seems that the tip links add a certain extra springiness to the hair bundle in the small range of motions over which the transduction channels are changing between closed and open states. If nonlinear distortions of basilar membrane vibrations arise from the properties of the hair bundle, then it is likely that hair cells can indeed influence basilar membrane motion, thereby accounting for the cochlea’s extreme sensitivity. When we hear difference tones, we may be paying the price in distortion for an exquisitely fast and sensitive transduction mechanism. References JARAMILLO, F., V. S. MARKIN AND A. J. HUDSPETH (1993) Auditory illusions and the single hair cell. Nature 364: 527–529. PLANCHART, A. E. (1960) A study of the theories of Giuseppe Tartini. J. Music Theory 4: 32–61. ROBLES, L., M. A. RUGGERO AND N. C. RICH (1991) Two-tone distortion in the basilar membrane of the cochlea. Nature 439: 413–414. sensory cells of the inner ear, are the most likely candidates for driving this active process. The motion of the traveling wave initiates sensory transduction by displacing the hair cells that sit atop the basilar membrane. Because these structures are anchored at different positions, the vertical component of the traveling wave is translated into a shearing motion between the basilar membrane and the overlying tectorial membrane (Figure 12.6). This motion bends the tiny processes, called stereocilia, that protrude from the apical ends of the hair cells, leading to voltage changes across the hair cell membrane. How the bending of stereocilia leads to receptor potentials in hair cells is considered in the following section. Hair Cells and the Mechanoelectrical Transduction of Sound Waves The hair cell is an evolutionary triumph that solves the problem of transforming vibrational energy into an electrical signal. The scale at which the The Auditor y System 295 (A) Resting position Pivot points for tectorial and basilar membranes are offset Tectorial membrane Inner hair cell Outer hair cells Basilar membrane (B) Sound-induced vibration Shear force Upward phase Shear force Downward phase Figure 12.6 Movement of the basilar membrane creates a shearing force that bends the stereocilia of the hair cells. The pivot point of the basilar membrane is offset from the pivot point of the tectorial membrane, so that when the basilar membrane is displaced, the tectorial membrane moves across the tops of the hair cells, bending the stereocilia. 296 Chapter Twelve (A) (B) (D) (C) Figure 12.7 The structure and function of the hair bundle in vestibular and cochlear hair cells. The vestibular hair bundles shown here resemble those of cochlear hair cells, except for the presence of the kinocilium, which disappears in the mammalian cochlea shortly after birth. (A) The hair bundle of a guinea pig vestibular hair cell. This view shows the increasing height leading to the kinocilium (arrow). (B) Cross section through the vestibular hair bundle shows the 9 + 2 array of microtubules in the kinocilium (top), which contrasts with the simpler actin filament structure of the stereocilia. (C) Scanning electron micrograph of a guinea pig cochlear outer hair cell bundle viewed along the plane of mirror symmetry. Note the graded lengths of the stereocilia, and the absence of a kinocilium. (D) Tip links that connect adjacent stereocilia are believed to be the mechanical linkage that opens and closes the transduction channel. (A from Lindeman, 1973; B from Hudspeth, 1983; C from Pickles, 1988; D from Fain, 2003.) hair cell operates is truly amazing: At the limits of human hearing, hair cells can faithfully detect movements of atomic dimensions and respond in the tens of microseconds! Furthermore, hair cells can adapt rapidly to constant stimuli, thus allowing the listener to extract signals from a noisy background. The hair cell is a flask-shaped epithelial cell named for the bundle of hairlike processes that protrude from its apical end into the scala media. Each hair bundle contains anywhere from 30 to a few hundred hexagonally arranged stereocilia, with one taller kinocilium (Figure 12.7A). Despite their names, only the kinocilium is a true ciliary structure, with the characteristic two central tubules surrounded by nine doublet tubules that define cilia (Figure 12.7B). The function of the kinocilium is unclear, and in the cochlea of humans and other mammals it actually disappears shortly after birth (Figure 12.7C). The stereocilia are simpler, containing only an actin cytoskeleton. Each stereocilium tapers where it inserts into the apical membrane, forming a hinge about which each stereocilium pivots (Figure 12.7D). The stereocilia are graded in height and are arranged in a bilaterally symmetric fashion (in vestibular hair cells, this plane runs through the kinocilium). Displacement of the hair bundle parallel to this plane toward the tallest stereocilia depolarizes the hair cell, while movements parallel to this plane toward the shortest stereocilia cause hyperpolarization. In contrast, displacements perpendicular to the plane of symmetry do not alter the hair cell’s membrane potential. The hair bundle movements at the threshold of hearing are approximately 0.3 nm, about the diameter of an atom of gold. Hair cells can convert the displacement of the stereociliary bundle into an electrical potential in as little as 10 microseconds; as described below, such speed is required for the accurate localization of the source of the sound. The need for microsecond resolution places certain constraints on the transduction mechanism, ruling out the rela- The Auditor y System 297 tively slow second messenger pathways used in visual and olfactory transduction (see Chapters 7, 10, and 14); a direct, mechanically gated transduction channel is needed to operate this quickly. Evidently the filamentous structures that connect the tips of adjacent stereocilia, known as tip links, directly open cation-selective transduction channels when stretched, allowing K+ ions to flow into the cell (see Figure 12.7D). As the linked stereocilia pivot from side to side, the tension on the tip link varies, modulating the ionic flow and resulting in a graded receptor potential that follows the movements of the stereocilia (Figures 12.8 and 12.9). The tip link model also explains why only deflections along the axis of the hair bundle activate transduction channels, since tip links join adjacent stereocilia along the axis directed toward the tallest stereocilia (see also Box B in Chapter 13). The exquisite mechanical sensitivity of the stereocilia also presents substantial risks: high intensity sounds can shear off the hair bundle, resulting in profound hearing deficits. Because human stereocilia, unlike those in fishes and birds, do not regenerate such damage is irreversible. The small number of hair cells (a total of about 30,000 in a human, or 15,000 per ear) further compounds the sensitivity of the inner (A) (B) Hyperpolarization Depolarization K+ K+ K+ K+ Depolarization Nucleus Nucleus Ca2+ Ca2+ channel Vesicles Vesicles Transmitter Afferent nerve To brain Ca2+ Transmitter Afferent nerve To brain Figure 12.8 Mechanoelectrical transduction mediated by hair cells. (A,B) When the hair bundle is deflected toward the tallest stereocilium, cationselective channels open near the tips of the stereocilia, allowing K+ ions to flow into the hair cell down their electrochemical gradient (see text on next page for the explanation of this peculiar situation). The resulting depolarization of the hair cell opens voltage-gated Ca2+ channels in the cell soma, allowing calcium entry and release of neurotransmitter onto the nerve endings of the auditory nerve. (After Lewis and Hudspeth, 1983) 298 Chapter Twelve (A) Displacement 300 500 0° 0° 700 90° 90° 0° 900 Time 1000 (B) 15 a.c. component d.c. 2000 component 10 Membrane potential Figure 12.9 Mechanoelectrical transduction mediated by vestibular hair cells. (A) Vestibular hair cell receptor potentials (bottom three traces; blue) measured in response to symmetrical displacement (top trace; yellow) of the hair bundle about the resting position, either parallel (0°) or orthogonal (90°) to the plane of bilateral symmetry. (B) The asymmetrical stimulus/response (xaxis/y-axis) function of the hair cell. Equivalent displacements of the hair bundle generate larger depolarizing responses than hyperpolarizing responses because most transduction channels are closed “at rest” (i.e., 0 µm). (C) Receptor potentials generated by an individual hair cell in the cochlea in response to pure tones (indicated in Hz, right). Note that the hair cell potential faithfully follows the waveform of the stimulating sinusoids for low frequencies (< 3kHz), but still responds with a DC offset to higher frequencies. (A after Shotwell et al., 1981; B after Hudspeth and Corey, 1977; C after Palmer and Russell, 1986.) 0° 25 mV Receptor potentials (mV) (C) 90° 3000 5 4000 0 5000 –5 –10 –2 –1 0 1 2 Hair bundle displacement (µm) 0 10 20 30 40 Time (ms) 50 60 70 ear to environmental and genetic insults. An important goal of current research is to identify the stem cells and factors that could contribute to the regeneration of human hair cells, thus affording a possible therapy for some forms of sensorineural hearing loss. Understanding the ionic basis of hair cell transduction has been greatly advanced by intracellular recordings made from these tiny structures. The hair cell has a resting potential between –45 and –60 mV relative to the fluid that bathes the basal end of the cell. At the resting potential, only a small fraction of the transduction channels are open. When the hair bundle is displaced in the direction of the tallest stereocilium, more transduction channels open, causing depolarization as K+ enters the cell. Depolarization in turn opens voltage-gated calcium channels in the hair cell membrane, and the resultant Ca2+ influx causes transmitter release from the basal end of the cell onto the auditory nerve endings (Figure 12.8A,B). Such calcium-dependent exocytosis is similar to chemical neurotransmission elsewhere in the central and peripheral nervous system (see Chapters 5 and 6); thus the hair cell has become a useful model for studying calcium-dependent transmitter release. Because some transduction channels are open at rest, the receptor potential is biphasic: Movement toward the tallest stereocilia depolarizes the cell, while move- The Auditor y System 299 ment in the opposite direction leads to hyperpolarization. This situation allows the hair cell to generate a sinusoidal receptor potential in response to a sinusoidal stimulus, thus preserving the temporal information present in the original signal up to frequencies of around 3 kHz (Figure 12.9). Hair cells still can signal at frequencies above 3 kHz, although without preserving the exact temporal structure of the stimulus: the asymmetric displacement-receptor current function of the hair cell bundle is filtered by the cell’s membrane time constant to produce a tonic depolarization of the soma, augmenting transmitter release and thus exciting VIIIth nerve terminals. The high-speed demands of mechanoelectrical transduction have resulted in some impressive ionic specializations within the inner ear. An unusual adaptation of the hair cell in this regard is that K+ serves both to depolarize and repolarize the cell, enabling the hair cell’s K+ gradient to be largely maintained by passive ion movement alone. As with other epithelial cells, the basal and apical surfaces of the hair cell are separated by tight junctions, allowing separate extracellular ionic environments at these two surfaces. The apical end (including the stereocilia) protrudes into the scala media and is exposed to the K+-rich, Na+-poor endolymph, which is produced by dedicated ion pumping cells in the stria vascularis (Figure 12.10). The basal end of the hair cell body is bathed in the same fluid that fills the scala tympani, the perilymph, which resembles other extracellular fluids in that it is K+poor and Na+-rich. In addition, the compartment containing endolymph is about 80 mV more positive than the perilymph compartment (this difference is known as the endocochlear potential), while the inside of the hair cell is about 45 mV more negative than the perilymph (and 125 mV more negative than the endolymph). The resulting electrical gradient across the membrane of the stereocilia (about 125 mV; the difference between the hair cell resting potential and the endocochlear potential) drives K+ through open transduc- Tunnel of Corti Endolymph High K+ 80 mV Spiral ganglion Tectorial membrane Outer hair cells Scala vestibuli Scala media Stria vascularis Scala tympani Perilymph Low K+ 0 mV Inner hair cells –45 mV Basilar membrane Figure 12.10 The stereocilia of the hair cells protrude into the endolymph, which is high in K+ and has an electrical potential of +80 mV relative to the perilymph. 300 Chapter Twelve tion channels into the hair cell, even though these cells already have a high internal K+ concentration. K+ entry via the transduction channels electrotonically depolarizes the hair cell, opening voltage-gated Ca2+ and K+ channels located in the membrane of the hair cell soma (see Box B in Chapter 13). The opening of somatic K+ channels favors K+ efflux, and thus repolarization; the efflux occurs because the perilymph surrounding the basal end is low in K+ relative to the cytosol, and because the equilibrium potential for K+ is more negative than the hair cell’s resting potential (EKBasal ≈ –85 mV). Repolarization of the hair cell via K+ efflux is also facilitated by Ca2+ entry. In addition to modulating the release of neurotransmitter, Ca2+ entry opens Ca2+-dependent K+ channels, which provide another avenue for K+ to enter the perilymph. Indeed, the interaction of Ca2+ influx and Ca2+-dependent K+ efflux can lead to electrical resonances that enhance the tuning of response properties within the inner ear (also explained in Box B in Chapter 13). In essence, the hair cell operates as two distinct compartments, each dominated by its own Nernst equilibrium potential for K+; this arrangement ensures that the hair cell’s ionic gradient does not run down, even during prolonged stimulation. At the same time, rupture of Reissner’s membrane, which normally separates the scalae media and vestibuli, or compounds such as ethacrynic acid (see Box A), which selectively poison the ion-pumping cells of the stria vascularis, can cause the endocochlear potential to dissipate, resulting in a sensorineural hearing deficit. In short, the hair cell exploits the different ionic milieus of its apical and basal surfaces to provide extremely fast and energy-efficient repolarization. Two Kinds of Hair Cells in the Cochlea The cochlear hair cells in humans consist of one row of inner hair cells and three rows of outer hair cells (see Figure 12.4). The inner hair cells are the actual sensory receptors, and 95% of the fibers of the auditory nerve that project to the brain arise from this subpopulation. The terminations on the outer hair cells are almost all from efferent axons that arise from cells in the superior olivary complex. A clue to the significance of this efferent pathway was provided by the discovery that an active process within the cochlea, as mentioned, influences basilar membrane motion. First, it was found that the cochlea actually emits sound under certain conditions. These otoacoustical emissions can be detected by placing a sensitive microphone at the eardrum and monitoring the response after briefly presenting a tone or click, and provide a useful means to assess cochlear function in the newborn (this test is now done routinely to rule out congenital deafness). Such emissions can also occur spontaneously, especially in certain pathological states, and are thus one potential source of tinnitus (ringing in the ears). These observations clearly indicate that a process within the cochlea is capable of producing sound. Second, stimulation of the crossed olivocochlear bundle, which supplies efferent input to the outer hair cells, can broaden VIIIth nerve tuning curves. Third, the high sensitivity notch of VIIIth nerve tuning curves is lost when the outer hair cells are selectively inactivated. Finally, isolated outer hair cells contract and expand in response to small electrical currents, thus providing a potential source of energy to drive an active process within the cochlea. Thus, it seems likely that the outer hair cells sharpen the frequency-resolving power of the cochlea by actively contracting and relaxing, thus changing the stiffness of the tectorial membrane at particular locations. This active The Auditor y System 301 process explains the nonlinear vibration of the basilar membrane at low sound intensities (see Box D). Tuning and Timing in the Auditory Nerve The rapid response time of the transduction apparatus allows the membrane potential of the hair cell to follow deflections of the hair bundle up to moderately high frequencies of oscillation. In humans, the receptor potentials of certain hair cells and the action potentials of their associated auditory nerve fiber can follow stimuli of up to about 3 kHz in a one-to-one fashion. Such real-time encoding of stimulus frequency by the pattern of action potentials in the auditory nerve is known as the “volley theory” of auditory information transfer. Even these extraordinarily rapid processes, however, fail to follow frequencies above 3 kHz (see Figure 12.9). Accordingly, some other mechanism must be used to transmit auditory information at higher frequencies. The tonotopically organized basilar membrane provides an alternative to temporal coding, namely a “labeled-line” coding mechanism. In this case, frequency information is specified by preserving the tonotopy of the cochlea at higher levels in the auditory pathway. Because the auditory nerve fibers associate with the inner hair cells in approximately a one-to-one ratio (although several or more VIIIth nerve fibers synapse on a single hair cell), each auditory nerve fiber transmits information about only a small part of the audible frequency spectrum. As a result, auditory nerve fibers related to the apical end of the cochlea respond to low frequencies, and fibers that are related to the basal end respond to high frequencies (see Figure 12.5). The limitations of specific fibers can be seen in electrophysiological recordings of responses to sound (Figure 12.11). These threshold functions are called tuning curves, and the lowest threshold of the tuning curve is called the characteristic frequency. Since the topographical order of the characteristic frequency of neurons is retained throughout the system, information about frequency is also preserved. Cochlear implants (see Box C) exploit the tonotopic organization of the cochlea, and particularly its eighth nerve afferents, to roughly recreate the patterns of VIIIth nerve activity elicited by sounds. In patients with damaged hair cells, such implants can effectively bypass the impaired transduction apparatus, and thus restore some degree of auditory function. The other prominent feature of hair cells—their ability to follow the waveform of low-frequency sounds—is also important in other more subtle aspects of auditory coding. As mentioned earlier, hair cells have biphasic response properties. Because hair cells release transmitter only when depolarized, auditory nerve fibers fire only during the positive phases of low-frequency sounds (see Figure 12.11). The resultant “phase locking” that results provides temporal information from the two ears to neural centers that compare interaural time differences. The evaluation of interaural time differences provides a critical cue for sound localization and the perception of auditory “space.” That auditory space can be perceived is remarkable, given that the cochlea, unlike the retina, cannot represent space directly. A final point is that VIIIth nerve activity patterns are not simply faithful neural replicas of the auditory stimulus itself. Indeed, William Bialek and his colleagues at Princeton University have shown that the VIIIth nerve in the bullfrog encodes conspecific mating calls more efficiently than artificial sounds with similar frequency and amplitude characteristics. Thus both animal and human studies support the idea that the auditory periphery is optimized to 302 Chapter Twelve (A) (B) Frequency (kHz) −20 1.0 10.0 −40 −80 −20 −60 −80 −80 −20 Apex: Low frequencies Base: High frequencies −80 Cochlea −20 Basilar membrane Cranial nerve VIII −80 −20 (C) Spikes/second Threshold intensity (relative dB) required to stimulate unit above spontaneous firing rate 0.1 −80 −20 −80 0.1 1.0 Frequency (kHz) 10.0 Figure 12.11 Response properties of auditory nerve fibers. (A) Frequency tuning curves of six different fibers in the auditory nerve. Each graph plots, across all frequencies to which the fiber responds, the minimum sound level required to increase the fiber’s firing rate above its spontaneous firing level. The lowest point in the plot is the weakest sound intensity to which the neuron will respond. The frequency at this point is called the neuron’s characteristic frequency. (B) The frequency tuning curves of auditory nerve fibers superimposed and aligned with their approximate relative points of innervation along the basilar membrane. (In the side view schematic, the basilar membrane is represented as a black line within the unrolled cochlea.) (C) Temporal response patterns of a low-frequency axon in the auditory nerve. The stimulus waveform is indicated beneath the histograms, which show the phase-locked responses to a 50-ms tone pulse of 260 Hz. Note that the spikes are all timed to the same phase of the sinusoidal stimulus. (A after Kiang and Moxon, 1972; C after Kiang, 1984.) The Auditor y System 303 transmit species-typical vocal sounds, rather than simply transmitting all sounds equally well to central auditory areas. How Information from the Cochlea Reaches Targets in the Brainstem A hallmark of the ascending auditory system is its parallel organization. This arrangement becomes evident as soon as the auditory nerve enters the brainstem, where it branches to innervate the three divisions of the cochlear nucleus. The auditory nerve (the major component of cranial nerve VIII) comprises the central processes of the bipolar spiral ganglion cells in the cochlea (see Figure 12.4); each of these cells sends a peripheral process to contact one inner hair cell and a central process to innervate the cochlear nucleus. Within the cochlear nucleus, each auditory nerve fiber branches, sending an ascending branch to the anteroventral cochlear nucleus and a descending branch to the posteroventral cochlear nucleus and the dorsal cochlear nucleus (Figure 12.12). The tonotopic organization of the cochlea is maintained in the three parts of the cochlear nucleus, each of which contains different populations of cells with quite different properties. In addition, the patterns of termination of the auditory nerve axons differ in density and type; thus, there are several opportunities at this level for transformation of the information from the hair cells. Integrating Information from the Two Ears Just as the auditory nerve branches to innervate several different targets in the cochlear nuclei, the neurons in these nuclei give rise to several different pathways (see Figure 12.12). One clinically relevant feature of the ascending projections of the auditory brainstem is a high degree of bilateral connectivity, which means that damage to central auditory structures is almost never manifested as a monaural hearing loss. Indeed, monaural hearing loss strongly implicates unilateral peripheral damage, either to the middle or inner ear, or to the VIIIth nerve itself (see Box C). Given the relatively byzantine organization already present at the level of the auditory brainstem, it is useful to consider these pathways in the context of their functions. The best-understood function mediated by the auditory brainstem nuclei, and certainly the one most intensively studied, is sound localization. Humans use at least two different strategies to localize the horizontal position of sound sources, depending on the frequencies in the stimulus. For frequencies below 3 kHz (which can be followed in a phase-locked manner), interaural time differences are used to localize the source; above these frequencies, interaural intensity differences are used as cues. Parallel pathways originating from the cochlear nucleus serve each of these strategies for sound localization. The human ability to detect interaural time differences is remarkable. The longest interaural time differences, which are produced by sounds arising directly lateral to one ear, are on the order of only 700 microseconds (a value given by the width of the head divided by the speed of sound in air, about 340 m/s). Psychophysical experiments show that humans can actually detect interaural time differences as small as 10 microseconds; two sounds presented through earphones separated by such small interaural time differences are perceived as arising from the side of the leading ear. This sensitivity translates into accuracy for sound localization of about 1 degree. 304 Chapter Twelve Figure 12.12 Diagram of the major auditory pathways. Although many details are missing from this diagram, two important points are evident: (1) the auditory system entails several parallel pathways, and (2) information from each ear reaches both sides of the system, even at the level of the brainstem. Cerebrum Primary auditory cortex Medial geniculate complex of the thalamus Rostral midbrain Inferior colliculus Caudal midbrain Nucleus of lateral leminiscus Ponsmidbrain junction Mid-pons Superior olive Cochlear nuclei Dorsal Rostral medulla Posteroventral Auditory nerve Cochlea Spiral ganglion Anteroventral The Auditor y System 305 How is timing in the 10 microseconds range accomplished by neural components that operate in the millisecond range? The neural circuitry that computes such tiny interaural time differences consists of binaural inputs to the medial superior olive (MSO) that arise from the right and left anteroventral cochlear nuclei (Figure 12.13; see also Figure 12.12). The medial superior olive contains cells with bipolar dendrites that extend both medially and laterally. The lateral dendrites receive input from the ipsilateral anteroventral cochlear nucleus, and the medial dendrites receive input from the contralateral anteroventral cochlear nucleus (both inputs are excitatory). As might be expected, the MSO cells work as coincidence detectors, responding when both excitatory signals arrive at the same time. For a coincidence mechanism to be useful in localizing sound, different neurons must be maximally sensitive to different interaural time delays. The axons that project from the anteroventral cochlear nucleus evidently vary systematically in length to create delay lines. (Remember that the length of an axon divided by its conduction velocity equals the conduction time.) These anatomical differences compensate for sounds arriving at slightly different times at the two ears, so that the resultant neural impulses arrive at a particular MSO neuron simultaneously, making each cell especially sensitive to sound sources in a particular place. The mechanisms enabling MSO neurons to function as coincidence detectors at the microsecond level are still poorly understood, but certainly reflect one of the more impressive biophysical specializations in the nervous system. Sound localization perceived on the basis of interaural time differences requires phase-locked information from the periphery, which, as already Loudspeaker 1 Sound reaches left ear first Figure 12.13 Diagram illustrating how the MSO computes the location of a sound by interaural time differences. A given MSO neuron responds most strongly when the two inputs arrive simultaneously, as occurs when the contralateral and ipsilateral inputs precisely compensate (via their different lengths) for differences in the time of arrival of a sound at the two ears. The systematic (and inverse) variation in the delay lengths of the two inputs creates a map of sound location: In this model, E would be most sensitive to sounds located to the left, and A to sounds from the right; C would respond best to sounds coming from directly in front of the listener. (After Jeffress, 1948.) 3 Sound reaches right ear a little later Left ear Longer path to neuron E Cochlea and cochlear nucleus 2 Action potential begins traveling toward MSO 1 Right ear leading neuron MSO 2 A 3 5 D C B 4 3 4 Action potential from right ear begins traveling toward MSO 5 4 Left ear leading neuron E 2 1 Right ear Shorter path to neuron E 5 Action potentials converge on an MSO neuron that responds most strongly if their arrival is coincident Cochlea and cochlear nucleus 306 Chapter Twelve emphasized, is available to humans only for frequencies below 3 kHz. (In barn owls, the reigning champions of sound localization, phase locking occurs at up to 9 kHz.) Therefore, a second mechanism must come into play at higher frequencies. At frequencies higher than about 2 kHz, the human head begins to act as an acoustical obstacle because the wavelengths of the sounds are too short to bend around it. As a result, when high-frequency sounds are directed toward one side of the head, an acoustical “shadow” of lower intensity is created at the far ear. These intensity differences provide a second cue about the location of a sound. The circuits that compute the position of a sound source on this basis are found in the lateral superior olive (LSO) and the medial nucleus of the trapezoid body (MNTB) (Figure 12.14). Excitatory axons project directly from the ipsilateral anteroventral cochlear nucleus to the LSO (as well as to the MSO; see Figure 12.13). Note that the LSO also receives inhibitory input from the contralateral ear, via an inhibitory neuron in the MNTB. This excitatory/inhibitory interaction Speaker (B) 1 Stronger stimulus to left ear excites left LSO 2 This stimulus also inhibits right LSO via MNTB interneuron Net excitation to higher centers LSO Net inhibition MNTB Output of LSO (A) Left LSO output 70 40 20 Left > right Right LSO output 0 −20 −40 −70 Right > left Relative loudness Section from pons 3 Excitation from left side is greater than inhibition from right side, resulting in net excitation to higher centers 4 Inhibition from left side is greater than excitation from right side, resulting in net inhibition on right and no signal to higher centers Figure 12.14 Lateral superior olive neurons encode sound location through interaural intensity differences. (A) LSO neurons receive direct excitation from the ipsilateral cochlear nucleus; input from the contralateral cochlear nucleus is relayed via inhibitory interneurons in the MNTB. (B) This arrangement of excitation–inhibition makes LSO neurons fire most strongly in response to sounds arising directly lateral to the listener on the same side as the LSO, because excitation from the ipsilateral input will be great and inhibition from the contralateral input will be small. In contrast, sounds arising from in front of the listener, or from the opposite side, will silence the LSO output, because excitation from the ipsilateral input will be minimal, but inhibition driven by the contralateral input will be great. Note that LSOs are paired and bilaterally symmetrical; each LSO only encodes the location of sounds arising on the same side of the body as its location. The Auditor y System 307 results in a net excitation of the LSO on the same side of the body as the sound source. For sounds arising directly lateral to the listener, firing rates will be highest in the LSO on that side; in this circumstance, the excitation via the ipsilateral anteroventral cochlear nucleus will be maximal, and inhibition from the contralateral MNTB minimal. In contrast, sounds arising closer to the listener’s midline will elicit lower firing rates in the ipsilateral LSO because of increased inhibition arising from the contralateral MNTB. For sounds arising at the midline, or from the other side, the increased inhibition arising from the MNTB is powerful enough to completely silence LSO activity. Note that each LSO only encodes sounds arising in the ipsilateral hemifield; it therefore takes both LSOs to represent the full range of horizontal positions. In summary, there are two separate pathways—and two separate mechanisms—for localizing sound. Interaural time differences are processed in the medial superior olive, and interaural intensity differences are processed in the lateral superior olive. These two pathways are eventually merged in the midbrain auditory centers. Monaural Pathways from the Cochlear Nucleus to the Lateral Lemniscus The binaural pathways for sound localization are only part of the output of the cochlear nucleus. This fact is hardly surprising, given that auditory perception involves much more than locating the position of the sound source. A second major set of pathways from the cochlear nucleus bypasses the superior olive and terminates in the nuclei of the lateral lemniscus on the contralateral side of the brainstem (see Figure 12.12). These particular pathways respond to sound arriving at one ear only and are thus referred to as monaural. Some cells in the lateral lemniscus nuclei signal the onset of sound, regardless of its intensity or frequency. Other cells in the lateral lemniscus nuclei process other temporal aspects of sound, such as duration. The precise role of these pathways in processing temporal features of sound is not yet known. As with the outputs of the superior olivary nuclei, the pathways from the nuclei of the lateral lemniscus converge at the midbrain. Integration in the Inferior Colliculus Auditory pathways ascending via the olivary and lemniscal complexes, as well as other projections that arise directly from the cochlear nucleus, project to the midbrain auditory center, the inferior colliculus. In examining how integration occurs in the inferior colliculus, it is again instructive to turn to the most completely analyzed auditory mechanism, the binaural system for localizing sound. As already noted, space is not mapped on the auditory receptor surface; thus the perception of auditory space must somehow be synthesized by circuitry in the lower brainstem and midbrain. Experiments in the barn owl, an extraordinarily proficient animal at localizing sounds, show that the convergence of binaural inputs in the midbrain produces something entirely new relative to the periphery—namely, a computed topographical representation of auditory space. Neurons within this auditory space map in the colliculus respond best to sounds originating in a specific region of space and thus have both a preferred elevation and a preferred horizontal location, or azimuth. Although comparable maps of auditory space have not yet been found in mammals, humans have a clear perception of 308 Chapter Twelve both the elevational and azimuthal components of a sound’s location, suggesting that we have a similar auditory space map. Another important property of the inferior colliculus is its ability to process sounds with complex temporal patterns. Many neurons in the inferior colliculus respond only to frequency-modulated sounds, while others respond only to sounds of specific durations. Such sounds are typical components of biologically relevant sounds, such as those made by predators, or intraspecific communication sounds, which in humans include speech. The Auditory Thalamus Despite the parallel pathways in the auditory stations of the brainstem and midbrain, the medial geniculate complex (MGC) in the thalamus is an obligatory relay for all ascending auditory information destined for the cortex (see Figure 12.12). Most input to the MGC arises from the inferior colliculus, although a few auditory axons from the lower brainstem bypass the inferior colliculus to reach the auditory thalamus directly. The MGC has several divisions, including the ventral division, which functions as the major thalamocortical relay, and the dorsal and medial divisions, which are organized like a belt around the ventral division. In some mammals, the strictly maintained tonotopy of the lower brainstem areas is exploited by convergence onto MGC neurons, generating specific responses to certain spectral combinations. The original evidence for this statement came from research on the response properties of cells in the MGC of echolocating bats. Some cells in the so-called belt areas of the bat MGC respond only to combinations of widely spaced frequencies that are specific components of the bat’s echolocation signal and of the echoes that are reflected from objects in the bat’s environment. In the mustached bat, where this phenomenon has been most thoroughly studied, the echolocation pulse has a changing frequency (frequency-modulated, or FM) component that includes a fundamental frequency and one or more harmonics. The fundamental frequency (FM1) has low intensity and sweeps from 30 kHz to 20 kHz. The second harmonic (FM2) is the most intense component and sweeps from 60 kHz to 40 kHz. Note that these frequencies do not overlap. Most of the echoes are from the intense FM2 sound, and virtually none arise from the weak FM1, even though the emitted FM1 is loud enough for the bat to hear. Apparently, the bat assesses the distance to an object by measuring the delay between the FM1 emission and the FM2 echo. Certain MGC neurons respond when FM2 follows FM1 by a specific delay, providing a mechanism for sensing such frequency combinations. Because each neuron responds best to a particular delay, the population of MGC neurons encodes a range of distances. Bat sonar illustrates two important points about the function of the auditory thalamus. First, the MGC is the first station in the auditory pathway where selectivity for combinations of frequencies is found. The mechanism responsible for this selectivity is presumably the ultimate convergence of inputs from cochlear areas with different spectral sensitivities. Second, cells in the MGC are selective not only for frequency combinations, but also for specific time intervals between the two frequencies. The principle is the same as that described for binaural neurons in the medial superior olive, but in this instance, two monaural signals with different frequency sensitivity coincide, and the time difference is in the millisecond rather than the microsecond range. In summary, neurons in the medial geniculate complex receive convergent inputs from spectrally and temporally separate pathways. This complex, by The Auditor y System 309 virtue of its convergent inputs, mediates the detection of specific spectral and temporal combinations of sounds. In many species, including humans, varying spectral and temporal cues are especially important features of communication sounds. It is not known whether cells in the human medial geniculate are selective to combinations of sounds, but the processing of speech certainly requires both spectral and temporal combination sensitivity. The Auditory Cortex The ultimate target of afferent auditory information is the auditory cortex. Although the auditory cortex has a number of subdivisions, a broad distinction can be made between a primary area and peripheral, or belt, areas. The primary auditory cortex (A1) is located on the superior temporal gyrus in the temporal lobe and receives point-to-point input from the ventral division of the medial geniculate complex; thus, it contains a precise tonotopic map. The belt areas of the auditory cortex receive more diffuse input from the belt areas of the medial geniculate complex and therefore are less precise in their tonotopic organization. The primary auditory cortex (A1) has a topographical map of the cochlea (Figure 12.15), just as the primary visual cortex (V1) and the primary somatic sensory cortex (S1) have topographical maps of their respective sensory (A) 8 0 00 H z 1 6 ,0 0 0 Hz 40 00 Hz 2000 Hz 1000 Hz 500 Hz Primary auditory cortex Corresponds to base of cochlea Corresponds to apex of cochlea Secondary auditory cortex (belt areas) (B) Frontal and parietal lobes removed Lateral sulcus Secondary auditory cortex Right hemisphere Wernicke’s Primary area auditory cortex Wernicke’s area Left hemisphere Figure 12.15 The human auditory cortex. (A) Diagram showing the brain in left lateral view, including the depths of the lateral sulcus, where part of the auditory cortex occupying the superior temporal gyrus normally lies hidden. The primary auditory cortex (A1) is shown in blue; the surrounding belt areas of the auditory cortex are in red. The primary auditory cortex has a tonotopic organization, as shown in this blowup diagram of a segment of A1 (right). (B) Diagram of the brain in left lateral view, showing locations of human auditory cortical areas related to processing speech sounds in the intact hemisphere. Right: An oblique section (plane of dashed line) shows the cortical areas on the superior surface of the temporal lobe. Note that Wernicke’s area, a region important in comprehending speech, is just posterior to the primary auditory cortex. 310 Chapter Twelve Box E 0.5 0 –0.5 100 Frequency (kHz) Most natural sounds are complex, meaning that they differ from the pure tones or clicks that are frequently used in neurophysiological studies of the auditory system. Rather, natural sounds are tonal: they have a fundamental frequency that largely determines the “pitch” of the sound, and one or more harmonics of different intensities that contribute to the quality or “timbre” of a sound. The frequency of a harmonic is, by definition, a multiple of the fundamental frequency, and both may be modulated over time. Such frequency-modulated (FM) sweeps can rise or fall in frequency, or change in a sinusoidal or some other fashion. Occasionally, multiple nonharmonic frequencies may be simultaneously present in some communication or musical sounds. In some sounds, a level of spectral splatter or “broadband noise” is embedded within tonal or frequency modulated sounds. The variations in the sound spectrum are typically accompanied by a modulation of the amplitude envelop of the complex sound as well. All of these features can be visualized by performing a spectrographic analysis. How does the brain represent such complex natural sounds? Cognitive studies of complex sound perception provide some understanding of how a large but limited number of neurons in the brain can dynamically represent an infinite variety of natural stimuli in the sensory Amplitude (V) Representing Complex Sounds in the Brains of Bats and Humans Harmonics of fb0 bUFM 60 Noisy fSFM fb0 20 fb0 0 20 40 Time (ms) 60 (A) Amplitude envelope (above) and spectrogram (below) of a composite syllable emitted by mustached bats for social communication. This composite consists of two simple syllables, a fixed Sinusoidal FM (fSFM) and a bent Upward FM (bUFM) that emerges from the fSFM after some overlap. Each syllable has its own fundamental (fa0 and fb0) and multiple harmonics. (Courtesy of Jagmeet Kanwal.) environment of humans and other animals. In bats, specializations for processing complex sounds are apparent. Studies in echolocating bats show that both communication and echolocation sounds (Figure A) are processed not only within some of the same areas, but also within the same neurons in the auditory cortex. In humans, multiple modes of processing are also likely, given the large overlap within the superior and middle temporal gyri in the temporal lobe for the repre- sentation of different types of complex sounds. Asymmetrical representation is another common principle of complex sound processing that results in lateralized (though largely overlapping) representations of natural stimuli. Thus, speech sounds that are important for communication are lateralized to the left in the belt regions of the auditory cortex, whereas environmental sounds that are important for reacting to and recogniz- epithelia. Unlike the visual and somatic sensory systems, however, the cochlea has already decomposed the acoustical stimulus so that it is arrayed tonotopically along the length of the basilar membrane. Thus, A1 is said to comprise a tonotopic map, as do most of the ascending auditory structures between the cochlea and the cortex. Orthogonal to the frequency axis of the tonotopic map is a striped arrangement of binaural properties. The neurons in one stripe are excited by both ears (and are therefore called EE cells), while the neurons in the next stripe are excited by one ear and inhibited by the other ear (EI cells). The EE and EI stripes alternate, an arrangement that is reminiscent of the ocular dominance columns in V1 (see Chapter 11). The Auditor y System 311 Speech Left Environmental Left Music Left (B) Top: Reconstructed functional magnetic resonance images of BOLD contrast signal change (average for 8 subjects) showing significant (p < 0.001) activation elicited by speech, environmental, and musical sounds on surface views of the left versus the right side of the human brain. Bottom: Bar graphs showing the total significant activation to each category of complex sounds in the core and belt areas of the auditory cortex for the left versus the right side. (Courtesy of Jagmeet Kanwal.) References Right Right Sum of voxel values Speech Right Environmental Music 20 12 12 10 6 6 0 Left Right Left Right Core Belt 0 Left Right Left Right Core Belt ing aspects of the auditory environment are represented in each hemisphere (Figure B). Musical sounds that can either motivate us to march in war or to relax and meditate when coping with physical and emotional stress are highly lateralized to the right in the belt regions of the auditory cortex. The extent of lateralization for speech and possibly music may 0 Left Right Left Right Core Belt vary with sex, age, and training. In some species of bats, mice, and primates, processing of natural communication sounds appears to be lateralized to the left hemisphere. In summary, natural sounds are complex and their representation within the sensory cortex tends to be asymmetric across the two hemispheres. The auditory cortex obviously does much more than provide a tonotopic map and respond differentially to ipsi- and contralateral stimulation. Although the sorts of sensory processing that occur in the auditory cortex are not well understood, they are likely to be important to higher-order processing of natural sounds, especially those used for communication (Box E; see also Chapter 26). One clue about such processing comes from work in marmosets, a small neotropical primate with a complex vocal repertoire. The primary auditory cortex and belt areas of these animals are indeed organized tonotopically, but also contain neurons that are strongly responsive to spectral combinations that characterize certain vocalizations. The responses EHRET, G. (1987) Left hemisphere advantage in the mouse brain for recognizing ultrasonic communication calls. Nature 325: 249–251. ESSER, K.-H., C. J. CONDON, N. SUGA AND J. S. KANWAL (1997) Syntax processing by auditory cortical neurons in the FM-FM area of the mustached bat, Pteronotus parnellii. Proc. Natl. Acad. Sci. USA 94: 14019–14024. HAUSER, M. D. AND K. ANDERSSON (1994) Left hemisphere dominance for processing vocalizations in adult, but not infant, rhesus monkeys: Field experiments. Proc. Natl. Acad. Sci. USA 91: 3946-3948. KANWAL, J. S., J. KIM AND K. KAMADA (2000) Separate, distributed processing of environmental, speech and musical sounds in the cerebral hemispheres. J. Cog. Neurosci. (Supp.): p. 32. KANWAL, J. S., J. S. MATSUMURA, K. OHLEMILLER AND N. SUGA (1994) Acoustic elements and syntax in communication sounds emitted by mustached bats. J. Acous. Soc. Am. 96: 1229–1254. KANWAL, J. S. AND N. SUGA (1995) Hemispheric asymmetry in the processing of calls in the auditory cortex of the mustached bat. Assoc. Res. Otolaryngol. 18: 104. 312 Chapter Twelve of these neurons to the tonal stimuli do not accurately predict their responses to the spectral combinations, suggesting that, in accord with peripheral optimization, cortical processing is in part dedicated to detecting particular intraspecific vocalizations. Another clue about the role of the primary auditory cortex in the processing of intraspecific communication sounds comes from work in echolocating bats. Consistent with the essential role that echolocation plays in the survival of these crepuscular animals, certain regions of the bat primary auditory cortex, like those described in the MGC, are tuned in a systematic manner to the delays between frequency modulated pulses and their echoes, thus providing information about target distance and velocity. These delay-tuned neurons can exhibit highly specific responses to intraspecific communication calls, suggesting that the same cortical neurons can serve these two distinct auditory functions (see Box E). Evidently the general ability of the mammalian auditory cortex to detect certain spectral and temporal combinations of natural sounds has been exploited in bats to serve sonar-mediated navigation, yielding these dual function neurons. Many of the dually specialized neurons are categorized as “combinationsensitive” neurons, i.e., neurons that show a nonlinear increase in their response magnitude when presented with a combination of tones and/or noise bands in comparison to the total magnitude of the response elicited by presenting each sound element separately. Combination-sensitive neurons are tuned to more than one frequency and are specialized to recognize complex species-specific sounds and extract information that is critical for survival. This sensitivity to combinations of simple sound elements appears to be a universal property of neurons for the perception of complex sounds by many animal species, such as frogs, birds bats and nonhuman primates. Therefore, combination-sensitive neurons most likely partake in the recognition of complex sounds in the human auditory cortex as well. Sounds that are especially important for intraspecific communication often have a highly ordered temporal structure. In humans, the best example of such time-varying signals is speech, where different phonetic sequences are perceived as distinct syllables and words (see Box A in Chapter 26). Behavioral studies in cats and monkeys show that the auditory cortex is especially important for processing temporal sequences of sound. If the auditory cortex is ablated in these animals, they lose the ability to discriminate between two complex sounds that have the same frequency components but which differ in temporal sequence. Thus, without the auditory cortex, monkeys cannot discriminate one conspecific communication sound from another. The physiological basis of such temporal sensitivity likely requires neurons that are sensitive to time-varying cues in communication sounds. Indeed, electrophysiological recordings from the primary auditory cortex of both marmosets and bats show that some neurons that respond to intraspecific communication sounds do not respond as strongly when the sounds are played in reverse, indicating sensitivity to the sounds’ temporal features. Studies of human patients with bilateral damage to the auditory cortex also reveal severe problems in processing the temporal order of sounds. It seems likely, therefore, that specific regions of the human auditory cortex are specialized for processing elementary speech sounds, as well as other temporally complex acoustical signals, such as music (Box B). Thus, Wernicke’s area, which is critical to the comprehension of human language, lies within the secondary auditory area (Figure 12.15; see also Chapter 26). The Auditor y System 313 Summary Sound waves are transmitted via the external and middle ear to the cochlea of the inner ear, which exhibits a traveling wave when stimulated. For highfrequency sounds, the amplitude of the traveling wave reaches a maximum at the base of the cochlea; for low-frequency sounds, the traveling wave reaches a maximum at the apical end. The associated motions of the basilar membrane are transduced primarily by the inner hair cells, while the basilar membrane motion is itself actively modulated by the outer hair cells. Damage to the outer or middle ear results in conductive hearing loss, while hair cell damage results in a sensorineural hearing deficit. The tonotopic organization of the cochlea is retained at all levels of the central auditory system. Projections from the cochlea travel via the eighth nerve to the three main divisions of the cochlear nucleus. The targets of the cochlear nucleus neurons include the superior olivary complex and nuclei of the lateral lemniscus, where the binaural cues for sound localization are processed. The inferior colliculus is the target of nearly all of the auditory pathways in the lower brainstem and carries out important integrative functions, such as processing sound frequencies and integrating the cues for localizing sound in space. The primary auditory cortex, which is also organized tonotopically, is essential for basic auditory functions, such as frequency discrimination and sound localization, and also plays an important role in processing of intraspecific communication sounds. The belt areas of the auditory cortex have a less strict tonotopic organization and also process complex sounds, such as those that mediate communication. In the human brain, the major speech comprehension areas are located in the zone immediately adjacent to the auditory cortex. Additional Reading Reviews COREY, D. P. AND A. J. HUDSPETH (1979) Ionic basis of the receptor potential in a vertebrate hair cell. Nature 281: 675–677. COREY, D.P. (1999) Ion channel defects in hereditary hearing loss. Neuron. 22(2):217-9. DALLOS, P. (1992) The active cochlea. J. Neurosci. 12: 4575–4585. GARCIA-ANOVEROS, J. AND D. P. COREY (1997) The molecules of mechanosensation. Ann. Rev. Neurosci. 20: 567–597. HEFFNER, H. E. AND R. S. HEFFNER (1990) Role of primate auditory cortex in hearing. In Comparative Perception, Volume II: Complex Signals. W. C. Stebbins and M. A. Berkley (eds.). New York: John Wiley. HUDSPETH, A. J. (1997) How hearing happens. Neuron 19: 947–950. HUDSPETH, A. J. (2000) Hearing and deafness. Neurobiol. Dis. 7: 511–514. HUDSPETH, A. J. AND M. KONISHI (2000) Auditory neuroscience: Development, transduction, and integration. Proc. Natl. Acad. Sci. USA 97: 11690–11691. HUDSPETH, A. J., Y. CHOE, A. D. MEHTA AND P. MARTIN (2000) Putting ion channels to work: Mechanoelectrical transduction, adaptation, and amplification by hair cells. Proc. Natl. Acad. Sci. USA 97: 11765-11772. KIANG, N. Y. S. (1984) Peripheral neural processing of auditory information. In Handbook of Physiology, Section 1: The Nervous System, Volume III. Sensory Processes, Part 2. J. M. Brookhart, V. B. Mountcastle, I. Darian-Smith and S. R. Geiger (eds.). Bethesda, MD: American Physiological Society. NEFF, W. D., I. T. DIAMOND AND J. H. CASSEDAY (1975) Behavioral studies of auditory discrimination. In Handbook of Sensory Physiology, Volumes V–II. W. D. Keidel and W. D. Neff (eds.). Berlin: Springer-Verlag. NELKEN, I. (2002) Feature detection by the auditory cortex. In Integrative Functions in the Mammalian Auditory Pathway, Springer Handbook of Auditory Research, Volume 15. D. Oertel, R. Fay and A. N. Popper (eds.). New York: Springer-Verlag, pp. 358–416. SUGA, N. (1990) Biosonar and neural computation in bats. Sci. Am. 262 (June): 60–68. Important Original Papers BARBOUR, D. L. AND X. WANG (2003) Contrast tuning in auditory cortex. Science. 299: 1073–1075. CRAWFORD, A. C. AND R. FETTIPLACE (1981) An electrical tuning mechanism in turtle cochlear hair cells. J. Physiol. 312: 377–412. FITZPATRICK, D. C., J. S. KANWAL, J. A. BUTMAN AND N. SUGA (1993) Combination-sensitive neurons in the primary auditory cortex of the mustached bat. J. Neurosci. 13: 931–940. COREY, D. P. AND A. J. HUDSPETH (1979) Ionic basis of the receptor potential in a vertebrate hair cell. Nature 281: 675–677. MIDDLEBROOKS, J. C., A. E. CLOCK, L. XU AND D. M. GREEN (1994) A panoramic code for sound location by cortical neurons. Science 264: 842–844. KNUDSEN, E. I. AND M. KONISHI (1978) A neural map of auditory space in the owl. Science 200: 795–797. JEFFRESS, L. A. (1948) A place theory of sound localization. J. Comp. Physiol. Psychol. 41: 35–39. 314 Chapter Twelve NELKEN, I., Y. ROTMAN AND O. BAR YOSEF (1999) Responses of auditory-cortex neurons to structural features of natural sounds. Nature 397: 154–157. SUGA, N., W. E. O’NEILL AND T. MANABE (1978) Cortical neurons sensitive to combinations of information-bearing elements of biosonar signals in the mustache bat. Science 200: 778–781. BÉKÉSY, G. (1960) Experiments in Hearing. New York: McGraw-Hill. (A collection of von Békésy’s original papers.) VON Books PICKLES, J. O. (1988) An Introduction to the Physiology of Hearing. London: Academic Press. YOST, W. A. AND G. GOUREVITCH (EDS.) (1987) Directional Hearing. Berlin: Springer Verlag. YOST, W. A. AND D. W. NIELSEN (1985) Fundamentals of Hearing. Fort Worth: Holt, Rinehart and Winston. Chapter 13 The Vestibular System Overview The vestibular system has important sensory functions, contributing to the perception of self-motion, head position, and spatial orientation relative to gravity. It also serves important motor functions, helping to stabilize gaze, head, and posture. The peripheral portion of the vestibular system includes inner ear structures that function as miniaturized accelerometers and inertial guidance devices, continually reporting information about the motions and position of the head and body to integrative centers in the brainstem, cerebellum, and somatic sensory cortices. The central portion of the system includes the vestibular nuclei, which make extensive connections with brainstem and cerebellar structures. The vestibular nuclei also directly innervate motor neurons controlling extraocular, cervical, and postural muscles. This motor output is especially important to stabilization of gaze, head orientation, and posture during movement. Although we are normally unaware of its functioning, the vestibular system is a key component in postural reflexes and eye movements. Balance, gaze stabilization during head movement, and sense of orientation in space are all adversely affected if the system is damaged. These manifestations of vestibular damage are especially important in the evaluation of brainstem injury. Because the circuitry of the vestibular system extends through a large part of the brainstem, simple clinical tests of vestibular function can be performed to determine brainstem involvement, even on comatose patients. The Vestibular Labyrinth The main peripheral component of the vestibular system is an elaborate set of interconnected chambers—the labyrinth—that has much in common, and is in fact continuous with, the cochlea (see Chapter 12). Like the cochlea, the labyrinth is derived from the otic placode of the embryo, and it uses the same specialized set of sensory cells—hair cells—to transduce physical motion into neural impulses. In the cochlea, the motion is due to airborne sounds; in the labyrinth, the motions transduced arise from head movements, inertial effects due to gravity, and ground-borne vibrations (Box A). The labyrinth is buried deep in the temporal bone and consists of the two otolith organs (the utricle and saccule) and three semicircular canals (Figure 13.1). The elaborate and tortuous architecture of these components explains why this part of the vestibular system is called the labyrinth. The utricle and saccule are specialized primarily to respond to linear accelerations of the head and static head position relative to the graviational axis, whereas the semicircular canals, as their shapes suggest, are specialized for responding to rotational accelerations of the head. 315 316 Chapter Thir teen Figure 13.1 The labyrinth and its innervation. The vestibular and auditory portions of the eighth nerve are shown; the small connection from the vestibular nerve to the cochlea contains auditory efferent fibers. General orientation in head is shown in Figure 12.3; see also Figure 13.8. Endolymphatic duct Ampullae Scarpa’s ganglion Utricle Vestibular part of cranial nerve VIII Facial nerve Auditory part of cranial nerve VIII Semicircular canals: Superior Posterior Horizontal Cochlea Saccule Canal reuniens The intimate relationship between the cochlea and the labyrinth goes beyond their common embryonic origin. Indeed, the cochlear and vestibular spaces are actually joined (see Figure 13.1), and the specialized ionic environments of the vestibular end organ parallel those of the cochlea. The membranous sacs within the bone are filled with fluid (endolymph) and are collectively called the membranous labyrinth. The endolymph (like the cochlear endolymph) is similar to intracellular solutions in that it is high in K+ and low in Na+. Between the bony walls (the osseous labyrinth) and the membranous labyrinth is another fluid, the perilymph, which is similar in composition to cerebrospinal fluid (i.e., low in K+ and high in Na+; see Chapter 12). The vestibular hair cells are located in the utricle and saccule and in three juglike swellings called ampullae, located at the base of the semicircular canals next to the utricle. Within each ampulla, the vestibular hair cells extend their hair bundles into the endolymph of the membranous labyrinth. As in the cochlea, tight junctions seal the apical surfaces of the vestibular hair cells, ensuring that endolymph selectively bathes the hair cell bundle while remaining separate from the perilymph surrounding the basal portion of the hair cell. Vestibular Hair Cells The vestibular hair cells, which like cochlear hair cells transduce minute displacements into behaviorally relevant receptor potentials, provide the basis for vestibular function. Vestibular and auditory hair cells are quite similar; a detailed description of hair cell structure and function has already been given in Chapter 12. As in the case of auditory hair cells, movement of the stereocilia toward the kinocilium in the vestibular end organs opens mechanically gated transduction channels located at the tips of the stereocilia, depolarizing the hair cell and causing neurotransmitter release onto (and excitation of) the vestibular nerve fibers. Movement of the stereocilia in the direction away from the kinocilium closes the channels, hyperpolarizing the hair cell and thus reducing vestibular nerve activity. The biphasic nature of the receptor potential means that some transduction channels are open in the absence of stimulation, with the result that hair cells tonically release The Vestibular System 317 transmitter, thereby generating considerable spontaneous activity in vestibular nerve fibers (see Figure 13.6). One consequence of these spontaneous action potentials is that the firing rates of vestibular fibers can increase or decrease in a manner that faithfully mimics the receptor potentials produced by the hair cells (Box B). Importantly, the hair cell bundles in each vestibular organ have specific orientations (Figure 13.2). As a result, the organ as a whole is responsive to displacements in all directions. In a given semicircular canal, the hair cells in the ampulla are all polarized in the same direction. In the utricle and saccule, a specialized area called the striola divides the hair cells into two populations with opposing polarities (Figure 13.2C; see also Figure 13.4C). The directional polarization of the receptor surfaces is a basic principle of organization in the vestibular system, as will become apparent in the following descriptions of the individual vestibular organs. The Otolith Organs: The Utricle and Saccule Displacements and linear accelerations of the head, such as those induced by tilting or translational movements (see Box A), are detected by the two otolith organs: the saccule and the utricle. Both of these organs contain a (A) Cross-sectional view Figure 13.2 The morphological polarization of vestibular hair cells and the polarization maps of the vestibular organs. (A) A cross section of hair cells shows that the kinocilia of a group of hair cells are all located on the same side of the hair cell. The arrow indicates the direction of deflection that depolarizes the hair cell. (B) View looking down on the hair bundles. (C) In the ampulla located at the base of each semicircular canal, the hair bundles are oriented in the same direction. In the sacculus and utricle, the striola divides the hair cells into populations with opposing hair bundle polarities. (B) Top view Direction of depolarizing deflection Kinocilium Stereocilia Hair cells Supporting cells Nerve fibers (C) Ampulla of superior canal Superior Anterior Striola Posterior Medial Anterior Inferior Utricular macula Saccular macula Ampulla Striola Sacculus Posterior Utricle Lateral 318 Chapter Thir teen Box A Yaw: Rotation around z axis A Primer on Vestibular Navigation The function of the vestibular system can be simplified by remembering some basic terminology of classical mechanics. All bodies moving in a three-dimensional framework have six degrees of freedom: three of these are translational and three are rotational. The translational elements refer to linear movements in the x, y, and z axes (the horizontal and vertical planes). Translational motion in these planes (linear acceleration and static displacement of the head) is the primary concern of the otolith organs. The three degrees of rotational freedom refer to a body’s rotation relative to the x, y, and z axes and are commonly referred to as roll, pitch, and yaw. The semicircular canals are primarily responsible for sensing rotational accelerations around these three axes. Figure 13.3 Scanning electron micrograph of calcium carbonate crystals (otoconia) in the utricular macula of the cat. Each crystal is about 50 mm long. (From Lindeman, 1973.) Roll: Rotation around x axis z x y Pitch: Rotation around y axis sensory epithelium, the macula, which consists of hair cells and associated supporting cells. Overlying the hair cells and their hair bundles is a gelatinous layer; above this layer is a fibrous structure, the otolithic membrane, in which are embedded crystals of calcium carbonate called otoconia (Figures 13.3 and 13.4A). The crystals give the otolith organs their name (otolith is Greek for “ear stones”). The otoconia make the otolithic membrane considerably heavier than the structures and fluids surrounding it; thus, when the head tilts, gravity causes the membrane to shift relative to the sensory epithelium (Figure 13.4B). The resulting shearing motion between the otolithic membrane and the macula displaces the hair bundles, which are embedded in the lower, gelatinous surface of the membrane. This displacement of the hair bundles generates a receptor potential in the hair cells. A shearing motion between the macula and the otolithic membrane also occurs when the head undergoes linear accelerations (see Figure 13.5); the greater relative mass of the otolithic membrane causes it to lag behind the macula temporarily, leading to transient displacement of the hair bundle. The similar effects exerted on otolithic hair cells by certain head tilts and linear accelerations would be expected to render these different stimuli perceptually equivalent when visual feedback is absent, as occurs in the dark or when the eyes are closed. Nevertheless, evidence suggests that subjects can discriminate between these two stimulus categories, apparently through combined activity of the otolith organs and the semicircular canals. As already mentioned, the orientation of the hair cell bundles is organized relative to the striola, which demarcates the overlying layer of otoco- The Vestibular System 319 (A) (B) Striola Static tilt Otoconia Otolithic membrane, gelatinous layer Reticular membrane Gravitational force Supporting cells Hair cells Anter (C) ior Utricular macula l era Lat Striola Superior Saccular macula Anterior Utricular macula Saccular macula Figure 13.4 Morphological polarization of hair cells in the utricular and saccular maculae. (A) Cross section of the utricular macula showing hair bundles projecting into the gelatinous layer when the head is level. (B) Cross section of the utricular macula when the head is tilted. (C) Orientation of the utricular and saccular maculae in the head; arrows show orientation of the kinocilia, as in Figure 13.2. The saccules on either side are oriented more or less vertically, and the utricles more or less horizontally. The striola is a structural landmark consisting of small otoconia arranged in a narrow trench that divides each otolith organ. In the utricular macula, the kinocilia are directed toward the striola. In the saccular macula, the kinocilia point away from the striola. Note that, given the utricle and sacculus on both sides of the body, there is a continuous representation of all directions of body movement. nia (see Figure 13.4A). The striola forms an axis of mirror symmetry such that hair cells on opposite sides of the striola have opposing morphological polarizations. Thus, a tilt along the axis of the striola will excite the hair cells on one side while inhibiting the hair cells on the other side. The saccular macula is oriented vertically and the utricular macula horizontally, with a continuous variation in the morphological polarization of the hair cells Striola 320 Chapter Thir teen Box B Adaptation and Tuning of Vestibular Hair Cells Hair Cell Adaptation The minuscule movement of the hair bundle at sensory threshold has been compared to the displacement of the top of the Eiffel Tower by a thumb’s breadth! Despite its great sensitivity, the hair cell can adapt quickly and continuously to static displacements of the hair bundle caused by large movements. Such adjustments are especially useful in the otolith organs, where adaptation permits hair cells to maintain sensitivity to small linear and angular accelerations of the head despite the constant input from gravitational forces that are over a million times greater. In other receptor cells, such as photoreceptors, adaptation is accomplished by regulating the second messenger cascade induced by the initial transduction event. The hair cell has to depend on a different strategy, however, because there is no second messenger system between the initial transduction event and the subsequent receptor potential (as might be expected for receptors that respond so rapidly). Adaptation occurs in both directions in which the hair bundle displacement generates a receptor potential, albeit at (A) Kinocilium Force of displacement different rates for each direction. When the hair bundle is pushed toward the kinocilium, tension is initially increased in the gating spring. During adaptation, tension decreases back to the resting level, perhaps because one end of the gating spring repositions itself along the shank of the stereocilium. When the hair bundle is displaced in the opposite direction, away from the kinocilium, tension in the spring initially decreases; adaptation then involves an increase in spring tension. One theory is that a calcium-regulated motor such as a myosin ATPase climbs along actin filaments in the stereocilium and actively resets the tension in the transduction spring. During sustained depolarization, some Ca2+ enters through the transduction channel, along with K+. Ca2+ then causes the motor to spend a greater fraction of its time unbound from the actin, resulting in slippage of the spring down the side of the stereocilium. During sustained hyperpolarization (Figure A), Ca2+ levels drop below normal resting levels and the motor spends more of its time bound to the actin, thus climbing up the actin filaments and increasing the spring tension. As tension increases, some of the previously closed transduction channels open, admitting Ca2+ and thus slowing the motor’s progress until a balance is struck between the climbing and slipping of the motor. In support of this model, when internal Ca2+ is reduced artificially, spring tension increases. This model of hair cell adaptation presents an elegant molecular solution to the regulation of a mechanical process. Electrical Tuning Although mechanical tuning plays an important role in generating frequency selectivity in the cochlea, there are other mechanisms that contribute to this process in vestibular and auditory nerve cells. These other tuning mechanisms are especially important in the otolith organs, where, unlike the cochlea, there are no (A) Adaptation is explained in the gating spring model by adjustment of the insertion point of tips links. Movement of the insertion point up or down the shank of the stereocilium, perhaps driven by a Ca2+-dependent protein motor, can continually adjust the resting tension of the tip link. (After Hudspeth and Gillespie, 1994.) 1 Stereocilia deflected (leftward), slackening “springs,” which closes channels, resulting in a decrease of [Ca2+]i Actin filament Motor protein “walks” along actin Motor retensions gate spring Decreased Ca2+ Stereocilium Stereociliary pivot 2 Motor retensions “spring,” causing fraction of channels to reopen The Vestibular System 321 obvious macromechanical resonances to selectively filter and/or enhance biologically relevant movements. One such mechanism is an electrical resonance displayed by hair cells in response to depolarization: The membrane potential of a hair cell undergoes damped sinusoidal oscillations at a specific frequency in response to the injection of depolarizing current pulses (Figure B). The ionic mechanism of this process involves two major types of ion channels located in the membrane of the hair cell soma. The first of these is a voltage-activated Ca2+ conductance, which lets Ca2+ into the cell soma in response to depolarization, such as that generated by the transduction current. The second is a Ca2+-activated K+ conductance, which is triggered by the rise in internal Ca2+ concentration. These two currents produce an interplay of depolarization and repolarization that results in electrical resonance (Figure C). Activation of the hair cell’s calcium-activated K+ conductance occurs 10 to 100 times faster than that of similar currents in other cells. Such rapid kinetics allow this conductance to generate an electrical response that usually requires the fast properties of a voltagegated channel. Although a hair cell responds to hair bundle movement over a wide range of frequencies, the resultant receptor potential is largest at the frequency of electrical resonance. The resonance frequency represents the characteristic frequency of the hair cell, and transduction at that frequency will be most efficient. This electrical resonance has important implications for structures like the utricle and sacculus, which may encode a range of characteristic frequencies based on the different resonance frequencies of their constituent hair cells. Thus, electrical tuning in the otolith organs can generate enhanced tuning to biologically relevant frequencies of stimulation, even in the absence of macromechanical resonances within these structures. (B) (C) Voltage (mV) 15 References Assad, J. A. and D. P. Corey (1992) An active motor model for adaptation by vertebrate hair cells. J. Neurosci. 12: 3291–3309. CRAWFORD, A. C. AND R. FETTIPLACE (1981) An electrical tuning mechanism in turtle cochlear hair cells. J. Physiol. 312: 377–412. HUDSPETH, A. J. (1985) The cellular basis of hearing: The biophysics of hair cells. Science 230: 745–752. HUDSPETH, A. J. AND P. G. GILLESPIE (1994) Pulling strings to tune transduction: Adaptation by hair cells. Neuron 12: 1–9. LEWIS, R. S. AND A. J. HUDSPETH (1988) A model for electrical resonance and frequency tuning in saccular hair cells of the bull-frog, Rana catesbeiana. J. Physiol. 400: 275–297. LEWIS, R. S. AND A. J. HUDSPETH (1983) Voltage- and ion-dependent conductances in solitary vertebrate hair cells. Nature 304: 538–541. SHEPHERD, G. M. G. AND D. P. COREY (1994) The extent of adaptation in bullfrog saccular hair cells. J. Neurosci. 14: 6217–6229. 41 K+ enters stereocilia, depolarizes cell K+ 10 Knob on kinocilium K+ 0 −5 −10 Stereocilia 0 20 40 60 ms 80 100 120 140 Depolarization Current (nA) 1.0 0.5 Ca2+ 0 0 Voltage-gated Ca2+ channel 2 Ca2+ enters through voltage-gated channel 0 20 40 60 ms 80 100 120 140 K+ Electrical resonance 3 Ca2+ activates K+ channel; K+ exits cells, repolarizing cell Ca2+-dependent K+ channel (B) Voltage oscillations (upper trace) in an isolated hair cell in response to a depolarizing current injection (lower trace). (After Lewis and Hudspeth, 1983.) (C) Proposed ionic basis for electrical resonance in hair cells. (After Hudspeth, 1985.) 322 Chapter Thir teen located in each macula (as shown in Figure 13.4C, where the arrows indicate the direction of movement that produces excitation). Inspection of the excitatory orientations in the maculae indicates that the utricle responds to movements of the head in the horizontal plane, such as sideways head tilts and rapid lateral displacements, whereas the saccule responds to movements in the vertical plane (up–down and forward–backward movements in the sagittal plane). Note that the saccular and utricular maculae on one side of the head are mirror images of those on the other side. Thus, a tilt of the head to one side has opposite effects on corresponding hair cells of the two utricular maculae. This concept is important in understanding how the central connections of the vestibular periphery mediate the interaction of inputs from the two sides of the head (see the next section). How Otolith Neurons Sense Linear Forces The structure of the otolith organs enables them to sense both static displacements, as would be caused by tilting the head relative to the gravitational axis, and transient displacements caused by translational movements of the head. The mass of the otolithic membrane relative to the surrounding endolymph, as well as the otolithic membrane’s physical uncoupling from the underlying macula, means that hair bundle displacement will occur transiently in response to linear accelerations and tonically in response to tilting of the head. Therefore, both tonic and transient information can be conveyed by these sense organs. Figure 13.5 illustrates some of the forces produced by head tilt and linear accelerations on the utricular macula. These properties of hair cells are reflected in the responses of the vestibular nerve fibers that innervate the otolith organs. The nerve fibers have a Head tilt; sustained Backward Forward Upright No head tilt; transient Figure 13.5 Forces acting on the head and the resulting displacement of the otolithic membrane of the utricular macula. For each of the positions and accelerations due to translational movements, some set of hair cells will be maximally excited, whereas another set will be maximally inhibited. Note that head tilts produce displacements similar to certain accelerations. Forward acceleration Deceleration The Vestibular System 323 (A) Start tilt Figure 13.6 Response of a vestibular nerve axon from an otolith organ (the utricle in this example). (A) The stimulus (top) is a change in head tilt. The spike histogram shows the neuron’s response to tilting in a particular direction. (B) A response of the same fiber to tilting in the opposite direction. (After Goldberg and Fernandez, 1976.) End tilt Constant tilt 0 Discharge rate (spikes/s) 120 100 80 60 40 20 0 0 (B) 40 80 120 Time (s) Start tilt 160 200 160 200 End tilt 0 Discharge rate (spikes/s) Constant tilt 40 20 0 0 40 80 120 Time (s) steady and relatively high firing rate when the head is upright. The change in firing rate in response to a given movement can be either sustained or transient, thereby signaling either absolute head position or linear acceleration. An example of the sustained response of a vestibular nerve fiber innervating the utricle is shown in Figure 13.6. The responses were recorded from axons in a monkey seated in a chair that could be tilted for several seconds to produce a steady force. Prior to the tilt, the axon has a high firing rate, which increases or decreases depending on the direction of the tilt. Notice also that the response remains at a high level as long as the tilting force remains constant; thus, such neurons faithfully encode the static force being applied to the head (Figure 13.6A). When the head is returned to the original position, the firing level of the neurons returns to baseline value. Conversely, when the tilt is in the opposite direction, the neurons respond by decreasing their firing rate below the resting level (Figure 13.6B) and remain depressed as long as the static force continues. In a similar fashion, transient increases or decreases in firing rate from spontaneous levels signal the direction of linear accelerations of the head. The range of orientations of hair bundles within the otolith organs enables them to transmit information about linear forces in every direction 324 Chapter Thir teen the body moves (see Figure 13.4C). The utricle, which is primarily concerned with motion in the horizontal plane, and the saccule, which is concerned with vertical motion, combine to effectively gauge the linear forces acting on the head at any instant in three dimensions. Tilts of the head off the horizontal plane and translational movements of the head in any direction stimulate a distinct subset of hair cells in the saccular and utricular maculae, while simultaneously suppressing the responses of other hair cells in these organs. Ultimately, variations in hair cell polarity within the otolith organs produce patterns of vestibular nerve fiber activity that, at a population level, can unambiguously encode head position and the forces that influence it. The Semicircular Canals Cupula Ampulla Hair bundle Crista Membranous duct Hair cells Nerve fibers Figure 13.7 The ampulla of the posterior semicircular canal showing the crista, hair bundles, and cupula. The cupula is distorted by the fluid in the membranous canal when the head rotates. Whereas the otolith organs are primarily concerned with head translations and orientation with respect to gravity, the semicircular canals sense head rotations, arising either from self-induced movements or from angular accelerations of the head imparted by external forces. Each of the three semicircular canals has at its base a bulbous expansion called the ampulla (Figure 13.7), which houses the sensory epithelium, or crista, that contains the hair cells. The structure of the canals suggests how they detect the angular accelerations that arise through rotation of the head. The hair bundles extend out of the crista into a gelatinous mass, the cupula, that bridges the width of the ampulla, forming a fluid barrier through which endolymph cannot circulate. As a result, the relatively compliant cupula is distorted by movements of the endolymphatic fluid. When the head turns in the plane of one of the semicircular canals, the inertia of the endolymph produces a force across the cupula, distending it away from the direction of head movement and causing a displacement of the hair bundles within the crista (Figure 13.8A,B). In contrast, linear accelerations of the head produce equal forces on the two sides of the cupula, so the hair bundles are not displaced. Unlike the saccular and utricular maculae, all of the hair cells in the crista within each semicircular canal are organized with their kinocilia pointing in the same direction (see Figure 13.2C). Thus, when the cupula moves in the appropriate direction, the entire population of hair cells is depolarized and activity in all of the innervating axons increases. When the cupula moves in the opposite direction, the population is hyperpolarized and neuronal activity decreases. Deflections orthogonal to the excitatory–inhibitory direction produce little or no response. Each semicircular canal works in concert with the partner located on the other side of the head that has its hair cells aligned oppositely. There are three such pairs: the two pairs of horizontal canals, and the superior canal on each side working with the posterior canal on the other side (Figure 13.8C). Head rotation deforms the cupula in opposing directions for the two partners, resulting in opposite changes in their firing rates (Box C). Thus, the orientation of the horizontal canals makes them selectively sensitive to rotation in the horizontal plane. More specifically, the hair cells in the canal towards which the head is turning are depolarized, while those on the other side are hyperpolarized. For example, when the head accelerates to the left, the cupula is pushed toward the kinocilium in the left horizontal canal, and the firing rate of the relevant axons in the left vestibular nerve increases. In contrast, the cupula in the right horizontal canal is pushed away from the kinocilium, with a concomitant decrease in the firing rate of the related neurons. If the head movement is The Vestibular System 325 (B) (A) Cupula Cupula displacement Angular acceleration Ampulla Endolymph flow Semicircular canal Hair cells (C) Left and right horizontal canals Left anterior Right canal (AC) posterior canal (PC) Right anterior Left posterior canal (AC) canal (PC) Figure 13.8 Functional organization of the semicircular canals. (A) The position of the cupula without angular acceleration. (B) Distortion of the cupula during angular acceleration. When the head is rotated in the plane of the canal (arrow outside canal), the inertia of the endolymph creates a force (arrow inside the canal) that displaces the cupula. (C) Arrangement of the canals in pairs. The two horizontal canals form a pair; the right anterior canal (AC) and the left posterior canal (PC) form a pair; and the left AC and the right PC form a pair. to the right, the result is just the opposite. This push–pull arrangement operates for all three pairs of canals; the pair whose activity is modulated is in the plane of the rotation, and the member of the pair whose activity is increased is on the side toward which the head is turning. The net result is a system that provides information about the rotation of the head in any direction. How Semicircular Canal Neurons Sense Angular Accelerations Like axons that innervate the otolith organs, the vestibular fibers that innervate the semicircular canals exhibit a high level of spontaneous activity. As a result, they can transmit information by either increasing or decreasing their firing rate, thus more effectively encoding head movements (see above). The bidirectional responses of fibers innervating the hair cells of the semicircular canal have been studied by recording the axonal firing rates in a monkey’s 326 Chapter Thir teen Box C Throwing Cold Water on the Vestibular System Testing the integrity of the vestibular system can indicate much about the condition of the brainstem, particularly in comatose patients. Normally, when the head is not being rotated, the output of the nerves from the right and left sides are equal; thus, no eye movements occur. When the head is rotated in the horizontal plane, the vestibular afferent fibers on the side toward the turning motion increase their firing rate, while the afferents on the opposite side decrease their firing rate (Figures A and B). The net difference in firing rates then leads to slow movements of the eyes counter to the turning motion. This reflex response generates the slow component of a normal eye movement pattern called physiological nystagmus, which means “nodding” or oscillatory movements of the eyes (Figure B1). (The fast component is a saccade that resets the eye position; see Chapter 19.) Pathological nystagmus can occur if there is unilateral damage to the vestibular system. In this case, the silencing of the spontaneous output from the dam- aged side results in an unphysiological difference in firing rate because the spontaneous discharge from the intact side remains (Figure B2). The difference in firing rates causes nystagmus, even though no head movements are being made. Responses to vestibular stimulation are thus useful in assessing the integrity of the brainstem in unconscious patients. If the individual is placed on his or her back and the head is elevated to about 30° above horizontal, the horizontal semicircular canals lie in an almost vertical orientation. Irrigating one ear with cold water will then lead to spontaneous eye movements because convection currents in the canal mimic rotatory head movements away from the irrigated ear (Figure C). In normal individuals, these eye movements consist of a slow movement toward the irrigated ear and a fast movement away from it. The fast movement is most readily detected by the observer, and the significance of its direction can be kept in mind by using the (B) (1) Physiological nystagmus Head rotation Slow eye movement Fast eye movement (A) Right horizontal canal Left horizontal canal Right horizontal canal (A) View looking down on the top of a person’s head illustrates the fluid motion generated in the left and right horizontal canals, and the changes in vestibular nerve firing rates when the head turns to the right. (B) In normal individuals, rotating the head elicits physiological nystagmus (1), which consists of a slow eye movement counter to the direction of head turning. The slow component of the eye movements is due to the net differences in left and right vestibular nerve firing rates acting via the central circuit diagrammed in Figure 13.10. Spontaneous nystagmus (2), where the eyes move rhythmically from side to side in the absence of any head movements, occurs when one of the canals is damaged. In this situation, net differences in vestibular nerve firing rates exist even when the head is stationary because the vestibular nerve innervating the intact canal fires steadily when at rest, in contrast to a lack of activity on the damaged side. He Left horizontal canal ad turns Ampullae Axis of hair cells Primary vestibular afferents Fluid motion in horizontal ducts Body Increased firing Decreased firing (2) Spontaneous nystagmus Afferent fibers of nerve VIII Increase in firing Decrease in firing Baseline firing No firing The Vestibular System 327 (C) 1 Warm H2O irrigation 1 Cold H2O irrigation 2 Endolymph rises 2 Endolymph falls Left horizontal duct Right horizontal duct 3 Increased firing Gravity (horizontal canals of reclining patient are nearly vertical) mnemonic COWS (“Cold Opposite, Warm Same”). This same test can also be used in unconscious patients. In patients who are comatose due to dysfunction of both cerebral hemispheres but whose brainstem is intact, saccadic movements are no longer made and the response to (C) Caloric testing of vestibular function is possible because irrigating an ear with water slightly warmer than body temperature generates convection currents in the canal that mimic the endolymph movement induced by turning the head to the irrigated side. Irrigation with cold water induces the opposite effect. These currents result in changes in the firing rate of the associated vestibular nerve, with an increased rate on the warmed side and a decreased rate on the chilled side. As in head rotation and spontaneous nystagmus, net differences in firing rates generate eye movements. fourth, or sixth cranial nerves), or the peripheral nerves exiting these nuclei, vestibular responses are abolished (or altered, depending on the severity of the lesion). 3 Decreased firing cold water consists of only the slow movement component of the eyes to side of the irrigated ear (Figure D). In the presence of brainstem lesions involving either the vestibular nuclei themselves, the connections from the vestibular nuclei to oculomotor nuclei (the third, (D) Ocular reflexes in conscious patients (1) Normal (D) Caloric testing can be used to test the function of the brainstem in an unconscious patient. The figures show eye movements resulting from cold or warm water irrigation in one ear for (1) a normal subject, and in three different conditions in an unconscious patient: (2) with the brainstem intact; (3) with a lesion of the medial longitudinal fasciculus (MLF; note that irrigation in this case results in lateral movement of the eye only on the less active side); and (4) with a low brainstem lesion (see Figure 13.10). Ocular reflexes in unconscious patients (2) Brainstem intact (3) MLF lesion (bilateral) (4) Low brainstem lesion F S S Cold H2O S Cold H2O Cold H2O Cold H2O F S Warm H2O S Warm H2O S Warm H2O Warm H2O 328 Chapter Thir teen Acceleration Deceleration Constant velocity Discharge rate (spikes/s) 0 120 60 0 0 40 80 Time (s) 120 Figure 13.9 Response of a vestibular nerve axon from the semicircular canal to angular acceleration. The stimulus (top) is a rotation that first accelerates, then maintains constant velocity, and then decelerates the head. The axon increases its firing above resting level in response to the acceleration, returns to resting level during constant velocity, then decreases its firing rate below resting level during deceleration; these changes in firing rate reflect inertial effects on the displacement of the cupula. (After Goldberg and Fernandez, 1971.) vestibular nerve. Seated in a chair, the monkey was rotated continuously in one direction during three phases: an initial period of acceleration, then a periord of several seconds at constant velocity, and finally a period of sudden deceleration to a stop (Figure 13.9). The maximum firing rates observed correspond to the period of acceleration; the maximum inhibition corresponds to the period of deceleration. During the constant-velocity phase, the response adapts so that the firing rate subsides to resting level; after the movement stops, the neuronal activity decreases transiently before returning to the resting level. Neurons innervating paired semicircular canals have a complementary response pattern. Note that the rate of adaptation (on the order of tens of seconds) corresponds to the time it takes the cupula to return to its undistorted state (and for the hair bundles to return to their undeflected position); adaptation therefore can occur while the head is still turning, as long as a constant angular velocity is maintained. Such constant forces are rare in nature, although they are encountered on ships, airplanes, and space vehicles, where prolonged acceleratory arcs are sometimes described. Central Pathways for Stabilizing Gaze, Head, and Posture The vestibular end organs communicate via the vestibular branch of cranial nerve VIII with targets in the brainstem and the cerebellum that process much of the information necessary to compute head position and motion. As with the cochlear nerve, the vestibular nerves arise from a population of bipolar neurons, the cell bodies of which in this instance reside in the vestibular nerve ganglion (also called Scarpa’s ganglion; see Figure 13.1). The distal processes of these cells innervate the semicircular canals and the otolith organs, while the central processes project via the vestibular portion of cranial nerve VIII to the vestibular nuclei (and also directly to the cerebellum; Figure 13.10). The vestibular nuclei are important centers of integration, receiving input from the vestibular nuclei of the opposite side, as well as from the cerebellum and the visual and somatic sensory systems. Because vestibular and auditory fibers run together in the eighth nerve, damage to this structure often results in both auditory and vestibular disturbances. The central projections of the vestibular system participate in three major classes of reflexes: (1) helping to maintain equilibrium and gaze during movement, (2) maintaining posture, and (3) maintaining muscle tone. The first of these reflexes helps coordinate head and eye movements to keep gaze fixated on objects of interest during movements (other functions include protective or escape reactions; see Box D). The vestibulo-ocular reflex (VOR) in particular is a mechanism for producing eye movements that counter head movements, thus permitting the gaze to remain fixed on a particular point (Box C; see also Chapter 19). For example, activity in the left horizontal canal induced by leftward rotary acceleration of the head excites neurons in the left vestibular nucleus and results in compensatory eye movements to the right. This effect is due to excitatory projections from the vestibular nucleus to the contralateral nucleus abducens that, along with the oculomotor nucleus, help execute conjugate eye movements. For instance, horizontal movement of the two eyes toward the right requires contraction of the left medial and right lateral rectus muscles. Vestibular nerve fibers originating in the left horizontal semicircular canal project to the medial and lateral vestibular nuclei (see Figure 13.10). Excitatory fibers from the medial vestibular nucleus cross to the contralateral abducens nucleus, which has two outputs. One of these is a motor pathway The Vestibular System 329 Left eye Right eye Lateral rectus Medial rectus Lateral rectus Oculomotor nucleus Midbrain + − Medial longitudinal fasciculus Pons Abducens nucleus Scarpa’s ganglion Rostral medulla Medial vestibular nucleus that causes the lateral rectus of the right eye to contract; the other is an excitatory projection that crosses the midline and ascends via the medial longitudinal fasciculus to the left oculomotor nucleus, where it activates neurons that cause the medial rectus of the left eye to contract. Finally, inhibitory neurons project from the medial vestibular nucleus to the left abducens nucleus, directly causing the motor drive on the lateral rectus of the left eye to decrease and also indirectly causing the right medial rectus to relax. The consequence of these several connections is that excitatory input from the horizontal canal on one side produces eye movements toward the opposite side. Therefore, turning the head to the left causes eye movements to the right. In a similar fashion, head turns in other planes activate other semicircular canals, causing other appropriate compensatory eye movements. Thus, the VOR also plays an important role in vertical gaze stabilization in response to Figure 13.10 Connections underlying the vestibulo-ocular reflex. Projections of the vestibular nucleus to the nuclei of cranial nerves III (oculomotor) and VI (abducens). The connections to the oculomotor nucleus and to the contralateral abducens nucleus are excitatory (red), whereas the connections to ipsilateral abducens nucleus are inhibitory (black). There are connections from the oculomotor nucleus to the medial rectus of the left eye and from the adbucens nucleus to the lateral rectus of the right eye. This circuit moves the eyes to the right, that is, in the direction away from the left horizontal canal, when the head rotates to the left. Turning to the right, which causes increased activity in the right horizontal canal, has the opposite effect on eye movements. The projections from the right vestibular nucleus are omitted for clarity. 330 Chapter Thir teen the linear vertical head oscillations that accompany locomotion, and in response to vertical angular accelerations of the head, as can occur when riding on a swing. The rostrocaudal set of cranial nerve nuclei involved in the VOR (i.e., the vestibular, abducens, and oculomotor nuclei), as well as the VOR’s persistence in the unconscious state, make this reflex especially useful for detecting brainstem damage in the comatose patient (see Box C). Loss of the VOR can have severe consequences. A patient with vestibular damage finds it difficult or impossible to fixate on visual targets while the head is moving, a condition called oscillopsia (“bouncing vision”). If the damage is unilateral, the patient usually recovers the ability to fixate objects during head movements. However, a patient with bilateral loss of vestibular function has the persistent and disturbing sense that the world is moving when the head moves. The underlying problem in such cases is that information about head and body movements normally generated by the vestibular organs is not available to the oculomotor centers, so that compensatory eye movements cannot be made. Descending projections from the vestibular nuclei are essential for postural adjustments of the head, mediated by the vestibulo-cervical reflex (VCR), and body, mediated by the vestibulo-spinal reflex (VSR). As with the VOR, these postural reflexes are extremely fast, in part due to the small number of synapses interposed between the vestibular organ and the relevant motor neurons (Box D). Like the VOR, the VCR and the VSR are both compromised in patients with bilateral vestibular damage. Such patients exhibit diminished head and postural stability, resulting in gait deviations; they also have difficulty balancing. These balance defects become more pronounced in low light or while walking on uneven surfaces, indicating that balance normally is the product of vestibular, visual, and proprioceptive inputs. The anatomical substrate for the VCR involves the medial vestibular nucleus, axons from which descend in the medial longitudinal fasciculus to reach the upper cervical levels of the spinal cord (Figure 13.11). This pathway regulates head position by reflex activity of neck muscles in response to stimulation of the semicircular canals from rotational accelerations of the head. For example, during a downward pitch of the body (e.g., tripping), the superior canals are activated and the head muscles reflexively pull the head up. The dorsal flexion of the head initiates other reflexes, such as forelimb extension and hindlimb flexion, to stabilize the body and protect against a fall (see Chapter 16). The VSR is mediated by a combination of pathways, including the lateral and medial vestibulospinal tracts and the reticulospinal tract. The inputs from the otolith organs project mainly to the lateral vestibular nucleus, which in turn sends axons in the lateral vestibulospinal tract to the spinal cord (see Figure 13.11). These axons terminate monosynaptically on extensor motor neurons, and they disynaptically inhibit flexor motor neurons; the net result is a powerful excitatory influence on the extensor (antigravity) muscles. When hair cells in the otolith organs are activated, signals reach the medial part of the ventral horn. By activating the ipsilateral pool of motor neurons innervating extensor muscles in the trunk and limbs, this pathway mediates balance and the maintenance of upright posture. Decerebrate rigidity, characterized by rigid extension of the limbs, arises when the brainstem is transected above the level of the vestibular nucleus. Decerebrate rigidity in experimental animals is relieved when the vestibular nuclei are lesioned, underscoring the importance of the vestibular system to the maintenance of muscle tone. The tonic activation of extensor muscles in The Vestibular System 331 Lateral vestibular nucleus Cerebellum Mid-Pons Medial vestibular nucleus Rostral medulla Medial longitudinal fasciculus Lateral vestibulospinal tract Medial vestibulospinal tract Spinal cord Ventral horn Figure 13.11 Descending projections from the medial and lateral vestibular nuclei to the spinal cord underlie the VCR and VSR. The medial vestibular nuclei project bilaterally in the medial longitudinal fasciculus to reach the medial part of the ventral horns and mediate head reflexes in response to activation of semicircular canals. The lateral vestibular nucleus sends axons via the lateral vestibular tract to contact anterior horn cells innervating the axial and proximal limb muscles. Neurons in the lateral vestibular nucleus receive input from the cerebellum, allowing the cerebellum to influence posture and equilibrium. decerebrate rigidity suggests further that the vestibulospinal pathway is normally suppressed by descending projections from higher levels of the brain, especially the cerebral cortex (see also Chapter 16). Vestibular Pathways to the Thalamus and Cortex In addition to these several descending projections, the superior and lateral vestibular nuclei send axons to the ventral posterior nuclear complex of the thalamus, which in turn projects to two cortical areas relevant to vestibular 332 Chapter Thir teen Box D Mauthner Cells in Fish A primary function of the vestibular system is to provide information about the direction and speed of ongoing movements, ultimately enabling rapid, coordinated reflexes to compensate for both self-induced and externally generated forces. One of the most impressive and speediest vestibular-mediated reflexes is the tail-flip escape behavior of fish (and larval amphibians), a stereotyped response that allows a potential prey to elude its predators (Figure A; tap on the side of a fish tank if you want to observe the reflex). In response to a perceived risk, fish flick their tail and are thus propelled laterally away from the approaching threat. The circuitry underlying the tail-flip escape reflex includes a pair of giant medullary neurons called Mauthner cells, their vestibular inputs, and the spinal cord motor neurons to which the Mauthner cells project. (In most fish, there is one pair of Mauthner cells in a stereotypic location. Thus, these cells can be consistently visualized and studied from animal to animal.) Movements in the water, such as might be caused by an approaching predator, excite saccular hair cells in the vestibular labyrinth. These receptor potentials are transmitted via the central processes of vestibular ganglion cells in cranial nerve VIII to the two Mauthner cells in the brainstem. As in the vestibulo-spinal pathway in humans, the Mauthner cells project directly to spinal motor neurons. The small number of synapses intervening between the receptor cells and the motor neurons is one of the ways that this circuit has been optimized for speed by natural selection, an arrangement evident in humans as well. The large size of the Mauthner axons is another; the axons from these cells in a goldfish are about 50 µm in diameter. The optimization for speed and direction in the escape reflex also is reflected in the synapses vestibular nerve afferents make on each Mauthner cell (Figure B). These connections are electrical synapses that allow rapid and faithful transmission of the vestibular signal. An appropriate direction for escape is promoted by two features: (1) each Mauthner cell projects only to contralateral motor neurons; and (2) a local network of bilaterally projecting interneurons inhibits activity in the Mauthner cell away from the side on which the vestibular activity originates. In this way, the Mauthner cell on one side faithfully generates action potentials that command contractions of contralateral tail musculature, thus moving the fish out of the path of the oncoming predator. Conversely, the Mauthner cell on the opposite side is silenced by the local inhibitory network during the response (Figure C). (A) Bird’s-eye view of the sequential body orientations of a fish engaging in a tail-flip escape behavior, with time progressing from left to right. This behavior is largely mediated by vestibular inputs to Mauthner cells. (A) sensations (Figure 13.12). One of these cortical targets is just posterior to the primary somatosensory cortex, near the representation of the face; the other is at the transition between the somatic sensory cortex and the motor cortex (Brodmann’s area 3a; see Chapter 8). Electrophysiological studies of individual neurons in these areas show that the relevant cells respond to proprioceptive and visual stimuli as well as to vestibular stimuli. Many of these neurons are activated by moving visual stimuli as well as by rotation of the body (even with the eyes closed), suggesting that these cortical regions are involved in the perception of body orientation in extrapersonal space. Con- The Vestibular System 333 (B) (C) Right cranial nerve VIII Wave Time 2 Time 1 Midline Vestibular hair cells Right Mauthner cell Record Record Axon cap Lateral dendrite Left Mauthner cell Electrical synapse Cranial nerve VIII Right Mauthner cell Left cranial nerve VIII Left Mauthner cell Time 1 Record Right axon Left axon Record Time 2 Time (B) Diagram of synaptic events in the Mauthner cells of a fish in response to a disturbance in the water coming from the right. (C) Complementary responses of the right and left Mauthner cells mediating the escape response. Times 1 and 2 correspond to those indicated in Figure B. (After Furshpan and Furukuwa, 1962.) The Mauthner cells in fish are analogous to the reticulospinal and vestibulospinal pathways that control balance, posture, and orienting movements in mammals. The equivalent behavioral responses in humans are evident in a friendly game of tag, or more serious endeavors. References EATON, R. C., R. A. BOMBARDIERI AND D. L. MEYER (1977) The Mauthner-initiated startle response in teleost fish. J. Exp. Biol. 66: 65–81. FURSHPAN, E. J. AND T. FURUKAWA (1962) Intracellular and extracellular responses of the several regions of the Mauthner cell of the goldfish. J. Neurophysiol. 25:732–771. sistent with this interpretation, patients with lesions of the right parietal cortex suffer altered perception of personal and extra-personal space, as discussed in greater detail in Chapter 25. Summary The vestibular system provides information about the motion and position of the body in space. The sensory receptor cells of the vestibular system are located in the otolith organs and the semicircular canals of the inner ear. The JONTES, J. D., J. BUCHANAN AND S. J. SMITH (2000) Growth cone and dendrite dynamics in zebrafish embryos: Early events in synaptogenesis imaged in vivo. Nature Neurosci. 3: 231–237. O’MALLEY, D. M., Y. H. KAO AND J. R. FETCHO (1996) Imaging the functional organization of zebrafish hindbrain segments during escape behaviors. Neuron 17: 1145–1155. 334 Chapter Thir teen Postcentral gyrus Figure 13.12 Thalamocortical pathways carrying vestibular information. The lateral and superior vestibular nuclei project to the thalamus. From the thalamus, the vestibular neurons project to the vicinity of the central sulcus near the face representation. Sensory inputs from the muscles and skin also converge on thalamic neurons receiving vestibular input (see Chapter 9). Posterior parietal cortex (Area 5) (area 5) Vestibular Vestibular cortex cortex Region near face representation of SI Cerebrum Ventral posterior nucleus complex of the thalamus Lateral and superior vestibular nuclei Muscle and cutaneous afferents Pons otolith organs provide information necessary for postural adjustments of the somatic musculature, particularly the axial musculature, when the head tilts in various directions or undergoes linear accelerations. This information represents linear forces acting on the head that arise through static effects of gravity or from translational movements. The semicircular canals, in contrast, provide information about rotational accelerations of the head. This latter information generates reflex movements that adjust the eyes, head, and body during motor activities. Among the best studied of these reflexes are eye movements that compensate for head movements, thereby stabilizing the visual scene when the head moves. Input from all the vestibular organs is integrated with input from the visual and somatic sensory systems to provide perceptions of body position and orientation in space. The Vestibular System 335 Additional Reading Reviews BENSON, A. (1982) The vestibular sensory system. In The Senses, H. B. Barlow and J. D. Mollon (eds.). New York: Cambridge University Press. BRANDT, T. (1991) Man in motion: Historical and clinical aspects of vestibular function. A review. Brain 114: 2159–2174. FURMAN, J. M. AND R. W. BALOH (1992) Otolith-ocular testing in human subjects. Ann. New York Acad. Sci. 656: 431–451. GOLDBERG, J. M. (1991) The vestibular end organs: Morphological and physiological diversity of afferents. Curr. Opin. Neurobiol. 1: 229–235. GOLDBERG, J. M. AND C. FERNANDEZ (1984) The vestibular system. In Handbook of Physiology, Section 1: The Nervous System, Volume III: Sensory Processes, Part II, J. M. Brookhart, V. B. Mountcastle, I. Darian-Smith and S. R. Geiger (eds.). Bethesda, MD: American Physiological Society. HESS, B. J. (2001) Vestibular signals in self-orientation and eye movement control. News Physiolog. Sci. 16: 234–238. RAPHAN, T. AND B. COHEN. (2002) The vestibulo-ocular reflex in three dimensions. Exp. Brain Res. 145: 1–27. Important Original Papers GOLDBERG, J. M. AND C. FERNANDEZ (1971) Physiology of peripheral neurons innervating semicircular canals of the squirrel monkey, Parts 1, 2, 3. J. Neurophysiol. 34: 635–684. GOLDBERG, J. M. AND C. FERNANDEZ (1976) Physiology of peripheral neurons innervating otolith organs of the squirrel monkey, Parts 1, 2, 3. J. Neurophysiol. 39: 970–1008. LINDEMAN, H. H. (1973) Anatomy of the otolith organs. Adv. Oto.-Rhino.-Laryng. 20: 405–433. Books BALOH, R. W. AND V. HONRUBIA (2001) Clinical Neurophysiology of the Vestibular System, 3rd Ed. New York: Oxford University Press. BALOH, R. W. (1998) Dizziness, Hearing Loss, and Tinnitus. Philadelphia: F. A. Davis Company. Chapter 14 The Chemical Senses Overview Three sensory systems associated with the nose and mouth—olfaction, taste, and the trigeminal or general chemosensory system—are dedicated to the detection of chemicals in the environment. The olfactory system detects airborne molecules called odorants. In humans, odors provide information about food, self, other people, animals, plants, and many other aspects of the environment. Olfactory information can influence feeding behavior, social interactions and, in many animals, reproduction. The taste (or gustatory) system detects ingested, primarily water-soluble molecules called tastants. Tastants provide information about the quality, quantity, and safety of ingested food. Finally, the trigeminal chemosensory system provides information about irritating or noxious molecules that come into contact with skin or mucous membranes of the eyes, nose, and mouth. All three of these chemosensory systems rely on receptors in the nasal cavity, mouth, or on the face that interact with the relevant molecules and generate receptor and action potentials, thus transmitting information about chemical stimuli to appropriate regions of the central nervous system. The Organization of the Olfactory System From an evolutionary perspective, the chemical senses—particularly olfaction—are deemed the “oldest” sensory systems; nevertheless, they remain in many ways the least understood of the sensory modalities. The olfactory system (Figure 14.1) is the most thoroughly studied component of the chemosensory triad and processes information about the identity, concentration, and quality of a wide range of chemical stimuli that we associate with our sense of smell. These stimuli, called odorants, interact with olfactory receptor neurons found in an epithelial sheet—the olfactory epithelium— that lines the interior of the nose (Figure 14.1A,B). The axons arising from the receptor cells project directly to neurons in the olfactory bulb, which in turn sends projections to the pyriform cortex in the temporal lobe as well as other structures in the forebrain (Figure 14.1C). The olfactory system is thus unique among the sensory systems in that it does not include a thalamic relay from primary receptors en route to a neocortical (six-layered) region that processes the sensory information. Instead, the pyriform cortex is threelayered archicortex—considered to be phlogenetically older than the neocortex—and thus represents a specialized processing center dedicated to olfaction. Projections from the pyriform cortex relay olfactory information via the thalamus to association areas of the neocortex (see Figure 14.1C, D). The olfactory tract also projects to a number of other targets in the forebrain, including the hypothalamus and amygdala. The further processing that 337 338 Chapter Four teen (A) (B) Olfactory bulb Cribriform plate Olfactory bulb Cribriform plate Olfactory epithelium Nasal cavity Olfactory nerve Olfactory epithelium Airborne odors (C) (D) Olfactory tract Olfactory bulb Olfactory nerve (I) Olfactory receptors Olfactory bulb targets Orbitofrontal cortex Pyriform cortex Thalamus Olfactory tubercle Hypothalamus Amygdala Hippocampal formation Entorhinal cortex Olfactory Olfactory bulb tract Optic chiasm Olfactory tubercle Pyriform cortex Amygdala Entorhinal cortex Figure 14.1 Organization of the human olfactory system. (A) Peripheral and central components of the primary olfactory pathway. (B) Enlargement of region boxed in (A) showing the relationship between the olfactory epithelium (which contains the olfactory receptor neurons) and the olfactory bulb (the central target of olfactory receptor neurons). (C) Diagram of the basic pathways for processing olfactory information. (D) Central components of the olfactory system. (E) fMRI images showing focal activity in the regions of the olfactory bulb, pyriform cortex, and amygdala in a normal human being passively smelling odors. (From Savic et al., 2001.) (E) Orbitofrontal cortex Region of olfactory bulb Pyriform cortex Amygdala The Chemical Senses 339 occurs in these various regions identifies the odorant and initiates appropriate motor, visceral, and emotional reactions to olfactory stimuli. Despite its phylogenetic “age” and the unusual trajectory of olfactory information to the neocortex, the olfactory system abides by the basic principle that governs other sensory modalities: interactions with stimuli—in this case, airborne chemical odorants—at the periphery are transduced and encoded by receptors into electrical signals, which are then relayed to higher-order centers. Nevertheless, less is known about the central organization of the olfactory system than other sensory pathways. For example, the somatic sensory and visual cortices described in the preceding chapters all feature spatial maps of the relevant receptor surface, and the auditory cortex features frequency and other maps. Whether analogous maps of specific odorants (e.g., rose or pine) or odorant attributes (e.g., sweet or acrid) exist in the olfactory bulb or pyriform cortex is not yet known. Indeed, until recently it has been difficult to imagine what sensory qualities would be represented in an olfactory map, or what features might be processed in parallel as occurs in other sensory systems. Olfactory Perception in Humans In humans, olfaction is often considered the least acute of the senses, and a number of animals are obviously superior to humans in their olfactory abilities. This difference may reflect the larger number of olfactory receptor neurons (and odorant receptor molecules; see below) in the olfactory epithelium in many species and the proportionally larger area of the forebrain devoted to olfaction. In a 70-kg human, the surface area of the olfactory epithelium is approximately 10 cm2. In contrast, a 3-kg cat has about 20 cm2 of olfactory epithelium. Similarly, the relative size of the olfactory bulb and related structures versus the cortical hemispheres in a rodent or carnivore is quite large compared to that in humans. Humans are nonetheless quite good at detecting and identifying airborne molecules in the environment (Figure 14.2). For instance, the major aromatic constituent of bell pepper (2-isobutyl-3methoxypyrazine) can be detected at a concentration of 0.01 nM. However, the threshold concentrations for detection and identification of other odorants vary greatly. Ethanol, for example, cannot be identified until its concentration reaches approximately 2 mM. Small changes in molecular structure can also lead to large perceptual differences: The molecule D-carvone smells like caraway seeds, whereas L-carvone smells like spearmint. Since the number of odorants is very large, there have been several attempts to classify them in groups. The most widely used classification was developed in the 1950s by John Amoore, who divided odors into categories based on their perceived quality, molecular structure, and the fact that some people, called anosmics, have difficulty smelling one or another group. Amoore classified odorants as pungent, floral, musky, earthy, ethereal, camphor, peppermint, ether, and putrid; these categories are still used to describe odors, to study the cellular mechanisms of olfactory transduction, and to discuss the central representation of olfactory information. Nevertheless, this classification remains entirely empirical. A further complication in rationalizing the perception of odors is that their quality may change with concentration. For example, at low concentrations indole has a floral odor, whereas at higher concentrations it smells putrid. Despite these problems, the longevity of Amoore’s scheme makes clear that the olfactory system can identify odorant classes that have distinct perceptual qualities. Indeed, humans can per- 340 Chapter Four teen Figure 14.2 Chemical structure and human perceptual threshold for 12 common odorants. Molecules perceived at low concentrations are more lipid-soluble, whereas those with higher thresholds are more water-soluble. (After Pelosi, 1994.) O H O OH O O O Ethanol alcoholic 2 mM Ethyl acetate ethereal 0.06 mM Benzaldehyde bitter almond 0.3 µM 4-Hydroxyoctanoic acid lactone coconut 0.05 µM O O S Cl Cl O Cl O Pentadecalactone musky 7 nM Dimethylsulfide 5α-Androst-16-enputrid 3-one 5 nM urinous 0.6 nM O H CH Geosmin earthy 0.1 nM Anosmics Normal subjects 70 60 Frequency (%) N O N O 2-trans-6-cisNonadienal cucumber 0.07 nM β-Ionone violet 0.03 nM 2-Isobutyl-3methoxypyrazine bell pepper 0.01 nM ceive distinct odorant molecules as a particular identifiable smell. Thus, coconuts, violets, cucumbers, and bell peppers all have a unique odor generated by a specific molecule. Most naturally occurring odors, however, are blends of several odorant molecules, even though they are typically experienced as a single smell (such as the perceptions elicited by perfumes or the bouquet of a wine). Psychologists and neurologists have developed a variety of tests that measure the ability to detect common odors. Although most people are able to consistently identify a broad range of test odorants, others fail to identify one or more common smells (Figure 14.3). Such chemosensory deficits, called anosmias, are often restricted to a single odorant, suggesting that a specific element in the olfactory system, either an olfactory receptor gene (see below) or genes that control expression or function of a specific odorant receptor gene, is inactivated. Nevertheless, genetic analysis of anosmic individuals has yet to confirm this possibility. Anosmias often target perception of distinct, noxious odorants. About 1 person in 1000 is insensitive to butyl mercaptan, the foul-smelling odorant released by skunks. More serious is the inability to 90 80 2,3,6Trichloroanisole moldy 0.1 nM 50 40 30 20 10 0 0 1−2 3 4 5 Number correct 6−7 Figure 14.3 Anosmia is the inability to identify common odors. When subjects are presented with seven common odors (a test frequently used by neurologists), the vast majority of “normal” individuals can identify all seven odors correctly (in this case, baby powder, chocolate, cinnamon, coffee, mothballs, peanut butter, and soap). Some people, however, have difficulty identifying even these common odors. In this example, individuals previously identified as anosmics were presented with the same battery of odors, only a few could identify all of the odors (less than 15%), and more than half could not identify any of the odors. (After Cain and Gent, in Meiselman and Rivlin, 1986.) The Chemical Senses 341 Physiological and Behavioral Responses to Odorants In addition to olfactory perceptions, odorants can elicit a variety of physiological responses. Examples are the visceral motor responses to the aroma of appetizing food (salivation and increased gastric motility) or to a noxious smell (gagging and, in extreme cases, vomiting). Olfaction can also influence reproductive and endocrine functions. Women housed in single-sex dormitories, for instance, have menstrual cycles that tend to be synchronized, a phenomenon that appears to be mediated by olfaction. Volunteers exposed to gauze pads from the underarms of women at different stages of their menstrual cycles also tend to experience synchronized menses, and this synchronization can be disrupted by exposure to gauze pads from men. Olfaction also influences mother–child interactions. Infants recognize their mothers within hours after birth by smell, preferentially orienting toward their mothers’ breasts and showing increased rates of suckling when fed by their mother compared to being fed by other lactating females, or when presented experimentally with their mother’s odor versus that of an unrelated female (see Chapter 23). By the same token, mothers can discriminate their own infant’s odor when challenged with a range of odor stimuli from infants of similar age. In other animals, including many mammals, species-specific odorants called pheromones play important roles in behavior, by influencing social, reproductive, and parenting behaviors (Box A). In rats and mice, odorants thought to be pheromones are detected by G-protein-coupled receptors located at the base of the nasal cavity in distinct, encapsulated chemosensory structures called vomeronasal organs (VNOs). In many mammals, VNOs project to the accessory olfactory bulb, which in turn projects to the hypothalamus (where reproductive activity is generally regulated; see Chapter 29). VNOs are found bilaterally in only 8% of adult humans, and there is no clear 90 80 70 Percent correct detect hydrogen cyanide (1 in 10 people), which can be lethal, or ethyl mercaptan, the chemical added to natural gas to aid in the detection of gas leaks. The ability to identify odors normally decreases with age. If otherwise healthy subjects are challenged to identify a large battery of common odorants, people between 20 and 40 years of age can typically identify about 50–75% of the odors, whereas those between 50 and 70 correctly identify only about 30–45% (Figure 14.4). A more radically diminished or distorted sense of smell can accompany eating disorders, psychotic disorders (especially schizophrenia), diabetes, taking certain medications, and Alzheimer’s disease (all for reasons that remain obscure). Although the loss of human olfactory sensitivity is not usually a source of great concern, it can diminish the enjoyment of food and, if severe, can affect the ability to identify and respond appropriately to potentially dangerous odors such as spoiled food, smoke, or natural gas. The neural substrates for odor processing in humans includes all of the structures identified anatomically as part of the olfactory pathway: the olfactory bulb, pyriform and orbital cortices, amygdala and hypothalamus are all clearly activated by presentation of odorants in functional magnetic resonance images (fMRI) of normal human subjects (Figure 14.1E). Although fMRI cannot resolve differences in the local activity elicited by most individual odors, some clear distinctions have been seen that support corresponding behavioral observations. Furthermore, the decline in olfactory ability with age mentioned above is matched by a decline in the level of activity in olfactory regions of the aging human brain. 60 50 40 30 20 10 20 30 40 50 Age (years) 60 Figure 14.4 Normal decline in olfactory sensitivity with age. The ability to identify 80 common odorants declines markedly between 20 and 70 years of age. (After Murphy, 1986.) 70 342 Chapter Four teen Figure 14.5 Differential patterns of activation in the hypothalamus of normal human female (right) and male (left) subjects after exposure to an estrogen- or androgen-containing odor mix. (From Savic et al., 2001.) (A) Females Anterior hypothalamus (B) Males Posterior hypothalamus indication that these human structures have any significant function. The human genes encoding homologues of pheromone receptors expressed by VNO neurons in other mammals are mostly pseudogenes (i.e., the sequences have been mutated over the course of evolution so that these genes cannot be expressed). Thus, it is unlikely that human pheromone perception, if it exists, is mediated by the vomeronasal system, as is the case in other mammals. Nevertheless, recent observations suggest that exposure to androgen and estrogen-like compounds at concentrations below the level of conscious detection can elicit both behavioral responses and different patterns of brain activation in adult female and male human subjects (Figure 14.5). Thus, although most humans do not process pheromones by the vomeronasal system, other olfactory structures can evidently detect signals that may affect reproductive and other behaviors. The Olfactory Epithelium and Olfactory Receptor Neurons The transduction of olfactory information occurs in the olfactory epithelium, the sheet of neurons and supporting cells that lines approximately half of the nasal cavities. (The remaining surface is lined by respiratory epithelium, which lacks neurons and serves primarily as a protective surface.) The olfactory epithelium includes several cell types (Figure 14.6A). The most important of these is the olfactory receptor neuron, a bipolar cell that gives rise to a small-diameter, unmyelinated axon at its basal surface that transmits olfactory information centrally. At its apical surface, the receptor neuron gives rise to a single dendritic process that expands into a knoblike protrusion from which several microvilli, called olfactory cilia, extend into a thick layer of mucus. The mucus that lines the nasal cavity and controls the ionic milieu of the olfactory cilia is produced by secretory specializations (called Bowman’s glands) distributed throughout the epithelium. When the mucus layer becomes thicker, as during a cold, olfactory acuity decreases significantly. Two other cell classes, basal cells and sustentacular (supporting) cells, are also The Chemical Senses 343 (A) (B) Stimulation of olfactory cilia Stimulation of olfactory soma Record Record Cribriform plate Receptor cell axons (to olfactory bulb) Bowman’s gland Basal cell Pipette Odorant Dividing stem cell Mature receptor cell Developing receptor cell Odorant Supporting cell Mucus Olfactory cilia Odorants Odor Odor Membrane current (pA) Olfactory knob Cilia 0 −100 −200 −300 −400 0 1 2 3 4 5 6 present in the olfactory epithelium. This entire apparatus—mucus layer and epithelium with neural and supporting cells—is called the nasal mucosa. The superficial location of the nasal mucosa allows the olfactory receptor neurons direct access to odorant molecules. Another consequence, however, is that these neurons are exceptionally exposed. Airborne pollutants, allergens, microorganisms, and other potentially harmful substances subject the olfactory receptor neurons to more or less continual damage. Several mechanisms help maintain the integrity of the olfactory epithelium in the face of this trauma. The ciliated cells of the respiratory epithelium, a non-neural epithelium found at the most external aspect of the nasal cavity, warms and moistens the inspired air. In addition, glandular cells throughout the epithelium secrete mucus, which traps and neutralizes potentially harmful agents. In both the respiratory and olfactory epithelium, immunoglobulins are secreted into the mucus, providing an initial defense against harmful antigens, and the sustentacular cells contain enzymes (cytochrome P450s and others) that catabolize organic chemicals and other potentially damaging molecules that enter the nasal cavity. The ultimate solution to this problem, however, is to replace olfactory receptor neurons in a normal cycle of degeneration and regeneration. In rodents, the entire population of olfactory neurons is renewed every 6 to 8 weeks. This feat is accomplished by maintaining among the basal cells a population of precursors (stem cells) that divide to give rise to new receptor neurons (see Figure 14.6A). This naturally occurring regeneration of olfactory receptor cells provides an opportunity to investigate how neural precursor cells can successfully produce new neurons and reconstitute function in the mature central nervous system, a topic of broad clinical interest. Recent evidence suggests that many of the signaling molecules that influence neuronal differentiation, axon outgrowth, and synapse formation during development elsewhere in the nervous system (see Chapters 21 and 22) perform similar functions for regenerating olfactory 0 1 2 3 4 5 6 Time (s) Figure 14.6 Structure and function of the olfactory epithelium. (A) Diagram of the olfactory epithelium showing the major cell types: olfactory receptor neurons and their cilia, sustentacular cells (that detoxify potentially dangerous chemicals), and basal cells. Bowman’s glands produce mucus. Nerve bundles of unmyelinated neurons and blood vessels run in the basal part of the mucosa (called the lamina propria). Olfactory receptor neurons are generated continuously from basal cells. (B) Generation of receptor potentials in response to odors takes place in the cilia of receptor neurons. Thus, odorants evoke a large inward (depolarizing) current when applied to the cilia (left), but only a small current when applied to the cell body (right). (A after Anholt, 1987; B after Firestein et al., 1991.) 344 Chapter Four teen Box A Olfaction, Pheromones, and Behavior in the Hawk Moth Olfactory information guides essential behaviors in virtually all species. The importance of olfactory cues in reproductive behaviors has been particularly well characterized in the hawk moth, Manduca sexta. In Manduca, males identify potential mates by following a plume of pheromones exuded by the female. Similarly, the female uses an olfactory cue—a molecule made by tobacco plants—to identify an appropriate site to lay eggs. These olfactory functions in the moths are sexually dimorphic: Only males respond to female pheromones, and only females detect the olfactory stimulus from the tobacco plant needed for egg-laying. These abilities are mediated by an olfactory system that shares some remarkable similarities with mammalian systems. Male and female moths have different olfactory receptor cells (and associated structures) on their antennae which generate receptor potentials in Male Antennal nerve response to the female-specific pheromones or the tobacco plant odorants. These peripheral receptors project to olfactory recipient structures that are reminiscent of the mammalian olfactory bulb (see figure). The target structure in the moth—called the antennal lobe—is comprised of an array of iterated circuits that are referred to as glomeruli and are surprisingly similar in both structure and function to glomeruli in the mammalian olfactory bulb. In males, the antennal receptor neurons sensitive to the female pheromone project to a distinct subset of glomeruli called the macroglomerular complex. These glomeruli are specifically active in the presence of female pheromone and, if absent, prevent any behavioral response to the female scent. Finally, the development of these sexually dimorphic central circuits is controlled by the periphery. If a male antennae is transplanted to a genotypically female moth, a macroglomerular com- Female Macroglomerular complex Glomeruli Glomeruli Antennal lobe neurons Male and female olfactory glomeruli in the antennal lobe are specialized for odorant-mediated behaviors. The male-specific macroglomerular complex (MCG) is essential for processing the female pheromone. plex develops in the antennal lobe. The female-specific pheromone has been identified, as have several receptor molecules specifically associated with the male or female olfactory pathway, respectively. Not surprisingly, pheromone receptors in the male are members of a special class of seven transmembrane odorant receptors found in other invertebrates and vertebrates. The matching of identified glomeruli with receptor cells expressing specific receptor molecules may be a general rule in olfactory systems. If so, the neurobiology of a sexually dimorphic olfactory behavior in the moth provides an ideal model system in which to study chemosensory processing of specific odorants. References FARKAS, S. R. AND H. H. SHOREY (1972) Chemical trial following by flying insects: A mechanism for orientation to a distant odor source. Science 178: 67–68. MATSUMOTO, S. G. AND J. G. HILDEBRAND (1981) Olfactory mechanisms in the moth Manduca sexta: Response characteristics and morphology of central neurons in the antennal lobe. Proc. Roy. Soc. London B. 213: 249–277. SCHNEIDERMAN, A. M., S. G. MATSUMOTO AND J. G. HILDEBRAND (1982) Trans-sexually grafted antennae influence development of sexually dimorphic neurons in moth brain. Nature 298: 844–846. SCHNEIDERMAN, A. M., J. G. HILDEBRAND, M. M. BRENNAN AND J. H. TUMLINSON (1986) Trans-sexually grafted antennae alter pheromone-directed behavior in a moth. Nature 323: 801–803. STRAUSFELD, N. J. AND J. G. HILDEBRAND (1999) Olfactory systems: Common design, uncommon origin. Curr. Opin. Neurobiol. 9: 634–639. The Chemical Senses 345 receptor neurons in the adult. Understanding how the new olfactory receptor neurons differentiate into functional neurons, extend axons to the brain, and reestablish appropriate functional connections is obviously relevant to stimulating the regeneration of functional connections elsewhere in the brain after injury or disease (see Chapter 24). The Transduction of Olfactory Signals The cellular and molecular machinery for olfactory transduction is located in the cilia of olfactory receptor neurons (see Figure 14.6B). Despite their external appearance, olfactory cilia do not have the cytoskeletal features of motile cilia (the so-called 9 + 2 arrangement of microtubules). Odorant transduction begins with odorant binding to specific receptors on the external surface of cilia. Binding may occur directly, or by way of proteins in the mucus (called odorant binding proteins) that sequester the odorant and are thought to shuttle it to the receptor. Several additional steps then generate a receptor potential by opening ion channels. In mammals, the principal pathway involves cyclic nucleotide-gated ion channels, similar to those found in rod photoreceptors (see Chapter 10). The olfactory receptor neurons express an olfactory-specific G-protein (Golf), which activates an olfactory-specific adenylate cyclase (Figure 14.7A). Both of these proteins are restricted to the knob and cilia, consistent with the idea that odor transduction occurs in these portions of the olfactory receptor neuron. Stimulation of odorant receptor molecules leads to an increase in cyclic AMP (cAMP) which opens channels that permit Na+ and Ca2+ entry (mostly Ca2+), thus depolarizing the neuron. This depolarization, amplified by a Ca2+-activated Cl– current, is conducted passively from the cilia to the axon hillock region of the olfactory receptor neuron, where action potentials are generated and transmitted to the olfactory bulb. Olfactory receptor neurons are especially efficient at extracting a signal from chemosensory noise. Fluctuations in the cAMP concentration in an olfactory receptor neuron could, in theory, cause the receptor cell to be activated in the absence of odorants. Such nonspecific responses do not occur, however, because the cAMP-gated channels are blocked at the resting potential by the high Ca2+ and Mg2+ concentrations in mucus. To overcome this (A) Figure 14.7 Olfactory transduction and olfactory receptor molecules. (A) Odorants in the mucus bind directly (or are shuttled via odorant binding proteins) to one of many receptor molecules located in the membranes of the cilia. This association activates an odorantspecific G-protein (Golf) that, in turn, activates an adenylate cyclase, resulting in the generation of cyclic AMP (cAMP). One target of cAMP is a cation-selective channel that, when open, permits the influx of Na+ and Ca2+ into the cilia, resulting in depolarization. The ensuing increase in intracellular Ca2+ opens Ca2+-gated Cl– channels that provide most of the depolarization of the olfactory receptor potential. The receptor potential is reduced in magnitude when cAMP is broken down by specific phosphodiesterases to reduce its concentration. At the same time, Ca2+ complexes with calmodulin (Ca2+-CAM) and binds to the channel, reducing its affinity for cAMP. Finally, Ca2+ is extruded through the Ca2+/Na+ exchange pathway. (B) The generic structure of putative olfactory odorant receptors. These proteins have seven transmembrane domains, plus a variable cell surface region and a cytoplasmic tail that interacts with Gproteins. As many as 1000 genes encode proteins of similar inferred structure in several mammalian species, including humans. Each gene presumably encodes an odorant receptor that detects a particular set of odorant molecules. (Adapted from Menini, 1999.) (B) Odorant molecule Active Na+/ Ca2+ channel Ca2+-gated Cl− channel Cl− Receptor Active adenylate Ca2+ protein cyclase Na+ γ β Active G-protein GTP Ca2+ ATP cAMP (Second messenger) Na+ N Ca2+ cAMP Golf Na+/Ca2+ exchanger Cl– Ca2+-CAM C Variable amino acids Conserved amino acids 346 Chapter Four teen voltage-dependent block, several channels must be opened at once. This requirement ensures that olfactory receptor neurons fire only in response to stimulation by odorants. Moreover, changes in the odorant concentration change the latency of response, the duration of the response, and/or the firing frequency of individual neurons, each of which provides additional information about the environmental circumstances to the central stations in the system. Finally, like other sensory receptors, olfactory neurons adapt in the continued presence of a stimulus. Adaptation is apparent subjectively as a decreased ability to identify or discriminate odors during prolonged exposure (e.g., decreased awareness of being in a “smoking” room at a hotel as more time is spent there). Physiologically, olfactory receptor neurons indicate adaptation by a reduced rate of action potentials in response to the continued presence of an odorant. Adaptation occurs because of (1) increased Ca2+ binding by calmodulin, which decreases the sensitivity of the channel to cAMP; and (2) the extrusion of Ca2+ through the activation of Na+/Ca2+ exchange proteins, which reduces the depolarizing potential from Ca2+ activated Cl– channels. Odorant Receptors Figure 14.8 Odorant receptor genes. (A) The number of genes that encode odorant receptors in C. elegans, Drosophila, mice, and humans are indicated in the appropriate boxes. In each instance, we see the seven transmembrane domains characteristic of G-protein-coupled receptors (dark regions); however, the comparative size of each domain, plus the intervening cytoplasmic or cell surface domains, varies from species to species. In addition, splice sites (arrowheads) reflect introns in the genomic sequences of the two invertebrates; in contrast, the genes for mammalian odorant receptors lack introns. (B) The distribution of odorant receptor genes in the human genome. The relative number of genes is indicated by the green bar extending toward the right from each chromosome. There are odorant receptor genes on each human chromosome except for chromosomes 20, 22, and the Y chromosome. (A after Dryer, 2000; B after Mombaerts, 2001.) ▼ Olfactory receptor molecules (Figure 14.7B) are homologous to a large family of other G-protein-linked receptors that includes β-adrenergic receptors, muscarinic acetylcholine receptors, and the photopigments rhodopsin and the cone opsins. In all invertebrates and vertebrates examined thus far, odorant receptor proteins have seven membrane-spanning hydrophobic domains, potential odorant binding sites in the extracellular domain of the protein, and the usual ability to interact with G-proteins at the carboxyl terminal region of their cytoplasmic domain. The amino acid sequences for these molecules also show substantial variability, particularly in regions that code for the membrane-spanning domains. The specificity of olfactory signal transduction is presumably the result of this molecular variety of odorant receptor molecules present in the nasal epithelium. The numbers of odorant receptor genes in humans and other animals also varies widely (Figure 14.8A). Analysis of the complete human genome sequence has idenfied approximately 950 odorant receptor genes. In rodents (the mouse has been the animal of choice for such studies because of its well-established genetics), genome analysis has identified about 1500 different odorant receptor genes. Thus, in mammals, odorant receptors are the largest known gene family, representing between 3 and 5% of all genes. (A) C. elegans TM1 TM2 TM3 TM4 TM5 TM6 TM7 1000 D. melanogaster TM1 TM2 TM3 TM4 TM5 TM6 TM7 60 Mammal TM1 TM2 TM3 TM4 TM5 TM6 TM7 Mouse 1500 (B) 20 13 21 6 1 14 7 22 15 X 2 16 8 17 3 9 18 19 10 4 11 5 12 Y Human 950 348 Chapter Four teen (A) (B) Figure 14.9 Odorant receptor gene expression. (A) Individual olfactory receptor neurons labeled immunohistochemically with the olfactory marker protein OMP (green label; OMP is selective for all olfactory receptor neurons) and the olfactory receptor neuron-specific adenylyl cyclase III (red label) that is limited to cilia. The labels are in register with the segregation of signal transduction components to this domain. (B) The distribution of OMP-expressing olfactory receptor neurons in the nasal epithelium of an adult mouse, demonstrated with a reporter transgene. The protuberances oriented diagonally from left to right represent individual turbinates within the olfactory epithelium. The remaining bony and soft-tissue structures of the nose have been dissected away. (C) The distribution of olfactory receptor neurons expressing the I7 odorant receptor. These cells are restricted to a distinct domain or zone in the epithelium. The inset photo shows that odorant receptor-expressing cells are indeed cilia-bearing olfactory receptor neurons. (D) Olfactory receptor neurons expressing the M81 odorant receptor are limited to a zone that is completely distinct from that of the I7 receptor. (A courtesy of C. Balmer and A. LaMantia; B–D from Bozza et al., 2002.) (C) (D) Additional sequence analysis of human and mouse odorant receptor genes, however, suggests that many—around 60% in human and 20% in mouse— are not transcribed. Thus, the numbers of functional odorant receptor proteins are estimated to be around 400 in humans and 1200 in mice. Similar analysis of complete genome sequences from the worm C. elegans and the fruit fly D. melanogaster indicate that there are approximately 1000 odorant receptor genes in the worm, but only about 60 in the fruit fly. The significance of these quite different numbers of odorant receptor genes is not known. Due to the large number of odorant receptor genes, expression in olfactory receptor neurons has only been confirmed for a limited subset (Figure 14.9). Messenger RNAs for different odorant receptor genes are expressed in subsets of olfactory neurons that occur in bilaterally symmetric patches of olfactory epithelium. Additional evidence for odorant receptor gene expression comes from molecular genetic experiments where reporter proteins like β-galactosidase or green fluorescent protein have been inserted into odorant receptor gene loci. In these experiments (done primarily in mice and fruit flies) expression of the reporter protein is limited to individual olfactory receptor neurons and their processes in distinct regions of the olfactory epithelium. Genetic as well as cell biological analysis shows that each olfactory receptor neuron expresses only one or at most a few odorant receptor genes. Thus, different odors must activate molecularly and spatially distinct subsets of olfactory receptor neurons. Olfactory Coding Like other sensory receptor cells, individual olfactory receptor neurons are sensitive to a subset of stimuli. Presumably, depending on the particular olfactory receptor molecules they express, some olfactory receptor neurons exhibit marked selectivity to a particular chemical stimulus, whereas others are activated by a number of different odorant molecules (Figure 14.10A). In addition, olfactory receptor neurons can exhibit different thresholds for a particular odorant. That is, receptor neurons that are inactive at concentrations sufficient to stimulate some neurons are activated when exposed to higher concentrations of the same odorant. These characteristics suggest The Chemical Senses 349 why the perception of an odor can change as a function of its concentration (Figure 14.10B). The relationship between physiological tuning of single olfactory receptor neurons and chemical specificity of single odorant receptor molecules remains unclear. At present, there is only one mammalian odorant receptor molecule, the I7 receptor, whose ligand specificity has been evaluated. This receptor has a high affinity for the aldehyde n-octanal, as well as some affinity for related molecules. While most of the molecular analysis has been done in rodents, humans can perceive n-octanal—it smells like freshly cut grass. Thus, it is possible that ligands for other individual odorant receptors eventually will be found, and these ligands will correspond to perceptually relevant odors. How olfactory receptor neurons represent the identity and concentration of a given odorant is a complex issue that is unlikely to be explained solely by the properties of the primary receptor neurons. Nevertheless, neurons with specific receptors are located in particular parts of the olfactory epithelium. As in other sensory systems, the topographical arrangement of receptor neurons expressing distinct odorant receptor molecules is referred to as space coding, although the meaning of this phrase in the olfactory system is much less clear than in vision, where a topographical map correlates with visual space. The coding of olfactory information also has a temporal dimension. Sniffing, for instance, is a periodic event that elicits trains of action potentials and synchronous activity in populations of neurons. Information conveyed by timing is called temporal coding and occurs in a variety of species (Box B). The contribution of spatial or temporal coding to olfactory perception is not yet known. Figure 14.10 Responses of olfactory receptor neurons to selected odorants. (A) Neuron 1 responds similarly to three different odorants. In contrast, neuron 2 responds to only one of these odorants. Neuron 3 responds to two of the three stimuli. The responses of these receptor neurons were recorded by whole-cell patch clamp recording; downward deflections represent inward currents measured at a holding potential of –55 mV. (B) Responses of a single olfactory receptor neuron to changes in the concentration of a single odorant, isoamyl acetate. The upper trace in each panel (red) indicates the duration of the odorant stimulus; the lower trace the neuronal response. The frequency and number in each panel of action potentials increases as the odorant concentration increases. (A after Firestein, 1992; B after Getchell, 1986.) (B) (A) Neuron 1 CINEOLE ISOAMYL ACETATE ACETOPHENONE Background 0 Odorant on Odorant off – 800 3.6 × 10–7 M Membrane current (pA) Neuron 2 0 – 400 Neuron 3 9.0 × 10–7 M 0 1.8 × 10–6 M – 600 0 Stimulus on 4 6 Stimulus off 0 on 4 off Time (s) 6 0 on 4 off 6 0 1 2 3 4 5 Time (s) 6 7 350 Chapter Four teen Box B Temporal “Coding” of Olfactory Information in Insects Most studies of olfaction in mammals have emphasized the spatial patterns of receptors in the nose and glomeruli in the bulb that are activated by specific odorants. However, beginning with Edgar Adrian’s study of the hedgehog olfactory bulb in 1942, odor-induced temporal oscillations have been described in species as diverse as turtles and primates. A variety of functions have been proposed for these oscillatory phenomena, including identification of odor type and perception of odor intensity. Gilles Laurent and colleagues at California Institute of Technology have recently found that olfaction in insects does show an important temporal component related to behavior. By recording intracellularly from neurons in the antennal lobe in crickets (a structure analogous to the olfactory bulb in mammals; see also Box A) and extracellularly in the mushroom body (analogous to the mammalian pyriform cortex), they found that the projection neurons in the antennal lobe (corresponding to mammalian mitral cells) respond to a given odor with a variety of temporal patterns that differ from odor to odor but are reproducible for the same odor. The figure here shows a schematic representation of these temporal aspects of the odor response of four such projection neurons. The upper panel shows a local field potential recording from the mushroom body (MB) that represents the synaptic activity of many neurons. During presentation of the odor, a pattern of activity is generated by the synchronized firing of many projection neurons. Interestingly, this oscillation at 20–30 Hz is independent of the odor. Each small sphere in the lower panels represents the state of one of the four neurons before, during, and after the application of an odorant. White balls represents a silent or inhibited state, blue balls an active but unsynchronized state, and orange balls an active and synchroOdor off Antennal lobe neurons Field membrane potential (MB) Odor on 1 2 3 4 Time (ms) Temporal coding of olfactory information in insects. (From Laurent et al., 1996.) nized state. The figure shows that at different times during the odor presentation, various neurons are in synchrony and thus contribute at different times to the field potential recorded in the mushroom body. Desynchronizing the neurons has the effect of eliminating the 20–30 Hz oscillation. Desynchronization does not modify the insects’ responses to odors, but eliminates their ability to distinguish among similar odors. These observations suggest that coherent firing among neurons is an important component of olfactory processing in this species, and raise the possibility that temporal coding is a more important aspect of mammalian olfaction than has so far been imagined. References ADRIAN, E. D. (1942) Olfactory reactions in the brain of the hedgehog. J. Physiol. (Lond.) 100: 459–473. FREEMAN, W. J. AND K. A. GRADJSKI (1987) Relation of olfactory EEG to behavior: Factor analysis. Behav. Neurosci. 101: 766–777. KAY L. M. AND G. LAURENT (1999) Odor- and context-dependent modulation of mitral cell activity in behaving rats. Nature Neurosci. 2: 1003–1009. LAM, Y.-W., L. B. COHEN, M. WACHOWIAK AND M. R. ZOCHOWSKI (2000) Odors elicit three different oscillations in the turtle olfactory bulb. J. Neurosci. 202: 749–762. LAURENT, G. (1999) A systems perspective on early olfactory coding. Science 286: 723–728. LAURENT, G., M. WEHR AND H. DAVIDOWITZ (1996) Temporal representation of odors in an olfactory network. J. Neurosci. 15: 3837–3847. STOPFER, M. AND G. LAURENT (1999) Shortterm memory in olfactory network dynamics. Nature 402: 664–668. The Olfactory Bulb Transducing and relaying odorant information centrally from olfactory receptor neurons are only the first steps in processing olfactory signals. As the olfactory receptor axons leave the olfactory epithelium, they coalesce to form a large number of bundles that together make up the olfactory nerve The Chemical Senses 351 (cranial nerve I). Each olfactory nerve projects ipsilaterally to the olfactory bulb on that side, which lies on the ventral anterior aspect of the ipsilateral forebrain. The most distinctive feature of the olfactory bulb is an array of more or less spherical accumulations of neuropil 100–200 µm in diameter called glomeruli, which lie just beneath the surface of the bulb and receive the primary olfactory axons (Figure 14.11A–C). Although a remarkable feature of the mammalian olfactory bulb, glomeruli are found in central nervous system regions that process olfaction in most animals, including insects like Drosophila (Figure 14.11A inset) and the moth (see Box A). In vertebrates, the olfactory bulb comprises several cell and neuropil layers that receive, process, and relay olfactory information. Within each glomerulus, the axons of the receptor neurons contact the apical dendrites of mitral cells, which are the principal projection neurons of the olfactory bulb. The cell bodies of the mitral cells are located in a distinct layer deep to the olfactory glomeruli and, in adults, extend a primary dendrite into a single glomerulus, where the dendrite gives rise to an elaborate tuft of branches onto which the primary olfactory axons synapse (Figure 14.11B and D). Each glomerulus in the mouse (where glomerular connectivity has been studied quantitatively) includes the apical dendrites of approximately 25 mitral cells, which receive innervation from approximately 25,000 olfactory receptor axons. This degree of convergence presumably serves to increase the sensitivity of mitral cells to ensure odor detection, and to increase the signal strength by averaging out uncorrelated noise. Each glomerulus also includes dendritic processes from two other classes of local circuit neurons: tufted cells and periglomerular cells (approximately 50 tufted cells and 25 periglomerular cells contribute to each glomerulus). Although it is generally assumed that these neurons sharpen the sensitivity of individual glomeruli, their function is unclear. Finally, granule cells, which constitute the innermost layer of the vertebrate olfactory bulb, synapse primarily on the basal dendrites of mitral cells within the external plexiform layer (Figure 14.11C,D). These cells lack an identifiable axon, and instead make dendrodendritic synapses on mitral cells. Granule cells are thought to establish local lateral inhibitory circuits as well as participating in synaptic plasticity in the olfactory bulb. In addition, olfactory granule cells and periglomerular cells are among the few classes of neurons in the forebrain that can be replaced throughout life. Granule cell precursors are maintained in a region outside of the olfactory bulb, within the cells that line the ventricles of the cortical hemispheres called the anterior subventricular zone (see Chapter 24). The neural stem cells in this region can give rise to postmitotic cells that then migrate through a region referred to as the rostral migratory stream that spans the territory between the frontal cortex and the olfactory bulb. Once these migrating neurons arrive in the olfactory bulb, they can differentiate into either mature granule or periglomerular cells. Although it is tempting to speculate about the functional significance of this ongoing generation of granule cells, little is known about how these newly generated cells influence olfactory function or odor perception. Bilaterally symmetrical subsets of glomeruli in the olfactory bulb (Figure 14.11E) receive input from olfactory receptor neurons that express distinct odorant receptor molecules. Thus, there is a special zone-to-zone projection between individual glomeruli in the olfactory bulb and groups of olfactory receptor neurons. As already mentioned, however, there is no obvious systematic representation in this arrangement as there is, for example, in the somatic sensory or visual systems. Rather, there is an affinity between widely distributed cells in the olfactory epithelium and a limited ensemble 352 Chapter Four teen (A) (B) (C) Glomeruli External plexiform layer Mitral cell layer Internal plexiform layer Granule cell layer (D) (E) Granule cells Lateral olfactory tract to olfactory cortex Mitral cell Tufted cell Periglomerular (C) cell Glomerulus Cribriform plate Axons Axons of olfactory receptor cells Olfactory receptor cells Olfactory epithelium ▲ The Chemical Senses 353 Figure 14.11 The organization of the mammalian olfactory bulb. (A) When the bulb is viewed from its dorsal surface (visualized here in a living mouse in which the overlying bone has been removed), olfactory glomeruli can be seen. The dense accumulation of dendrites and synapses that constitute glomeruli are stained here with a vital fluorescent dye that recognizes neuronal processes. The inset shows a similar arrangement of glomeruli in the mushroom body (the equivalent of the olfactory bulb) in Drosophila. (B) Among the major neuronal components of each glomerulus are the apical tufts of mitral cells, which project to the pyriform cortex and other bulb targets (see Figure 14.1C). In this image of a coronal section through the bulb, they have been labeled retrogradely by placing the lipophilic tracer Di-I in the lateral olfactory tract. (C) The cellular structure of the olfactory bulb, shown in a Nissl-stained coronal section. The five layers of the bulb are indicated. The glomerular layer includes the tufts of mitral cells, the axon terminals of olfactory receptor neurons, and periglomerular cells that define the margins of each glomerulus. The external plexiform layer is made up of lateral dendrites of mitral cells, cell bodies and lateral dendrites of tufted cells, and dendrites of granule cells that make dendrodendritic synapses with the other dendritic elements. The mitral cell layer is defined by the cell bodies of mitral cells, and mitral cell axons are found in the internal plexiform layer. Finally, granule cell bodies are densely packed into the granule cell layer. (D) Diagram of the laminar and circuit organization of the olfactory bulb, shown in a cutaway view from its medial surface. Olfactory receptor cell axons synapse with mitral cell apical dendritic tufts and periglomerular cell processes within glomeruli. Granule cells and mitral cell lateral dendrites constitute the major synaptic elements of the external plexiform layer. (E) Axons from olfactory receptor neurons that express a particular odorant receptor gene converge on a small subset of bilaterally symmetrical glomeruli. These glomeruli, indicated in the boxed area in the upper panel, are shown at higher magnification in the lower panel. The projections from the olfactory epithelium have been labeled by a reporter transgene inserted by homologous recombination (“knocked in”) into the genetic locus that encodes the particular receptor. (A from LaMantia et al., 1992; B,C from Pomeroy et al., 1990; E from Mombaerts et al., 1996.) of target glomeruli. This arrangement suggests that individual glomeruli respond specifically (or at least selectively) to distinct odorants. Many investigations have confirmed the selective (but not uniquely specific) responsiveness of glomeruli to particular odorants using electrophysiological methods, voltage-sensitive dyes, and, most recently, intrinsic signals that depend on blood oxygenation (Figure 14.12). Such studies have also shown that increasing the odorant concentration increases the activity of individual glomeruli, as well as the number of glomeruli activated. While the exact mechanism by which these distributed patterns of activity represent odor quality and concentration remains unclear, one useful metaphor is to consider the sheet of glomeruli in the olfactory bulb as a bank of lights on a movie marquee: the spatial distribution of active and inactive glomeruli provides a message that is unique for a given odorant at a particular concentration. Central Projections of the Olfactory Bulb Glomeruli in the olfactory bulb are the sole target of olfactory receptor neurons, and thus the only relay—via the axons of mitral and tufted cells—for olfactory information from the periphery to the rest of the brain. The mitral cell axons form a bundle—the lateral olfactory tract—that projects to the accessory olfactory nuclei, the olfactory tubercle, the entorhinal cortex, and portions of the amygdala (see Figure 14.1A). The major target of the olfactory tract is the three-layered pyriform cortex in the ventromedial aspect of 354 Chapter Four teen 0.001% 0.01% Amyl actetate concentration Figure 14.12 Glomerular activity recorded by optical imaging (see Box C in Chapter 11). Dorsal surface of the olfactory bulb in a living rat monitored as increasing concentrations of amyl acetate are presented to the animal. The higher the concentration, the more intense the activity in the particular glomeruli that respond to the odor. The column at left shows the entire dorsal surface of the olfactory bulb; the column at right shows a higher magnification of the individual glomeruli (indicated by the box in the left-hand column). (From Rubin and Katz, 1999.) 0.1% 1% 5% 10% 100% the temporal lobe near the optic chiasm. Neurons in pyriform cortex respond to odors, and mitral cell inputs from glomeruli receiving odorant receptorspecific projections remain partially segregated. The further processing that occurs in this region, however, is not well understood. The axons of pyramidal cells in the pyriform cortex project in turn to several thalamic and hypothalamic nuclei and to the hippocampus and amygdala. Some neurons from pyriform cortex also innervate a region in the orbitofrontal cortex comprising multimodal neurons that respond to olfactory and gustatory stimuli. Information about odors thus reaches a variety of forebrain regions, allowing olfactory cues to influence cognitive, visceral, emotional, and homeostatic behaviors The Organization of the Taste System The taste system, acting in concert with the olfactory and trigeminal systems, indicates whether food should be ingested. Once in the mouth, the chemical constituents of food interact with receptors on taste cells located in epithelial specializations called taste buds in the tongue. The taste cells transduce these stimuli and provide additional information about the identity, concentration, and pleasant or unpleasant quality of the substance. This information also prepares the gastrointestinal system to receive food by causing salivation and swallowing (or gagging and regurgitation if the substance is unpleasant). Information about the temperature and texture of food is transduced and relayed from the mouth via somatic sensory receptors from the trigeminal and other sensory cranial nerves to the thalamus and somatic sensory cortices (see Chapters 8 and 9). Of course, food is not simply The Chemical Senses 355 eaten for nutritional value; “taste” also depends on cultural and psychological factors. How else can one explain why so many people enjoy consuming hot peppers or bitter-tasting liquids such as beer? Like the olfactory system, the taste system includes both peripheral receptors and a number of central pathways (Figure 14.13). Taste cells (the peripheral receptors) are found in taste buds distributed on the dorsal surface of the tongue, soft palate, pharynx, and the upper part of the esophagus (Figure 14.13A; see also Figure 14.14). These cells make synapses with primary sensory axons that run in the chorda tympani and greater superior petrosal branches of the facial nerve (cranial nerve VII), the lingual branch of the glossopharyngeal nerve (cranial nerve IX), and the superior laryngeal branch (A) Gustatory cortex (frontal operculum) Ventral posterior medial nucleus of thalamus VII IX X Nucleus of the solitary tract Gustatory cortex (insula) Tongue Larynx Ventral posterior medial nucleus of thalamus (B) VPM of thalamus Solitary nucleus of brainstem Insula and frontal cortex Axons from the nucleus of the solitary tract Hypothalamus Amygdala Taste buds Cranial (ant. two-thirds of tongue) nerve VII Taste buds Cranial (post. one-third of tongue) nerve IX Taste buds (epiglottis) Cranial nerve X Figure 14.13 Organization of the human taste system. (A) Drawing on the left shows the relationship between receptors in the oral cavity and upper alimentary canal, and the nucleus of the solitary tract in the medulla. The coronal section on the right shows the VPM nucleus of the thalamus and its connection with gustatory regions of the cerebral cortex. (B) Diagram of the basic pathways for processing taste information. 356 Chapter Four teen of the vagus nerve (cranial nerve X) to innervate the taste buds in the tongue, palate, epiglottis, and esophagus, respectively (see Appendix A for a review of the cranial nerves). The central axons of these primary sensory neurons in the respective cranial nerve ganglia project to rostral and lateral regions of the nucleus of the solitary tract in the medulla (Figure 14.13B), which is also known as the gustatory nucleus of the solitary tract complex (recall that the posterior region of the solitary nucleus is the main target of afferent visceral sensory information related to the sympathetic and parasympathetic divisions of the visceral motor system; see Chapter 20). The distribution of these cranial nerves and their branches in the oral cavity is topographically represented along the rostral–caudal axis of the rostral portion of the gustatory nucleus; the terminations from the facial nerve are rostral, the glossopharyngeal are in the mid-region, and those from the vagus nerve are more caudal in the nucleus. Integration of taste and visceral sensory information is presumably facilitated by this arrangement. The caudal part of the nucleus of the solitary tract also receives innervation from subdiaphragmatic branches of the vagus nerve, which control gastric motility. Interneurons connecting the rostral and caudal regions of the nucleus represent the first interaction between visceral and gustatory stimuli. This close relationship of gustatory and visceral information makes good sense, since an animal must quickly recognize if it is eating something that is likely to make it sick, and respond accordingly. Axons from the rostral (gustatory) part of the solitary nucleus project to the ventral posterior complex of the thalamus, where they terminate in the medial half of the ventral posterior medial nucleus. This nucleus projects in turn to several regions of the cortex, including the anterior insula in the temporal lobe and the operculum of the frontal lobe. There is also a secondary cortical taste area in the caudolateral orbitofrontal cortex, where neurons respond to combinations of visual, somatic sensory, olfactory, and gustatory stimuli. Interestingly, when a given food is consumed to the point of satiety, specific orbitofrontal neurons in the monkey diminish their activity to that tastant, suggesting that these neurons are involved in the motivation to eat (or not to eat) particular foods. Finally, reciprocal projections connect the nucleus of the solitary tract via the pons to the hypothalamus and amygdala (see Figure 14.13B). These projections presumably influence appetite, satiety, and other homeostatic responses associated with eating (recall that the hypothalamus is the major center governing homeostasis; see Chapter 20). Taste Perception in Humans Most taste stimuli are nonvolatile, hydrophilic molecules soluble in saliva. Examples include salts such as NaCl needed for electrolyte balance; essential amino acids such as glutamate needed for protein synthesis; sugars such as glucose needed for energy; and acids such as citric acid that indicate the palatability of various foods (oranges, in the case of citrate). Bitter-tasting molecules, including plant alkaloids like atropine, quinine, and strychnine, indicate foods that may be poisonous. Placing bitter compounds in the mouth usually deters ingestion unless one “acquires a taste” for the substance, as for the quinine in tonic water. The taste system encodes information about the quantity as well as the identity of stimuli. In general, the higher the stimulus concentration, the greater the perceived intensity of taste. Threshold concentrations for most ingested tastants are quite high, however. For example, the threshold concentration for citric acid is about 2 mM; for salt (NaCl), 10 mM; and for The Chemical Senses 357 sucrose, 20 mM. (Recall that the perceptual threshold for some odorants is as low as 0.01 nM.) Because the body requires substantial concentrations of salts and carbohydrates, taste cells may respond only to relatively high concentrations of these essential substances in order to promote an adequate intake. Clearly, it is advantageous for the taste system to detect potentially dangerous substances (e.g., bitter-tasting plant compounds that may be noxious or poisonous) at much lower concentrations. Thus the threshold concentration for quinine is 0.008 mM, and for strychnine 0.0001 mM. As in olfaction, gustatory sensitivity declines with age. Adults tend to add more salt and spices to food than children. The decreased sensitivity to salt can be problematic for older people with electrolyte and/or fluid balance problems. Unfortunately, a safe and effective substitute for NaCl has not yet been developed. There is a common misconception that sweet is perceived at the tip of the tongue, salt along its posterolateral edges, sour along the mediolateral edges, and bitter on the back of the tongue. This arrangement was initially proposed in 1901 by Deiter Hanig, who measured taste thresholds for NaCl, sucrose, quinine, and hydrochloric acid (HCl). Hanig never said that other regions of the tongue were insensitive to these chemicals, but only indicated which regions were most sensitive. People missing the anterior part of their tongue (or who have facial nerve lesions) can still taste sweet and salty stimuli. In fact, all of these tastes can be detected over the full surface the tongue (Figure 14.14A). However, different regions of the tongue do have different thresholds. Because the tip of the tongue is most responsive to sweet-tasting compounds, and because these compounds produce pleasurable sensations, information from this region activates feeding behaviors such as mouth movements, salivary secretion, insulin release, and swallowing. In contrast, responses to bitter compounds are greatest on the back of the tongue. Activation of this region by bitter-tasting substances elicits protrusion of the tongue and other protective reactions that prevent ingestion. Sour-tasting compounds elicit grimaces, puckering responses, and massive salivary secretion to dilute the tastant. Based on general agreement across cultures, there are five perceptually distinct categories of taste: salt, sour, sweet, umami (from the Japanese word for delicious, umami refers to savory tastes, including monosodium glutamate and other amino acids), and bitter. However, there are obvious limitations to this classification. People experience a variety of taste sensations in addition to these five, including astringent (cranberries and tea), pungent (hot peppers and ginger), fat, starchy, and various metallic tastes, to name only a few. In addition, mixtures of chemicals may elicit entirely new taste sensations. But even though the “taste code” defined by the five primary taste classes is not yet fully understood, these tastes correspond to distinct classes of receptors in subsets of taste cells. Thus, taste perception is closely linked to the molecular biology of taste transduction. Idiosyncratic Responses to Tastants Taste responses vary among individuals. For example, many people (about 30–40% of the U.S. population) cannot taste the bitter compound phenylthiocarbamide (PTC) but can taste molecules such as quinine and caffeine that also produce bitter sensations. Indeed, humans can be divided into two groups with quite different thresholds for bitter compounds containing the N—C=S group found in PTC. The difference between these individuals is the presence of a single autosomal gene (Ptc) with a dominant (tasters) and 358 Chapter Four teen (A) (B) Papilla Bitter Circumvallate papillae (cranial nerve IX) Sour Foliate papilla Sweet/ umami Taste bud Trench Fungiform papillae (cranial nerve VII) Salty (C) (D) Taste pore Taste cells Synapse Gustatory Basal cell Microvilli afferent axons Figure 14.14 Taste buds and the peripheral innervation of the tongue. (A) Distribution of taste papillae on the dorsal surface of the tongue. Different responses to sweet, salty, sour, and bitter tastants recorded in the three cranial nerves that innervate the tongue and epiglottis are indicated at left. (B) Diagram of a circumvallate papilla showing location of individual taste buds. (C) Light micrograph of a taste bud. (D) Diagram of a taste bud, showing various types of taste cells and the associated gustatory nerves. The apical surface of the receptor cells have microvilli that are oriented toward the taste pore. (C from Ross, Rommell and Kaye, 1995.) a recessive (non-tasters) allele. Interestingly, people who are extremely sensitive to PTC or its analogues—so-called called “supertasters”—have more taste buds than normal and tend to avoid certain foods such as grapefruit, green tea, and broccoli, all of which contain bitter-tasting compounds. Thus, an individual’s genetic makeup with respect to taste receptors has implications for diet, and even health. The relationship between taste perception and the molecular character of tastants is also variable. A number of quite different compounds taste sweet to humans. These include saccharides (glucose, sucrose, and fructose), organic anions (saccharin), amino acids (aspartame, or Nutrasweet®), L-phenyalanine methyl ester, and proteins (monellin and thaumatin). People can distinguish among different sweeteners, and some find saccharin to have a bitter-tasting component. One reason for such discrimination is that some of these compounds activate separate receptors. For example, saccharides activate cAMP pathways, whereas nonsac- The Chemical Senses 359 charide sweeteners such as amino acids activate IP3 pathways. Thus the perceptual experience of “sweet” encompasses much more than the taste of sucrose. It can be elicited by various sensory transduction mechanisms, and may generate sensory qualities different from those generated by sucrose. Taste sensitivity for salt also relies on a number of mechanisms. Not all salts, or even all monovalent chloride salts, activate the same pathway. Psychophysical studies have shown that amiloride, a diuretic that blocks Na+ entry through amiloride-sensitive Na+ channels, decreases the taste intensity of NaCl and LiCl, but not KCl. Although LiCl tastes salty, it cannot be used as a substitute for NaCl because it has profound effects on the central nervous system—clinically, LiCl is used to treat bipolar disorders. Sodium succinate, NH4Cl, and CsCl do not taste exclusively salty. Indeed, CsCl has a bitter or salty-bitter taste that probably arises from the inhibition of K+ channels. Additional evidence for a distinct receptor for NaCl comes from developmental studies. Infants up to 4 months old can distinguish between water and sucrose (and lactose), water and acid, and water and bitter tastants, but they cannot distinguish between water and a 0.2 M NaCl solution. Thus, either the receptor for Na+ has not yet been expressed, or, if expressed, it is not yet functional. Infants between the ages of 4 and 6 months, however, can discriminate between NaCl solutions and water, and children can detect the full salty taste of NaCl at about 4 years of age. Clearly, a given individual’s perception of tastants results from many idiosyncracies of the taste system. These idiosyncracies may underlie personal preferences and aversions that lead to individual variation in ingestive behaviors (eating and drinking). The French aphorism chacun à son goût (“each to his own taste”) reflects not only individual preferences but the biology of the taste-sensing system. The Organization of the Peripheral Taste System Approximately 4000 taste buds in humans are distributed throughout the oral cavity and upper alimentary canal. Taste buds are about 50 mm wide at their base and approximately 80 mm long, each containing 30 to 100 taste cells (the sensory receptor cells), plus a few basal cells (Figure 14.14B–D). About 75% percent of all taste buds are found on the dorsal surface of the tongue in small elevations called papillae (see Figure 14.14A). There are three types of papillae: fungiform (which contain about 25% of the total number of taste buds), circumvallate (which contain 50% of the taste buds), and foliate (which contain 25%). Fungiform papillae are found only on the anterior two-thirds of the tongue; the highest density (about 30/cm2) is at the tip. Fungiform papillae have a mushroom-like structure (hence their name) and typically have about 3 taste buds at their apical surface. There are 9 circumvallate papillae arranged in a chevron at the rear of the tongue. Each consists of a circular trench containing about 250 taste buds along the trench walls. Two foliate papillae are present on the posterolateral tongue, each having about 20 parallel ridges with about 600 taste buds in their walls. Thus, chemical stimuli on the tongue first stimulate receptors in the fungiform papillae and then in the foliate and circumvallate papillae. Tastants subsequently stimulate scattered taste buds in the pharynx, larynx, and upper esophagus. Taste cells in individual taste buds (see Figure 14.14C,D) synapse with primary afferent axons from branches of three cranial nerves: the facial (VII), glossopharyngeal (IX), and vagus (X) nerves (see Figure 14.13). The taste cells in fungiform papillae on the anterior tongue are innervated exclusively by the 360 Chapter Four teen chorda tympani branch of the facial nerve; in circumvallate papillae, the taste cells are innervated exclusively by the lingual branch of the glossopharyngeal nerve; and in the palate they are innervated by the greater superior petrosal branch of the facial nerve. Taste buds of the epiglottis and esophagus are innervated by the superior laryngeal branch of the vagus nerve. The initiating events of chemosensory transduction occur in the taste cells, which have receptors on microvilli that emerge from the apical surface of the taste cell (see Figure 14.14D and 14.15). The apical surfaces of individual taste cells in taste buds are clustered in a small opening (about 1 mm) near the surface of the tongue called a taste pore. The synapses that relay the receptor activity are made onto the afferent axons of the various cranial nerves at the basal surface. Like olfactory receptor neurons (and presumably for the same reasons), taste cells have a lifetime of only about 2 weeks and are normally regenerated from basal cells. Taste Receptors and the Transduction of Taste Signals The major perceptual categories of taste—salty, sour, sweet, umami, and bitter—are represented by five distinct classes of taste receptors. These receptors are found in the apical microvilli of taste cells. Salty and sour tastes are primarily elicited by ionic stimuli such as the positively charged ions in salts (like Na+ from NaCl), or the H+ in acids (acetic acid, for example, which gives vinegar its sour taste). These ions in salty and sour tastants initiate sensory transduction via specific ion channels: the amiloride-sensitive Na+ channel for salty tastes, and an H+-sensitive, cation-selective channel for sour (Figures 14.15 and 14.16). The receptor potential generated by the positive inward current carried either by Na+ for salty or H+ for sour depolarizes the taste cell. This initial depolarization leads to the activation of voltagegated Na+ channels in the basolateral aspect of the taste cell. This additional depolarization activates voltage-gated Ca2+ channels, leading to the release of neurotransmitter from the basal aspect of the taste cell and the activation of action potentials in ganglion cell axons (Figure 14.15). In humans and other mammals, sweet and amino acid (umami) receptors are heteromeric G-protein-coupled receptors that share a common seventransmembrane receptor subunit called T1R3, which is paired with the T1R2 seven-transmembrane receptor for perception of sweet, or with the T1R1 receptor for amino acids (Figure 14.16). The T1R2 and T1R1 receptors are expressed in different subsets of taste cells, indicating that there are, respectively, sweet- and amino acid-selective cells in the taste buds (see Figure 14.17). Upon binding sugars or other sweet stimuli, the T1R2/T1R3 receptor initiates a G-protein-mediated signal transduction cascade that leads to the activation of the phospholipase C isoform PLCβ2, leading in turn to increased concentrations of inositol triphosphate (IP3) and to the opening of TRP channels (specifically the TRPM5 channel), which depolarizes the taste cell via increased intracellular Ca2+. Similarly, the T1R1/T1R3 receptor is broadly tuned to the 20 standard L-amino acids found in proteins (but not to their D-amino acid enantiomers). Transduction of amino acid stimuli via the T1R1/T1R3 receptor also reflects G-protein-coupled intracellular signaling leading to PLCβ2-mediated activation of the TRPM5 channel and depolarization of the taste cell (see Figure 14.16). Another family of G-protein-coupled receptors known as the T2R receptors transduce bitter tastes. There are approximately 30 T2R subtypes en- The Chemical Senses 361 Salt, acids (sour) Sweet, bitter, umami (amino acid) Receptor Ion channel Apical domain G-protein Second messengers TRPM5 channel Endoplasmic reticulum Na+ channel Depolarization Na+ K+ channel Basolateral domain Ca2+ K+ Ca2+ Ca2+ channel Ca2+ Transmitter release Primary sensory neuron Serotonin receptor Action potential coded by 30 genes in humans and other mammals, and multiple T2R subtypes are expressed in single taste cells. Nevertheless, T2R receptors are not expressed in the same taste cells as T1R1, 2, and 3 receptors. Thus, the receptor cells for bitter tastants are presumably a distinct class. Although the transduction of bitter stimuli relies on a similar mechanism to that for sweet and amino acid tastes, a taste cell-specific G-protein, gustducin, is found primarily in T2R-expressing taste cells and apparently contributes to the transduction of bitter tastes. The remaining steps in bitter transduction are similar to those for sweet and amino acids: PLCβ2-mediated activation of TRPM5 channels depolarizes the taste cell, resulting in the release of neurotransmitter at the synapse between the taste cell and sensory ganglion cell axon. Figure 14.15 Basic components of sensory transduction in taste cells. Taste cells are polarized epithelial cells with an apical and a basolateral domain separated by tight junctions. Tastant-transducing channels (salt and sour) and Gprotein-coupled receptors (sweet, amino acid, and bitter) are limited to the apical domain. Intracellular signaling components that are coupled to taste receptor molecules (G-proteins and various second messenger-related molecules) are also enriched in the apical domain. Voltage-regulated Na+, K+, and Ca2+ channels that mediate release of neurotransmitter from presynaptic specializations at the base of the cell onto terminals of peripheral sensory afferents are limited to the basolateral domain, as is endoplasmic reticulum that also modulates intracellular Ca2+ concentration and contributes to the release of neurotransmitter. The neurotransmitter serotonin, among others, is found in taste cells, and serotonin receptors are found on the sensory afferents. Finally, the TRPM5 channel, which facilitates G-protein-coupled receptor-mediated depolarization, is expressed in taste cells. Its localization to apical versus basal domains is not yet known. 362 Chapter Four teen Figure 14.16 Molecular mechanisms of taste transduction via ion channels and G-protein-coupled receptors. Cation selectivity of the amiloride-sensitive Na+ versus the H+-sensitive proton channel provides the basis for specificity of salt and sour tastes. In each case, positive current via the cation channel leads to depolarization of the cell. For sweet, amino acid (umami), and bitter tastants, different classes of G-protein-coupled receptors mediate transduction. For sweet tastants, heteromeric complexes of the T1R2 and T1R3 receptors transduce stimuli via a PLCβ2-mediated, IP3-dependent mechanism that leads to activation of the TRPM5 Ca2+ channel. For amino acids, heteromeric complexes of T1R1 and T1R3 receptors transduce stimuli via the same PLCβ2/IP3/TRPM5-dependent mechanism. Bitter tastes are transduced via a distinct set of G-protein-coupled receptors, the T2R receptor subtypes.The details of T2R receptors are less well established; however, they apparently associate with the taste cellspecific G-protein gustducin, which is not found in sweet or amino acid receptor-expressing taste cells. Nevertheless, stimulus-coupled depolarization for bitter tastes relies upon the same PLCβ2/IP3/TRPM5-dependent mechanism used for sweet and amino acid taste transduction. Salt Acids (sour) Amiloride-sensitive Na+ channel H+-sensitive cation channel Na+ H+ Sweet Amino acids (umami) T1R2 T1R3 T1R1 TRPM5 channel T1R3 Ca2+ β α γ α G-protein β γ α PLCβ2 IP3 G-protein Bitter TRPM5 channel T2R α β Ca2+ γ Gustducin α PLCβ2 IP3 Neural Coding in the Taste System In the taste system, neural coding refers to the way that the identity, concentration, and “hedonic” (pleasurable or aversive) value of tastants is represented in the pattern of action potentials relayed to the brain. Neurons in the taste system (or in any other sensory system) might be specifically “tuned” to respond with a maximal change in electrical activity to a single taste stimulus. Such tuning is thought to rely on specificity at the level of the receptor cells, as well as on the maintenance of separate channels for the relay of this information from the periphery to the brain. This sort of coding scheme is referred to as a labeled line code, since responses in specific cells presumably correspond to distinct stimuli. The segregated expression of sweet, amino acid, and bitter receptors in different taste cells (Figure 14.17) is consistent with labeled line coding. The Chemical Senses 363 The results of molecular genetic experiments in mice are consistent with a labeled line code. Initial support came from studies in which the genes that specify the sweet and amino acid heteromeric receptors (T1R2 and T1R1) were inactivated in mice. Such mice lack behavioral responses to a broad range of sweet or amino acid stimuli, depending on the gene that has been inactivated. Moreover, recordings of electrical activity in the relevant branches of cranial nerves VII, IX, or X showed that action potentials in response to sweet or amino acid stimuli were lost in parallel with the genetic mutation and behavioral change. Finally, these deficits in transduction and perception were unchanged at a broad range of concentrations, indicating that the molecular specificity of each receptor is quite rigid—the remaining receptors could not respond, even at high concentrations of sweet or amino acid stimuli. These observations suggest that sweet and amino acid transduction and perception depend on labeled lines from the periphery. Bitter taste proved harder to analyze because of the larger number of T2R bitter receptors. To circumvent this challenge, Charles Zuker, Nicholas Ryba and colleagues took advantage of the shared aspects of intracellular signaling for sweet, amino acid, and bitter tastes (see Figure 14.16). Thus, if the genes for either the TRPM5 channel or PLCβ2 are inactivated, behavioral and physiological responses to sweet, amino acid, and bitter stimuli are abolished while salty and sour perception (and the related physiological responses) remain (Figure 14.17). To evaluate whether taste cells expressing the T2R family of receptors provide a labeled line for bitter tastes, PLCβ2 was selectively reexpressed in T2R-expressing taste cells in a PLCb2 mutant mouse. Thus, in these mice, only the taste cells that normally express the T2R subset of taste cells (which expresses most of the T2R receptors in concert) can now tranduce taste signals. If these cells provide a labeled line for bitter tastes, the “rescued” mice (i.e., those expressing PLCβ2 in T2R cells) should regain their perceptual and physiological responses to bitter taste, but not to sweet or amino acid tastes. This was indeed the result of the experiment—behavioral and physiological responses to bitter tastes, but not sweet or amino acid tastes, were restored to normal levels (see Figure 14.17). Evidently, taste coding for sweet, amino acid, and bitter—as judged by taste perception and the related neural activity in peripheral nerves—reflects labeled lines established by the identity of the taste receptor proteins and the subsets of taste cells that express them. These observations support the labeled line hypothesis for primary tastes; however, they do not provide a full account of how either primary or complex tastes are represented in patterns of neural activity in central stations of the taste system (e.g., the solitary nucleus, the thalamus, or the insular cortex). Indeed, little is known about the representation of taste information in the CNS, either at the level of recordings from individual cells or the representation of tastes across an ensemble of neurons in relevant areas of the brainstem, thalamus, or cortex. Trigeminal Chemoreception The third of the major chemosensory systems, the trigeminal chemosensory system, consists of polymodal nociceptive neurons and their axons in the trigeminal nerve (cranial nerve V) and, to a lesser degree, nociceptive neurons whose axons run in the glossopharyngeal and vagus nerves (IX and X) (see Appendix A). These neurons and their associated endings are typically 364 Chapter Four teen (A) (B) (C) Sweet (T1R2) Umami (T1R1) Bitter (T2Rs) Lingual epithelium Taste cells (D) (E) Behavioral response (relative to water) Wild type 6 4 TRPM5–/– 2 30 100 300 Sucrose (mM) 8 Wild type 6 4 TRPM5–/– 2 0 1000 (G) 3 10 30 Glutamate (mM) TRPM5–/– 1.2 0.8 Wild type 0.4 0.0 100 (H) 10 0.01 0.1 1.0 Quinine (mM) 8 Wild type 6 4 PLC –/– b2 T2R– rescue 10 6 4 2 0 0 100 300 Sucrose (mM) 1000 Wild type 8 2 30 10 (I) 12 Behavioral response (relative to water) Behavioral response (relative to water) 10 Behavioral response (relative to water) Behavioral response (relative to water) 8 Behavioral response (relative to water) 12 10 0 (F) PLCb2–/– T2R–rescue PLCb2–/– 1.2 0.8 T2R–rescue 0.4 Wild type 3 10 30 Glutamate (mM) 100 0.0 0.01 0.1 1.0 Quinine (mM) 10 activated by chemicals classified as irritants, including air pollutants (e.g., sulfur dioxide), ammonia (smelling salts), ethanol (liquor), acetic acid (vinegar), carbon dioxide (in soft drinks), menthol (in various inhalants sensation; see Box A in Chapter 9), and capsaicin (the compound in hot chili peppers that elicits the characteristic burning sensation). Irritant-sensitive polymodal nociceptors alert the organism to potentially harmful chemical stimuli that have been ingested, respired, or come in contact with the face, and are closely tied to the trigeminal pain system discussed in Chapter 9. Trigeminal chemosensory information from the face, scalp, cornea, and mucous membranes of the oral and nasal cavities is relayed via the three major sensory branches of the trigeminal nerve: the ophthalmic, maxillary, ▲ The Chemical Senses 365 Figure 14.17 Specificity in peripheral taste coding supports the labeled line hypothesis. (A–C) Sweet (A), amino acid (B), and bitter (C) receptors are expressed in different subsets of taste cells. (D–E) The gene for the TRPM5 channel can be inactivated, or “knocked out,” in mice (TRPM5–/–) and behavioral responses measured with a taste preference test. The mouse is presented with two drinking spouts, one with water and the other with a tastant; behavioral responses are measured as the frequency of licking of the two spouts. For pleasant tastes like sweet (sucrose; D) or umami (glutamate; E) control mice lick the spout with the tastant more frequently, and higher concentrations of tastant leads to increased response (blue lines). In TRPM5–/– mice, this behavioral response (i.e., a preference for the tastant versus water) is eliminated at all concentrations (red lines). (F) For an aversive tastant like bitter quinine, control mice prefer water. This behavioral response—which is initially low—is further diminished with higher quinine concentrations (blue line). Inactivation of TRPM5 also eliminates this behavioral response, regardless of tastant concentration (red line). (G–I) When the PLCb2 gene is knocked out, behavioral response to (G) sucrose, (H) glutamate, and (I) quinine are eliminated (red lines). When PLCb2 is re-expressed only in T2R-expressing taste cells, behavioral responses to sucrose and glutamate are not rescued (dotted green lines in G and H); however, the behavioral response to quinine is restored to normal levels (compare the blue and dotted green lines in I). (After Zhang et al., 2003.) and mandibular (Figure 14.18). The central target of these afferent axons is the spinal component of the trigeminal nucleus, which relays this information to the ventral posterior medial nucleus of the thalamus and thence to the somatic sensory cortex and other cortical areas that process facial irritation and pain (see Chapter 9). Many compounds classified as irritants can also be recognized as odors or tastes; however, the threshold concentrations for trigeminal chemoreception are much higher than those for olfaction or taste. When potentially irritating compounds are presented to people who have lost their sense of smell, perceptual thresholds are found to be approximately 100 times higher than those of normal subjects who perceive the compounds as odors (Figure 14.19). Similar differences occur in identifying chemicals as tastes rather than irritants. Thus, 0.1 M NaCl has a salty taste, but 1.0 M NaCl is perceived as an irritant. Another common irritant is ethanol. When placed on the tongue at moderate temperatures and high concentrations—as in drinking vodka “neat”—ethanol produces a burning sensation. A variety of physiological responses mediated by the trigeminal chemosensory system are triggered by exposure to irritants. These include increased salivation, vasodilation, tearing, nasal secretion, sweating, decreased respiratory rate, and bronchoconstriction. Consider, for instance, the experience that follows the ingestion of capsaicin (see Box A in Chapter 9). These reactions are generally protective in that they dilute the stimulus (tearing, salivation, sweating) and prevent inhaling or ingesting more of it. The receptors for irritants are primarily on the terminal branches of polymodal nociceptive neurons, as described for the pain and temperature systems in Chapter 9. Although these receptors respond to many of the same stimuli as olfactory receptor neurons (e.g., aldehydes, alcohols), they are probably not activated by the same mechanism; for instance, the G-proteincoupled receptors for odorants are found only in olfactory receptor neurons. With the exception of capsaicin and acidic stimuli, both of which activate cation-selective ion channels, little is known about the transduction mechanisms for irritants, or their central processing. Trigeminal ganglion Mandibular nerve Ophthalmic nerve Ethmoid nerve (nose) Ciliary nerves (cornea) Maxillary nerve Lingual nerve (tongue) Inferior alveolar nerve (teeth) Figure 14.18 Diagram of the branches of the trigeminal nerve that innervate the oral, nasal, and ocular cavities. The chemosensitive structures innervated by each trigeminal branch are indicated in parentheses. 366 Chapter Four teen 10 5 10 4 Anosmics 10 3 Threshold (ppm) Figure 14.19 Perceptual thresholds in anosmic and normal subjects for related organic chemicals. In anosmics, these chemicals are only detected as irritants at relatively high concentrations (indicated here in parts per million, ppm); in normal subjects, they are first detected at much lower concentrations as odors. The numbers 1–8 stand for the aliphatic alcohols from methanol to 1-octanol. Perceptual thresholds for three additional common irritants—phenylethyl alcohol (PEA), pyridine (Pyr), and menthol (Men)—are shown at the far right. (After Commetto-Muniz and Cain, 1990.) 10 2 10 1 10 0 Normal subjects 10−1 10−2 10−3 1 2 3 4 5 6 Carbon chain length 7 8 PEA Pyr Men Summary The chemical senses—olfaction, taste, and the trigeminal chemosensory system—all contribute to sensing airborne or soluble molecules from a variety of sources. Humans and other mammals rely on this information for behaviors as diverse as attraction, avoidance, reproduction, feeding, and avoiding potentially dangerous circumstances. Receptor neurons in the olfactory epithelium transduce chemical stimuli into neuronal activity via the stimulation of G-protein-linked receptors; this interaction leads to elevated levels of second messengers such as cAMP, which in turn open cation-selective channels. These events generate receptor potentials in the membrane of the olfactory receptor neuron, and ultimately action potentials in the afferent axons of these cells. Taste receptor cells, in contrast, use a variety of mechanisms for transducing chemical stimuli. These include ion channels that are directly activated by salts and amino acids, and G-protein-linked receptors that activate second messengers. For both smell and taste, the spatial and temporal patterns of action potentials provide information about the identity and intensity of chemical stimuli. The trigeminal chemosensory system responds to irritants by means of mechanisms that are less well understood. Each of the approximately 10,000 odors that humans recognize (and an undetermined number of tastes and irritant molecules) is evidently encoded by the activity of a distinct population of receptor cells in the nose, tongue, and oral cavity. Olfaction, taste, and trigeminal chemosensation all are relayed via specific pathways in the central nervous system. Receptor neurons in the olfactory system project directly to the olfactory bulb. In the taste system, information is relayed centrally by cranial sensory ganglion cells to the solitary nucleus in the brainstem. In the trigeminal chemosensory system, information is relayed via trigeminal ganglion cell projections to the spinal trigeminal nucleus in the brainstem. Each of these structures project in turn to many sites in the brain that process chemosensory information in ways that give rise to some of the most sublime pleasures that humans experience. The Chemical Senses 367 Additional Reading Reviews BUCK, L. B. (2000) The molecular architecture of odor and pheromone sensing in mammals. Cell 100: 611–618. ERICKSON, R. P. (1985) Definitions: A matter of taste. In Taste, Olfaction, and the Central Nervous System. D. W. Pfaff (ed.). New York: Rockefeller University Press, p. 129. HERNESS, M. S. AND T. A. GILBERTSON (1999) Cellular mechanisms of taste transduction. Annu. Rev. Physiol. 61: 873–900. HILDEBRAND, J. G. AND G. M. SHEPHERD (1997) Mechanisms of olfactory discrimination: Converging evidence for common principles across phyla. Annu. Rev. Neurosci. 20: 595–631. KRUGER, L. AND P. W. MANTYH (1989) Gustatory and related chemosensory systems. In Handbook of Chemical Neuroanatomy, Vol. 7, Integrated Systems of the CNS, Part II. A. Björkland, T. Hökfelt and L. W. Swanson (eds.). New York: Elsevier Science, pp. 323–410. LAURENT, G. (1999) A systems perspective on early olfactory coding. Science 286: 723–728. LINDEMANN, B. (1996) Taste reception. Physiol. Rev. 76: 719–766. MENINI, A. (1999) Calcium signaling and regulation in olfactory neurons. Curr. Opin. Neurobiol. 9: 419–426. YAMAMOTO, T., T. NAGAI, T. SHIMURA AND Y. YASOSHIMA (1998) Roles of chemical mediators in the taste system. Jpn. J. Pharmacol. 76: 325–348. ZUTALL, F. AND T. LEINDERS-ZUTALL (2000) The cellular and molecular basis of odor adaptation. Chem. Senses 25: 473–481. Important Original Papers ADLER, E., M. A. HOON, K. L. MUELLER, J. CHRANDRASHEKAR, N. J. P. RYBA AND C. S. ZUCKER (2000) A novel family of mammalian taste receptors. Cell 100: 693–702. ASTIC, L. AND D. SAUCIER (1986) Analysis of the topographical organization of olfactory epithelium projections in the rat. Brain Res. Bull. 16(4): 455–462. AVANET, P. AND B. LINDEMANN (1988) Amilorideblockable sodium currents in isolated taste receptor cells. J. Memb. Biol. 105: 245–255. BUCK, L. AND R. AXEL (1991) A novel multigene family may encode odorant receptors: A molecular basis for odor recognition. Cell 65: 175–187. CATERINA, M. J. AND 8 OTHERS (2000) Impaired nociception and pain sensation in mice lacking the capsaicin receptor. Science 288: 306–313. CHAUDHARI, N., A. M. LANDIN AND S. D. ROPER (2000) A metabotropic glutamate receptor variant functions as a taste receptor. Nature Neurosci. 3: 113–119. GRAZIADEI, P. P. C. AND G. A. MONTI-GRAZIADEI (1980) Neurogenesis and neuron regeneration in the olfactory system of mammals. III. Deafferentation and reinnervation of the olfactory bulb following section of the fila olfactoria in rat. J. Neurocytol. 9: 145–162. KAY, L. M. AND G. LAURENT (2000) Odor- and context-dependent modulation of mitral cell activity in behaving rats. Nature Neurosci. 2: 1003–1009. MALNIC, B., J. HIRONO, T. SATO AND L. B. BUCK (1999) Combinatorial receptor codes for odors. Cell 96: 713–723. MOMBAERTS, P. AND 7 OTHERS (1996) Visualizing an olfactory sensory map. Cell 87: 675–686. NELSON, G., M. A. HOON, J. CHANDRASHEKAR, Y. ZHANG, N. J. P. RYBA AND C. S. ZUKER (2001) Mammalian sweet taste receptors. Cell 106: 381–390. NELSON, G. AND 6 OTHERS. (2002) An aminoacid taste receptor. Nature 416: 199–202. ROLLS, E. T. AND L. L. BAYLIS (1994) Gustatory, olfactory and visual convergence within primate orbitofrontal cortex. J. Neurosci. 14: 5437–5452. SCHIFFMAN, S. S., E. LOCKHEAD AND F. W. MAES (1983) Amiloride reduces taste intensity of salts and sweeteners. Proc. Natl. Acad. Sci. USA 80: 6136–6140. VASSAR, R., S. K. CHAO, R. SITCHERAN, J. M. NUNEZ, L. B. VOSSHALL AND R. AXEL (1994) Topographic organization of sensory projections to the olfactory bulb. Cell 79: 981–991. WONG, G. T., K. S. GANNON AND R. F. MARGOLSKEE (1996) Transduction of bitter and sweet taste by gustducin. Nature 381: 796–800. ZHANG, Y. AND 7 OTHERS. (2003) Coding of sweet, bitter, and umami tastes: Different receptor cells sharing similar signaling pathways. Cell 112: 293–301. ZHAO, G. Q. AND 6 OTHERS (2003) The receptors for mammalian sweet and umami taste. Cell 115: 255–266. Books BARLOW, H. B. AND J. D. MOLLON (1989) The Senses. Cambridge: Cambridge University Press, Chapters 17–19. DOTY, R. L. (ED.) (1995) Handbook of Olfaction and Gustation. New York: Marcel Dekker. FARBMAN, A. I. (1992) Cell Biology of Olfaction. New York: Cambridge University Press. GETCHELL, T. V., L. M. BARTOSHUK, R. L. DOTY AND J. B. SNOW, JR. (1991) Smell and Taste in Health and Disease. New York: Raven Press. SIMON, S. A. AND S. D. ROPER (1993) Mechanisms of Taste Transduction. Boca Raton, FL: CRC Press. Movement and Its Central Control III UNIT III MOVEMENT AND ITS CENTRAL CONTROL Fluorescence photomicrograph showing motor axons (green) and neuromuscular synapses (orange) in transgenic mice that have been genetically engineered to express fluorescent proteins. (Courtesy of Bill Snider and Jeff Lichtman.) 15 16 17 18 19 20 Lower Motor Neuron Circuits and Motor Control Upper Motor Neuron Control of the Brainstem and Spinal Cord Modulation of Movement by the Basal Ganglia Modulation of Movement by the Cerebellum Eye Movements and Sensory Motor Integration The Visceral Motor System Movements, whether voluntary or involuntary, are produced by spatial and temporal patterns of muscular contractions orchestrated by the brain and spinal cord. Analysis of these circuits is fundamental to an understanding of both normal behavior and the etiology of a variety of neurological disorders. This unit considers the brainstem and spinal cord circuitry that make elementary reflex movements possible, as well as the circuits that organize the intricate patterns of neural activity responsible for more complex motor acts. Ultimately, all movements produced by the skeletal musculature are initiated by “lower” motor neurons in the spinal cord and brainstem that directly innervate skeletal muscles; the innervation of visceral smooth muscles is separately organized by the autonomic divisions of the visceral motor system. The lower motor neurons are controlled directly by local circuits within the spinal cord and brainstem that coordinate individual muscle groups, and indirectly by “upper” motor neurons in higher centers that regulate those local circuits, thus enabling and coordinating complex sequences of movements. Especially important are circuits in the basal ganglia and cerebellum that regulate the upper motor neurons, ensuring that movements are performed with spatial and temporal precision. Specific disorders of movement often signify damage to a particular brain region. For example, clinically important and intensively studied neurodegenerative disorders such as Parkinson’s disease, Huntington’s disease, and amyotrophic lateral sclerosis result from pathological changes in different parts of the motor system. Knowledge of the various levels of motor control is essential for understanding, diagnosing, and treating these diseases. Chapter 15 Lower Motor Neuron Circuits and Motor Control Overview Skeletal (striated) muscle contraction is initiated by “lower” motor neurons in the spinal cord and brainstem. The cell bodies of the lower neurons are located in the ventral horn of the spinal cord gray matter and in the motor nuclei of the cranial nerves in the brainstem. These neurons (also called α motor neurons) send axons directly to skeletal muscles via the ventral roots and spinal peripheral nerves, or via cranial nerves in the case of the brainstem nuclei. The spatial and temporal patterns of activation of lower motor neurons are determined primarily by local circuits located within the spinal cord and brainstem. Descending pathways from higher centers comprise the axons of “upper” motor neurons and modulate the activity of lower motor neurons by influencing this local circuitry. The cell bodies of upper motor neurons are located either in the cortex or in brainstem centers, such as the vestibular nucleus, the superior colliculus, and the reticular formation. The axons of the upper motor neurons typically contact the local circuit neurons in the brainstem and spinal cord, which, via relatively short axons, contact in turn the appropriate combinations of lower motor neurons. The local circuit neurons also receive direct input from sensory neurons, thus mediating important sensory motor reflexes that operate at the level of the brainstem and spinal cord. Lower motor neurons, therefore, are the final common pathway for transmitting neural information from a variety of sources to the skeletal muscles. Neural Centers Responsible for Movement The neural circuits responsible for the control of movement can be divided into four distinct but highly interactive subsystems, each of which makes a unique contribution to motor control (Figure 15.1). The first of these subsystems is the local circuitry within the gray matter of the spinal cord and the analogous circuitry in the brainstem. The relevant cells include the lower motor neurons (which send their axons out of the brainstem and spinal cord to innervate the skeletal muscles of the head and body, respectively) and the local circuit neurons (which are the major source of synaptic input to the lower motor neurons). All commands for movement, whether reflexive or voluntary, are ultimately conveyed to the muscles by the activity of the lower motor neurons; thus these neurons comprise, in the words of the great British neurophysiologist Charles Sherrington, the “final common path” for movement. The local circuit neurons receive sensory inputs as well as descending projections from higher centers. Thus, the circuits they form provide much of the coordination between different muscle groups that is 371 372 Chapter Fifteen Figure 15.1 Overall organization of neural structures involved in the control of movement. Four systems—local spinal cord and brainstem circuits, descending modulatory pathways, the cerebellum, and the basal ganglia— make essential and distinct contributions to motor control. DESCENDING SYSTEMS Upper Motor Neurons Motor Cortex Planning, initiating, and directing voluntary movements BASAL GANGLIA Gating proper initiation of movement Brainstem Centers Basic movements and postural control CEREBELLUM Sensory motor coordination Local circuit neurons Lower motor neuron integration Motor neuron pools Lower motor neurons SPINAL CORD AND BRAINSTEM CIRCUITS Sensory inputs SKELETAL MUSCLES essential for organized movement. Even after the spinal cord is disconnected from the brain in an experimental animal such as a cat, appropriate stimulation of local spinal circuits elicits involuntary but highly coordinated limb movements that resemble walking. The second motor subsystem consists of the upper motor neurons whose cell bodies lie in the brainstem or cerebral cortex and whose axons descend to synapse with the local circuit neurons or, more rarely, with the lower motor neurons directly. The upper motor neuron pathways that arise in the cortex are essential for the initiation of voluntary movements and for complex spatiotemporal sequences of skilled movements. In particular, descending projections from cortical areas in the frontal lobe, including Brodmann’s area 4 (the primary motor cortex), the lateral part of area 6 (the lateral premotor cortex), and the medial part of area 6 (the medial premotor cortex) are essential for planning, initiating, and directing sequences of voluntary movements. Upper motor neurons originating in the brainstem are responsible for regulating muscle tone and for orienting the eyes, head, and body with respect to vestibular, somatic, auditory, and visual sensory information. Their contributions are thus critical for basic navigational movements, and for the control of posture. The third and fourth subsystems are complex circuits with output pathways that have no direct access to either the local circuit neurons or the lower motor neurons; instead, they control movement by regulating the activity of the upper motor neurons. The third and larger of these subsystems, the cerebellum, is located on the dorsal surface of the pons (see Chapter 1). The cerebellum acts via its efferent pathways to the upper motor neurons as a servomechanism, detecting the difference, or “motor error,” between an intended movement and the movement actually performed (see Chapter 19). The cerebellum uses this information about discrepancies to Lower Motor Neuron Circuits and Motor Control 373 mediate both real-time and long-term reductions in these motor errors (the latter being a form of motor learning). As might be expected from this account, patients with cerebellar damage exhibit persistent errors in movement. The fourth subsystem, embedded in the depths of the forebrain, consists of a group of structures collectively referred to as the basal ganglia (see Chapter 1). The basal ganglia suppress unwanted movements and prepare (or “prime”) upper motor neuron circuits for the initiation of movements. The problems associated with disorders of basal ganglia, such as Parkinson’s disease and Huntington’s disease, attest to the importance of this complex in the initiation of voluntary movements (see Chapter 17). Despite much effort, the sequence of events that leads from volitional thought to movement is still poorly understood. The picture is clearest, however, at the level of control of the muscles themselves. It therefore makes sense to begin an account of motor behavior by considering the anatomical and physiological relationships between lower motor neurons and the muscle fibers they innervate. Motor Neuron–Muscle Relationships By injecting individual muscle groups with visible tracers that are transported by the axons of the lower motor neurons back to their cell bodies, the lower motor neurons that innervate each of the body’s skeletal muscles can be seen in histological sections of the ventral horns of the spinal cord. Each lower motor neuron innervates muscle fibers within a single muscle, and all the motor neurons innervating a single muscle (called the motor neuron pool for that muscle) are grouped together into rod-shaped clusters that run parallel to the long axis of the cord for one or more spinal cord segments (Figure 15.2). An orderly relationship between the location of the motor neuron pools and the muscles they innervate is evident both along the length of the spinal cord and across the mediolateral dimension of the cord, an arrangement that in effect provides a spatial map of the body’s musculature. For example, the motor neuron pools that innervate the arm are located in the cervical enlargement of the cord and those that innervate the leg in the lumbar enlargement (see Chapter 1). The mapping, or topography, of motor neuron pools in the mediolateral dimension can be appreciated in a cross section through the cervical enlargement (the level illustrated in Figure 15.3). Thus, neurons that innervate the axial musculature (i.e., the postural muscles of the trunk) are located medially in the cord. Lateral to these cell groups are motor neuron pools innervating muscles located progressively more laterally in the body. Neurons that innervate the muscles of the shoulders (or pelvis, if one were to look at a similar section in the lumbar enlargement; see Figure 15.2) are the next most lateral group, whereas those that innervate the proximal muscles of the arm (or leg) are located laterally to these. The motor neuron pools that innervate the distal parts of the extremities, the fingers or toes, lie farthest from the midline. This spatial organization provides clues about the functions of the descending upper motor neuron pathways described in the following chapter; some of these pathways terminate primarily in the medial region of the spinal cord, which is concerned with postural muscles, whereas other pathways terminate more laterally, where they have access to the lower motor neurons that control movements of the distal parts of the limbs, such as, the toes and the fingers. Two types of lower motor neuron are found in these neuronal pools. Small g motor neurons innervate specialized muscle fibers that, in combina- 374 Chapter Fifteen (A) (B) Medial gastrocnemius injection (C) Soleus injection Dorsal horn L7 S1 Ventral horn Lower motor neurons Figure 15.2 Organization of lower motor neurons in the ventral horn of the spinal cord demonstrated by labeling of their cell bodies following injection of a retrograde tracer in individual muscles. Neurons were identified by placing a retrograde tracer into the medial gastrocnemius or soleus muscle of the cat. (A) Section through the lumbar level of the spinal cord showing the distribution of labeled cell bodies. Lower motor neurons form distinct clusters (motor pools) in the ventral horn. Spinal cord cross sections (B) and a reconstruction seen from the dorsal surface (C) illustrate the distribution of motor neurons innervating individual skeletal muscles in both axes of the cord. The cylindrical shape and distinct distribution of different pools are especially evident in the dorsal view of the reconstructed cord. The dashed lines in (C) represent individual lumbar and sacral spinal cord segments. (After Burke et al., 1977.) tion with the nerve fibers that innervate them, are actually sensory receptors called muscle spindles (see Chapter 8). The muscle spindles are embedded within connective tissue capsules in the muscle, and are thus referred to as intrafusal muscle fibers (fusal means capsular). The intrafusal muscle fibers are also innervated by sensory axons that send information to the brain and spinal cord about the length and tension of the muscle. The function of the γ motor neurons is to regulate this sensory input by setting the intrafusal muscle fibers to an appropriate length (see the next section). The second type of lower motor neuron, called a motor neurons, innervates the extrafusal muscle fibers, which are the striated muscle fibers that actually generate the forces needed for posture and movement. Although the following discussion focuses on the lower motor neurons in the spinal cord, comparable sets of motor neurons responsible for the control of muscles in the head and neck are located in the brainstem. The latter neurons are distributed in the eight motor nuclei of the cranial nerves in the medulla, pons, and midbrain (see Appendix A). Somewhat confusingly, but quite appropriately, these motor neurons in the brainstem are also called lower motor neurons. Lower Motor Neuron Circuits and Motor Control 375 Proximal muscles Distal muscles Figure 15.3 Somatotopic organization of lower motor neurons in a cross section of the ventral horn at the cervical level of the spinal cord. Motor neurons innervating axial musculature are located medially, whereas those innervating the distal musculature are located more laterally. The Motor Unit Most mature extrafusal skeletal muscle fibers in mammals are innervated by only a single α motor neuron. Since there are by far more muscle fibers than motor neurons, individual motor axons branch within muscles to synapse on many different fibers that are typically distributed over a relatively wide area within the muscle, presumably to ensure that the contractile force of the motor unit is spread evenly (Figure 15.4). In addition, this arrangement reduces the chance that damage to one or a few α motor neurons will significantly alter a muscle’s action. Because an action potential generated by a (A) (B) Motor neuron in spinal cord Figure 15.4 The motor unit. (A) Diagram showing a lower motor neuron in the spinal cord and the course of its axon to its target muscle. (B) Each motor neuron synapses with multiple fibers within the muscle. The motor neuron and the fibers it contacts define the motor unit. Cross section through the muscle shows the relatively diffuse distribution of muscle fibers (red dots) contacted by the motor neuron. Muscle fibers innervated by a single motor neuron 376 Chapter Fifteen Figure 15.5 Comparison of the force and fatigability of the three different types of motor units. In each case, the response reflects stimulation of a single motor neuron. (A) Change in muscle tension in response to a single motor neuron action potential. (B) Tension in response to repetitive stimulation of the motor neurons. (C) Response to repeated stimulation at a level that evokes maximum tension. The ordinate represents the force generated by each stimulus. Note the strikingly different fatigue rates. (After Burke et al., 1974.) motor neuron normally brings to threshold all of the muscle fibers it contacts, a single α motor neuron and its associated muscle fibers together constitute the smallest unit of force that can be activated to produce movement. Sherrington was again the first to recognize this fundamental relationship between an α motor neuron and the muscle fibers it innervates, for which he coined the term motor unit. Both motor units and the α motor neurons themselves vary in size. Small α motor neurons innervate relatively few muscle fibers and form motor units that generate small forces, whereas large motor neurons innervate larger, more powerful motor units. Motor units also differ in the types of muscle fibers that they innervate. In most skeletal muscles, the smaller motor units comprise small “red” muscle fibers that contract slowly and generate relatively small forces; but, because of their rich myoglobin content, plentiful mitochondria, and rich capillary beds, such small red fibers are resistant to fatigue (these units are also innervated by relatively small α motor neurons). These small units are called slow (S) motor units and are especially important for activities that require sustained muscular contraction, such as the maintenance of an upright posture. Larger α motor neurons innervate larger, pale muscle fibers that generate more force; however, these fibers have sparse mitochondria and are therefore easily fatigued. These units are called fast fatigable (FF) motor units and are especially important for brief exertions that require large forces, such as running or jumping. A third class of motor units has properties that lie between those of the other two. These fast fatigue-resistant (FR) motor units are of intermediate size and are not quite as fast as FF units. They generate about twice the force of a slow motor unit and, as the name implies, are substantially more resistant to fatigue (Figure 15.5). These distinctions among different types of motor units indicate how the nervous system produces movements appropriate for different circumstances. In most muscles, small, slow motor units have lower thresholds for activation than the larger units and are tonically active during motor acts that require sustained effort (standing, for instance). The thresholds for the large, fast motor units are reached only when rapid movements requiring great force are made, such as jumping. The functional distinctions between (A) (B) (C) 60 100 Fast fatigable Fast fatigable Fast fatigable Percent maximum force 50 Grams of force 40 30 Fast fatigueresistant Fast fatigueresistant 20 0 Fast fatigue-resistant 100 0 Slow 100 10 Slow Slow 0 0 50 100 150 200 250 Time (ms) 300 0 500 1000 Time (ms) 1500 0 0 2 4 Time (min) 6 60 Lower Motor Neuron Circuits and Motor Control 377 the various classes of motor units also explain some structural differences among muscle groups. For example, a motor unit in the soleus (a muscle important for posture that comprises mostly small, slow units) has an average innervation ratio of 180 muscle fibers for each motor neuron. In contrast, the gastrocnemius, a muscle that comprises both small and larger units, has an innervation ratio of ~1000–2000 muscle fibers per motor neuron, and can generate forces needed for sudden changes in body position. More subtle variations are present in athletes on different training regimens. Thus, muscle biopsies show that sprinters have a larger proportion of powerful but rapidly fatiguing pale fibers in their leg muscles than do marathoners. Other differences are related to the highly specialized functions of particular muscles. For instance, the eyes require rapid, precise movements but little strength; in consequence, extraocular muscle motor units are extremely small (with an average innervation ratio of only 3!) and have a very high proportion of muscle fibers capable of contracting with maximal velocity. The Regulation of Muscle Force Increasing or decreasing the number of motor units active at any one time changes the amount of force produced by a muscle. In the 1960s, Elwood Henneman and his colleagues at Harvard Medical School found that progressive increases in muscle tension could be produced by progressively increasing the activity of axons that provide input to the relevant pool of lower motor neurons. This gradual increase in tension results from the recruitment of motor units in a fixed order according to their size. By stimulating either sensory nerves or upper motor pathways that project to a lower motor neuron pool while measuring the tension changes in the muscle, Henneman found that in experimental animals only the smallest motor units in the pool are activated by weak synaptic stimulation. When synaptic input increases, progressively larger motor units that generate larger forces are recruited: As the synaptic activity driving a motor neuron pool increases, low threshold S units are recruited first, then FR units, and finally, at the highest levels of activity, the FF units. Since these original experiments, evidence for the orderly recruitment of motor units has been found in a variety of voluntary and reflexive movements. As a result, this systematic relationship has come to be known as the size principle. An illustration of how the size principle operates for the motor units of the medial gastrocnemius muscle in the cat is shown in Figure 15.6. When the animal is standing quietly, the force measured directly from the muscle tendon is only a small fraction (about 5%) of the total force that the muscle can generate. The force is provided by the S motor units, which make up about 25% of the motor units in this muscle. When the cat begins to walk, larger forces are necessary: locomotor activities that range from slow walking to fast running require up to 25% of the muscle’s total force capacity. This additional need is met by the recruitment of FR units. Only movements such as galloping and jumping, which are performed infrequently and for short periods, require the full power of the muscle; such demands are met by the recruitment of the FF units. Thus, the size principle provides a simple solution to the problem of grading muscle force: The combination of motor units activated by such orderly recruitment optimally matches the physiological properties of different motor unit types with the range of forces required to perform different motor tasks. The frequency of the action potentials generated by motor neurons also contributes to the regulation of muscle tension. The increase in force that 378 Chapter Fifteen 100 80 Percent maximum force Figure 15.6 The recruitment of motor neurons in the cat medial gastrocnemius muscle under different behavioral conditions. Slow (S) motor units provide the tension required for standing. Fast fatigue-resistant (FR) units provide the additional force needed for walking and running. Fast fatigable (FF) units are recruited for the most strenuous activities, such as jumping. (After Walmsley et al., 1978.) Jump Fast fatigable 60 Gallop 40 Run 20 0 Figure 15.7 The effect of stimulation rate on muscle tension. (A) At low frequencies of stimulation, each action potential in the motor neuron results in a single twitch of the related muscle fibers. (B) At higher frequencies, the twitches sum to produce a force greater than that produced by single twitches. (C) At a still higher frequency of stimulation, the force produced is greater, but individual twitches are still apparent. This response is referred to as unfused tetanus. (D) At the highest rates of motor neuron activation, individual twitches are no longer apparent (a condition called fused tetanus). (A) Force Single muscle twitches (5 Hz) Fast fatigueresistant Walk Stand 0 25 50 75 Percent of motor neuron pool recruited Slow 100 occurs with increased firing rate reflects the summation of successive muscle contractions: The muscle fibers are activated by the next action potential before they have had time to completely relax, and the forces generated by the temporally overlapping contractions are summed (Figure 15.7). The lowest firing rates during a voluntary movement are on the order of 8 per second (Figure 15.8). As the firing rate of individual units rises to a maximum of about 20–25 per second in the muscle being studied here, the amount of force produced increases. At the highest firing rates, individual muscle fibers are in a state of “fused tetanus”—that is, the tension produced in individual motor units no longer has peaks and troughs that correspond to the individual twitches evoked by the motor neuron’s action potentials. Under normal conditions, the maximum firing rate of motor neurons is less than that required for fused tetanus (see Figure 15.8). However, the asynchronous firing of different lower motor neurons provides a steady level of input to the muscle, which causes the contraction of a relatively constant number of motor units and averages out the changes in tension due to contractions and relaxations of individual motor units. All this allows the resulting movements to be executed smoothly. (B) Temporal summation (20 Hz) (C) (D) Unfused tetanus (80 Hz) Fused tetanus (100 Hz) Lower Motor Neuron Circuits and Motor Control 379 Figure 15.8 Motor units recorded transcutaneously in a muscle of the human hand as the amount of voluntary force produced is progressively increased. Motor units (represented by the lines between the dots) are initially recruited at a low frequency of firing (8 Hz); the rate of firing for each unit increases as the subject generates more and more force. (After Monster and Chan, 1977.) Unit firing rate (Hz) 24 20 16 12 8 4 1 5 10 50 100 Voluntary force (grams) 500 1000 The Spinal Cord Circuitry Underlying Muscle Stretch Reflexes The local circuitry within the spinal cord mediates a number of sensory motor reflex actions. The simplest of these reflex arcs entails a sensory response to muscle stretch, which provides direct excitatory feedback to the motor neurons innervating the muscle that has been stretched (Figure 15.9). As already mentioned, the sensory signal for the stretch reflex originates in muscle spindles, the sensory receptors embedded within most muscles (see the previous section and Chapter 8). The spindles comprise 8–10 intrafusal fibers arranged in parallel with the extrafusal fibers that make up the bulk of the muscle (Figure 15.9A). Large-diameter sensory fibers, called Ia afferents, are coiled around the central part of the spindle. These afferents are the largest axons in peripheral nerves and, since action potential conduction velocity is a direct function of axon diameter (see Chapters 2 and 3), they mediate very rapid reflex adjustments when the muscle is stretched. The stretch imposed on the muscle deforms the intrafusal muscle fibers, which in turn initiate action potentials by activating mechanically gated ion channels in the afferent axons coiled around the spindle. The centrally projecting branch of the sensory neuron forms monosynaptic excitatory connections with the α motor neurons in the ventral horn of the spinal cord that innervate the same (homonymous) muscle and, via local circuit neurons, forms inhibitory connections with the α motor neurons of antagonistic (heteronymous) muscles. This arrangement is an example of what is called reciprocal innervation and results in rapid contraction of the stretched muscle and simultaneous relaxation of the antagonist muscle. All of this leads to especially rapid and efficient responses to changes in the length or tension in the muscle (Figure 15.9B). The excitatory pathway from a spindle to the α motor neurons innervating the same muscle is unusual in that it is a monosynaptic reflex; in most cases, sensory neurons from the periphery do not contact the lower motor neuron directly but exert their effects through local circuit neurons. This monosynaptic reflex arc is variously referred to as the “stretch,” “deep tendon,” or “myotatic” reflex, and it is the basis of the knee, ankle, jaw, biceps, or triceps responses tested in a routine neurological examination. The tap of the reflex hammer on the tendon stretches the muscle and therefore excites an afferent volley of activity in the Ia sensory axons that innervate the muscle spindles. The afferent volley is relayed to the α motor neurons in the brainstem or spinal cord, and an efferent volley returns to the muscle (see Figure 1.5). Since muscles are always under some degree of 380 Chapter Fifteen (A) Muscle spindle (B) α Motor neuron γ Motor neuron Spindle afferent (Ia sensory neuron) α Motor neuron Ia sensory neuron Muscle spindle Homonymous muscle Synergist Antagonist Capsule surrounding spindle Passive stretch (C) Descending facilitation and inhibition α Motor neuron Disturbance (addition of liquid to glass) Force required to hold glass Muscle Length change in muscle fiber Load − Inhibited + Increase spindle afferent discharge Resistance Spindle receptor ▲ Lower Motor Neuron Circuits and Motor Control 381 Figure 15.9 Stretch reflex circuitry. (A) Diagram of muscle spindle, the sensory receptor that initiates the stretch reflex. (B) Stretching a muscle spindle leads to increased activity in Ia afferents and an increase in the activity of α motor neurons that innervate the same muscle. Ia afferents also excite the motor neurons that innervate synergistic muscles, and inhibit the motor neurons that innervate antagonists (see also Figure 1.5). (C) The stretch reflex operates as a negative feedback loop to regulate muscle length. stretch, this reflex circuit is normally responsible for the steady level of tension in muscles called muscle tone. Changes in muscle tone occur in a variety of pathological conditions, and it is these changes that are assessed by examination of tendon reflexes. In terms of engineering principles, the stretch reflex arc is a negative feedback loop used to maintain muscle length at a desired value (Figure 15.9C). The appropriate muscle length is specified by the activity of descending upper motor neuron pathways that influence the motor neuron pool. Deviations from the desired length are detected by the muscle spindles, since increases or decreases in the stretch of the intrafusal fibers alter the level of activity in the sensory axons that innervate the spindles. These changes lead in turn to adjustments in the activity of the α motor neurons, returning the muscle to the desired length by contracting the stretched muscle and relaxing the opposed muscle group, and by restoring the level of spindle activity to what it was before. The smaller γ motor neurons control the functional characteristics of the muscle spindles by modulating their level of excitability. As was described earlier, when the muscle is stretched, the spindle is also stretched and the rate of discharge in the afferent fibers increased. When the muscle shortens, however, the spindle is relieved of tension, or “unloaded,” and the sensory axons that innervate the spindle might therefore be expected to fall silent during contraction. However, they remain active. The γ motor neurons terminate on the contractile poles of the intrafusal fibers, and the activation of these neurons causes intrafusal fiber contraction—in this way maintaining the tension on the middle (or equatorial region) of the intrafusal fibers where the sensory axons terminate. Thus, co-activation of the α and γ motor neurons allows spindles to function (i.e., send information centrally) at all muscle lengths during movements and postural adjustments. The Influence of Sensory Activity on Motor Behavior The level of γ motor neuron activity often is referred to as γ bias, or gain, and can be adjusted by upper motor neuron pathways as well as by local reflex circuitry. The larger the gain of the stretch reflex, the greater the change in muscle force that results from a given amount of stretch applied to the intrafusal fibers. If the gain of the reflex is high, then a small amount of stretch applied to the intrafusal fibers will produce a large increase in the number of α motor neurons recruited and a large increase in their firing rates; this in turn leads to a large increase in the amount of tension produced by the extrafusal fibers. If the gain is low, a greater stretch is required to generate the same amount of tension in the extrafusal muscle fibers. In fact, the gain of the stretch reflex is continuously adjusted to meet different functional requirements. For example, while standing in a moving bus, the gain of the stretch reflex can be modulated by upper motor neuron pathways to com- 382 Chapter Fifteen (A) α Motor neuron activation without γ (B) α Motor neuron activation with γ Stimulate Stimulate Extrafusal muscle fibers Intrafusal muscle fibers Record Stimulate α motor neuron Stimulate α motor neuron Record Stimulate Spindle afferent Spindle afferent Stimulate γ motor neuron Record Record Ia response ‘‘filled in’’ Contraction Figure 15.10 The role of γ motor neuron activity in regulating the responses of muscle spindles. (A) When α motor neurons are stimulated without activation of γ motor neurons, the response of the Ia fiber decreases as the muscle contracts. (B) When both α and γ motor neurons are activated, there is no decrease in Ia firing during muscle shortening. Thus, the γ motor neurons can regulate the gain of muscle spindles so they can operate efficiently at any length of the parent muscle. (After Hunt and Kuffler, 1951.) Afferent activity Afferent activity Muscle force Muscle force Contraction pensate for the variable changes that occur as the bus stops and starts or progresses relatively smoothly. During voluntary movements, α and γ motor neurons are often co-activated by higher centers to prevent muscle spindles from being unloaded (Figure 15.10). In addition, the level of γ motor neuron activity can be modulated independently of α activity if the context of a movement requires it. In general, the baseline activity level of γ motor neurons is high if a movement is relatively difficult and demands rapid and precise execution. For example, recordings from cat hindlimb muscles show that γ activity is high when the animal has to perform a difficult movement such as walking across a narrow beam. Unpredictable conditions, as when the animal is picked up or handled, also lead to marked increases in γ activity and greatly increased spindle responsiveness. Gamma motor neuron activity, however, is not the only factor that sets the gain of the stretch reflex. The gain also depends on the level of excitability of the α motor neurons that serve as the efferent side of this reflex loop. Thus, in addition to the influence of descending upper motor neuron projections, other local circuits in the spinal cord can change the gain of the stretch reflex by excitation or inhibition of either α or γ motor neurons. Other Sensory Feedback That Affects Motor Performance Another sensory receptor that is important in the reflex regulation of motor unit activity is the Golgi tendon organ. Golgi tendon organs are encapsu- Lower Motor Neuron Circuits and Motor Control 383 lated afferent nerve endings located at the junction of a muscle and tendon (Figure 15.11A; see also Table 9.1). Each tendon organ is innervated by a single group Ib sensory axon (the Ib axons are slightly smaller than the Ia axons that innervate the muscle spindles). In contrast to the parallel arrangement of extrafusal muscle fibers and spindles, Golgi tendon organs are in series with the extrafusal muscle fibers. When a muscle is passively stretched, most of the change in length occurs in the muscle fibers, since they are more elas- (A) MUSCLE PASSIVELY STRETCHED (B) (1) Muscle Spindles Muscle fibers Capsule (2) Golgi Tendon Organs Muscle stretched Muscle stretched Intrafusal muscle fibers Extrafusal muscle fibers Record Record Ib afferent neuron Spindle afferent GTO Axon Golgi tendon organ afferent Golgi tendon organ Afferent activity Collagen fibrils Afferent activity Stretch Stretch Muscle length Muscle length Tendon MUSCLE ACTIVELY CONTRACTED Figure 15.11 Comparison of the function of muscle spindles and Golgi tendon organs. (A) Golgi tendon organs are arranged in series with extrafusal muscle fibers because of their location at the junction of muscle and tendon. (B) The two types of muscle receptors, the muscle spindles (1) and the Golgi tendon organs (2), have different responses to passive muscle stretch (top) and active muscle contraction (bottom). Both afferents discharge in response to passively stretching the muscle, although the Golgi tendon organ discharge is much less than that of the spindle. When the extrafusal muscle fibers are made to contract by stimulation of their motor neurons, however, the spindle is unloaded and therefore falls silent, whereas the rate of Golgi tendon organ firing increases. (B after Patton, 1965.) (1) Muscle Spindles (2) Golgi Tendon Organs Muscle contracted Muscle contracted Record Record Stimulate Stimulate Stimulate α motor neuron Stimulate α motor neuron Spindle afferent Golgi tendon organ afferent Afferent activity Shorten Afferent activity Shorten Muscle length Muscle length 384 Chapter Fifteen Box A Locomotion in the Leech and the Lamprey All animals must coordinate body movements so they can navigate successfully in their environment. All vertebrates, including mammals, use local circuits in the spinal cord (central pattern generators) to control the coordinated movements associated with locomotion. The cellular basis of organized locomotor activity, however, has been most thoroughly studied in an invertebrate, the leech, and a simple vertebrate, the lamprey. Both the leech and the lamprey lack peripheral appendages for locomotion possessed by many vertebrates (limbs, flippers, fins, or their equivalent). Furthermore, their bodies comprise repeating muscle segments (as well as repeating skeletal elements in the lamprey). Thus, in order to move through the water, both animals must coordinate the movement of each segment. They do this by orchestrating a sinusoidal displace- ment of each body segment in sequence, so that the animal is propelled forward through the water. The leech is particularly well-suited for studying the circuit basis of coordinated movement. The nervous system in the leech consists of a series of interconnected segmental ganglia, each with motor neurons that innervate the corresponding segmental muscles (Figure A). These segmental ganglia facilitate electrophysiological studies, because there is a limited number of neurons in each and each neuron has a distinct identity. The neurons can thus be recognized and studied from animal to animal, and their electrical activity correlated with the sinusoidal swimming movements. A central pattern generator circuit coordinates this undulating motion. In the leech, the relevant neural circuit is an ensemble of sensory neurons, interneurons, and motor neurons repeated in each (A) LEECH Posterior sucker (B) LEECH Head EMGV Ventral cell Dorsal muscle Flattener muscle Segmental ganglion Ventral muscle EMGD Dorsal cell To muscles (A) The leech propels itself through the water by sequential contraction and relaxation of the body wall musculature of each segment. The segmental ganglia in the ventral midline coordinate swimming, each ganglion containing a population of identified neurons. (B) Electrical recordings from the ventral (EMGV) and dorsal (EMGD) muscles in the leech and the corresponding motor neurons show a reciprocal pattern of excitation for the dorsal and ventral muscles of a given segment. segmental ganglion that controls the local sequence of contraction and relaxation in each segment of the body wall musculature (Figure B). The sensory neurons detect the stretching and contraction of the body wall associated with the sequential swimming movements. Dorsal and ventral motor neurons in the circuit provide innervation to dorsal and ventral muscles, whose phasic contractions propel the leech forward. Sensory information and motor neuron signals are coordinated by interneurons that fire rhythmically, setting up phasic patterns of activity in the dorsal and ventral cells that lead to sinusoidal movement. The intrinsic swimming rhythm is established by a variety of membrane conductances that mediate periodic bursts of suprathreshold action potentials followed by well-defined periods of hyperpolarization. The lamprey, one of the simplest vertebrates, is distinguished by its clearly Lower Motor Neuron Circuits and Motor Control 385 segmented musculature and by its lack of bilateral fins or other appendages. In order to move through the water, the lamprey contracts and relaxes each muscle segment in sequence (Figure C), which produces a sinusoidal motion, much like that of the leech. Again, a central pattern generator coordinates this sinusoidal movement. Unlike the leech with its segmental ganglia, the lamprey has a continuous spinal cord that innervates its muscle segments. The lamprey spinal cord is simpler than that of other vertebrates, and several classes of identified neurons occupy stereotyped positions. This orderly arrangement again facilitates the identification and analysis of neurons that constitute the central pattern generator circuit. (C) LAMPREY In the lamprey spinal cord, the intrinsic firing pattern of a set of interconnected sensory neurons, interneurons and motor neurons establishes the pattern of undulating muscle contractions that underlie swimming (Figure D). The patterns of connectivity between neurons, the neurotransmitters used by each class of cell, and the physiological properties of the elements in the lamprey pattern generator are now known. Different neurons in the circuit fire with distinct rhythmicity, thus controlling specific aspects of the swim cycle (Figure E). Particularly important are reciprocal inhibitory connections across the midline that coordinate the pattern generating circuitry on each side of the spinal cord. This circuitry in the lamprey thus provides a basis for understanding the cir- Duration of EMG activity in each segmental muscle Anterior Posterior 1 swim cycle (D) LAMPREY Brainstem inputs Record cuits that control locomotion in more complex vertebrates. These observations on pattern generating circuits for locomotion in relatively simple animals have stimulated parallel studies of terrestrial mammals in which central pattern generators in the spinal cord also coordinate locomotion. Although different in detail, terrestrial locomotion ultimately relies on the sequential movements similar to those that propel the leech and the lamprey through aquatic environments, and intrinsic physiological properties of spinal cord neurons that establish rythmicity for coordinated movement. References GRILLNER, S., D. PARKER AND A. EL MANIRA (1998) Vertebrate locomotion: A lamprey perspective. Ann. N.Y. Acad. Sci. 860: 1–18. MARDER, E. AND R. M. CALABRESE (1996) Principles of rhythmic motor pattern generation. Physiol. Rev. 76: 687–717. STENT, G. S., W. B. KRISTAN, W. O. FRIESEN, C. A. ORT, M. POON AND R. M. CALABRESE (1978) Neural generation of the leech swimming movement. Science 200: 1348–1357. (C) In the lamprey, as this diagram indicates, the pattern of activity across segments is also highly coordinated. (D) The elements of the central pattern generator in the lamprey have been worked out in detail, providing a guide to understanding homologous circuitry in more complex spinal cords. (E) As in the leech, different patterns of electrical activity in lamprey spinal neurons (neurons ED and LV in this example) correspond to distinct periods in the sequence of muscle contractions related to the swim cycle. (E) LAMPREY EV IV ID ED Sensory inputs Sensory inputs MV LV LD Midline Ventral root Record ED MD LV Dorsal root 386 Chapter Fifteen tic than the fibrils of the tendon. When a muscle actively contracts, however, the force acts directly on the tendon, leading to an increase in the tension of the collagen fibrils in the tendon organ and compression of the intertwined sensory receptors. As a result, Golgi tendon organs are exquisitely sensitive to increases in muscle tension that arise from muscle contraction but, unlike spindles, are relatively insensitive to passive stretch (Figure 15.11B). The Ib axons from Golgi tendon organs contact inhibitory local circuit neurons in the spinal cord (called Ib inhibitory interneurons) that synapse, in turn, with the α motor neurons that innervate the same muscle. The Golgi tendon circuit is thus a negative feedback system that regulates muscle tension; it decreases the activation of a muscle when exceptionally large forces are generated and this way protects the muscle. This reflex circuit also operates at reduced levels of muscle force, counteracting small changes in muscle tension by increasing or decreasing the inhibition of α motor neurons. Under these conditions, the Golgi tendon system tends to maintain a steady level of force, counteracting effects that diminish muscle force (such as fatigue). In short, the muscle spindle system is a feedback system that monitors and maintains muscle length, and the Golgi tendon system is a feedback system that monitors and maintains muscle force. Like the muscle spindle system, the Golgi tendon organ system is not a closed loop. The Ib inhibitory interneurons also receive synaptic inputs from a variety of other sources, including cutaneous receptors, joint receptors, muscle spindles, and descending upper motor neuron pathways (Figure 15.12). Acting in concert, these inputs regulate the responsiveness of Ib interneurons to activity arising in Golgi tendon organs. Ib inhibitory interneuron Ib afferent Descending pathways Motor neuron Figure 15.12 Negative feedback regulation of muscle tension by Golgi tendon organs. The Ib afferents from tendon organs contact inhibitory interneurons that decrease the activity of α motor neurons innervating the same muscle. The Ib inhibitory interneurons also receive input from other sensory fibers, as well as from descending pathways. This arrangement prevents muscles from generating excessive tension. Flexor muscle Golgi tendon organ + Extensor muscle Lower Motor Neuron Circuits and Motor Control 387 Flexion Reflex Pathways So far, the discussion has focused on reflexes driven by sensory receptors located within muscles or tendons. Other reflex circuitry mediates the withdrawal of a limb from a painful stimulus, such as a pinprick or the heat of a flame. Contrary to what might be imagined given the speed with which we are able to withdraw from a painful stimulus, this flexion reflex involves several synaptic links (Figure 15.13). As a result of activity in this circuitry, stimulation of nociceptive sensory fibers leads to withdrawal of the limb from the source of pain by excitation of ipsilateral flexor muscles and reciprocal inhibition of ipsilateral extensor muscles. Flexion of the stimulated limb is also accompanied by an opposite reaction in the contralateral limb (i.e., the contralateral extensor muscles are excited while flexor muscles are inhibited). This crossed extension reflex provides postural support during withdrawal of the affected limb from the painful stimulus. Like the other reflex pathways, local circuit neurons in the flexion reflex pathway receive converging inputs from several different sources, including other spinal cord interneurons and upper motor neuron pathways. Although the functional significance of this complex pattern of connectivity is unclear, changes in the character of the reflex following damage to descending pathways provides some insight. Under normal conditions, a noxious stimulus is required to evoke the flexion reflex; following damage to descending pathways, however, other types of stimulation, such as squeezing a limb, can sometimes produce the same response. This observation suggests that the descending projections to the spinal cord modulate the responsiveness of the local circuitry to a variety of sensory inputs. Spinal Cord Circuitry and Locomotion The contribution of local circuitry to motor control is not, of course, limited to reflexive responses to sensory inputs. Studies of rhythmic movements such as locomotion and swimming in animal models (Box A) have demonstrated that local circuits in the spinal cord called central pattern generators are fully capable of controlling the timing and coordination of such complex patterns of movement, and of adjusting them in response to altered circumstances (Box B). A good example is locomotion (walking, running, etc.). The movement of a single limb during locomotion can be thought of as a cycle consisting of two phases: a stance phase, during which the limb is extended and placed in contact with the ground to propel humans or other bipeds forward; and a swing phase, during which the limb is flexed to leave the ground and then brought forward to begin the next stance phase (Figure 15.14A). Increases in the speed of locomotion reduce the amount of time it takes to complete a cycle, and most of the change in cycle time is due to shortening the stance phase; the swing phase remains relatively constant over a wide range of locomotor speeds. In quadrupeds, changes in locomotor speed are also accompanied by changes in the sequence of limb movements. At low speeds, for example, there is a back-to-front progression of leg movements, first on one side and then on the other. As the speed increases to a trot, the movements of the right forelimb and left hindlimb are synchronized (as are the movements of the left forelimb and right hindlimb). At the highest speeds (gallop), the movements of the two front legs are synchronized, as are the movements of the two hindlimbs (Figure 15.14B). Given the precise timing of the movement of individual limbs and the coordination among limbs that are required in this process, it is natural to Cutaneous afferent fiber from nociceptor (Aδ) Motor neuron Extensor muscle + + Extensor muscle Flexor muscle Stimulated leg flexes to withdraw Cutaneous receptor Opposite leg extends to support Figure 15.13 Spinal cord circuitry responsible for the flexion reflex. Stimulation of cutaneous receptors in the foot (by stepping on a tack in this example) leads to activation of spinal cord local circuits that withdraw (flex) the stimulated extremity and extend the other extremity to provide compensatory support. 388 Chapter Fifteen Box B The Autonomy of Central Pattern Generators: Evidence from the Lobster Stomatogastric Ganglion A principle that has emerged from studies of central pattern generators is that rhythmic patterns of firing elicit complex motor responses without need of ongoing sensory stimulation. A good example is the behavior mediated by a small group of nerve cells in lobsters and other crustaceans called the stomatogastric ganglion (STG) that controls the muscles of the gut (Figure A). This ensemble of 30 motor neurons and interneurons in the lobster is perhaps the most completely characterized neural circuit known. Of the 30 cells, defined subsets are essential for two distinct rhythmic movements: gastric mill movements that mediate grinding of food by “teeth” in the lobster’s foregut, and pyloric movements that propel food into the hindgut. Phasic firing patterns of the motor neurons and interneurons of the STG are directly correlated with these two rhythmic move- (A) Somatogastric ganglion ments. Each of the relevant cells has now been identified based on its position in the ganglion, and its electrophysiological and neuropharmacological properties characterized (Figures B and C). Patterned activity in the motor neurons and interneurons of the ganglion begins only if the appropriate neuromodulatory input is provided by sensory axons that originate in other ganglia. Depending upon the activity of the sensory axons, neuronal ensembles in the STG produce one of several characteristic rhythmic firing patterns. Once activated, however, the intrinsic membrane properties of identified cells within the ensemble sustain the rhythmicity of the circuit in the absence of further sensory input. Another key fact that has emerged from this work is that the same neurons can participate in different programmed motor activities, as circumstances Motor nerve Dorsal dilator muscle (B) VD IC PY (C) Control Pilocarpine Gastric mill Cardiac sac Pylorus Serotonin Constrictor muscles Proctolin Brain Esophageal Esophagus ganglion Commissural ganglion Pyloric dilator muscle (A) Location of the lobster stomatogastric ganglion in relation to the gut. (B) Subset of identified neurons in the stomatogastric ganglion that generates gastric mill and pyloric activity. The abbreviations indicate individual identified neurons, all of which project to different pyloric muscles (except the AB neuron, which is an interneuron). (C) Recording from one of the neurons, the lateral pyloric or LP neuron, in this circuit showing the different patterns of activity elicited by several neuromodulators known to be involved in the normal synaptic interactions in this ganglion. PD AB Dopamine LP Record Lower Motor Neuron Circuits and Motor Control 389 demand. For example, the subset of neurons producing gastric mill activity overlaps the subset that generates pyloric activity. This economic use of neuronal subsets has not yet been described in the central pattern generators of mammals, but seems likely to be a feature of all such circuits. References HARTLINE, D. K. AND D. M. MAYNARD (1975) Motor patterns in the stomatogastric ganglion of the lobster, Panulirus argus. J. Exp. Biol. 62: 405–420. MARDER, E. AND R. M. CALABRESE (1996) Principles of rhythmic motor pattern generation. Physiol. Rev. 76: 687–717. assume that locomotion is accomplished by higher centers that organize the spatial and temporal activity patterns of the individual limbs. However, following transection of the spinal cord at the thoracic level, a cat’s hindlimbs will still make coordinated locomotor movements if the animal is supported and placed on a moving treadmill (Figure 15.14C). Under these conditions, the speed of locomotor movements is determined by the speed of the treadmill, suggesting that the movement is nothing more than a reflexive response to stretching the limb muscles. This possibility is ruled out, however, by experiments in which the dorsal roots are also sectioned. Although the speed of walking is slowed and the movements are less coordinated than under normal conditions, appropriate locomotor movements are still observed. These and other observations in experimental animals show that the basic rhythmic patterns of limb movement during locomotion are not dependent on sensory input; nor are they dependent on input from descending projections from higher centers. Rather, each limb appears to have its own central pattern generator responsible for the alternating flexion and extension of the limb during locomotion (see Box B). Under normal conditions, the central pattern generators for the limbs are variably coupled to each other by additional local circuits in order to achieve the different sequences of movements that occur at different speeds. Although some locomotor movements can also be elicited in humans following damage to descending pathways, these are considerably less effective than the movements seen in the cat. The reduced ability of the transected spinal cord to mediate rhythmic stepping movements in humans presumably reflects an increased dependence of local circuitry on upper motor neuron pathways. Perhaps bipedal locomotion carries with it requirements for postural control greater than can be accommodated by spinal cord circuitry alone. Whatever the explanation, the basic oscillatory circuits that control such rhythmic behaviors as flying, walking, and swimming in many animals also play an important part in human locomotion. The Lower Motor Neuron Syndrome The complex of signs and symptoms that arise from damage to the lower motor neurons of the brainstem and spinal cord is referred to as the “lower motor neuron syndrome.” In clinical neurology, this constellation of problems must be distinguished from the “upper motor neuron syndrome” that results from damage to the descending upper motor neuron pathways (see Chapter 16 for a discussion of the signs and symptoms associated with damage to upper motor neurons). SELVERSTON, A. I., D. F. RUSSELL AND J. P. MILLER (1976) The stomatogastric nervous system: Structure and function of a small neural network. Progress in Neurobiology 7: 215–290. 390 Chapter Fifteen Swing (A) E3 LH RH LF F Stance E1 Flexion E2 E3 F Extension RF Flexors EMG Extensors (B) LH LF RH RF LH LF RH RF LH LF RH RF LH LF RH RF (C) WALK Level of transection of spinal cord Extensors Flexors TROT Stance Swing PACE GALLOP Time Figure 15.14 The cycle of locomotion for terrestrial mammals (a cat in this instance) is organized by central pattern generators. (A) The step cycle, showing leg flexion (F) and extension (E) and their relation to the swing and stance phases of locomotion. EMG indicates electromyographic recordings. (B) Comparison of the stepping movements for different gaits. Brown bars, foot lifted (swing phase); gray bars, foot planted (stance phase). (C) Transection of the spinal cord at the thoracic level isolates the hindlimb segments of the cord. The hindlimbs are still able to walk on a treadmill after recovery from surgery, and reciprocal bursts of electrical activity can be recorded from flexors during the swing phase and from extensors during the stance phase of walking. (After Pearson, 1976.) Damage to lower motor neuron cell bodies or their peripheral axons results in paralysis (loss of movement) or paresis (weakness) of the affected muscles, depending on the extent of the damage. In addition to paralysis and/or paresis, the lower motor neuron syndrome includes a loss of reflexes (areflexia) due to interruption of the efferent (motor) limb of the sensory motor reflex arcs. Damage to lower motor neurons also entails a loss of muscle tone, since tone is in part dependent on the monosynaptic reflex arc that links the muscle spindles to the lower motor neurons (see also Box D in Chapter 16). A somewhat later effect is atrophy of the affected muscles due to denervation and disuse. The muscles involved may also exhibit fibrillations and fasciculations, which are spontaneous twitches characteristic of single denervated muscle fibers or motor units, respectively. These phenomena arise from changes in the excitability of denervated muscle fibers in the case of fibrillation, and from abnormal activity of injured α motor neurons in the case of fasciculations. These spontaneous contractions can be readily recognized in an electromyogram, providing an especially helpful clinical tool in diagnosing lower motor neuron disorders (Box C). Lower Motor Neuron Circuits and Motor Control 391 Box C Amyotrophic Lateral Sclerosis Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease that affects an estimated 0.05% of the population in the United States. It is also called Lou Gehrig’s disease, after the New York Yankees baseball player who died of the disorder in 1936. ALS is characterized by the slow but inexorable degeneration of α motor neurons in the ventral horn of the spinal cord and brainstem (lower motor neurons), and of neurons in the motor cortex (upper motor neurons). Affected individuals show progressive weakness due to upper and/or lower motor neuron involvement, wasting of skeletal muscles due to lower motor neuron involvement, and usually die within 5 years of onset. Sadly, these patients are condemned to watch their own demise, since the intellect remains intact. No available therapy effectively prevents the inexorable progression of this disease. Approximately 10% of ALS cases are familial, and several distinct familial forms have been identified. An autosomal dominant form of familial ALS (FALS) is caused by mutations of the gene that encodes the cytosolic antioxidant enzyme copper/zinc superoxide dismutase (SOD1). Mutations of SOD1 account for roughly 20% of families with FALS. A rare autosomal recessive, juvenile-onset form is caused by mutations in a protein called alsin, a putative GTPase regulator. Another rare type of FALS consists of a slowly progressive, autosomal dominant, lower motor neuron disease without sensory symptoms, with onset in early adulthood; this form is caused by mutations of a protein named dynactin. How these mutant genes lead to the phenotype of motor neuron disease is uncertain. Defects of axonal transport have long been hypothesized to cause ALS. Evidence for this cause is that transgenic mice with mutant SOD1 exhibit defects in slow axonal transport early in the course of the disease, and that dynactin binds to microtubules and thus that mutant dynactin may modify axonal transport along microtubules. However, whether defective axonal transport is the cellular mechanism by which these Summary Four distinct but highly interactive motor subsystems—local circuits in the spinal cord and brainstem, descending upper motor neuron pathways that control these circuits, the basal ganglia, and the cerebellum—all make essential contributions to motor control. Alpha motor neurons located in the spinal cord and in the cranial nerve nuclei in the brainstem directly link the nervous system and muscles, with each motor neuron and its associated muscle fibers constituting a functional entity called the motor unit. Motor units vary in size, amount of tension produced, speed of contraction, and degree of fatigability. Graded increases in muscle tension are mediated by both the orderly recruitment of different types of motor units and an increase in motor neuron firing frequency. Local circuitry involving sensory inputs, local circuit neurons, and α and γ motor neurons are especially important in the reflexive control of muscle activity. The stretch reflex is a monosynaptic circuit with connections between sensory fibers arising from muscle spindles and the α motor neurons that innervate the same or syner- mutant proteins lead to motor neuron disease remains to be clearly established. Despite these uncertainties, demonstration that mutations of each of these three genes can cause familial ALS has given scientists valuable clues about the molecular pathogenesis of at least some forms of this tragic disorder. References ADAMS, R. D. AND M. VICTOR (2001) Principles of Neurology, 7th Ed. New York: McGrawHill, pp. 1152–1158. HADANO, S. AND 20 OTHERS (2001) A gene encoding a putative GTPase regulator is mutated in familial amyotrophic lateral sclerosis 2. Nature Genetics 29: 166–173. PULS, I. AND 13 OTHERS (2003) Mutant dynactin in motor neuron disease. Nature Genetics 33: 455–456, ROSEN, D. R. AND 32 OTHERS (1993) Mutations in Cu/Zn superoxide dismutase gene are associated with familial amyotrophic lateral sclerosis. Nature 362: 59–62. YANG, Y. AND 16 OTHERS (2001) The gene encoding alsin, a protein with three guaninenucleotide exchange factor domains, is mutated in a form of recessive amyotrophic lateral sclerosis. Nature Genetics 29: 160–165. 392 Chapter Fifteen gistic muscles. Gamma motor neurons regulate the gain of the stretch reflex by adjusting the level of tension in the intrafusal muscle fibers of the muscle spindle. This mechanism sets the baseline level of activity in α motor neurons and helps to regulate muscle length and tone. Other reflex circuits provide feedback control of muscle tension and mediate essential functions such as the rapid withdrawal of limbs from painful stimuli. Much of the spatial coordination and timing of muscle activation required for complex rhythmic movements such as locomotion are provided by specialized local circuits called central pattern generators. Because of their essential role in all of these circuits, damage to lower motor neurons leads to paralysis of the associated muscle and to other changes, including the loss of reflex activity, the loss of muscle tone, and eventually muscle atrophy. Additional Reading Reviews BURKE, R. E. (1981) Motor units: Anatomy, physiology and functional organization. In Handbook of Physiology, V. B. Brooks (ed.). Section 1: The Nervous System. Volume 1, Part 1. Bethesda, MD: American Physiological Society, pp. 345–422. BURKE, R. E. (1990) Spinal cord: Ventral horn. In The Synaptic Organization of the Brain, 3rd Ed. G. M. Shepherd (ed.). New York: Oxford University Press, pp. 88–132. GRILLNER, S. AND P. WALLEN (1985) Central pattern generators for locomotion, with special reference to vertebrates. Annu. Rev. Neurosci. 8: 233–261. HENNEMAN, E. (1990) Comments on the logical basis of muscle control. In The Segmental Motor System, M. C. Binder and L. M. Mendell (eds.). New York: Oxford University Press, pp. 7–10. HENNEMAN, E. AND L. M. MENDELL (1981) Functional organization of the motoneuron pool and its inputs. In Handbook of Physiology, V. B. Brooks (ed.). Section 1: The Nervous System. Volume 1, Part 1. Bethesda, MD: American Physiological Society, pp. 423–507. LUNDBERG, A. (1975) Control of spinal mechanisms from the brain. In The Nervous System, Volume 1: The Basic Neurosciences. D. B. Tower (ed.). New York: Raven Press, pp. 253–265. PATTON, H. D. (1965) Reflex regulation of movement and posture. In Physiology and Biophysics, 19th Ed., T. C. Rugh and H. D. Patton (eds.). Philadelphia: Saunders, pp. 181–206. PEARSON, K. (1976) The control of walking. Sci. Amer. 235 (Dec.): 72–86. PROCHAZKA, A., M. HULLIGER, P. TREND AND N. DURMULLER (1988) Dynamic and static fusimotor set in various behavioral contexts. In Mechanoreceptors: Development, Structure, and Function. P. Hnik, T. Soulup, R. Vejsada and J. Zelena (eds.). New York: Plenum, pp. 417–430. SCHMIDT, R. F. (1983) Motor systems. In Human Physiology. R. F. Schmidt and G. Thews (eds.). Berlin: Springer Verlag, pp. 81–110. Important Original Papers BURKE, R. E., D. N. LEVINE, M. SALCMAN AND P. TSAIRES (1974) Motor units in cat soleus muscle: Physiological, histochemical, and morphological characteristics. J. Physiol. (Lond.) 238: 503–514. BURKE, R. E., P. L. STRICK, K. KANDA, C. C. KIM AND B. WALMSLEY (1977) Anatomy of medial gastrocnemius and soleus motor nuclei in cat spinal cord. J. Neurophysiol. 40: 667–680. HENNEMAN, E., E. SOMJEN, AND D. O. CARPENTER (1965) Excitability and inhibitability of motoneurons of different sizes. J. Neurophysiol. 28: 599–620. HUNT, C. C. AND S. W. KUFFLER (1951) Stretch receptor discharges during muscle contraction. J. Physiol. (Lond.) 113: 298–315. LIDDELL, E. G. T. AND C. S. SHERRINGTON (1925) Recruitment and some other factors of reflex inhibition. Proc. R. Soc. London 97: 488–518. LLOYD, D. P. C. (1946) Integrative pattern of excitation and inhibition in two-neuron reflex arcs. J. Neurophysiol. 9: 439–444. MONSTER, A. W. AND H. CHAN (1977) Isometric force production by motor units of extensor digitorum communis muscle in man. J. Neurophysiol. 40: 1432–1443. WALMSLEY, B., J. A. HODGSON AND R. E. BURKE (1978) Forces produced by medial gastrocnemius and soleus muscles during locomotion in freely moving cats. J. Neurophysiol. 41: 1203–1215. Books BRODAL, A. (1981) Neurological Anatomy in Relation to Clinical Medicine, 3rd Ed. New York: Oxford University Press. SHERRINGTON, C. (1947) The Integrative Action of the Nervous System, 2nd Ed. New Haven: Yale University Press. Chapter 16 Overview The axons of upper motor neurons descend from higher centers to influence the local circuits in the brainstem and spinal cord that organize movements by coordinating the activity of lower motor neurons (see Chapter 15). The sources of these upper motor neuron pathways include several brainstem centers and a number of cortical areas in the frontal lobe. The motor control centers in the brainstem are especially important in ongoing postural control. Each center has a distinct influence. Two of these centers, the vestibular nuclear complex and the reticular formation, have widespread effects on body position. Another brainstem center, the red nucleus, controls movements of the arms; also in the brainstem, the superior colliculus contains upper motor neurons that initiate orienting movements of the head and eyes. The motor and “premotor” areas of the frontal lobe, in contrast, are responsible for the planning and precise control of complex sequences of voluntary movements. Most upper motor neurons, regardless of their source, influence the generation of movements by directly affecting the activity of the local circuits in the brainstem and spinal cord (see Chapter 15). Upper motor neurons in the cortex also control movement indirectly, via pathways that project to the brainstem motor control centers, which, in turn, project to the local organizing circuits in the brainstem and cord. A major function of these indirect pathways is to maintain the body’s posture during cortically initiated voluntary movements. Descending Control of Spinal Cord Circuitry: General Information Some insight into the functions of the different sources of the upper motor neurons is provided by the way the lower motor neurons and local circuit neurons—the ultimate targets of the upper motor neurons—are arranged within the spinal cord. As described in Chapter 15, lower motor neurons in the ventral horn of the spinal cord are organized in a somatotopic fashion: The most medial part of the ventral horn contains lower motor neuron pools that innervate axial muscles or proximal muscles of the limbs, whereas the more lateral parts contain lower motor neurons that innervate the distal muscles of the limbs. The local circuit neurons, which lie primarily in the intermediate zone of the spinal cord and supply much of the direct input to the lower motor neurons, are also topographically arranged. Thus, the medial region of the intermediate zone of the spinal cord gray matter contains the local circuit neurons that synapse with lower motor neurons in the medial part of the ventral horn, whereas the lateral regions of the intermediate zone contain local neurons that synapse primarily with lower motor neurons in the lateral ventral horn. 393 Upper Motor Neuron Control of the Brainstem and Spinal Cord 394 Chapter Sixteen Commissural axons Longdistance local circuit neurons Shortdistance local circuit neurons Motor nuclei (to limb muscles) Motor nuclei (to axial muscles) Figure 16.1 Local circuit neurons that supply the medial region of the ventral horn are situated medially in the intermediate zone of the spinal cord gray matter and have axons that extend over a number of spinal cord segments and terminate bilaterally. In contrast, local circuit neurons that supply the lateral parts of the ventral horn are located more laterally, have axons that extend over a few spinal cord segments, and terminate only on the same side of the cord. Descending pathways that contact the medial parts of the spinal cord gray matter are involved primarily in the control of posture; those that contact the lateral parts are involved in the fine control of the distal extremities. The patterns of connections made by local circuit neurons in the medial region of the intermediate zone are different from the patterns made by those in the lateral region, and these differences are related to their respective functions (Figure 16.1). The medial local circuit neurons, which supply the lower motor neurons in the medial ventral horn, have axons that project to many spinal cord segments; indeed, some project to targets along the entire length of the cord. Moreover, many of these local circuit neurons also have axonal branches that cross the midline in the commissure of the spinal cord to innervate lower motor neurons in the medial part of the contralateral hemicord. This arrangement ensures that groups of axial muscles on both sides of the body act in concert to maintain and adjust posture. In contrast, local circuit neurons in the lateral region of the intermediate zone have shorter axons that typically extend fewer than five segments and are predominantly ipsilateral. This more restricted pattern of connectivity underlies the finer and more differentiated control that is exerted over the muscles of the distal extremities, such as that required for the independent movement of individual fingers during manipulative tasks. Differences in the way upper motor neuron pathways from the cortex and brainstem terminate in the spinal cord conform to these functional distinctions between the local circuits that organize the activity of axial and distal muscle groups. Thus, most upper motor neurons that project to the medial part of the ventral horn also project to the medial region of the intermediate zone; the axons of these upper motor neurons have collateral branches that terminate over many spinal cord segments, reaching medial cell groups on both sides of the spinal cord. The sources of these projections are primarily the vestibular nuclei and the reticular formation (see next section); as their terminal zones in the medial spinal cord gray matter suggest, they are concerned primarily with postural mechanisms (Figure 16.2). In contrast, descending axons from the motor cortex generally terminate in lateral parts of the spinal cord gray matter and have terminal fields that are restricted to only a few spinal cord segments (Figure 16.3). These corticospinal pathways are primarily concerned with precise movements involving more distal parts of the limbs. Two additional brainstem structures, the superior colliculus and the red nucleus, also contribute upper motor neuron pathways to the spinal cord (rubro means red; the adjective is derived from the rich capillary bed that gives the nucleus a reddish color in fresh tissue). The axons arising from the superior colliculus project to medial cell groups in the cervical cord, where they influence the lower motor neuron circuits that control axial musculature of the neck (see Figure 16.2). These projections are particularly important in generating orienting movements of the head (the role of the superior colliculus in the generation of head and eye movements is covered in detail in Chapter 19). The red nucleus projections are also limited to the cervical level of the cord, but these terminate in lateral regions of the ventral horn and intermediate zone (see Figure 16.2). The axons arising from the red nucleus participate together with lateral corticospinal tract axons in the control of the arms. The limited distribution of rubrospinal projections may seem surprising, given the large size of the red nucleus in humans. In fact, Figure 16.2 Descending projections from the brainstem to the spinal cord. Pathways that influence motor neurons in the medial part of the ventral horn originate in the reticular formation, vestibular nucleus, and superior colliculus. Those that influence motor neurons that control the proximal arm muscles originate in the red nucleus and terminate in more lateral parts of the ventral horn. (A) COLLICULOSPINAL TRACT (B) RUBROSPINAL TRACT Superior colliculus Red nucleus Cervical spinal cord (D) VESTIBULOSPINAL TRACTS (C) RETICULOSPINAL TRACT Pontine and medullary reticular formation Cervical spinal cord Lateral and medial vestibular nuclei 396 Chapter Sixteen (B) INDIRECT CORTICAL PROJECTIONS (A) DIRECT CORTICAL PROJECTIONS Primary motor cortex Primary somatic sensory cortex Primary motor cortex Medial and lateral premotor cortex Primary somatic sensory cortex Cerebrum Brainstem Reticular formation Corticoreticulospinal tract Pyramidal decussation Lateral corticospinal tract Spinal cord Medial and lateral premotor cortex ▲ Upper Motor Neuron Control of the Brainstem and Spinal Cord 397 Figure 16.3 Direct and indirect pathways from the motor cortex to the spinal cord. Neurons in the motor cortex that supply the lateral part of the ventral horn (A) to initiate movements of the distal limbs also terminate on neurons in the reticular formation (B) to mediate postural adjustments that support the movement. The reticulospinal pathway terminates in the medial parts of the ventral horn, where lower motor neurons that innervate axial muscles are located. Thus, the motor cortex has both direct and indirect routes by which it can influence the activity of spinal cord neurons. the bulk of the red nucleus in humans is a subdivision that does not project to the spinal cord at all, but relays information from the cortex to the cerebellum (see Chapter 18). Motor Control Centers in the Brainstem: Upper Motor Neurons That Maintain Balance and Posture As described in Chapter 13, the vestibular nuclei are the major destination of the axons that form the vestibular division of the eighth cranial nerve; as such, they receive sensory information from the semicircular canals and the otolith organs that specifies the position and angular acceleration of the head. Many of the cells in the vestibular nuclei that receive this information are upper motor neurons with descending axons that terminate in the medial region of the spinal cord gray matter, although some extend more laterally to contact the neurons that control the proximal muscles of the limbs. The projections from the vestibular nuclei that control axial muscles and those that influence proximal limb muscles originate from different cells and take different routes (called the medial and lateral vestibulospinal tracts). Other upper motor neurons in the vestibular nuclei project to lower motor neurons in the cranial nerve nuclei that control eye movements (the third, fourth, and sixth cranial nerve nuclei). This pathway produces the eye movements that maintain fixation while the head is moving (see Chapters 13 and 19). The reticular formation is a complicated network of circuits located in the core of the brainstem that extends from the rostral midbrain to the caudal medulla and is similar in structure and function to the intermediate gray matter in the spinal cord (see Figure 16.4 and Box A). Unlike the welldefined sensory and motor nuclei of the cranial nerves, the reticular formation comprises clusters of neurons scattered among a welter of interdigitating axon bundles; it is therefore difficult to subdivide anatomically. The neurons within the reticular formation have a variety of functions, including cardiovascular and respiratory control (see Chapter 20), governance of myriad sensory motor reflexes (see Chapter 15), the organization of eye movements (see Chapter 19), regulation of sleep and wakefulness (see Chapter 27), and, most important for present purposes, the temporal and spatial coordination of movements. The descending motor control pathways from the reticular formation to the spinal cord are similar to those of the vestibular nuclei; they terminate primarily in the medial parts of the gray matter where they influence the local circuit neurons that coordinate axial and proximal limb muscles (see Figure 16.2). Both the vestibular nuclei and the reticular formation provide information to the spinal cord that maintains posture in response to environmental (or selfinduced) disturbances of body position and stability. As expected, the vestibular nuclei make adjustments in posture and equilibrium in response to infor- 398 Chapter Sixteen Box A The Reticular Formation If one were to exclude from the structure of the brainstem the cranial nerve nuclei, the nuclei that provide input to the cerebellum, the long ascending and descending tracts that convey explicit sensory and motor signals, and the structures that lie dorsal and lateral to the ventricular system, what would be left is a central core region known as the tegmentum (Latin for “covering structure”), so named because it “covers” the ventral part of the brainstem. Scattered among the diffuse fibers that course through the tegmentum are small clusters of neurons that are collectively known as the reticular formation. With few exceptions, these clusters of neurons are difficult to recognize as distinct nuclei in standard histological preparations. Indeed, the modifying term reticular (“like a net”) was applied to this loose collection of neuronal clusters because the early neurohistologists envisioned these neurons as part of a sparse network of diffusely connected cells that extends from the intermediate gray regions of the cervical spinal cord to the lateral regions of the hypothalamus and certain nuclei along the midline of the thalamus. These early anatomical concepts were influenced by lesion experiments in animals and clinical observations in human patients made in the 1930s and 40s. These studies showed that damage to the upper brainstem tegmentum produced coma, suggesting the existence of a neural system in the midbrain and rostral pons that supported normal conscious brain states and transitions between sleep and wakefulness. These ideas were articulated most influentially by G. Moruzzi and H. Magoun when they proposed a “reticular activating system” to account for these functions and the critical role of the brainstem reticular formation. Current evidence generally supports the notion of an activating function of the rostral reticular formation; however, neuroscientists now recognize the complex interplay of a variety of neurochemical systems (with diverse post synaptic effects) comprising distinct cell clusters in the rostral tegmentum, and a myriad of other functions performed by neuronal clusters in more caudal parts of the reticular formation. Thus, with the advent of more precise means of demonstrating anatomical connections, as well as more sophisticated means of identifying neurotransmitters and the activity patterns of individual neurons, the concept of a “sparse network” engaged in a common function is now obsolete. Nevertheless, the term reticular formation remains, as does the daunting challenge of understanding the anatomical complexity and functional heterogeneity of this complex brain region. Fortunately, two simplfying generalizations can be made. First, the functions of the different clusters of neurons in the reticular formation can be grouped into two broad categories: modulatory functions and premotor functions. Second, the modulatory functions are primarily found in the rostral sector of the reticular formation, whereas the premotor functions are localized in more caudal regions. Several clusters of large (“magnocellular”) neurons in the midbrain and rostral pontine reticular formation participate—together with certain diencephalic nuclei—in the modulation of conscious states (see Chapter 27). These effects are accomplished by long-range, diencephalic projections of cholinergic neurons near the superior cerebellar peduncle, as well as the more widespread forebrain projections of noradrenergic neurons in the locus coeruleus and serotenergic neurons in the raphe nuclei. Generally speaking, these biogenic amine neurotransmitters function as neuromodulators (see Chapter 6) that alter the membrane potential and thus the firing patterns of thalamocortical and cortical neurons (the details of these effects are explained in Chapter 27). Also included in this category are the dopaminergic systems of the ventral midbrain that modulate cortico-striatal interactions in the basal ganglia (see Chapter 17) and the responsiveness of neurons in the prefrontal cortex and limbic forebrain (see Chapter 28). However, not all modulatory projections from the rostral reticular formation are directed toward the forebrain. Although not always considered part of the reticular formation, it is helpful to include in this functional group certain neuronal columns in the periaqueductal gray (surrounding the cerebral aqueduct) that project to the dorsal horn of the spinal cord and modulate the transmission of nociceptive signals (see Chapter 9). Reticular formation neurons in the caudal pons and medulla oblongata generally serve a premotor function in the sense that they intergate feedback sensory signals with executive commands from upper motor neurons and deep cerebellar nuclei and, in turn, organize the efferent activities of lower visceral motor and certain somatic motor neurons in the brainstem and spinal cord. Examples of this functional category include the smaller (“parvocellular”) neurons that coordinate a broad range of motor activities, including the gaze centers discussed in Chapter 19 and local circuit neurons near the somatic motor and branchiomotor nuclei that organize mastication, facial expressions, and a variety of reflexive orofacial behaviors such as sneezing, hiccuping, yawning, and swallowing. In addition, there are “autonomic centers” that organize the efferent activities of specific pools of primary visceral motor neurons. Included in this subgroup are distinct clusters of neurons in the ventral-lateral medulla that generate respiratory rhythms, and others that regulate the cardioinhibitory Upper Motor Neuron Control of the Brainstem and Spinal Cord 399 Mesencephalic and rostral pontine reticular formation Modulates forebrain activity Midbrain Pons Caudal pontine and medullary reticular formation Premotor coordination of lower somatic and visceral motor neuronal pools Medulla Midsagittal view of the brain showing the longitudinal extent of the reticular formation and highlighting the broad functional roles performed by neuronal clusters in its rostral (blue) and caudal (red) sectors. output of neurons in the nucleus ambiguus and the dorsal motor nucleus of the vagus nerve. Still other clusters organize more complex activities that require the coordination of both somatic motor and visceral motor outflow, such as gagging and vomiting, and even laughing and crying. One set of neuronal clusters that does not fit easily into this rostral-caudal framework is the set of neurons that give rise to the reticulospinal projections. As described in the text, these neurons are distributed in both rostral and caudal sectors of the reticular formation and they give rise to long-range projections mation from the inner ear. Direct projections from the vestibular nuclei to the spinal cord ensure a rapid compensatory response to any postural instability detected by the inner ear (see Chapter 13). In contrast, the motor centers in the reticular formation are controlled largely by other motor centers in the cortex or brainstem. The relevant neurons in the reticular formation initiate adjustments that stabilize posture during ongoing movements. The way the upper motor neurons of the reticular formation maintain posture can be appreciated by analyzing their activity during voluntary movements. Even the simplest movements are accompanied by the activation of muscles that at first glance seem to have little to do with the primary purpose of the movement. For example, Figure 16.5 shows the pattern of muscle activity that occurs as a subject uses his arm to pull on a handle in response to an auditory tone. Activity in the biceps muscle begins about 200 that innervate lower motor neuronal pools in the medial ventral horn of the spinal cord. The reticulospinal inputs serve to modulate the gain of segmental reflexes involving the muscles of the trunk and proximal limbs and to initiate certain stereotypical patterns of limb movement. In summary, the reticular formation is best viewed as a heterogeneous collection of distinct neuronal clusters in the brainstem tegmentum that either modulate the excitability of distant neurons in the forebrain and spinal cord or coordinate the firing patterns of more local lower motor neuronal pools engaged in reflexive or stereotypical somatic motor and visceral motor behavior. References BLESSING, W. W. (1997) Inadequate frameworks for understanding bodily homeostasis. Trends Neurosci. 20: 235–239 HOLSTEGE, G., R. BANDLER AND C. B. SAPER (EDS.) (1996) Progress in Brain Research, Vol. 107. Amsterdam: Elsevier. LOEWY, A. D. AND K. M. SPYER (EDS.) (1990) Central Regulation of Autonomic Functions. New York: Oxford. MASON, P. (2001) Contributions of the medullary raphe and ventromedial reticular region to pain modulation and other homeostatic functions. Annu. Rev. Neurosci. 24: 737–777. MORUZZI, G. AND H. W. MAGOUN (1949) Brain stem reticular formation and activation of the EEG. EEG Clin. Neurophys. 1: 455–476. 400 Chapter Sixteen Superior colliculus Mesencephalic reticular formation 1 Medial lemniscus Midbrain Substantia nigra Pontine reticular formation 2 Lower pons Hypoglossal nucleus Middle medulla 1 2 3 Figure 16.4 The location of the reticular formation in relation to some other major landmarks at different levels of the brainstem. Neurons in the reticular formation are scattered among the axon bundles that course through the medial portion of the midbrain, pons, and medulla (see Box A). Fourth ventricle Abducens nucleus Middle cerebellar peduncle Medial lemniscus 3 Cerebral peduncle Corticospinal fibers Dorsal motor nucleus of vagus Medullary reticular formation Medial lemniscus Inferior olive Medullary pyramid ms after the tone. However, as the records show, the contraction of the biceps is accompanied by a significant increase in the activity of a proximal leg muscle, the gastrocnemius (as well as many other muscles not monitored in the experiment). In fact, contraction of the gastrocnemius muscle begins well before contraction of the biceps. These observations show that postural control entails an anticipatory, or feedforward, mechanism (Figure 16.6). As part of the motor plan for moving the arm, the effect of the impending movement on body stability is “evaluated” and used to generate a change in the activity of the gastrocnemius muscle. This change actually precedes and provides postural support for the movement of the arm. In the example given in Figure 16.5, contraction of the biceps would tend to pull the entire body forward, an action that is opposed by the contraction of the gastrocnemius muscle. In short, this feedforward mechanism “predicts” the resulting disturbance in body stability and generates an appropriate stabilizing response. The importance of the reticular formation for feedforward mechanisms of postural control has been explored in more detail in cats trained to use a forepaw to strike an object. As expected, the forepaw movement is accompanied by feedforward postural adjustments in the other legs to maintain the animal upright. These adjustments shift the animal’s weight from an even distribution over all four feet to a diagonal pattern, in which the weight is carried mostly by the contralateral, nonreaching forelimb and the ipsilateral hindlimb. Lifting of the forepaw and postural adjustments in the other limbs can also be Upper Motor Neuron Control of the Brainstem and Spinal Cord 401 induced in an alert cat by electrical stimulation of the motor cortex. After pharmacological inactivation of the reticular formation, however, electrical stimulation of the motor cortex evokes only the forepaw movement, without the feedforward postural adjustments that normally accompany them. The results of this experiment can be understood in terms of the fact that the upper motor neurons in the motor cortex influence the spinal cord circuits by two routes: direct projections to the spinal cord and indirect projections to brainstem centers that in turn project to the spinal cord (see Figure 16.3). The reticular formation is one of the major destinations of these latter projections from the motor cortex; thus, cortical upper motor neurons initiate both the reaching movement of the forepaw and also the postural adjustments in the other limbs necessary to maintain body stability. The forepaw movement is initiated by the direct pathway from the cortex to the spinal cord (and possibly by the red nucleus as well), whereas the postural adjustments are mediated via pathways from the motor cortex that reach the spinal cord indirectly, after an intervening relay in the reticular formation (the corticoreticulospinal pathway). Further evidence for the contrasting functions of the direct and indirect pathways from the motor cortex and brainstem to the spinal cord comes from experiments carried out by the Dutch neurobiologist Hans Kuypers, who examined the behavior of rhesus monkeys that had the direct pathway to the spinal cord transected at the level of the medulla, leaving the indirect descending upper motor neuron pathways to the spinal cord via the brainstem centers intact. Immediately after the surgery, the animals were able to use axial and proximal muscles to stand, walk, run, and climb, but they had great difficulty using the distal parts of their limbs (especially their hands) independently of other body movements. For example, the monkeys could cling to the cage but were unable to reach toward and pick up food with their fingers; rather, they used the entire arm to sweep the food toward them. After several weeks, the animals recovered some independent use of their hands and were again able to pick up objects of interest, but this action still involved the concerted closure of all of the fingers. The ability to make independent, fractionated movements of the fingers, as in opposing the movements of the fingers and thumb to pick up an object, never returned. These observations show that following damage to the direct corticospinal pathway at the level of the medulla, the indirect projections from the motor cortex via the brainstem centers (or from brainstem centers alone) are capable of sustaining motor behavior that involves primarily the use of proximal muscles. In contrast, the direct projections from the motor cortex to the spinal cord provide the speed and agility of movements, enabling a higher degree of precision in fractionated finger movements than is possible using the indirect pathways alone. Central command Feedforward for anticipated postural instability Limb movement Postural adjustment Postural instability Feedback for unanticipated postural instability Biceps EMG 0 100 300 Time (ms) Tone 500 Gastrocnemius EMG 0 100 300 Time (ms) Tone 500 Figure 16.5 Anticipatory maintenance of body posture. At the onset of a tone, the subject pulls on a handle, contracting the biceps muscle. To ensure postural stability, contraction of the gastrocnemius muscle precedes that of the biceps. EMG refers to the electromyographic recording of muscle activity. Figure 16.6 Feedforward and feedback mechanisms of postural control. Feedforward postural responses are “preprogrammed” and typically precede the onset of limb movement (see Figure 16.4). Feedback responses are initiated by sensory inputs that detect postural instability. 402 Chapter Sixteen Selective damage to the corticospinal tract (i.e., the direct pathway) in humans is rarely seen in the clinic. Nonetheless, this evidence in nonhuman primates showing that direct projections from the cortex to the spinal cord are essential for the performance of discrete finger movements helps explain the limited recovery in humans after damage to the motor cortex or to the internal capsule. Immediately after such an injury, such patients are typically paralyzed. With time, however, some ability to perform voluntary movements reappears. These movements, which are presumably mediated by the brainstem centers, are crude for the most part, and the ability to perform discrete finger movements such as those required for writing, typing, or buttoning typically remains impaired. The Corticospinal and Corticobulbar Pathways: Upper Motor Neurons That Initiate Complex Voluntary Movements The upper motor neurons in the cerebral cortex reside in several adjacent and highly interconnected areas in the frontal lobe, which together mediate the planning and initiation of complex temporal sequences of voluntary movements. These cortical areas all receive regulatory input from the basal ganglia and cerebellum via relays in the ventrolateral thalamus (see Chapters 17 and 18), as well as inputs from the somatic sensory regions of the parietal lobe (see Chapter 8). Although the phrase “motor cortex” is sometimes used to refer to these frontal areas collectively, more commonly it is restricted to the primary motor cortex, which is located in the precentral gyrus (Figure 16.7). The primary motor cortex can be distinguished from the adjacent “premotor” areas both cytoarchitectonically (it is area 4 in Brodmann’s nomenclature) and by the low intensity of current necessary to elicit movements by electrical stimulation in this region. The low threshold for eliciting movements is an indicator of a relatively large and direct pathway from the primary area to the lower motor neurons of the brainstem and spinal cord. This section and the next focus on the organization and functions of the primary motor cortex and its descending pathways, whereas the subsequent section addresses the contributions of the adjacent premotor areas. The pyramidal cells of cortical layer V (also called Betz cells) are the upper motor neurons of the primary motor cortex. Their axons descend to the brainstem and spinal motor centers in the corticobulbar and corticospinal tracts, passing through the internal capsule of the forebrain to enter the cerebral peduncle at the base of the midbrain (Figure 16.8). They then (A) Lateral view Lateral premotor cortex Figure 16.7 The primary motor cortex and the premotor area in the human cerebral cortex as seen in lateral (A) and medial (B) views. The primary motor cortex is located in the precentral gyrus; the premotor area is more rostral. Medial premotor cortex (B) Medial view Primary motor cortex Medial premotor cortex Primary motor cortex Upper Motor Neuron Control of the Brainstem and Spinal Cord 403 Figure 16.8 The corticospinal and corticobulbar tracts. Neurons in the motor cortex give rise to axons that travel through the internal capsule and coalesce on the ventral surface of the midbrain, within the cerebral peduncle. These axons continue through the pons and come to lie on the ventral surface of the medulla, giving rise to the pyramids. Most of these pyramidal fibers cross in the caudal part of the medulla to form the lateral corticospinal tract in the spinal cord. Those axons that do not cross form the ventral corticospinal tract. Cortex Internal capsule Red nucleus Corticospinal and corticobulbar tracts Midbrain Cerebral peduncle Trigeminal motor nucleus (V) Middle pons Pontine fiber bundles Hypoglossal nucleus (XII) Middle medulla Corticobulbar collaterals to reticular formation Pyramid Pyramidal decussation Caudal medulla Ventral corticospinal tract Lateral corticospinal tract Spinal cord Lower motor neuron 404 Chapter Sixteen Box B Patterns of Facial Weakness and Their Importance for Localizing Neurological Injury The signs and symptoms pertinent to the cranial nerves and their nuclei are of special importance to clinicians seeking to pinpoint the neurological lesions that produce motor deficits. An especially instructive example is provided by the muscles of facial expression. It has long been recognized that the distribution of facial weakness provides important localizing clues indicating whether the underlying injury involves lower motor neurons in the facial motor nucleus (and/or their axons in the facial nerve) or the inputs that govern these neurons, which arise from upper motor neurons in the cerebral cortex. Damage to the facial motor nucleus or its nerve affects all the muscles of facial expression on the side of the lesion (lesion C in the figure); this is expected given the intimate anatomical and functional linkage between lower motor neurons and skeletal muscles. A pattern of impairment that is more difficult to explain accompanies unilateral injury to the motor areas in the lateral frontal lobe (primary motor cortex, lateral premotor cortex), as occurs strokes that involve the middle cerebral artery (lesion A in the figure). Most patients with such injuries have difficulty controlling the contralateral muscles around the mouth but retain the ablility to symmetrically raise their eyebrows, wrinkle their forehead, and squint. Until recently, it was assumed that this pattern of inferior facial paresis with superior facial sparing could be attributed to (presumed) bilateral projections from the face representation in the primary motor cortex to the facial motor nucleus; in this conception, the intact ipsilateral corticobulbar projections were considered sufficient to motivate the contractions of the superior muscles of the face. However, recent tract-tracing studies in non-human primates have sug- gested a different explanation. These studies demonstrate two important facts that clarify the relations among the face representations in the cerebral cortex and the facial motor nucleus. First, the corticobulbar projections of the primary motor cortex are directed predominantly toward the lateral cell columns in the contralateral facial motor nucleus, which control the movements of the perioral musculature. Thus, the more dorsal cell columns in the facial motor nucleus that innervate superior facial muscles do not receive significant input from the primary motor cortex. Second, these dorsal cell columns are governed by an acces- Face representation in cingulate motor area Face representation in right primary motor cortex A B Upper motor neuron lesion Pons Facial nucleus Facial nerve Lower motor neuron lesion C Weakness of inferior facial muscles Weakness of superior and inferior facial muscle Organization of projections from cerebral cortex to the facial motor nucleus and the effects of upper and lower motor neuron lesions. Upper Motor Neuron Control of the Brainstem and Spinal Cord 405 sory motor area in the anterior cingulate gyrus, a cortical region that is associated with emotional processing (see Chapter 28). Therefore, a better interpretation is that strokes involving the middle cerebral artery spare the superior aspect of the face because the relevant upper motor neurons are in the cingulum, which is supplied by the anterior cerebral artery. An additional puzzle has also been resolved by these studies. Strokes involving the anterior cerebral artery or subcortical lesions that interrupt the corticobul- bar projection (lesion B in the figure) seldom produce significant paresis of the superior facial muscles. Superior facial sparing in these situations may arise because this cingulate motor area sends descending projections through the corticobulbar pathway that bifuracte and innervate dorsal facial motor cell columns on both sides of the brainstem. Thus, the superior muscles of facial expression are controlled by symmetrical inputs from the cingulate motor areas in both hemispheres. run through the base of the pons, where they are scattered among the transverse pontine fibers and nuclei of the pontine gray matter, coalescing again on the ventral surface of the medulla where they form the medullary pyramids. The components of this upper motor neuron pathway that innervate cranial nerve nuclei, the reticular formation, and the red nucleus (that is, the corticobulbar tract) leave the pathway at the appropriate levels of the brainstem (see Figure 16.8 and Box B ). At the caudal end of the medulla, most, but not all, of the axons in the pyramidal tract cross (or “decussate”) to enter the lateral columns of the spinal cord, where they form the lateral corticospinal tract. A smaller number of axons enters the spinal cord without crossing; these axons, which comprise the ventral corticospinal tract, terminate either ipsilaterally or contralaterally, after crossing in the midline (via spinal cord commissure). The ventral corticospinal pathway arises primarily from regions of the motor cortex that serve axial and proximal muscles. The lateral corticospinal tract forms the direct pathway from the cortex to the spinal cord and terminates primarily in the lateral portions of the ventral horn and intermediate gray matter (see Figures 16.3 and 16.8). The indirect pathway to lower motor neurons in the spinal cord runs, as already described, from the motor cortex to two of the sources of upper motor neurons in the brainstem: the red nucleus and the reticular formation. In general, the axons to the reticular formation originate from the parts of the motor cortex that project to the medial region of the spinal cord gray matter, whereas the axons to the red nucleus arise from the parts of the motor cortex that project to the lateral region of the spinal cord gray matter. Functional Organization of the Primary Motor Cortex Clinical observations and experimental work dating back a hundred years or more have provided a reasonably coherent picture of the functional organization of the motor cortex. By the end of the nineteenth century, experimental work in animals by the German physiologists G. Theodor Fritsch and Eduard Hitzig had shown that electrical stimulation of the motor cortex elicits contractions of muscles on the contralateral side of the body. At about the same time, the British neurologist John Hughlings Jackson surmised that the motor cortex contains a complete representation, or map, of the body’s musculature. References JENNY, A. B. AND C. B. SAPER (1987) Organization of the facial nucleus and corticofacial projection in the monkey: A reconsideration of the upper motor neuron facial palsy. Neurology 37: 930–939. KUYPERS, H. G. J. M. (1958) Corticobulbar connexions to the pons and lower brainstem in man. Brain 81: 364–489. MORECRAFT, R. J., J. L. LOUIE, J. L. HERRICK AND K. S. STILWELL-MORECRAFT (2001) Cortical innervation of the facial nucleus in the nonhuman primate: A new interpretation of the effects of stroke and related subtotal brain trauma on the muscles of facial expression. Brain 124: 176–208. 406 Chapter Sixteen Jackson reached this conclusion from his observation that the abnormal movements during some types of epileptic seizures “march” systematically from one part of the body to another. For instance, partial motor seizures may start with abnormal movements of a finger, progress to involve the entire hand, then the forearm, the arm, the shoulder, and, finally, the face. This early evidence for motor maps in the cortex was confirmed shortly after the turn of the nineteenth century when Charles Sherrington published his classical maps of the organization of the motor cortex in great apes, using focal electrical stimulation. During the 1930s, one of Sherrington’s students, the American neurosurgeon Wilder Penfield, extended this work by demonstrating that the human motor cortex also contains a spatial map of the body’s musculature. By correlating the location of muscle contractions with the site of electrical stimulation on the surface of the motor cortex (the same method used by Sherrington), Penfield mapped the representation of the muscles in the precentral gyrus in over 400 neurosurgical patients (Figure 16.9). He found that this motor map shows the same disproportions observed in the somatic sensory maps in the postcentral gyrus (see Chapter 8). Thus, the musculature used in tasks requiring fine motor control (such as movements of the face and hands) occupies a greater amount of space in the (A) Central sulcus Primary motor cortex Figure 16.9 Topographic map of the body musculature in the primary motor cortex. (A) Location of primary motor cortex in the precentral gyrus. (B) Section along the precentral gyrus, illustrating the somatotopic organization of the motor cortex. The most medial parts of the motor cortex are responsible for controlling muscles in the legs; the most lateral portions are responsible for controlling muscles in the face. (C) Disproportional representation of various portions of the body musculature in the motor cortex. Representations of parts of the body that exhibit fine motor control capabilities (such as the hands and face) occupy a greater amount of space than those that exhibit less precise motor control (such as the trunk). (B) Corticospinal tract Shoulder Arm Head Hand Digits Thumb Neck Eyes Nose Face Lips Jaw Tongue Throat Corticobulbar tract Genitalia Trunk Leg Feet Toes (C) Upper Motor Neuron Control of the Brainstem and Spinal Cord 407 map than does the musculature requiring less precise motor control (such as that of the trunk). The behavioral implications of cortical motor maps are considered in Boxes C and D. The introduction in the 1960s of intracortical microstimulation (a more refined method of cortical activation) allowed a more detailed understanding of motor maps. Microstimulation entails the delivery of electrical currents an order of magnitude smaller than those used by Sherrington and Penfield. By passing the current through the sharpened tip of a metal microelectrode inserted into the cortex, the upper motor neurons in layer V that project to lower motor neuron circuitry can be stimulated focally. Although intracortical stimulation generally confirmed Penfield’s spatial map in the motor cortex, it also showed that the finer organization of the map is rather different than most neuroscientists imagined. For example, when microstimulation was combined with recordings of muscle electrical activity, even the smallest currents capable of eliciting a response initiated the excitation of several muscles (and the simultaneous inhibition of others), suggesting that organized movements rather than individual muscles are represented in the map (see Box C ). Furthermore, within major subdivisions of the map (e.g., arm, forearm, or finger regions), a particular movement could be elicited by stimulation of widely separated sites, indicating that neurons in nearby regions are linked by local circuits to organize specific movements. This interpretation has been supported by the observation that the regions responsible for initiating different movements overlap substantially. About the same time that these studies were being undertaken, Ed Evarts and his colleagues at the National Institutes of Health were pioneering a technique in which implanted microelectrodes were used to record the electrical activity of individual motor neurons in awake, behaving monkeys. In these experiments, the monkeys were trained to perform a variety of motor tasks, thus providing a means of correlating neuronal activity with voluntary movements. Evarts and his group found that the force generated by contracting muscles changed as a function of the firing rate of upper motor neurons. Moreover, the firing rates of the active neurons often changed prior to movements involving very small forces. Evarts therefore proposed that the primary motor cortex contributes to the initial phase of recruitment of lower motor neurons involved in the generation of finely controlled movements. Additional experiments showed that the activity of primary motor neurons is correlated not only with the magnitude, but also with the direction of the force produced by muscles. Thus, some neurons show progressively less activity as the direction of movement deviates from the neuron’s “preferred direction.” A further advance was made in the mid-1970s by the introduction of spike-triggered averaging (Figure 16.10). By correlating the timing of the cortical neuron’s discharges with the onset times of the contractions generated by the various muscles used in a movement, this method provides a way of measuring the influence of a single cortical motor neuron on a population of lower motor neurons in the spinal cord. Recording such activity from different muscles as monkeys performed wrist flexion or extension demonstrated that the activity of a number of different muscles is directly facilitated by the discharges of a given upper motor neuron. This peripheral muscle group is referred to as the “muscle field” of the upper motor neuron. On average, the size of the muscle field in the wrist region is two to three muscles per upper motor neuron. These observations confirmed that single upper motor neurons contact several lower motor neuron pools; the results are also consistent with the general conclusion that movements, rather than individual muscles, 408 Chapter Sixteen Box C What Do Motor Maps Represent? Electrical stimulation studies carried out by the neurosurgeon Wilder Penfield and his colleagues in human patients (and by Sherrington and later Clinton Woolsey and his colleagues in experimental animals) clearly demonstrated a systematic map of the body’s musculature in the primary motor cortex (see text). The fine structure of this map, however, has been a continuing source of controversy. Is the map in the motor cortex a “piano keyboard” for the control of individual muscles, or is it a map of movements, in which specific sites control multiple muscle groups that contribute to the generation of particular actions? Initial experiments implied that the map in the motor cortex is a fine-scale representation of individual muscles. Thus, stimulation of small regions of the map activated single muscles, suggesting that vertical columns of cells in the motor cortex were responsible for controlling the actions of particular muscles, much as columns in the somatic sensory map are thought to analyze particular types of stimulus information (see Chapter 8). More recent studies using anatomical and physiological techniques, however, have shown that the map in the motor cortex is far more complex than a columnar representation of particular muscles. Individual pyramidal tract axons are now known to terminate on sets of spinal motor neurons that innervate different muscles. This relationship is evident even for neurons in the hand representation of the motor cortex, the region that controls the most discrete, fractionated movements. Furthermore, cortical microstimulation experiments have shown that contraction of a single muscle can be evoked by stimulation over a wide region of the motor cortex (about 2–3 mm in macaque monkeys) in a complex, mosaic fashion. It seems likely that horizontal connections within the motor cortex and local circuits in the spinal cord create ensembles of neurons that coordinate the pattern of firing in the population of ventral horn cells that ultimately generate a given movement. Thus, while the somatotopic maps in the motor cortex generated by early studies are correct in their overall topography, the fine structure of the map is far more intricate. Unraveling these details of motor maps still holds the key to understanding how patterns of activity in the motor cortex generate a given movement. References BARINAGA, M. (1995) Remapping the motor cortex. Science 268: 1696–1698. LEMON, R. (1988) The output map of the primate motor cortex. Trends Neurosci. 11: 501–506. PENFIELD, W. AND E. BOLDREY (1937) Somatic motor and sensory representation in the cerebral cortex of man studied by electrical stimulation. Brain 60: 389–443. SCHIEBER, M. H. AND L. S. HIBBARD (1993) How somatotopic is the motor cortex hand area? Science 261: 489–491. WOOLSEY, C. N. (1958) Organization of somatic sensory and motor areas of the cerebral cortex. In Biological and Biochemical Bases of Behavior, H. F. Harlow and C. N. Woolsey (eds.). Madison, WI: University of Wisconsin Press, pp. 63–81. are encoded by the activity of the upper motor neurons in the cortex (see Box C ). Finally, the relative amount of activity across large populations of neurons appears to encode the direction of visually-guided movements. Thus, the direction of movements in monkeys could be predicted by calculating a “neuronal population vector” derived simultaneously from the discharges of upper motor neurons that are “broadly tuned” in the sense that they discharge prior to movements in many directions (Figure 16.11). These observations showed that the discharges of individual upper motor neurons cannot specify the direction of an arm movement, simply because they are tuned too broadly; rather, each arm movement must be encoded by the concurrent discharges of a large population of such neurons. The Premotor Cortex A complex mosaic of interconnected frontal lobe areas that lie rostral to the primary motor cortex also contributes to motor functions (see Figure 16.7). The upper motor neurons in this premotor cortex influence motor behavior Upper Motor Neuron Control of the Brainstem and Spinal Cord 409 Figure 16.10 The influence of single cortical upper motor neurons on muscle activity. (A) Diagram illustrates the spike triggering average method for correlating muscle activity with the discharges of single upper motor neurons. (B) The response of a thumb muscle (bottom trace) follows by a fixed latency the single spike discharge of a pyramidal tract neuron (top trace). This technique can be used to determine all the muscles that are influenced by a given motor neuron (see text). (After Porter and Lemon, 1993.) (A) Detection of postspike facilitation Record Recording from cortical motor neuron Primary motor cortex Spinal motor neuron Spikes of single cortical motor neuron Electromyograph (EMG) Rectified EMG Trigger averager input Rectifier Spike-triggered averaging (B) Postspike facilitation by cortical motor neuron Cortical motor neuron spike n = 9000 spikes Spike-triggered average of EMG Time (ms) 410 Chapter Sixteen Box D Sensory Motor Talents and Cortical Space Are special sensory motor talents, such as the exceptional speed and coordination displayed by talented athletes, ballet dancers, or concert musicians visible in the structure of the nervous system? The widespread use of noninvasive brain imaging techniques (see Box A in Chapter 1) has generated a spate of studies that have tried to answer this and related questions. Most of these studies have sought to link particular sensory motor skills to the amount of brain space devoted to such talents. For example, a study of professional violinists, cellists, and classical guitarists purported to show that representations of the “fingering” digits of the left hand in the right primary somatic sensory cortex are larger than the corresponding representations in nonmusicians. Although such studies in humans remain controversial (the techniques are only semiquantitative), the idea that greater motor talents (or any other ability) will be reflected in a greater amount of brain space devoted to that task makes good sense. In particular, comparisons across species show that special talents are invariably based on commensurately sophisticated brain circuitry, which means more neurons, more synaptic contacts between neurons, and more supporting glial cells—all of which occupy more space within the brain. The size and proportion of bodily representations in the primary somatic sensory and motor cortices of various animals reflects species-specific nuances of mechanosensory discrimination and motor control. Thus, the representations of the paws are disproportionately large in the sensorimotor cortex of raccoons; rats and mice devote a great deal of cortical space to representations of their prominent facial whiskers; and a large fraction of the sensorimotor cortex of the star-nosed mole is given over to representing the elaborate nasal appendages that provide critical mechanosensory information for this burrowing species. The link between behavioral competence and the allocation of space is equally apparent in animals in which a particular ability has diminished, or has never developed fully, during the course of evolution. Nevertheless, it remains uncertain how—or if—this principle applies to variations in behavior among members of the same species, including humans. For example, there does not appear to be any average hemisphere asymmetry in the allocation of space in either the primary sensory or motor area, as measure