Progress in Neurobioloyy Vol. 17, pp. 203 to 282, 1981 Printed in Great Britain. All rights reserved 0301-0082/81/040203-80140.00/0 Copyright O Pergamon Press Ltd SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS" LONG-TERM REGULATION OF MOTONEURON FUNCTION ALAN D. GRINNELLand ALBERT A. HERRERA* Dept. of Physiology and Jerry Lewis Neuromuscular Research Center, Unirersity of California, Los Angeles, California, 90024, U.S,A. (Received 31 July 1981) Contents Abbreviations 1. Introduction 2. The development of appropriate neuromuscular connections 2.1 Outline of events in development 2.2 Specificity of neuromuscular connections 2.2.1. Neuromuscular development in the chick 2.2.1.1. Specificity of nerve-muscle matching 2.2.1.2. Axonal pathway recognition 2.2.2. Neuromuscular development in amphibians 2.2.3. Neuromuscular development in mammals 2.2.4. Summary 2.3. Neuronal death during development 2.3.1. Quantitative dependence on target size 2.3.2. Evidence for dependence on synaptic function in the periphery 2.3.3. Evidence against peripheral determination of cell death 2.3.4. Summary 2.4 Synapse elimination 2.4.1. Generality of the phenomenon 2.4.2. Timing and extent of synapse elimination 2.4.3. Specificity of synapse elimination and changes in motor unit size 2.4.4. Partial denervation and the role of competition in synapse elimination 2.4.5. Polyneuronal innervation by regenerating neonatal neurons, and its elimination 2.4.6. Effects of altered activity on synapse elimination 2.4.7. Summary 3. Long-term plasticity in adult neuromuscular junctions 3.1 Synaptic effectiveness 3.2, Synaptic remodelling 3.3, Effects of decreased or increased use 3.4, Seasonal and hormonal effects 3.4.1. "Winter frogs" and hibernating mammals 3.4.2. Effects of testosterone 3.4.3. Effects of epinephrine and norepinephrine 3.4.4. Presynaptic effects of acetyicholine 3.4.5. Other hormones 3.5. Effects of aging 3.6. Effects of disease 3.6.1. Myasthenia gravis 3.6.2. Eaton-Lambert and other myasthenic syndromes 3.6.3. Murine motor endplate disease 3.6.4. Wobbler mice 3.6.5. Murine muscular dystrophy 3.6.6. Muscular dystrophy in the chicken 3.7. Summary 4. Responses of adult motoneurons to experimental manipulation 4.1. Sprouting 4.1.1. Responses to partial denervation 4.1.2. Response to synaptic block 4.1.3. Effect of muscle activity on sprouting 4.1.4. Terminal and collateral sprouting as responses to different signals 204 204 205 205 206 207 207 208 211 211 212 212 213 213 215 216 217 217 217 218 220 221 222 223 224 225 227 227 228 228 230 231 232 234 234 235 235 236 237 237 237 238 239 239 240 240 240 241 242 * Present address: Dept. of Biological Sciences, University of Southern California, Los Angeles, California, 90007, U,S.A. J.P.N. 17/4--A 203 204 ALAN D. GRINNELL AND ALBERT A. HERRERA 4.1.5. Postulated mechanisms for the regulation of sprouting 4.1.5.1. Peripheral sprouting factors 4.1.5.2. Central regulatory mechanisms and contralateral sprouting 4.1.6. Summary 4.2. Axotomy and the dependence of adult motoneurons on the periphery 4.2.1. Metabolic responses 4.2.2. Changes in motoneuron electrical properties 4.2.3. Changes in synaptic circuitry in the spinal cord 4.2.4. Effects on neuroglia 4.2.5. Role of s e n s o r y a x o n s in the axotomy response 4.2.6. Altered neuromuscular activity and peripheral regulatory substances 4.2.7. Summary 4.3. Experimental alteration of synaptic effectiveness 4.3.1. The dependence of synaptic effectiveness on motor unit size 4.3.2. Synaptic effectiveness can be altered by contralateral denervation 4.3.3. Summary 5. Regulation of motoneuron function during regeneration 5.1. Selective reinnervation of old endplate sites 5.2. Competitive interactions and specificity of reinnervation 5.2.1. Specific reinnervation of different muscle fiber types 5.2.1.1. Mammals 5.2.1.2. Birds 5.2.1.3. Lower vertebrates 5.2.2. Reinnervation of muscles of similar fiber type 5.2.2.1. Where correct and foreign nerves have equal opportunity 5.2.2.2. Competition between foreign nerves in frogs 5.2.2.3. Displacement of sprouted collaterals 5.2.2.4. Mechanisms of competitive displacement in lower vertebrates 5.2.2.5. Competitive interaction between synapses on adult mammalian muscle fibres 5.2.3. Summary 6. Conclusions Acknowledgements References 242 242 244 245 245 245 245 246 247 247 248 249 249 249 250 252 252 252 254 255 255 255 255 256 256 258 258 259 262 263 264 265 265 Abbreviations ACh = acetylcholine; AChE = acetylcholinesterase; AChR = acetylcholine receptor; ALD = anterior latissimus dorsi; a-BTX = alpha-bungarotoxin; BoTX = botulinum toxin; CAT ~ choline acetyltransferase; CNS = central nervous system; CP = cutaneous pectoris; D F P = diisopropylfluorophosphate; dTC = d tubocurarine; E = excitatory; EDL = extensor digitorum longus; ED II = extensor of the second digit; E P P = endplate potential; EPSP = excitatory postsynaptic potential; H R P = horseradish peroxidase; I = inhibitory; I0 = inferior oblique; L4 = 4th lumbar spinal nerve; L5 = 5th lumbar spinal nerve; m E P P = miniature endplate potential; M N = motoneuron; PLD = posterior latissimus dorsi; SI = 1st sacral spinal nerve; $2 = 2nd sacral spinal nerve; SCG = superior cervical ganglion; SE = synapse elimination; SO = superior oblique; TTX = tetrodotoxin. 1. Introduction That the adult vertebrate neuromuscular system is highly specific in its connections and function is indisputable. There is less agreement about how it acquires this specificity. Until recently, in fact, it was widely held that growing motor axons formed connections with the first uninnervated muscle fibers they encountered during development, and that the nervous system performed the delicate task of making appropriate central connections. Regeneration of neuromuscular connections in most cases is conspicuously non-specific. Moreover, neuromuscular junctions are usually viewed as being unchanging, all-or-none synapses that are interesting mainly as the best available subject for study of chemical synaptic mechanisms. On the contrary, research of the past several years has established that remarkable specificity of connections is achieved by a combination of active pathway selection and recognition of appropriate target muscles, that specific regeneration of connections occurs in many cases, and that these "standard chemical synapses" are capable of extreme forms of plasticity. Moreover, both specificity of connections and synaptic plasticity can be .studied more easily in neuromuscular systems that in most central pathways. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 205 This potential has not gone unnoticed. There has been an explosion of research dealing with these phenomena in recent years. Much of this work has already been well reviewed (Harris, 1974; Purves, 1976b; Vrbov/t, Gordon and Jones, 1978; Lomo and Jansen, 1980; Landmesser, 1980; Mark, 1980; Brown, Holland and Hopkins, 1981; Dennis, 1981). We will not attempt a full historical review of aspects of the subject that have recently received such treatment, Even more sensibly, we have elected to leave almost entirely to others the enormous subjects of neuronal ("trophic") regulation of muscle properties (Guth, 1968; Gutmann, 1976; Harris, 1974; Rosenthal, 1977; Lomo, 1976; Purves, 1976b; Fambrough, 1979; Lomo and Jansen, 1980) and the processes of neuromuscular synapse formation in vitro (Shimada and Fischman, 1973; Kidokoro et al., 1975; Nelson, 1975; Patrick et al., 1978). In some instances, there are close parallels between studies of autonomic ganglia and those of muscle, some of which we will cite, without pretending to a thorough review of the ganglionic literature (for which see Purves, 1976b; Purves and Lichtman, 1978). Rather, our aim is to provide a systematic review of vertebrate neuromuscular specificity and plasticity viewed from the perspective of the motoneuron, emphasizing the variety and effectiveness of factors that regulate motoneuron survival, connectivity and function. 2. The Development of Appropriate Neuromuscular Connections 2.1. OUTLINE OF EVENTS IN DEVELOPMENT Our understanding of the development of neuromuscular junctions has come mostly from studies of avian and amphibian embryos. In the chick embryo (21 day incubation time), regional determination of the spinal cord (cervical, thoracic, lumbar, etc.) is achieved by day 2, before differentiation of spinal motoneurons (MNs), which begins on day 3 and proceeds in a rostro-caudal sequence. At a given segmental level of the cord, medially located MNs appear first, more lateral ones later, while myoblasts of the same segment "condense" from embryonic mesenchyme to form myotubes and myofibers in a proximo-distal sequence (Hollyday and Hamburger, 1977). The total number of MNs is present by about 5 1/2 to 6 days (Hamburger, 1977). The proliferation and differentiation of this population of MNs progresses according to programmed genetic instructions and is not dependent on the presence of target tissues (Hamburger, 1958), sensory input (Wenger, 1951), or descending input (Hamburger, 1946). MN differentiation is rapid and nearly synchronous for the segments innervating any given limb (Hollyday and Hamburger, 1977). Immediately after MN differentiation, axons exit the cord via cranial and segmental nerves and travel via stereotyped pathways to the primitive muscle masses. Nerves arrive at the periphery about the time myotubes are forming, well before the muscle masses have subdivided into individual muscles (Landmesser and Morris, 1975; Landmesser, 1978b). Axonal outgrowth and myelination are accelerated by chronic low frequency spinal cord stimulation (Toutant, et al., 1980), but the role of normal activity in these processes has not been established. Synapse formation can occur very quickly once the growing axons contact their target tissue, as functional synapses are seen almost as soon as the axons can be found in the chick limb bud (Landmesser and Morris, 1975; Landmesser, 1978b). In Xenopus laevis embryos, functional transmission can occur within minutes of contact, before there is any morphological sign of synaptic differentiation (Blackshaw and Warner, 1976; Kullberg et al., 1977). Moreover, motor axons can be induced to form functional synapses long before they would in the normal course of development. A section of fully developed tail muscle transplanted in place of the hind limb bud in a frog tadpole can be functionally innervated by the developing sciatic nerve as soon as it reaches the periphery, when only myoblasts would normally be present in the limb bud (Letinsky, 1974a). Quantal contents are low and decrease with repetitive stimulation in such precocious junctions, but basic release properties appear normal (Letinsky, 1974b). 206 ALAN D. GRINNELL AND ALBERT A. HERRERA Almost as soon as the total complement of MNs has appeared, and at the time when they first begin to make connections peripherally, the MNs become dependent on their target tissue. In normal development, there is a precipitous loss of MNs (see Section 2.3. below). Indeed, it is now clear that overproduction of neurons, followed by massive cell death, is characteristic of many, if not all, parts of the developing nervous system (Cowan, 1973, 1978). In the anterior horn of the spinal cord, this cell death results in the formation of discrete populations (pools) of MNs in characteristic locations, each of which innervates a specific muscle or small set of muscles (Romanes, 1964; Sz6kely and Cz6h, 1967; Cruce, 1974; Landmesser, 1978a). At the same time, the individual muscles are differentiating from the primitive muscle masses. Less is known about the timing of central sensory connections, but they apparently form significantly later than the peripheral motor connections. Specific interneuronal connections, on the other hand, appear before sensory input becomes effective, at least in the chick lumbar spinal cord. At stage 31 (7 days), synergistic muscles spontaneously contract in synchrony, out of phase with antagonistic muscle contraction (Bekoff, et al., 1975; Bekoff, 1976). Sensory stimulation becomes effective in eliciting reflexes at about stage 34 (8 days; see Landmesser, 1980). Despite the massive neuronal cell death, and lasting well after it has ceased, there is continued growth and branching of motor axons innervating the muscle. In all vertebrate skeletal muscles studied, this results in extensive p0!yneurona! innervation of individual fibers, usually at the same endplate site. This multiple innervation reaches its maximum extent at approximately the time of birth, and is followed, during early postnatal life, by elimination of most of these synaptic inputs (see Section 2.5). In the mature animal, these processes result in a set of highly specific connections between identifiable MN pools and their associated muscles, with each muscle fiber innervated by a characteristic number of axons (usually one), at one or a small number of endplate sites at stereotyped locations. 2.2. SPECIFICITY OF NEUROMUSCULAR CONNECTIONS This basic description still leaves unanswered the question of how the MNs become matched to their appropriate muscles during development. Are the growing axons actively guided to their correct destinations? How accurate is this process? Are mistakes made, and how are they corrected? Why do most of the MNs that differentiate ultimately die? And what accounts for the widespread elimination of supernumerary synapses at later stages? Several hypotheses have been proposed to account for the development of the mature pattern of neuromuscular connections. The motor axons might grow out and make connections with muscles randomly, following which each muscle somehow specifies the connections that its MNs will make in the CNS (myotypic specification) (Weiss, 1936). The existence of discrete pools of MNs matched to specific muscles makes this highly unlikely; but it could still be the case if connections are made in a tightly programmed developmental sequence wherein the destinations of outgrowing axons are determined not by recognition by specific pathways or targets, but by the timing of their emergence from the cord, by their topographic relationships to each other, and by mechanical cues produced by successive elements encountered in the developing limb (H0rder, 1978). Important as these factors may be, evidence obtained in the past few years strongly supports the view that there is true matching specificity between MNs and target muscles, and chemospecific recognition of appropriate pathways by growing axons as well. Our increased understanding of developmental specificity can be attributed in large measure to the introduction of the techniques of retrograde and orthograde axonal transport of horseradish peroxidase (HRP) (Saito and Zacks, 1969; Lamb, 1976; Landmesser, 1978a,b; Hollyday, et al., 1977). Different methods of HRP staining allow one to SPECIFICITY AND PLASTICITYOF NEUROMUSCULAR CONNECTIONS 207 identify selectively (a) the cell bodies of neurons innervating a localized region of HRPinjected muscle, (b) the pathways followed by axons of neurons near a site of HRP injection in the spinal cord, or (c) the pathways of axons that have been damaged peripherally in the presence of HRP. 2.2.1. Neuromuscular development in the chick 2.2.1.1. Specificity of nerve-muscle matching The most careful analysis of neuromuscular development has been done in the chick embryo, especially by Landmesser and her collaborators, building on the classical work of Hamburger (1939, 1958, 1975b, 1977; Hamburger and Levi-Montalcini, 1949). Landmesser and Morris (1975) made careful measurements of muscle contraction and muscle nerve activity in response to stimulation of different spinal nerves. They found that, at the earliest developmental stage at which they could elicit contraction (stage 28, 5 1/2 days), the pattern of contraction, the electrical activity of the nerve branches, and the absence of axon reflexes suggested that the connections were as specific as in the mature animal. A similar electrophysiological study of the distribution of axons from different spinal nerves innervating developing chick wings that had deletions or truncations showed that, whatever the missing portion of the developing wing, the nerve branching pattern was normal for the remaining portion, and the remaining muscles were still innervated specifically by the normal distribution of spinal nerves (Stifling and Summerbell, 1977). The axons whose normal targets were absent appeared to be missing, but this was not determined directly. These findings of specific innervation have been strongly reinforced by the application of retrograde HRP labeling. After mapping MN-muscle connections in the fully developed chick hindlimb (Landmesser, 1978a), Landmesser (1978b) mapped projections in the same way at stage 28, shortly after the first functional neuromuscular connections are made (about stage 26) and before the period of cell death (Hamburger, 1975b; ChuWang and Oppenheim, 1978b). There was no significant difference in projection from that seen in the mature chick cord. At most, a small percentage of axons could have been outside their normal adult projection and gone undetected. Pettigrew, Lindeman, and Bennett (1979) claim that such aberrant projections are present in the chick wing at stage 26, but are subsequently lost. Using HRP-labeling and electrophysiological recording, they reported that between stages 26 and 29 the limb is innervated by spinal segments 12-17, at stage 29 by segments 14-16, and at stage 31 and thereafter by only segments 14 and 15. The discrepancy between this finding and those of Landmesser cannot be explained with certainty. It is unlikely that the basic mechanisms of development differ in the wing and leg. Moreover, recent research by Lance-Jones and Landmesser (1981a), in which both orthograde and retrograde HRP transport were used to trace axonal pathways, shows convincingly that from the time the spinal nerves first approach the limb (stage 23), axons of any given segmental nerve follow a discrete and consistent pathway to the appropriate muscles, with no evidence of outgrowth to limb areas not innervated in the adult. Other recent work also supports the conclusion that there are few if any projection errors. Taking advantage of the observation that neuromuscular block with curare or ct-neurotoxin during development prevents most cell death (Pittman and Oppenheim, 1978), Oppenheim (1981) treated chick embryos in this way between days 5 and 10, from before synapse formation through the period of normal cell death. Most of the neurons survived, and HRP injections at day 10 showed that the projection patterns were the same with and without cell death, i.e. there did not seem to be widespread projections that were normally lost during cell death. Consequently, the possibility must be entertained that the findings of Bennett et al. (1979), implying diffuse innervation in the stage 26 wing, may be due to unanticipated spread of HRP from the site of injection 208 ALAN D, GRINNELL AND ALBERT A. HERRERA and electrophysiological pickup of electrical activity from a larger region of muscle than expected (see Lance-Jones and Landmesser, 1981a). Thus MNs appear to be specified accurately for particular target muscles before synaptic contact is made. This specification is presumably based on their locaton within the developing spinal cord. In this regard, it is noteworthy that the relative position of MNs in the mature chick spinal cord is well correlated with the position of their axonal terminations in the primary muscle masses, but not with the final position of the mature muscles they innervate (Landmesser, 1978a). Medially situated MNs project to the ventral muscle mass, lateral MNs to the dorsal muscle mass. Within each muscle mass, more anteriorly-located MNs project to the more proximal musculature. 2.2.1.2. Axonal pathway recognition Developing chick motor axons follow specific and stereotyped pathways to their appropriate muscles (Lance-Jones and Landmesser, 1981a). In principle, this might result from orderly timing of axonal outgrowth, continuing interactions between adjacent axons to maintain topographic relationships, and non-specific guidance of growth by the environment at successive stages of development (Horder, 1978). However, careful examination of axonal pathways in embryos that have had portions of the spinal cord or limb experimentally altered (Lance-Jones and Landmesser, 1980a,b, 1981b), provide convincing evidence for active chemospecific axonal recognition of appropriate pathways and targets. Using retrograde HRP transport to trace the axons innervating different muscles, Lance-Jones and Landmesser (1981a) found that axons of the 4-5 motoneuron pools emerging from the cord in any given segmental nerve tended to be extensively intermingled throughout the nerve. Only as they approach the cranial and ischiatic plexuses at the base of the limb, and pass through them into the limb, do the axons destined for any given muscle become spatially segregated. Thus the axons of a given muscle do not grow in a parallel array and maintain topographic relationship with each other in a coherent bundle. Instead, axons cross each other and move past large numbers of other axons on the way to their targets. These findings appear incompatible with exclusive determination of axonal pathways by the spatio-temporal factors proposed by Hordor (1978), although they do not preclude important non-specific control of nerve pathways in the limb itself. There is good evidence for such local control of nerve branching patterns within the limb. When a chick wing bud was rotated 180: distal to the base of the limb at stage 19, reversing both the anterior-posterior and dorsal-ventral axes, approaching axons from spinal nerves 14, 15, and 16 formed an apparently normal plexus at the base of the wing, and emerged from it in the normal pattern of dorsal and ventral nerves. On growing into the rotated portion of the limb, which was the opposite part of the limb than normal, these nerves branched in a pattern characteristic of that part of the limb, and innervated inappropriate muscles (Stirling and Summerbell, 1979). The gradual convergence of axons from a given motor pool during their transit through the distal spinal nerve, plexus, and proximal nerve trunk is a critical and fascinating aspect of specificity. A number of studies involving early embryonic manipulations have helped elucidate the rules followed by these axons. Lance-Jones and Landmesser (1980a) showed, for example, that when the MNs that would normally innervate the proximal limb musculature were removed at early stages by spinal cord deletions, axons destined for more distal muscles grew past the uninnervated muscles on the way to their appropriate targets (see also Castro, 1963). Thus even when a large proportion of the normal number of axons was absent, others found their way through the plexus and to their appropriate muscles successfully. These axons must have ignored any non-specific cues that normally guide growing axons to proximal muscles, as well as any attractive influence expressed by uninnervated muscle, to find their way to specific appropriate targets. In other experiments, Landmesser and her colleagues (Morris, 1978; Lance-Jones and Landmesser, 1980b, ]981b) have used orthograde and retrograde transport of HRP, and SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 209 electromyographic recording, to trace fiber pathways after early anterior-posterior rotations of the chick lumbosacral spinal cord, limb transplantations, or addition of supernumerary limbs. Within the reversed spinal segments, M N pools developed with the same size, shape and topographical position as normal, but in the mirror-image locations. If relatively small shifts were made in the segmental position of the hind limb, or only 3-4 spinal segments were rotated, spinal nerves could enter the normal plexus, albeit from an abnormal direction. Such displaced axons found new pathways through the plexus to enter the limb in the correct nerve branch and reinnervated their appropriate muscles (see Fig. 1). With larger limb displacements, or large spinal cord rotations, and in supernumerary limbs where the limb was innervated by abnormal spinal nerves, axons typically did not show any ordered pathway through the plexus, and innervated inappropriate muscles (Lance-Jones and Landmesser, 1981b). Interestingly, even in these cases, the nerve plexuses and limb nerves were grossly normal in appearance, emphasizing that the overall nerve branching pattern may in fact be determined primarily by non-specific morphogenetic factors. With respect to these factors, Bennett, Davey, and U~bel (i980), on the basis of an electron microscopic analysis of axonal growth and plexus formation in stage 22-26 chick embryos, suggest that nerve pathways in the limb and plexus are formed by attraction of growing axons toward sequentially differentiating muscle masses, by something analogous to nerve growth factor. On the other hand, with such limb rotations, or with large displacements of spinal cord or •target, individual M N pools innervate many muscles. The diffuse nature of innervation in these cases indicates that the morphogenetic determinants that govern nerve location and limb branching pattern cannot respecify MNs or direct them to appropriate targets (Lance-Jones and Landmesser, 1981b). Lance-Jones and Landmesser (1981b) made the additional important observation that in some cases, even when a spinal nerve did not find its way to the apprapriate plexus, and so entered the limb in an abnormal position, many of the axons from a given M N pool diverged from this abnormal pathway and found their way to the appropriate muscle via an entirely aberrant projection. When a muscle was innervated by a novel nerve branch of this type, the branch contained only appropriate axons. Based on these experiments, it can be concluded that growing motor axons can recognize and actively respond to guidance cues that are present in the region of the plexuses and the base of the developing limb. The cues are apparently local, although whether they are in the extracellular matrix, the pre-muscle mesenchyme, or some other cell type is not known. When an axon enters the region of its normal plexus, it apparently finds signals that guide it to the correct pathway, sometimes via aberrant nerve branches in the base of the limb. However, if the axon is sufficiently displaced so that it enters a different plexus, its further growth is not specific, although it does run in nerves whose distribution is characteristic for the limb region it has entered. If an axon enters the limb in the appropriate nerve trunk, it finds a sequence of powerful cues that guide it to the appropriate muscle, even if spinal cord deletions have left other muscles along its path uninnervated. (It would be of interest to combine spinal deletions with removal of specific muscles to see if axons would innervate inappropriate muscles in the absence of their normal target.) If axons enter the wrong part of the limb, they grow along pathways characteristic of that part of the limb, and innervate inappropiate muscles, apparently randomly. There is some evidence, on the other hand, that a secondary form of specificity may exist in cases where the displacement of spinal cord segments or target tissue is so great that few if any axons can find their appropriate targets. If the lumbar and brachial segments of the chick spinal cord are interchanged early in development, the limbs become innervated with peripheral and interneuronal connections that cause the wing to move like a leg and vice versa (Straznicky, 1963). In studies of the innervation patterns of supernumerary hind limbs innervated by inappropriate spinal segments, Morris (1978) and Hollyday et al. (1977) found consistencies that implied either that the limb was respecifying the MNs (Morris, 1978), or that, in the absence of the appropriate muscle for 210 ALAN D. GRINNELL AND ALBERTA. HERRERA Experimental Control st=~ 23~ LS 1 s~ 2s i T7 LSI ,k.x.. s t ~ 241 2 3 4 5 6 78 3 s~ 2 3 4 5 6 7 8 2 1 T7 4 5 6 7 8 2s, 3 2 1 T7 4 5 6 7 8 I~ S, rtoaus-" ?;F FIG. 1. Novel pathway selection by chick motor axons to their correct targets following reversal of a section of the spinal cord containing the 7th thoracic and first three lumbar segments. Motoneurons of segments T7 and LI were labelled by direct HRP injection into the cord, and their axons traced at stages before (top) and after (below) the onset of motoneuron cell death and muscle mass cleavage. Note that in the experimental {cord reversed) embryos, most of the axons follow highly abnormal pathways to reach the correct muscle nerves, while avoiding incorrect branches. Calibration bar = 255 pm. (From Lance-Jones and Landmesser, 1980b). each nerve, axons made specific second choices (Hollyday et al., 1977). Thus the relative" cranio--caudal position of motor pools supplying individual muscles throughout the limb was conserved and MNs innervating specific muscles occupied a similar medial to lateral position in the cord in both normal and supernumerary limbs. In a more recent analysis of axonal pathways in similar supernumerary limb preparations, Lance-Jones and Landmesser (1981b) confirmed this type of organization, but emphasized the lack of patterned innervation from inappropriate cord segments. Thoracic segments innervate the limb very sparsely and innervation of individual muscles is often diffuse. However, many instances were found of aberrant nerve branches bringing axons from the nearest region of the plexus of the normal limb to their appropriate muscles in the supernumerary limb. Thus the signals guiding the axons to their appropriate muscles appear to be very strong; in the absence of the correct signal, connections show little order and many permanent "erroneous" projections are formed. There is already a large and exciting field of research directed at understanding possible mechanisms by which axons recognize appropriate pathways. We will not review this work here, but can summarize some of the concepts that are emerging: There is a tendency for axons with similar origins and destinations to fasciculate and remain associated as they grow to their targets. It is also clear that growing axons will follow nonspecific mechanical cues. However, especially where axons diverge, sort out, and find abnormal routes back to normal pathways (as occurs in the neuromuscular system), it appears necessary to invoke some form of chemoaffinity guidance (Sperry, 1963). It can be postulated that each neuron responds in positive or negative ways to a number of morphogenetic gradients in order to choose the pathway of highest affinity from those available. Considerable progress has already been made in demonstrating differential affinities among different cell types in several systems, and in identifying neuron- or target cell-specific surface molecules, mostly glycoproteins and glycolipids (see reviews in SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 211 Moscona, 1974; Barondes, 1976; Pfenninger and Mayli6-Pfenninger, 1978, 1981; Gottlieb and Glaser, 1980). It is of particular interest that different monoclonal antibodies have already been made that are specific to different neurons in the leech ganglion (Zipser and McKay, 1981). Application of this powerful technique to vertebrate nerve and muscle may provide important clues to mechanisms of specificity (see McKay et al., 1981 for a general review of techniques and recent progress). 2.2.2. Neuromuscular development in amphibians The pools of MNs innervating different muscles are less discretely organized in amphibians than in the chick (Sz6kely and Cz6h, 1967). Consequently, it is more difficult to map specific projections, or to detect developmental errors. However, Lamb (1976), who initiated the use of H R P retrograde transport for this purpose, found that the first motor axons that grow into the developing hind limb of Xenopus (stage 51 ; see Prestige, 1973) all send branches into muscle mesenchyme at the base of the limb, although there is no evidence that functional connections are present in all cases. Soon thereafter (stage 53), as more distal musculature differentiates, the projections from caudal parts of the cord are lost in the proximal muscles and are found in the distal muscles. Lamb (1977) showed that the early, inappropriate projections were lost due to cell death rather than to withdrawal and relocation of branches, while subsequent connections showed few if any apparent projection errors. In a similar study of hind limb innervation in axolotls, even more diffuse initial innervation was described by McGrath and Bennett (1979). At the earliest stages examined (hind limb flexor muscle mass 1.5 mm long), spinal nerves 15 and 17 both projected to all areas of the flexor mass as judged by the presence of endplate potentials (EPPs). By the time the flexor muscle mass had grown to 5.5 mm (of an adult length of 22 ram), the innervation pattern was almost adult with most of the musculature receiving input from only one or the other of these two nerves. These findings suggest widespread projections by each spinal nerve, with subsequent refinement of connections based on survival of only the appropriately matched axons. However, careful interpretation is needed, and some additional information would be useful. No mention is made, for example, of the possibility that electrotonic coupling between muscle cells at early stages might be increasing the apparent area of each projection (see Kullberg et ai., 1977; Dennis, 1971; Dennis et al., 1981). Moreover, the loss of "erroneous" projections might be due to real elimination, by cell death, of axons that project to the wrong part of the muscle mass, or it might be due simply to the interruption of branches that extend beyond an axon's "correct" territory when cleavage planes form in the muscle mass (Landmesser, 1980). Disappearance of the erroneous projections coincides approximately with the time of cleavage of the mass into component muscles (Sullivan, 1962; Dunlap, 1966). 2.2.3. Neuromuscular development in mammals The accuracy of initial innervation pattern has been little studied in mammals, with the exception of a careful study of rat intercostal innervation by Dennis and his colleagues (Dennis and Harris, 1979; Dennis et al., 1981). They found that the vast majority of axons projected to the appropriate segmental muscle. However, occasionally one motor unit of an intercostal muscle was found to be driven by an axon branch from the nerve that predominantly innervated the next more posterior intercostal muscle. At 17 days of gestation, at least one such branch was seen in about 40~o of the intercostals examined. By the 20th day, the incidence has fallen to about 10~, and there is no further change with age. The loss of these anterior projections occurs with the same time course and to approximately the same extent as synapse elimination in the "appropriate" muscle (see Section 2.4.2.), so it cannot safely be interpreted as representing specific elimination of 212 ALAN D. GRINNELLAND ALBERTA. HERRERA developmental errors (Dennis et al., 1981). Moreover, since the intercostals normally contract together, it is questionable whether aberrant branches are in fact "errors", in any sense. 2.2.4. Summary Viewing all of these results, it appears valid to conclude that there is, indeed, matching specificity between developing motoneurons and target muscles. Motoneurons, and/or their axons, are genetically programmed to recognize not only the appropriate target muscle fibers, but to follow actively a succession of cues guiding them to the correct part of the limb. There is little convicing evidence for "erroneous" neuromuscular projections or connections in the normally developing chick. With large displacements of spinal segments or target tissue, on the other hand, inappropriate connections form and survive indefinitely. In frogs, most connections develop appropriately, but there are some early connections that appear to be inappropriate and do not survive the period of cell death. It is possible that the phenomenon in frogs represents a useful peripheral model for similar elimination of developmental errors in the central nervous system (CNS), for example in the projections of axons from the isthmo-optic nucleus (Clarke and Cowan, 1976), and the gradual segregation of binocular inputs to visual centers (Hubel, Wiesel and LeVay, 1977; Rakic, 1977; Innocenti et al., 1977; LeVay et al., 1978; Shatz and Stryker, 1978; Hubel, 1978). In the CNS, the elimination of mistakes is accomplished in some cases by cell death (perhaps due to lack of appropriate feedback from postsynaptic cells, or appropriate functional reinforcement), in other cases apparently by migration of postsynaptic neurons or retraction of terminals. On the other hand, it is important to emphasize that most neuromuscular connections appear to be made correctly, and it may not be valid to describe the few abnormal early neuromuscular projections that have been seen as "errors". The earliest differentiating neurons in the ventral spinal cord may have a developmental role unrelated to final connectivity. They might, for example, provide pathfinder axons whose functions is to find the correct route to the base of the limb, and be programmed to die after follower axons have gone on from there to find their way to the appropriate targets using cues built into more distal structures. This would necessarily represent only a small fraction of the total cell death, since nearly half the motoneurons in the developing chick cord die (Hamburger, 1975a; Chu-Wang and Oppenheim, 1978b). Early axons might also play an important role in trophically facilitating and guiding differentiation of muscle independent of either a pathfinder function or the traditional function of coordinated driving of specific muscles. 2.3. NEURONAL DEATH DURING DEVELOPMENT During a brief period soon after the full complement of MNs has differentiated and while peripheral synapse formation is in progress, a high percentage of the MNs die: 80-90~o in amphibians (Hughes, 1968; Prestige, 1967), 40-75 ~o in mice (Harris-Flanagan, 1969) and about 4 0 ~ in chicks (Hamburger, 1958, 1975a). There have been two principal functions postulated for the vast overproduction of neurons and their subsequent death: (1) the elimination of developmental errors (Hughes, 1968) and (2) provision of sufficient redundancy to guarantee that all target cells are adequately innervated (Hamburger, 1939; Hamburger and Levi-Montalcini, 1949; Hamburger, 1977). To these can be added the possibility that large numbers of immature neurons are necessary in some unknown way for developmental induction or trophic interaction either within the nervous system or in relation to target tissues (Prestige, 1970). In frog embryos at least, cell death does eliminate "erroneous" projections as described in the last section. However, in chick embryos, and perhaps in mammals as well, there are remarkably few aberrant projections, while large numbers of apparently correctly SPECIFICITY AND PLASTICITYOF NEUROMUSCULAR CONNECTIONS 213 projecting neurons die (Landmesser, 1978; Oppenheim, 1981). In embryos with rotated cords, many incorrectly projecting MNs survive even when the muscles they innervate also receive correct projections (Lance-Jones and Landmesser, 1981b). Such quantitative arguments urge another function for cell death. 2.3.1. Quantitative dependence on target size The redundancy hypothesis has received much support. A frequent tenet of this hypothesis is that the cells that die are those that do not make connections successfully in the periphery. Hamburger and Levi-Montalcini (1949) concluded that death in sensory ganglia was restricted mainly to young cells that had not innervated the periphery, and the same may be true of motor connections. It was shown early in the century (Shorey, 1909) that amputation of a limb in a developing chick embryo results in abnormally great loss of both MNs and sensory ganglion cells. Extensive work by Detwiler (see review, 1933) on amphibian embryos, and by Hamburger and his colleagues (Hamburger, 1975a,b, 1977; Hollyday and Hamburger, 1976; see also Chu-Wang and Oppenheim, 1978a,b) on chick embryos, established that there is a clear quantitative relationship between the amount of target tissue and the number of surviving neurons. Removal of a chick or salamander limb results in almost total disappearance of MNs from the corresponding ventral horn segments. If a supernumerary limb is implanted, there is a less dramatic (up to 25-30~) increase in the number of surviving MNs, and maturation is accelerated (Hughes and Tschumi, 1958; Prestige, 1967, 1970; Hollyday and Hamburger, 1977). Interestingly, in amphibians, one characteristic population of small ventral horn cells, possibly MNs that have not established functional contact with the periphery, disappears within 2-3 days after amputation (if this is done during the time of naturally occurring cell death). Another population of larger cells also shows increased total cell death, but only after a period of weeks or months (Prestige, 1970). In this respect they act more like mature MNs. No such distinction between MNs has been drawn in the chick. 2.3.2. Evidence for dependence on synaptic function in the periphery Although these findings are consistent with the redundancy hypothesis, it is clearly an oversimplification to say that only those cells die that do not successfully innervate their target tissue. As was noted above, some of the MNs that disappear during amphibian development are among the first to reach the periphery (Lamb, 1976, 1977; see also McGrath and Bennett, 1979). In general, HRP-labeling studies have shown that most MNs project to their targets at the time of cell death (Clarke and Cowan, 1976; Landmesser, 1978b; Chu-Wang and Oppenheim, 1978b). On the other hand, it cannot be concluded with certainty that many of the neurons that die have made functional synapses in the periphery, since axons will pick up HRP whether or not they have formed functional synapses (Heaton, 1977) and synapses are difficult to demonstrate morphologically at early stages. Neuronal cell death also plays an important role in autonomic ganglia (see Purves and Lichtman, 1978, for review). Here too, it has been shown that many cells that die have already made peripheral connections (Pilar, et al., 1980), and that ganglion cells, like MNs, can be "saved" by increasing the amount of peripheral target tissue (Narayanan and Narayanan, 1978) or by partial denervation to reduce the number of ganglion cells projecting to the target in the first place (Pilar et al., 1980). In the latter case, a much higher percentage of the remaining neurons survive than would normally have been the case for that population, and they show abnormally fast maturation as well (in axon diameter, conduction velocity, and degree of glial ensheathment). Thus in the ciliary ganglion, the competition that results in the death of some neurons delays the maturation of those that survive (Landmesser, 1979). The evidence for peripheral regulation of neuron survival, during the period of normal cell death, is compelling. This dependence develops only at the time of peripheral synapse formation (Hughes and Tschumi, 1958; Prestige, 1967), and the timing and extent of MN 214 ALAN D. GRINNELL AND ALBERT A. HERRERA cell death are dependent on synaptic function and muscle activity. Pittman and Oppenheim (1978; see also Oppenheim and Majors-Willard, 1978) found that block of synaptic action either presynaptically (botulinum toxin, BoTX) or postsynaptically (curare or ct-neurotoxin from the cobra Naja naja) resulted in much reduced cell death, a result confirmed by Laing and Prestige (1978), Creazzo and Sohal (1978) and Olek (1980). Most interestingly, this rescue is apparently permanent if the paralysis extends beyond the normal period of cell death. Somehow, the normal musculature is then able to maintain an increased number of MNs. Synaptic block after the period of cell death had no effect on the number of MNs remaining, while a partial dose, which lost its effect during the cell death period, caused a delay in cell death until function resumed, at which time there was accelerated loss until the number of MNs fell to normal levels (Pittman and Oppenheim, 1978) or even below normal (Laing and Prestige, 1978). lit should be noted that BoTX results in muscle atrophy associated with profuse motor innervation, many endplates, and high levels of choline acetyltransferase (CAT), all consistent with a decrease in cell death (Giacobini-Robecchi el al., 1975); while Naja nigricolis cobra toxin causes muscle atrophy associated with severe loss of evident innervation (Giacobini et al., 1973), reflecting a very different mode of action on the MNs, despite the similar effect on cell death.] The action of these toxins is discussed further below, in the context of nerve terminal sprouting. Interestingly, fl-bungarotoxin, which blocks transmitter release by a mechanism different than that of BoTX, appears to cause enhanced MN death (Olek, 1980). It is known to cause degeneration of nerve terminals and cell death even in mature animals, however, so this effect may have obscured any tendency for increased survival due to reduced neuromuscular activity. It must always be borne in mind that each of these toxins could have a variety of effects, most of which are not known. The coincidence of cell death with peripheral innervation and the dependence on target mass suggests that MNs compete for some essential factor. Many laboratories are now searching for a t r o p h i c substance that promotes growth and survival of MNs, analogous to the profound effects of nerve growth factor on sympathetic and dorsal root ganglion neurons (Black, et al., 1972; Campenot, 1977; Letourneau, 1978; Gunderson and Barrett, 1979; Ebendal et al., 1980). There have been promising developments in this direction. Most noteworthy is the demonstration that media conditioned by contact with heart or skeletal muscle cells contain one or more substances that promote maintenance and neurite outgrowth from cultured ciliary ganglion cells (Nishi and Berg, 1977, 1979; Bennett and Nurcombe, 1979; McClennan and Hendry, 1980; Ebendal et al., 1980) or explants of chick spinal cord, retina, or optic tectum (Obata and Tanaka, 1980). Two separate components have been obtained from chick embryo eye extracts that affect ciliary ganglion cells in culture. One (MW 2 × l04) stimulates growth of neurons but not production of CAT; the other (MW 5 × l04) increases CAT activity per neuron, but does not infuence growth (Nishi and Berg, 1981). Whether or not such substances are involved during the period of cell death, it is generally believed that certain cells are saved by physiological support at the synapse and others are lost, either because they do not receive sufficient support or because they are actively forced to withdraw from synaptic contacts by virtue of some change brought about by activity of another nerve terminal. Upon considering these descriptions of the regulation of MN death, there seem to be some puzzling (and doubtless ultimately informative) inconsistencies. Before peripheral synapses are made, proliferation of MNs and outgrowth of axons are independent of the periphery. At a certain time, however, these neurons become highly vulnerable,, and appear to die if they do not compete successfully for something associated with effectiveness at the periphery. Ablation of a limb causes death of most of the MNs destined to innervate that limb. Even in the presence of the limb, there is massive cell death if the MNs are prevented mechanically from reaching it (Hughes and Tschumi, 1958). On the other hand, blocking synaptic function, which might appear to be equivalent to preventing innervation in the first place, results in much reduced cell death. And if this block extends beyond the period of normal cell death, MNs that would normally die appar- SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 215 ently survive indefinitely (Pittman and Oppenheim, 1978). One could speculate that with time the neurons simply lose their dependence on the peripheral trophic influence and/or susceptibility to competitive influences from other axons, or that the muscle cells become capable of maintaining more inputs [which seems unlikely in view of the atrophy and retarded differentiation seen in such hyperinnervated muscles (Giacobini et al., 1973; Giacobini-Robecchi et al., 1975; Pittman and Oppenheim, 1978; Creazzo and Sohal, 1978)]. 2.3.3. Evidence against peripheral determination of cell death These are not the only inconsistencies. There is now evidence that the fundamental limit to how many neurons will survive may not be set by the size of the peripheral target, or peripheral competition, after all. Lamb (1980), using Xenopus larvae, ablated one hindlimb before the period of cell death, and directed projections from both sides of the spinal cord into the remaining limb bud. The ipsilateral cord innervated the limb in normal fashion, with normal amounts of cell death. However, the contralateral side of the cord also innervated the limb, in most cases nearly as extensively as it would a normal limb. There was no reciprocal relationship apparent between the extent of contralateral and ipsilateral innervation. The limb was not hypertrophied, although the muscles were innervated by more than the usual number of axons. Injection of H R P into the gastrocnemius in several animals showed that this muscle received innervation from the appropriate pools of MNs on both sides of the cord. No information is available concerning the density of innervation, motor unit size, or the possibility of contralateral innervation of other muscles. However, the clear implication of these interesting results is that the competition that is normally responsible for the death of certain MNs does not operate between corresponding populations on the two sides of the cord. Alternatively, we may be overestimating the role of competition and underestimating the precision of peripheral connections. If specificity requirements are so exact that MNs of a given pool cannot survive unless they innervate a certain part of their appropriate muscle, then Lamb's (1980) results could still be explained in terms of peripheral regulation. The limit may not be how many neurons can innervate a muscle, but how many axons from either side of the cord find the correct region of an appropriate muscle. Lamb (1979) also showed that in Xenopus the knee flexor muscles receive projections from most of the lumbar segments of the ventral column before cell death, but not thereafter. The more caudal MNs projecting to these muscles die or lose those axon branches. When the more rostrally located MNs are ablated at early stages, so that the caudal MNs are the only ones innervating these muscles, their projections still disappear during cell death, leaving the knee flexors uninnervated. This result, also, is diflicult to explain on the grounds of peripheral competition, but seems to require some overriding influence of specificity or central interaction. Lamb (1979, 1980) emphasizes the possibility that MNs may be competing for afferent connections or blood supply in the spinal cord. There are cases in other systems where the survival of neurons, especially during development, seems to depend on the integrity of afferent inputs (Levi-Montalcini, 1949; Clarke and Cowan, 1976; Sohal, 1977). In the case of MNs, it is probable that the important afferents would be from interneurons, which are functional during much of the period of cell death (Bekoff, 1976), rather than primary afferents, which do not become effective until near the end of the cell death period (Landmesser, 1980). On the other hand, the arrival of primary afferents may somehow help trigger the phenomenon. It should also be borne in mind that all of these processes may be strongly regulated by hormones. In an amphibian embryo, for example, hypophysectomy essentially halts development (and cell death), while thyroxin accelerates the rate of both (Beaudoin, 1956; Race, 1961). Interestingly, the accelerating effect of thyroxin in a hypophysectomized tadpole is significantly delayed on the side where a limb has been amputated. It is possible that thyroxin, in accelerating cell death on the normal side, does not act directly on MNs, but rather increases the rate of normal differentiation and death of those cells that are inappropriate or suffer a competitive disadvantage, either centrally or peripherally. 216 ALAN D. GRINNELLAND ALBERTA. HERRERA R E M O V A L OF SYNAPTIC BLOCK I 100 -- SUSTAINED SYNAPTIC BLOCK i "... tr (,9 o9 z 0 nLu z (5) \\ - ..: _ _ 50-- INCREASED PERIPHERY NORMAL (4) (3) (1) \ \ \ "" 0 I B E G I N N I N G OF SYNAPSE FORMATION D E C R E A S E D P E R I P H E R Y OR MECHANICAL OBSTRUCTION (2) BIRTH TIME FIG. 2. Characteristic features of motoneuron (MN) cell death during development, based on experiments with frog, chick, and mouse embryos. Normally, beginning about the time of nervemuscle contact, the number of MNs rapidly declines. In chick and mouse embryos about 40-75~o of the MNs die within a few days (1). If the peripheral targets of the MNs are removed, or if motor axons are mechanically prevented from reaching their correct targets, almost all of the MNs may die (2). Adding target tissue significantly increases MN survival (3). Interestingly, if the activity of neuromuscular synapses is pharmacologically blocked during the period when cell death would normally occur, few MNs die (4). This may imply that during the critical time of MN cell death, competition between axons is important in determining survival. If the synaptic block is applied early, but removed during the period of cell death, subsequent MN loss occurs at a faster than normal rate, resulting in normal, or even subnormal numbers of neurons (5). Removal of synaptic block after the period of cell death does not affect MN numbers. Representative references: (1) Hamburger, 1975a, 1977; Prestige, 1967, 1970; Hughes, 1968; Harris-Flanagan, 1969. (2) Detwiler, 1933; Hamburger, 1958, 1975a, 1977: Hollyday and Hamburger, 1976: Chu-Wang and Oppenheim. 1977a.b; Hughes and Tschumi, 1958. (3) Hughes and Tschumi, 1958: Prestige, 1967, 1970; Hollyday and Hamburger, 1977. (4) Pittman and Oppenheim, 1978: Oppenhelm and Majors-Willard, 1978: Laing and Prestige, 1978; Creazzo and SohaL 1978: Olek, 1980. (5) Pittman and Oppenheim, 1978; Laing and Prestige, 1978. 2.3.4. Summary The cause of motoneuron (MN) cell death is not known. However, it is a dramatic phenomenon of great impOrtance in the morphogenesis of the central nervous system, and an extreme example of MN plasticity. Figure 2 summarizes many of the relevant findings. During the period of normal cell death, something determines whether a neuron will survive or die. Specificity may be involved, causing the selective death of inappropriately connected MNs. However, unless survival depends on innervation of the appropriate part of the correct muscle, this would appear to account for only a very minor component of cell death, since most cells that die have already projected to the appropriate muscle. For the vast majority of neurons, survival appears to depend on successful innervation of target tissue, except perhaps in the special case where all MNs are prevented from making functional connections. Many experiments, and the analogy with systems that depend on nerve growth factor, suggest that neurons compete for a trophic substance supplied by the periphery. However, the recent finding that Xenopus muscle can be hyperinnervated by MNs from both sides of the cord suggests that spinal interactions between MNs may be even more important than peripheral regulation. We are woefully ignorant of the development of synaptic interactions in the spinal cord, and SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 217 their possible trophic importance. These must not be neglected. However, it is difficult to see how competitive interaction within the cord could explain many of the results obtained with changes in peripheral target or neuromuscular synaptic block. Until more information is available, it is probably safer to suppose that MN death is the result of a number of influences including rejection of mis-matched projections, competition for influence on the target cells, and competition for afferents, space, or other factors in the spinal cord. 2.5. SYNAPSE ELIMINATION Even though cell death causes extensive changes in the innervation pattern of muscle, neuromuscular connectivity is Still far from its mature state. The total number of MNs is set, and they project selectively to the appropriate muscles. However, muscle fibers are still being added in the periphery, and there is profuse branching and remoclelling of individual axonal fields. Each muscle fiber acquires inputs from several different MNs, and any given MN may innervate a large fraction of the fibers in any given muscle. Then, characteristically beginning near the time of birth, there is a period of synapse elimination (SE), resulting in loss of most or all of the multiple inputs. The phenomenon of SE is a fascinating and important one that continues the developmental adjustment of the nervous system to its targets. In particular, through this process the quantitative match between MNs and periphery is refined into its adult pattern of motor unit organization. Since a comparable phenomenon of multiple innervation followed by SE can occur during neuromuscular regeneration, and competitive interaction between different axonal terminals on the same muscle fibers is probably involved in both, SE has important implications for mechanisms of regulation of synaptic efficacy in general. 2.4.1. Generality of the phenomenon SE was first described in mammalian muscle (Redfern, 1970), but appears now to be a common, if not general, characteristic of nervous system development. It is seen in chick (Bennett, Pettigrew and Taylor, 1973; Holt and Sohal, 1978; Srihari and Vrbov~, 1978; Renaud et al., 1978; Pockett et al., 1979) and amphibian (Bennett and Pettigrew, 1975; Letinsky and Morrison-Graham, 1980; Morrison-Graham, 1981) as well as mammalian muscle (see full discussion below). It is well documented in autonomic ganglia of mammals and birds (Lichtman, 1977, 1980; Lichtman and Purves, 1980; Lichtman et al., 1980; Purves and Lichtman, 1980a, b; Smolen and Raisman, 1980), in the synaptic input onto MNs (Conradi and Skoglund, 1969; Ronnevi and Conradi, 1974; Conradi and Ronnevi, 1977; Ronnevi, 1977), and in the cerebellum, where all but one climbing fiber input to Purkinje cells are eliminated (Crepel et al., 1976). It is possible that this form of peripheral synaptic remodelling will prove to be a good model for understanding the refinement of connections in the CNS during postnatal critical periods. The sharply defined ocular dominance organization of the primary visual cortex, for example, appears to be achieved by secondary elimination of initially widespread and overlapping projections from the lateral geniculate (Hubel and Wiesel, 1970; Hubel, et al., 1977; Rakic, 1977; LeVay et al., 1978; Shatz and Stryker, 1978; Hubel, 1978). This shaping of projections begins before birth, but is completed postnatally. Moreover, during the postnatal critical period one population or the other can be rendered ineffective reversibly by restricting visual experience, with correlated changes in projection pattern (Hubel et al., 1977; Movshon, 1976; Blasdel and Pettigrew, 1978). Since neuromuscular SE is also strongly influenced by levels and patterns of synaptic activity, it is of obvious interest to understand the mechanisms of SE in this accessible system, and what factors affect it. 2.4.2. Timing and extent of synapse elimination In mammalian muscles, up to 4-5 axons innervate each fiber at birth, or just before, and all but one are eliminated over the next few weeks (Redfern, 1970; Bagust, et al., 1973; Bennett and Pettigrew, 1974a; Benoit and Changeux, 1975; Brown, et al., 1976; 218 ALAN D. GRINNELL AND ALBERT A. HERRERA Jansen, Van Essen and Brown, 1975; Riley, 1977; Rosenthal and Taraskevich, 1977; Thompson, et al., 1979; O'Brien, et al., 1978; Betz et al., 1979, 1980b; Miyata and Yoshioka, 1980; Dennis, et al., 1981). Because of difficulties in resolving 3rd, 4th, and higher order inputs to a single endplate, most studies have documented only the rate of appearance of single innervation, and not the loss of higher order inputs. The study of Dennis et al. (1981) is an exception, showing that in rat intercostal muscles, the peak of polyneuronal innervation is reached at day 17 of gestation (with more than 3 inputs/ endplate), begins to decline at day 19, and is completely eliminated by postnatal day 11. In a systematic study of the timing of SE, Bixby and Van Essen (1979a) showed that the transition from multiple to single innervation happens at different times in different muscles in the rabbit, generally in a sequence corresponding to the order in which muscles differentiate, and at rates differing by up to 3X. The diaphragm undergoes SE about 1 week before the soleus, and at a slower rate. Other muscles are intermediate in both respects. This indicates that SE is not likely to be a response to some synchronous, body-wide signal. In the rat, with a. shorter gestation period, SE in the diaphragm and soleus are separated by only about 1 day, and occur at a rate 2-3X faster than in the rabbit (Bennett and Pettigrew, 1974a; Brown et al., 1976; Rosenthal and Taraskevich, 1977; Bixby and Van Essen, 1979a; Bixby, 1981). In chick development, SE is essentially complete before hatching (Srihari and Vrbov~i, 1978). It is probable that there is a prolonged period of dynamic change in the number of branches an axon forms and the terminals formed by each branch. This may be continuous from the period of cell death until SE is completed. It is generally agreed, however, that during the main period of SE there is no associated neuronal cell death or decrease in the number of axons innervating the muscle (Brown et al., 1976; Thompson and Jansen, 1977; Oppenheim and MajorsWillard, 1978; Prestige, 1967; but see Romanes, 1946). 2.4.3. Specificity o f synapse elimination, and changes in motor unit size As would be expected, motor units are very large and overlap extensively before SE (Bagust, et al., 1973; Brown et al., 1976). There is no difference between the number of motor units in the rat soleus before and after SE, but the mean motor unit size falls from about 23~0 of the muscle to the adult value of 4-5~o (Brown et al., 1976; Jansen et al., 1975). At 2-3 days after birth, there is more variability in the size of the motor units than in adult muscles: a range of about 10 fold in neonatal soleus, 3 fold in adults (Close, 1967; Brown et al., 1976). On the basis of early experiments in rats, it was concluded that SE affected predominately the larger motor units, with small motor units losing a smaller proportion of their terminals, if any (Brown et al., 1976; Jansen et al., 1975). However, in more extensive recent studies of SE in the rabbit soleus, it is clear that small units are not so favored; they may lose a higher proportion of their terminals than do the larger motor units (Gordon and Van Essen, 1981). Except for the claim (which may not prove to be true) that large motor units are more strongly affected during SE than are small motor units, there has been only one report implying that there may be specificity in which terminals are lost. Miyata and Yoshioka (1980) reported the selective elimination of terminals from axons of one of the two spinal nerves innervating the rats soleus, not the other. Using twitch tension measurements, they found that, at 6--9 days postnatally, spinal nerve L4 innervated virtually all of the soleus fibers, while L5 innervated only about 45-50~o. During SE, the percentage of fibers innervated by L4 dropped to about 50~o, and L5 continued to drive about 50~. Thus, in this much studied muscle, it is suggested that SE selectively affects the axons of L4. Miyata and Yoshioka propose several possible explanations for the selectivity, based on differences in motor unit size and timing of arrival of innervating axons. Most of these possibilities are experimentally testable, and may yield important insights into mechanisms of SE. On the other hand, rabbits again provide conflicting data. The rabbit soleus SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 219 is innervated by spinal nerves S1 and $2, with mean motor unit sizes on postnatal days 0 to 5 of 5.5 and 5 j°o of the whole muscle tension, respectively. H. Gordon and D. Van Essen (personal communication) find that motor units formed by both nerves shrink by approximately the same amount, to 1.8~ (S1) and 1.9~ ($2) by days 10 to 18. Morphological studies show a decrease in the number of preterminal axons from several to one during this period (Korneliussen and Jansen, 1976; Rosenthal and Taraskevich, 1977; Riley, 1977; Zelenfi et al., 1979). Although Rosenthal and Taraskevich (1977) reported seeing some degenerating terminals and apparent phagocytosis, most investigators have seen no evidence of degeneration of terminals during this process, leading to the suggestion that the eliminated terminals are being displaced and retracted (Korneliussen and Jansen, 1976; Riley, 1977, Bixby, 1981). B~xb-y0981), i-n a particularly thorough study of the phenomenon in rabbits, found no evidence for terminal degeneration. He also found that SE could not be correlated with myelination of preterminal axons, which had been suggested as one of the possible factors determining which terminals survive (Brown et al., 1976). Rosenthal and Taraskevich (1977) reported that the supernumerary EPP components appeared to disappear abruptly, rather than by a gradual decrease in quantal content or size (see also Miyata and Yoshioka, 1980). There have been few studies of the geometrical relationships of terminals during SE. In at least some cases, vacant postsynaptic gutters are seen at this time, implying that axonal processes have been eliminated from gutters that they alone occupied (Rosenthal and Taraskevich, 1977; Bixby, 1981). Usually, in fact, mammalian axonal terminals appear to occupy their own synaptic gutters even during the period of polyneuronal innervation (Kelly and Zacks, 1969; Korneliussen and Jansen, 1976). On the other hand, in frog tadpoles, there is preferential elimination of terminals that share gutters with other axonal branches (Morrison-Graham, 1981). On first examination of these findings, it could easily be concluded that most MNs lose the majority of their terminals during the period of SE. It is true that a large number of synapses are lost. However, as Purves and Lichtman (1980a) have emphasized, probably neither MNs nor other CNS neurons undergo a net reduction in synaptic area occupied during SE. In the CNS, there is typically a large increase in the number of synapses formed by a neuron, even during the period of SE (see, e.g., Cragg, 1975; Crain et al., 1973). In the rat submandibular gland, while the number of axons per cell drops from about 5 to 1, electron microscopic counts of boutons show that the number per cell increases by a mean of about 60°,, (Lichtman, 1977); and in the superior cervical ganglion (SCG), where the number of different axons contacting a cell drops from about 12 to 6 during SE, the absolute number of synapses per cell increases by about 8X (Black, et al., 1971; Smolen and Raisman, 1980). In the rat SCG (but not mouse or chick), despite this large increase in number of synaptic contacts, there is a permanent loss of axosomatic synapses (Smolen and Raisman, 1980). At neuromuscular junctions, individual terminals that do survive continue to increase in size and complexity during the period of SE and subsequent maturation (Nystrom, 1968; Bennett and Pettigrew, 1974a; Tuffery, 1971). Moreover, there may be an increase in numbers of muscle fibers after birth as well, so that many new synapses are forming (Chiakulis and Pauley, 1965; Kelly and Zacks, 1969a; Ontell and Dunn, 1978; Dennis et al., 1981). In the rat soleus, Chiakulis and Pauley (1965, but see Thompson and Jansen, 1977; Dennis, 1981) found approximately a 5000 increase in muscle fiber number after the first postnatal week, probably due both to differentiation of myoblasts and growth of satellite myotubes. Betz et al. (1979) reported a similar postnatal increase in the rat 4th deep lumbrical muscle and concluded that during much of the period of SE, de no~'o synapse formation on new fibers occurred at essentially the same rate as removal of synapses, while motor unit size stayed essentially constant. Subsequently, motor unit size decreased to the adult level, but with the increase in size of individual terminals it is probable that the total synaptic area maintained by a given MN does not decrease during this period. SE, therefore, is only a part of the overall process of synaptic rearrangement (Betz et al., 1979). lit is of interest that the process of postnatal muscle fiber proliferation in the rat lumbrical muscle is dependent on presence J.P.N. 174--B 220 ALAND. GRINNELLANDALBERTA. HERRERA of the nerve. In partial denervation experiments, Betz et al. (1980a) found that the number of muscle fibers added was a simple function of the number of motor units remaining.] 2.4.4. Partial denervation and the role of competition in synapse elimination Manipulations of maturing neuromuscular systems have provided more information about synapse elimination. It is clear, for example, that there is not simply a random withdrawal of a high percentage of branches by each axon. In their elegant studies of SE in the rat soleus. Brown et al. (1976) found no totally denervated fibers. The last synapse, for some reason, is never lost (or else is immediately replaced). They also partially denervated the soleus immediately postnatally by sectioning lumbar ventral root 5 (L5), leaving intact L4, which also innervates the muscle. This reduced the number of motor units from approximately 25 to 8 or fewer. Under these conditions there was a delay in SE in the remaining motor units, but eventually the phenomenon asserted itself, all polyneuronal innervation was lost. motor units shrank to approximately their normal mature size, and a significant fraction of the muscle fibers received no motor innervation at all (Brown et al.. 1976: Thompson and Jansen, 1977). From these studies came the conclusion that SE is a product of two processes: competition between different axons for survival on a given muscle fiber, and an intrinsic tendency to withdraw a large percentage of an axon's branches at this stage in development. Betz et al. (1980a) attempted to assess the relative importance of competition and intrinsic withdrawal in a different mammalian muscle preparation, the rat 4th deep lumbrical muscle. This muscle is innervated by two nerves, the lateral plantar nerve, which supplies most of the axons, and the sural nerve. In many cases, sectioning the lateral plantar nerve left only a single motor unit in the muscle, which therefore faced no competition. It was found in these cases that the mean size of these motor units, tested after the period of SE (4-5 weeks), was the same as before SE, whereas normally there is a 30~o decrease in number of fibers innervated. Betz and his colleagues concluded that, in the absence of competition, there is no intrinsic shrinking of motor units in this muscle. However, it is not entirely certain whether at greater intervals, some terminal withdrawal might be seen, since in the soleus SE is delayed by partial denervation and is not complete until 5-6 weeks of age (Brown et al., 1976). Nor was the possibility ruled out that the single motor units had expanded to larger size, followed by withdrawal to the size observed. However, in partially denervated soleus muscles of 1-5 day old rats, Thompson and Jansen (1977) saw no expansion of motor unit size, only delayed retraction. The sprouting behaviour of neonatal motor axons presents an interesting dilemma. As some synapses are withdrawn, nearby differentiating muscle fibers become innervated, presumably by sprouting of the same axons. However, in partially denervated neonatal mammalian muscles, there is little tendency for the axon to enlarge its arborization by sprouting to innervate denervated fibers. This contrasts with the behavior of axons in partially denervated adult mammalian muscle. In the partially denervated adult soleus, for example, the remaining axons enlarge their motor units in proportion to the degree of denervation, reaching a maximum of 4-5 times the normal size, and remain enlarged. leaving fibers denervated only if fewer than 5-7 axons survive out of an original population of about 25 (Thompson and Jansen, 1977: see also Section 4.1.1). The inability of neonatal axons to expand their fields significantly and innervate nearby denervated fibers could have many explanations. It might represent an intrinsic limit to branching and synapse formation, which each axon has already reached. Neonatal soleus motor units are, in fact, as large or larger than the maximally sprouted adult motor units. Alternatively, the sprouting signal produced by denervated immature fibers may not be as strong as that from newly differentiated fibers, or the neurons may be less responsive (see discussion of sprouting, below). SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 221 In frog tadpoles, partial denervation of the cutaneous pectoris muscle near the time of metamorphosis has been shown to produce extensive sprouting from the remaining intact axons (Morrison-Graham, 1981). Whether there is an overall increase in motor unit size is not known, however. 2.4.5. Polyneuronal innervation by regenerating neonatal neurons, and its elimination Interestingly, in some muscles (but not all), there is apparently a similar period during which reinnervation by a regenerating nerve is not successful. Betz et al. (1980b), for example, saw no regeneration of the lateral plantar nerve into the deep lumbrical muscle, even after 4-5 weeks; and Dennis and Harris (1980) found that internal intercostal axons, sectioned at birth or 1 week of age, regenerated but did not form functional endplates. Only after the animals had reached a minimum of 3 weeks of age, did functional synapses appear. The delay could be as long as 2 months. Animals denervated at 3 weeks of age or thereafter, on the other hand, characteristically showed successful reinnervation in about 7 days. Dennis and Harris (1980) suggest that the lack of adult-like metabolic changes following axotomy in hamster neonatal facial MNs (Kirshen and LaVelle, 1977) may reflect a general inability of neonatal mammalian MNs to regenerate synapses. On the other hand, Robbins et al. (1977) found successful reinervation of denervated neonatal gastrocnemius muscle by an implanted foreign nerve within 5 days, although reinnervation never exceeded more than about 10~ of the muscle. Moreover, Brown et al. (1976) found that if the rat soleus nerve was crushed on the first or second day after birth, there was rapid reinnervation, so that most fibers were functionally reinnervated within 2 weeks. The discrepancy between these results and those of Dennis and Harris (1980) is not easily explained, but nerve crush is both a less disruptive form of denervation than nerve section and leaves axons in a better position to find their way back to old endplate sites. The neonatal defficiency may be in factors facilitating regrowth to the appropriate sites. In frog tadpoles denervated at the time of metamorphosis, functional reinnervation is nearly complete within 7 days following nerve crush or 10 days following nerve section (Morrison-Graham, 1981). In the case of successful reinnervation of the rat soleus following nerve crush (Brown, et al., 1976) muscle fibers showed extensive polyneuronal innervation. This polyneuronal innervation was eliminated with almost the normal time course, despite the fact that the terminals had been present at most a few days, compared with normal development, where the synapses that compete with each other, and are lost, were mostly formed well before birth (Bennett and Pettigrew, 1974b). This suggests that SE (and perhaps the developmental state of the axon terminals) is determined more by the state of maturity of the muscle fibers or MN somas and their environment than by the age of the axon terminals (Brown et al., 1976). It is a curious and important fact that the multiple innervation seen during development in rat, chick fast twitch, and amphibian muscle, mostly occurs at the same endplate sites. Somehow, the initial innervation of a given fiber tends to prevent any further synapse formation on that fiber, except at that site. Even in muscle fibers that are normally innervated at more than one site, there appears to be a minimum distance between endplates, leading to speculation that each endplate prevents additional synapse formation on a certain expanse of membrane around it (Bennett and Pettigrew, 1974b, 1975). Polyneuronal innervation may still be seen at each site. These findings gain additional significance from experiments showing that regenerating axons tend to selectively reinnervate old endplate sites (Brown et al., 1976), much as they do in the adult, (see Section 5.1). Jansen and his colleagues have also studied reinnervation by a foreign nerve at the time of SE. As in the adult (see Section 5.2.2.5), successful innervation by a foreign nerve can occur only if the original innervation is removed (Jansen et al., 1973). If the original nerve is crushed, however, and the fibular nerve implanted into the soleus muscle far 222 ALAN D. GRINNELL AND ALBERT A. HERRERA from the original endplate region, it readily forms new endplates. In such experiments, Brown et al. (1976) showed frequent multiple endplate sites within 1 mm of each other. Thus regenerating axons do not rigidly adhere to the developmental rule that all axons innervating a given fiber must end at the same site. Nevertheless, the multiple sites were lost with only a slightly slower time course than in normal SE. The additional delay, of a few days (Thompson and Jansen, 1977) to a week (Brown et al., 1976), may have resulted from the greater separation between the terminals. After SE, some fibers still had multiple acetylcholine esterase (ACHE) sites, but these were almost all within 100/~m of each other, and were not correlated with multiple EPP components. These probably represented either vacated sites, or ones innervated by the same axon. The observation that separate endplates can survive if innervated by branches of the same axon suggests that the competition may not affect other branches of the same axon, or nearby terminals that are active simultaneously. Based on the importance of proximity in the mechanism of SE, and their observations in autonomic ganglia, Purves and his coworkers (Purves and Lichtman, 1980a,b; Purves and Hume, 1981) have pointed out an important correlation between the process of SE and dendritic structure. Neurons of the submandibular ganglion have a single dendrite and, after SE, it receives boutons from only one presynaptic axon (Lichtman, 1977). Cells of the SCG have several dendrites, and retain about the same number of axonal inputs (approximately 6) after SE (Black, et al., 1971; Smolen and Raisman, 1980). They speculate that a major role of dendrites in the CNS is to provide adequate separation between inputs from different axons so they can survive (Purves and Lichtman, 1980b). 2.4.6. Effects of altered activity on synapse elimination SE takes place during a time of greatly increased and coordinated muscle activity. Several experiments have tested the effects of altered synaptic function and muscle activity on SE. In general, manipulations that reduce the amount of activity delay SE, those that increase activity accelerate SE. Thus tenotomy (Benoit and Changeux, 1975), block with botulinum toxin (Brown et al., 1976), curarization (Srihari and Vrbov~i, 1978), deafferentation and spinal isolation (Zelemi et al., 1979" Miyata and Yoshioka. 1980), and tetrodotoxin (TTX)-block of nerve conduction (Thompson, et al., 1979) all delay SE. Curarization of a chick embryo during the 7th to 12th days of incubation resulted in as many as 12 axons innervating a single endplate in the anterior latissimus dorsi or posterior latissimus dorsi at day 18, when normally these muscles have the adult complement of 3 and 1 axonal inputs, respectively (Srihari and Vrbov~, 1978). All the extra inputs were nevertheless lost quickly when curare was removed. Curiously, chronic spinal cord stimulation for several days, starting at day 10 in chick embryos, results in increased terminal branching and numbers of junctions formed on the fibers of the posterior latissimus dorsi (Renaud et al., 1978). Accelerated SE has been shown to result from compensatory hypertrophy of a muscle synergistic to one that has been denervated (Zelen~ et al., 1979), or had chronic nerve stimulation (O'Brien, Purves and Vrbovfi, 1977; O'Brien, Ostberg and Vrbov~t, 1978). In certain cases, inactivity imposed by TTX nerve block (Thompson et al., 1979) or cordotomy (Miyata and Yoshioka, 1980) has been shown to cause an increase in polyneuronal innervation. This could be interpreted as due to axonal sprouting in the paralyzed muscle [Thompson et al., 1979). except that most evidence suggests that the ability of these neonatal nerves to sprout is very limited. Alternatively, Miyata and Yoshioka (1980) suggest that the increase in polyneuronal innervation can be quantitatively accounted for as a reactivation of synapses "eliminated" during approximately the 24 hours preceding cordotomy. The increase in polyneuronal innervation was seen within 1 day and there was no further increase after the second day, whereas sprouting in the adult is seen only after 3-4 days (cf. Brown and Ironton, 1977, 1978). Moreover, there was no increase in polyneuronal innervation after cordotomy if the rat was 1 month old. Thus it is possible that SE is not an irreversible process, SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 223 and that, for about 1 day, "eliminated" terminals remain nearby and are capable of quickly reinnervating a muscle. It is not clear how activity contributes to SE, but it cannot be considered surprising in view of the potent influences of activity on other neuromuscular properties: the distribution of acetylcholine (ACh) receptors on the muscle surface (for review see Lomo, 1976; Lomo and Westgaard, 1974, 1975; Fambrough, 1979), the ability or inability or implanted nerves to form synapses (Jansen et al., 1973; Frank et al., 1975), the regulatio n of axonal sprouting (see Sections 4.1.2, 4.1.3), the induction of AChE concentrations at newly formed synapses (Lomo and Slater, 1980b), and control of muscle protein and contractile properties (Lomo et al., 1974; Lomo, 1976). In many cases, interpretations of experiments are complicated by the possibility that the activity changes induced may be pathologically extreme. The only detailed attempt to explain SE, and the role of activity, has been put forward by Vrbovfi and her colleagues (Vrbovfi, et al., 1978; O'Brien, et al., 1978, 1980). Building on the earlier observations of Poberai, et al. (1972), they hypothesize that neutral proteolytic enzymes are released from muscle cell lysosomes in proportion to the degree of synaptic activation. These enzymes then act to destroy nerve terminals. Strong terminals, capable of rapid repair and growth, will survive, while weaker terminals will be unable to maintain their presence at the synapse. There is evidence for Ca 2*-induced release of proteolytic enzymes from the region of the endplate, for ACh-induced entry of Ca 2÷ into muscle fibers at the synapse, and for ultrastructural damage to the terminals when large amounts of Ca 2 ÷ enter the muscle (see full discussion and references in Vrbovfi et al., 1978 and O'Brien et al., 1978, 1980). However, it has not yet been established that this phenomenon is associated with SE, nor is it satisfactorily explained how this process could accomplish total removal of all but one synaptic input. Also, this hypothesis is not easily reconciled with the observation that sprouting occurs from terminals of intact axons in the presence of denervated muscle fibers, even when the intact axons are stimulate 4 at activity levels greater than normal (Brown and Holland, 1979). Nevertheless, careft, testing of this hypothesis should help advance our understanding of the mechanism of SE. Most studies of the effects of altered activity on SE have employed manipulations that affect all axons approximattely equally. However, it would be reasonable to expect that the most dramatic effects of altered activity might be seen when some axons are active, others inactive. In preliminary experiments using this approach, H. Gordon and D. Van Essen (personal communication) find that chronic TTX block of conduction in one of two nerves innervating the rabbit soleus causes a slight acceleration of SE in the terminals of the inactive nerve, It is noteworthy that while inactivity normally slows SE, in the presence of an active nerve the inactive nerve lost terminals faster than the active one. This suggests that competition between terminals is important, and that active terminals have a selective advantage. 2.4.7. Summary A qualitative summary of many of the observed characteristics of synapse elimination is presented in Fig. 3. Many questions remain unanswered. Most of what we know, however, is consistent with the view that synapse elimination (SE) is a manifestation of competition between axons that falls off sharply with distance and continues until only one input remains. This competition is superimposed on a general tendency for motor units to shrink, perhaps because each motoneuron (MN) can maintain only a certain maximum synaptic area. As each individual terminal increases in size and complexity, this increase is at the expense of other terminals. The phenomenon is not associated with cell death, and appears to depend on local synaptic feedback from muscle fibers, as judged by the significant asynchrony of timing in different muscles, the importance of synaptic activity in accelerating or delaying it, and the fact that it can operate- over distances as great as 1 mm if the terminals belong to different axons. It might be anticipated that the adult motor unit size distribution and innervation pattern would involve non-uniform loss of synapses during SE, but few studies have addressed this question. 224 ALAN D. GRINNELL AND ALBERT A. HERRERA 100 - • ,. • I e o o :" _o ~- .%~, • .o~. E > " z Z 3. 50-- "N • ~ frO ° J" • ::3 ~ LATE • "' z >- , • o / .,'~--" A C T I V I T Y BLOCK I I BIRTH HATCHING (chick) (mammal) TIME (6) )- F I G . 3. Characteristics of synapse elimination (SE) in vertebrate skeletal muscle. Most muscle fibers are innervated during development by two or more motor axons. SE begins before birth and continues until virtually all (fast twitch) muscle fibers receive only one input. SE occurs at significantly different times in different animal groups (see abscissa), and at different times and rates in different muscles of the same animal (curves 1 and 2). SE is delayed by partial denervation of a muscle or by uniform inactivity (3), and accelerated by uniform synaptic hyperactivity (4). Interestingly, inactive terminals on muscle fibers that are also innervated by active terminals show accelerated elimination (5). In some cases it has been observed that if synaptic block is initiated during the period of SE, there is a significant increase in polyneuronal innervation (6). This suggests that "eliminated" synapses can, for a brief time, be reestablished. Representative references: (1,2) Section 2.4.1, 2.4.2; Bennett and Pettigrew, 1974a; Brown eta/., 1976; Rosenthal and Tarskevich, 1977; Bixby and Van Essen, 1979a; Bixby, 1981; Srihari and Vrbov& 1978; Dennis et al., 1981. (3) Brown et al., 1976; Betz et al., 1980a; Benoit and Changeux, 1975; Srihari and Vrbovfi, 1978; Zelenfi et al., 1978; Thompson et al., 1977; Miyata and Yoshioka, 1980. (4) Zelenfi et aL, 1979; O'Brien et al., 1977, 1978. (5) H. Gordon and D. Van Essen, personal communication. (6) Thompson et al., 1979; Miyata and Yoshioka, 1980. There have been some intriguing claims of specificity, viz., that larger motor units selectively lose terminals, and that axons of only one of the two nerves to a muscle show SE. However, there are counter examples of both, and no conclusions about specificity seem warranted at present. Also, as will be seen below (Section 5.2.2.), in contrast to M N cell death, SE can occur even in the adult, probably by mechanisms similar to those operating during maturation. 3. Long-TermPlasticity in Adult Neuromuscular Junctions The foregoing discussion has dealt primarily with influences on MNs during development and maturation. Classically, the fully developed neuromuscular junction has been viewed as a simple relay synapse, acting as a highly reliable-but rigidly organized amplifier between the motor nerve spike and the muscle action potential. Long-term plasticity was thought to be restricted to the CNS. However, in recent years it has become evident that interesting, important forms of plasticity also exist at the neuromuscular junction, where they are easily accessible for study. In this review we will consider only long-term changes in neuromuscular function or connectivity in vertebrate twitch muscle fibers, changes with a time course measured in days or longer. We will not attempt to review the well known short-term forms of physiological plasticity, e.g., facilitation, depression, post-tetanic potentiation, etc., nor will we make more than occasional reference to the SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 225 voluminous literature on plasticity at invertebrate synapses (Kandel, 1976; Zucker, 1977), or in the vertebrate CNS and peripheral ganglia (Cotman, 1978; Lund, 1978; Tsukahara, 1981). The findings we will discuss indicate that neuromuscular synaptic function and morphology can be altered dramatically by a great variety of influences. Some of these have been discussed in previous reviews (Harris, 1974; Robbins, 1974, 1980). 3.1. SYNAPTIC EFFECTIVENESS The terms safety margin or safety factor are commonly used to describe the extent to which the EPP exceeds the minimum amplitude necessary to trigger a muscle fiber action potential. Estimates of safety margins at junctions in a variety of mammalian muscles range between 1.8 and 12. These were determined either by dividing average EPP amplitude by the depolarization necessary to reach threshold (Boyd and Martin, 1965; Gertler and Robbins, 1978; Harris and Ribchester, 1979a) or by quantifying the relationship between ACh receptor blockade and block of twitch tension (Paton and Waud, 1967; Waud, 1971; Barnard, et al., 1971; Chang, et al., 1975). Safety margins can also be estimated by the relationship between blockade of transmitter release with low Ca 2 ÷ concentration and reduction of twitch tension (Grinnell and Herrera, 1980a). As pointed out by Robbins (1974), the finding that mammalian junctions have such high safety factors must be borne in mind when comparing plasticity at the neuromuscular junction with plasticity at central synapses where the effectiveness of individual synaptic inputs is usually very much less. However, not all vertebrate neuromuscular junctions have high safety factors. The presence of some junctions on twitch fibers where transmission is subthreshold, i.e., safety margin < 1, has been reported for a variety of frog muscles, including the sartorius, extensor digitorum longus (EDL) IV, iliofibularis, and rectus abdominis (Fatt and Katz. 1951; Kuffler, 1952; Ralston and Libet, 1953; Hutter and Loewenstein, 1955; Wakabayashi and Iwasaki, 1962; Orkand, 1963; Forrester and Schmidt, 1970; Nastuk, 1971: Luff and Proske, 1976; Grinnell and Herrera, 1980a; Ridge and Thomson, 1980c). Frog cutaneous pectoris muscles, on the other hand, seem to be innervated exclusively by junctions where transmission is suprathreshold (Herrera and Grinnell, 1980a; Grinnell and Herrera, 1980a). Subthreshold junctions are also found in Xenop~ls pectoral muscle (Haimann et al., 1981a), carp muscle (Wakabayashi and Iwasaki, 1962), and chicken lateral gastrocnemius (Brown and Harvey, 1938). It is important to realize that safety margin depends on several presynaptic factors, e.g., transmitter release, which is itself time and activity dependent, and postsynaptic factors, e.g., receptor density, acetylcholine esterase (ACHE) activity, cable properties, resting potential, action potential threshold, and transmitter reversal potential. Each of these factors can be changed by a variety of conditons, with consequent changes in safety margin (see below). Several studies, for example, have demonstrated that transmitter release depends on endplate size. In the frog sartorius, both quantal content in low Ca2÷/high Mg 2÷ solution and spontaneous mEPP release are proportional to endplate area (Kuno, et al., 1971; Grinnell and Herrera, 1980a). Axolotl hindlimb muscles bathed in a high Mg 2÷ solution show a direct correlation between the value of the binomial release parameter n and synaptic size (Bennett and Raftos, 1977). Similar relationships apparently hold for X e n o p u s pectoral muscle (Angaut-Petit and Mallart, 1979) and mouse muscle (Harris and Ribchester, 1979a). The size of individual endplates in turn depends on muscle fiber size. This direct correlation has also been observed in muscles of the frog (Kuno et al., 1971 : Bennett and Pettigrew, 1975; Herrera and Grinnell, unpublished), chicken (Jedrezejczyk et al., 1973), mouse (Harris and Ribchester, 1979a), rat (Harris, 1954; Anzenbacher and Zenker, 1963; Granbacher, 1971; Korneliussen and Waerhaug, 1973), and cat (Nystrom, 226 ALAN D. GRINNELL AND ALBERT A. HERRERA 1968; Eldridge, et al., 1981), as well as in human muscles (CoOrs, 1955; Co6rs and Woolf, 1959). On the other hand, many of the morphological determinants of synaptic effectiveness vary widely in magnitude. In normal frog sartorius muscles, for example, randomly chosen endplates may show greater than 20-fold variation in nerve terminal length and there may be 7-fold variation in muscle fiber diameter (Herrera and Grinnelk unpublished observations). Given this variability, it is clear that further progress in the study of synaptic effectiveness may well depend on detailed correlations of function and structure. For this reason, in investigations of the factors regulating transmitter release capabilities of frog motor nerve terminals, we have found it important to routinely identify and measure single junctions from which recordings have been made in order to normalize transmitter release to nerve terminal size (Grinnell and Herrera, 1980a; Herrera and Grinnell, 1980a,b). Even when such corrections are made, however, some scatter still exists in the relationship between release and terminal size. This scatter may relate in part to the observation by Bennett and Lavidis (1979) that transmitter release may be highly non-uniform along the length of nerve terminals in toad muscle (but see DeCino, 1981). On the other hand, there are subtle forms of order within the variability that hint at the existence of interesting regulatory processes. In the singly innervated frog cutaneous pectoris muscle, larger muscle fibers tend to be innervated by larger nerve terminals. However, for fibers of any given size, there is a good deal of scatter in this relationship. Much of this scatter disappears if one determines the amount of transmitter released from these terminals. Within any given subpopulation of muscle fibers having approximately the same input impedance, there is an inverse relationship between terminal size and transmitter release per unit length, so that total release is approximately equalized (Nudell, 1981). Further evidence for regulation at this level is seen in the report by Weakley (1978) that EPPs recorded at different endplates on the same fiber were approximately equal in amplitude in tubocurarine (dTC) blocked frog sartorius muscles. The largest difference was only 3.8 fold. In contrast, in preparations blocked with a low CaZ+/high Mg 2+ Ringer, quantal contents at randomly sampled endplates in the same muscles showed up to 146 fold variation (average 49 foldt. He attributed the greater equivalence of EPPs along the length of single fibers to a muscle-to-motoneuron interaction controlling synaptic effectiveness. This comparison is perhaps not quite as striking as it first appears, since variation in quantal contents between endplates in low Ca z+blocked junctions is always more extreme than variation in tubocurarine-blocked preparations. This is expected due to the effects of non-linear summation (McLachlan and Martin, 1981) and the fact that the dependence of release on [Ca 2+] goes from a 4th power relationship at low [Ca z +] to a nearly linear one at normal [Ca 2 +] (Dodge and Rahamimoff, 1967). A more appropriate comparison would be with EPP amplitude variations at randomly sampled endplates in dTC-blocked muscles. In such measurements, we have found an average l l-fold maximum difference in EPP amplitudes at different endplates on the same muscle (range 4.1-25 fold, in 552 endplates, 15 musclesl (Herrera and Grinnell, unpublished). Thus, Weakley's conclusion is still supported, but the effect is less dramatic. Another clue to the existence of mechanisms for the regulation of synaptic effectiveness is the relationship between safety margin and the size of a MN's peripheral field (motor unit sizet. In normal frog sartorius muscles, the smallest motor units are composed primarily of junctions with low safety margins while larger motor units have more uniformly effective synapses (Grinnell and Herrera, 1980b; Grinnell and L. Trussell, unpublished). Frog neuromuscular junctions also show an interesting relationship between synaptic effectiveness and the degree of stretch of the muscle. It has long been known that passive stretch enhances the amplitude of the extracellularly recorded summed muscle action potential (Lamansky, 1870; Forbes, et al., 1923; Fulton, 1925; Libet, et al., 1951; Libet and Wright, 1952; Ralston and Libet, 1953). Modern techniques have shown that stretch SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 227 causes a reversible increase in EPP quantal content and miniature endplate potential (mEPP) frequency Watt and Katz, 1952; Hutter and Trautwein, 1956; Turkanis, 1973: Grinnell and Herrera, 1980a) and an increase in the rate of depression of repetitively evoked EPPs (Otsuka, et al., 1962). The effects are sufficiently pronounced to make it essential that length be controlled when studying neuromuscular function in frog muscle. Transmitter release from rat diaphragm neuromuscular junctions does not seem to be influenced by stretch (Galindo, 1971a; Turkanis, 19731. The mechanism by which stretch exerts its effect is not known, but it has been proposed that the difference in the effects of stretch on frog and rat junctions may relate to differences in nerve terminal geometry (Turkanis, 1973). Frog motor nerve terminals are elongated and run parallel with the long axis of the muscle fiber, while rat diaphragm terminals are convoluted within a smaller circular area. 3.2. SYNAPTICREMODELLING It was widely believed that the form of the normal adult neuromuscular junction was stable and unchanging until Barker and Ip (19661 reported that endplates in healthy cat and rabbit hindlimb muscles show signs of sprouting and degeneration. Their interpretation of this finding was that motor nerve terminals have a fixed life span, after which they are completely replaced. After re-examining the phenomenon, Tuffery (1971) concluded that sprouting and degeneration were not causally related, but instead were aspects of the normal aging process responsible for the increased endplate complexity and loss of muscle fibers seen in older animals. Synaptic remodelling in the normal adult has also been reported in rats (Cardasis and Padykula, 1979; Tweedle and Stephens, 1980, 1981; Courtney and Steinbach, unpublished data cited in Eldridge, et al., 1981), mice (Brown and Ironton, 1977a), and frogs (Wernig, et al., 1980a,b; Mallart et al., 1980a). The emerging view is that adult neuromuscular junctions are constantly being remodelled, that any "stable" form actually represents a state of equilibrium between growth and regression, and that this equilibrium can be shifted either direction by a variety of factors, such as aging, changes in use, seasonal changes, disease, and injury. 3.3. EFFECTSOF DECREASEDOR INCREASEDUSE Much attention has been focused on use-dependent plasticity at the neuromuscular junction because of its obvious potential as a model for plasticity in the central nervous system, e.g., memory and learning. Because of the technical difficulty of chronic nerve stimulation and the comparative ease of inducing paralysis, most studies have tested the effects of decreased rather than increased use. The older literature on use effects has been well reviewed (Sharpless, 1964; see also Robbins, 1974, 19801, so we will concentrate here on more recent reports. These findings are summarized in Table 1. It is apparent from Table 1 that decreased use can have profound efffects on the neuromuscular junction. The most consistently observed presynaptic effect of disuse is nerve terminal sprouting (see Section 4.11. Since sprouting is also seen in certain neuromuscular diseases where safety margin is impaired (see Section 3.6), it may well be that disuse-induced sprouting represents an adaptive response whose purpose is to help restore functional transmission and muscle activity. In this regard, it is interesting that there have been a few reports of increased quantal content and mEPP frequency, as well as one report of an apparent increase in safety margin at junctions in disused muscles (Table 1). The most striking postsynaptic effects are the appearance of extrajunctional ACh receptors and atrophy. Studies on the effects of increased use are few but intriguing. Snyder et al. (1973) found that increased use of the rat plantaris caused an increase in the activities of choline acetyltransferase and ACHE, but this may he a simple result of the hypertrophy they also observed. Vrbov~ and Wareham (1976) reported that 1 h of nerve stimulation in cat or 228 ALAN D. GRINNELL AND ALBERT A. HERRERA rat muscle caused an increase in ACh sensitivity and mEPP amplitude that persisted for 4-6 h. Gertler and Robbins (1978) increased activity in the rat EDL by stimulating the nerve in a tonic pattern 8 h/d for 20 d. This caused a slight (1.3X) but significant increase in quantal content at a variety of different stimulus frequencies. It is not clear if these findings are related to the activity-dependent increase in proteinase activity found by Poberal et al. (1972) at rat endplates. Further investigation will be necessary before it will be safe to generalize about the effects of increased use. There are many apparent conflicts in the literature on the effects of altered use (Table 1). Although some of these may relate to species or muscle differences, there remain disagreements which are not so easily explained. These conflicts may relate to the following important points of caution, many of which are paraphrased from Robbins' (1974) excellent discussion on the design of experiments on use-dependent effects: (1) Are the presumed changes in activity documented by direct observation? For example, one cannot safely assume that tenotomy always reduces muscle activity since its effects can be different for different muscles in the same species and for the same muscle in different species. (2) Are the changes in use pathologically extreme? Can it be reasonably assumed, for example, that the effects of total disuse (TTX implants) are simply a quantitative extension of responses that might occur in vivo with more modest use changes? (3) Do the drugs which are commonly used have important effects aside from their effect on activity'? For example, how many processes are disrupted by coichicine, and l~ow does BoTX work? (4) Has the importance of temporally patterned activity been considered? It is clear, for example, that nerve activity pattern can control a number of postsynaptic properties (reviewed by Lomo, 1976; Fambrough, 1976; Harris, 1981). (5) Have activitydependent changes been adequately described in terms of all the aspects of synaptic function known to control effectiveness'? It would be of great interest to know exactly which aspects of nerve terminal excitability, excitation-secretion coupling, and postsynaptic response are changed. Finally, it has been proposed that the changes seen in inactive muscles of hibernating animals (Albuquerque, et al., 1978) and inactive testosterone-sensitive muscles in castrated animals (Gutmann, et al., 1969; Vysko~il and Gutmann, 1969) may serve as adequate models for studying the effects of disuse. These data will be discussed below since it seems likely that these cases may involve direct metabolic or hormonal effects independent of changes in nerve activity (see Moravec, et al., 1973). 3.4. SEASONAL AND HORMONAL EFFECTS Aspects of synaptic function can show seasonal variation sometimes clearly mediated by hormonal changes. The following studies make it apparent that in addition to the well known postsynaptic ACh receptors, endplate structures may also bear receptors for a number of different chemical messengers (see Vizi, 1979). 3.4.1. " W i n t e r frogs" and hibernating mammals It is a common bit of scientific folklore that the most consistent results in experiments using English frogs are obtained in winter. Upon repetitive stimulation, junctions in sartorius muscles of winter frogs show increased facilitation and decreased depression (Otsuka, et al., 1962; Maeno, 1969), and transmission is much more easily blocked by raising Mg 2+ concentration (Braun, et al., 1966; Maeno, 1969), when compared with summer frogs. Increased facilitation and decreased depression is also seen in frogs maintained at low temperatures (Bowen and Merry, 1969). These results are consistent with the hypothesis of lower quantal release in winter muscles, but do not differentiate between physiological changes and alteration of terminal size. Otsuka et al. (1962) reported that the decreased depression in winter muscles could be at least partially reversed by feeding and overnight maintenance at warmer temperature. Winter sartorius junctions also show slight increases in ACh sensitivity around the junctional region (Feltz and Mallart, 1971) and a higher incidence of nerve terminal sprouting (Wernig et TABLE 1. EFFECTS OF ALTERED USE ON ENDPLATE POTENTIAL QUANTAL CONTENT MINIATURE ENDPLATE POTENTIAL FREQUENCY, MOTOR AXON SPROUTING, ENDPLATE DEGENERATION, ACETYLCHOLINE RECEPTOR DISTRIBUTION AS MEASURED BY ct-BuTX BINDING OR DEPOLARIZING RESPONSE TO ACETYLCHOLINE APPLICATION, RESTING POTENTIAL AND MUSCLE FIBER ATROPHY Decreased Use Manipulation tenotomy tenotomy tenotomy tenotomy tenotomy spinal isolation spinal isolation spinal isolation spinal isolation limb fixation limb fixation anesthetic nerve cuff anesthetic nerve cuff diptheria toxin botulinum toxin Tetrodotoxin Tetrodotoxin Tetrodotoxin Tetrodotoxin Tetrodotoxin Tetrodotoxin curare a-bungarotoxin ct-bungarotoxin ct-bungarotoxin Species rabbit rabbit cat cat rat cat rabbit rabbit cat rat rat rabbit rat rat rat rat rat rat rat rat mouse rat rat rat mouse Muscle soleus tibialis ant. soleus gastrocnemius soleus tenuissimus soleus tibialis ant. soleus soleus gastrocnemius tibialis ant. soleus soleus soleus soleus, edl soleus, edl soleus soleus, edl soleus soleus, peroneus diaphragm diaphragm soleus soleus Muscle Activity Quantal Content MEPP Freq. Sprouting d no ch no ch no ch End Plate Degen. AChR RP + - i,d no ch i no ch + i,d no ch d no ch no ch d d d d d d d d d + + i ** Other + + + + + + + + d d Atrophy i i i i,d no ch i i i i i i i no ch + + + + + * + + d + + no ch + i i i d Reference 1,2 1,2 3 3 4 5 2 2 6 7,8 9 10 4 4 11 12 13 11 14 15 16 17 17 11 18 rat rat rat soleus edl plantaris i i no ch 7' r" ,--I Z © re 7~ 0 Z Z z Increased Use nerve stimulation nerve stimulation tenotomy of synergists ,< i no ch i *** 19 20 9 Abbreviations: d = decrease, no ch = no change, i = increase, + = effect present, - = effect absent, * = decreased choline acetyltransferase (CAT) activity, ** = increased resistance to curare block, *** = increased CAT activity. References: (1) Vrbov~i, 1963: (2) Gwyn and Vrbovfi, 1966; (3J Nelson, 1969: (4) Lomo and Rosenthal, 1972: (5) Johns and Thesleff. 1961 : (6) Eldridge et al.. 1981 : (7) Fischbach and Robbins, 1971; (8J Robbins and Fischbach, 1971 ; (9) Snyder et al., 1973: (10) Robert and Oester, 1970:(11) Pestronk and Drachman, 1978; (12) Lavoie et al., 1976; (13) Pestronk et al., 1976; (14) Bray et al., 1979: (15) Snider and Harris, 1979: Brown and lronton, 1977a: (17) Berg and Hall, 1975; (18) Holland and Brown, 1980: (19) Vrbovli and Wareham, 1976; (20) Gertler and Robbins, 1978. ~D 230 ALAN D. GRINNELL AND ALBERT A. HERRERA A H RS IAT MEPP ACih~ ~ ~E FIG. 4. Physiological changes at the neuromuscular junction of awake (At and hibernating (HI animals. Upper part represents the nerve terminal, and lower part represents the post-synaptic membrane. Synthesis of transmitter (ST) represented by the arched arrow. The size of the reserve transmitter store (RS) in the nerve terminal is shown as open rectangles, and immediately available transmitter (IAT) as black rectangles. The relative miniature endplate potential frequencies (mEPP) in A and H are indicated by the sample recordings. ACh represents the iontophoretic acetyicholine pipette for estimating the length (black horizontal columns) and maximum sensitivity (black vertical columns) of the membrane to ACh. EPP represents the relative mean quantal content of the endplate potential in each condition. (After Moravec and Vyskofiil, 1976.) al., 1980a, b). The latter observation may be the morphological basis for the higher incidence of focal polyneuronal innervation reported in winter (180~,) compared with summer (9°Jo) frog sartorius muscles (Herrera and Grinnell, 1981). Hibernating animals undergo many metabolic changes to insure maintenance of body function during long periods of hypothermia and inactivity. Within the neuromuscular system, for example, the diaphragm must remain active. Neuromuscular junctions in hibernating animals show a variety of adaptive changes that further illustrate the plastic capabilities of this synapse. South (1961) has shown that transmission in several hibernating mammals is less sensitive to block by low temperature than in active animals. Figure 4, taken from a review by Moravec and Vysko6il (1976), summarizes the changes seen at neuromuscular junctions of hibernating golden hamsters. Within the nerve terminal, transmitter synthesis and the reserve store of transmitter are decreased (Melichar et al., 1973). While there is no change in the immediately available store of transmitter, EPP quantal content is reduced (Melichar et al., 1973) as well as mEPP frequency (Vysko~il, Moravec and Jansk~, 1971 ; Moravec et al., 1973; Vysko6il, 1976). The effect of mitochondrial poisons in increasing mEPP release is greater in hibernating muscles, suggesting that mitochondria in motor nerve terminals may sequester more Ca 2+ during hibernation (Vysko6il and Moravec, 1978). Postsynaptically, there is no change in resting potential (Moravec et al., 1973), TTX sensitivity, or tension generation (Deshpande, et al., 1976; Oliveira, et al., 1978; Albuquerque, et al., 1978; Vysko~il and Gutmann, 1977). There are conflicting results as to whether there are (Melichna et al., 1973; Vysko6il, Moravec and Jansk~, 1971; Moravec et al., 1973; Vysko~il, 1976; see also Lyman and O'Brien, 1969) or are not (Deshpande et al., 1976; Albuquerque et al., 1978) changes in ACh sensitivity and endplate currents. In summary, it appears that the MN is adapted to hibernation and the resulting inactivity by becoming more tolerant of low temperature and reducing transmitter synthesis and release to minimal levels. It is thus able to conserve metabolic energy (Moravec and Vysko~il, 1976) while insuring that functional transmission will still occur. 3.4.2. Effects o f testosterone There have been a number of reports that sexually active muscles and their MNs in male animals are under hormonal control by testosterone (reviewed by Arnold, 1981). SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 231 For example, castration causes a decrease in choline acetyltransferase (CAT) activity in the nerve supplying singing muscles in male zebra finches (Luine et al., 1980) and the nerves of a rat penile muscle, the levator ani (Gutmann, et al., 1969; Tu~ek, et al., 1976). Testosterone administration, on the other hand, increases CAT activity (Tu~ek et al., 1976a), mEPP frequency (Vysko6il and Gutmann, 1977), and endplate size (Tu~k et al., 1976a) and causes dramatic changes in nerve terminal ultrastructure (Hanzlikov~ and Gutmann, 19781 in the levator ani. Testosterone also increases the firing frequency of MNs innervating the muscles used by male Xenopus to clasp the female (Schneider et al., 1980). These observations suggest that androgens may influence synaptic transmission. Postsynaptically, castration and/or testosterone administration have profound effects on the size and contraction properties of male frog clasping muscles (Muller, et al., 1969; Melichna et al., 1972) and rat penile muscles (Venable, 1966; Gutmann et al., 1969; Tu~:ek et al., 1976a; Vysko~il and Gutmann, 19771. Cross-innervation experiments indicate that the levator ani muscle remains sensitive to testosterone even when innervated by a foreign nerve (Hanzlikovfi and Gutmann, 1972) and that the transplanted levator ani nerve cannot confer androgen sensitivity on a non-sexually active muscle (Hanzlikova and Gutmann, 19741. Thus the levator ani muscle itself is sensitive to the hormone. Nevertheless, there is suggestive evidence in this and other systems that the effects of androgens on MNs are direct, and not secondary effects mediated by the muscle. Both medullary MNs to vocalization muscles used in mate calling by Xenopus (Kelley et al., 1975; Kelley, 1980j and lumbar MNs to rat penile muscles (Breedlove and Arnold, 1980; see also Sar and Stumpf, 1977) have been shown to concentrate radioactively labelled androgens. These results demonstrate that major aspects of MN and muscle function can be controlled by circulating androgen levels. However, it remains to be explicitly demonstrated that androgens alter synaptic effectiveness in these systems. 3.4.3. Effects of epinephrine and norepinephrine It has been known for nearly 60 years that stimulation of the sympathetic innervation of fatigued frog muscle causes an increase in twitch tension as well as the expected vasoconstriction (Orbelli, 19231. Hutter and Loewenstein (1955) confirmed this result and showed an increase in EPP size, concluding that norepinephrine increased postsynaptic ACh sensitivity. Jenkinson, et al. (19681, on the other hand, showed that the primary effect of norepinephrine was to increase both nerve-evoked and spontaneous transmitter release. Norepinephrine had no effect on the response to iontophoretically applied ACh. In carp muscle, norepinephrine acts presynaptically to increase release, while epinephrine acts postsynaptically to increase ACh sensitivity (Hidaka and Kuriyama, 1969). In mammalian muscle these substances have similar effects. In hamster muscle, norepinephrine increases resting potential and EPP amplitude, as well as causing a positive voltage shift in the reversal potential for the endplate current (Moravec et al., 1973; Melichar et al., 1973). Norepinephrine also enhances release in rat muscle (Kuba, 1970). Epinephrine increases quantal content, mEPP frequency, input resistance, and action potential threshold in rat muscle (Krnjevic and Miledi, 1958; Kuba, 1970; Kuba and Tomita, 1971, 1972), but there is disagreement as to whether it does (Kuba, 1970) or does not (Krnjevic and Miledi, 1958) increase resting potential, mEPP amplitude, and ACh sensitivity. The action of adrenergic blocking agents indicates that presynaptic effects operate via an x-receptor and postsynaptic effects via a fl-receptor (Kuba, 1970). There is considerable controversy as to whether the action of catecholamines on motor nerve terminals involves cyclic adenosine 3'5'-monophosphate (Miyamoto and Breckenridge, 19741. 3.4.4. Presynaptic effects of acetylcholine Even before Koelle's suggestion that ACh released from nerve terminals may have presynaptic as well as postsynaptic action (Koelle, 1961), the characteristics, function, 232 ALAN D. GR1NNELL AND ALBERT A. HERRERA and even the existence of presynaptic ACh receptors (AChR) were subjects of much controversy. Although some efforts have been made to directly visualize these receptors, most of the literature deals with the effects of cholinergic agonists and antagonists and cholinesterase inhibitors on synaptic potentials, ACh release, and nerve terminal excitability. These diverse approaches have often yielded results that have been difficult to reconcile. However, since it has been clearly shown that physiological synaptic plasticity can have a presynaptic basis, it is important to consider whether modification of presynaptic AChR might be responsible for long-term alteration of the synaptic function. The subject has been well reviewed (Lilleheil and Naess, 1961; Koelle, 1962, 1971; Eccles, 1964; Riker and Okamoto, 1969; Katz, 1969; see especially Miyamoto, 1978). Daniels and Vogel (1975) reported the first effort to directly visualize presynaptic AChR, using an indirect immunoperoxidase technique to label ~-bungarotoxin (~t-BTX) bound to frog and mouse neuromuscular junctions. The presynaptic membrane was labelled but they pointed out that this result was inconclusive, since the peroxidase reaction product may have diffused to the presynaptic membrane from the more heavily labelled postsynaptic membrane. More convincing evidence for presynaptic AChR at newt, frog, and mouse neuromuscular junctions was obtained by Lentz, et al. (1977), using ~-BTX directly conjugated to horseradish peroxidase for better resolution. However, Matthews-Bellinger and Salpeter (1978) and Jones and Salpeter (1981), using 125I-ctBTX and electron microscopic autoradiography, concluded that there are essentially no AChRs on frog motor nerve terminals. All these methods presume, of course, that the putative presynaptic AChRs would bind ~-BTX with affinity and specificity similar to that of the postsynaptic receptors. AChE activity is also reportedly found on the presynaptic membrane at mouse junctions (Davis and Koelle, 1967: see also Nachmansohn, 1976). Finally, in a possibly related report, Freedman and Lentz (19801 found binding of peroxidase-conjugated :t-BTX to the axonal membrane at nodes of Ranvier in rat sciatic nerve. The many physiological studies testing for possible presynaptic effects of ACh and its agonists have recently been reviewed by Miyamoto (1978). We will therefore limit ourselves to a brief discussion of the reported effects of ACh itself on presynaptic function. In their initial description of mEPPs, Fatt and Katz (1952) mentioned that bath application of ACh sufficient to depolarize the frog sartorius endplate by a few mY did not appreciably alter mEPP frequency. The absence of an effect of ACh on mEPP frequency was also reported for the rat disphragm by Hubbard, et al. (1965). However, Blaber and Christ (1967) found that ACh could cause a small but significant increase in mEPP frequency in cat tenuissimus. Similarly, it was later reported by Katz and Miledi (1972) that ACh occasionally does increase mEPP frequency at frog endplates. Assuming that ACh depolarizes nerve terminals and raises their excitability (Barstad, 1962; Hubbard et al., 1965), Katz and Miledi (1972) proposed that the variability in the effect of ACh on mEPP frequency was due to variation in the initial level of the resting potential in the terminal. On the other hand, ACh seems to depress evoked transmitter release. In Mg: + blocked frog sartorius (Ciani and Edwards, 1963) and rat diaphragm (Hubbard et al., 1965), ACh caused a decrease in the quantal content of EPPs evoked at low frequency. As a sampling of the large body of literature detailing the effects of cholinergic antagonists and cholinesterase inhibitors on presynaptic function, we summarize some of the information on the presynaptic effects of d-tubocurarine and neostigmine in Table 2. Assuming all these effects operate via presynaptic AChRs, we include in Table 2 our impressions as to whether the evidence supports an excitatory (E) or inhibitory (I) role for the supposed receptor. The effects of neostigmine are complicated by the fact that it has a weak agonist action (Katz and Miledi, 1973; Kordas, et al., 1975). In spite of the many apparent conflicts in the above results, one can form several conclusions that serve as working hypotheses. Although there are contradictory results from :~-BTX binding studies, the majority of the ph2~siological evidence indicates that presynaptic AChRs do exist at the neuromuscular junction. The function of the receptors, however, may differ in different species. In the frog, the limited information available TABLE 2. PRESYNAPTIC EFFI-CTS OF I)-TUBOCURARINE AND NEOSTIGMINE AT THE NEUROMUSCULAR JUNCTION. SHOWN ARE EFFECTS ON UNBLOCKED QUANTAL CONTENT, QUANTAL CONTI'NT PARTIALLY BLOCKED WITH LOW C a 2 + / H I G H MG 2 ~, THE DEPRESSION OF REPETITIVELY EVOKED ENDPLATE POTENTIALS, MINIATURE ENDPLATE POTENTIAL FREQUENCY, BLOCKADE OF ANTn)ROMICALLY CONI)UCTING SPIKES IN MOTOR AXONS, AND A C H RELEASE. ALSO SHOWN IS OUR OPINION AS TO WHETHER THE RESULTS ARE CONSISTENT WITH AN EXCITATORY (E) OR INHIBITORY {1) PRESYNAPTIC ROLE FOR ACH. ABBREVIATIONS AS IN TABLE 1. Drug d-tubocurarine neostigmine Species rat frog rat rat cat rat rat, mouse, frog rat cat rat hamster rat frog rat frog rat cat rat rat guinea pig guinea pig frog rat rat rat rat cat frog rat Quantal Content Partially Unblocked Blocked Enhanced Depression MEPP Freq. Block of Antidromic Spikes ACh Release + + + + d d i or no ch E E E E E? E E + + no ch d d no ch i,d no ch E E --? d no ch no ch no ch no ch ~ + + + + + d d d Proposed Presynaptic Action of ACh i d no ch d d no oh': no ch i E E E E E? E E --I 1 E I I Reference Lilleheil & Naess, 1961 Otsuka, Endo & N o n o m u r a , 1962 Hubbard. Wilson & Miyamoto, 1969 Hubbard & Wilson, 1973 Blaber, 1970, 1973 Glavinovic, 1979 Magleby et al., 1981 Auerbach & Betz, 1971 Jacobs & Blaber, 1971 Galindo, 1971 Bauer, 1971 Gertler & Robbins, 1978 Martin, 1955 Beranek & Vyskocil, 1967 Bowen & Merry, 1969 Barstad, 1962 Standaert, 1964 Hubbard, Schmidt & Yokoda, 1965 C h a n g & Lee, 1966 Chang, et al., 1967 Beani, et al., 1964 Gergis, et al., 1971 Krnjevic & Mitchell, 1961 Fletcher & Forrester, 1970, 1975 Miledi et al. 1978 Hubbard, et al., 1965 Blaber & Christ, 1967 D u n c a n & Publicover, 1979 Wilson, 1980 ,.< Z ,--I ,.¢ z 7~ 234 ALAN D. GRINNELL AND ALBERT A. HERRERA indicates that the presynaptic receptors are muscarinic and act to inhibit transmitter release (Duncan and Publicover, 1979). In mammals, about which much more is known, nearly all of the reports on the effects of cholinergic drugs on transmitter release and nerve terminal excitability are consistent with the view that presynaptic AChRs are nicotinic and excitatory. The exact role of these receptors in generation of the EPP is not known however, nor is it clear if they function in long-term regulation of synaptic effectiveness. At frog neuromuscular junctions there is a continuous leak of ACh from the terminal such that a concentration on the order of 10-aM is maintained in the cleft (Katz and Miledi, 1977). It is possible that the changes induced in the nerve terminal membrane by this level of ACh may significantly affect transmitter release. A similar role of background ACh in long term control of the level of junctional AChE was suggested by Katz and Miledi (1977). 3.4.5. Other hormones Other chemical messengers may also influence the effectiveness of the neuromuscular junction. Vasoactive intestinal polypeptide has been reported to increase quantal content at frog neuromuscular junctions when applied either iontophoretically or in the bathing solution (Gold and Martin, 1980). Ganguly and Das (1979) have proposed that dopamine acts on a presynaptic muscarinic receptor to inhibit nerve evoked transmitter release at the rat neuromuscular junction. 3.5. EFFECTS OF AGING During the course of an individual's life, there are many changes at the neuromuscular junction which can be correlated with age. Because of its relatively short life span (2-3 years), most studies of the effects of aging on neuromuscular transmission have been done on the rat, although some information is available concerning other mammalian species and humans. The earlier literature has been reviewed (Gutmann, et al., 1968; Gutmann and Hanzlikovfi, 1972; Frolkis, et al., 1976). Although neuromuscular junctions show considerable morphological plasticity with age (for example, Gutmann and Hanzlikova, 1965, 1966; Harriman, et al., 1970; Tuffery, 1971; Tweedle and Stephens, 1980; Pestronk et al., 1980), we will focus here on recent reports of physiological changes. It should be realized, however, that in the absence of correlated morphology and physiology in the same endplates, it is not known whether some of the reported physiological changes are due simply to changes in synaptic morphology, or to changes in the intrinsic mechanisms of transmitter release and postsynaptic response. Only a few studies have documented the effects of age on the physiological properties of the rat neuromuscular junction. Further, it must be cautioned that directly comparing different studies may not be advisable due to differences in the age range of rats examined, their genetic background, and handling conditions. Nevertheless, several studies indicate that safety margins at junctions in older rat diaphragms are lower due to a decrease in quantal content (Kelly, 1978), more severe depression (Smith, 1979), and lower ACh content in the nerve terminals (Gibson and Smith, 1980). The gastrocnemius and gracilis muscles of old rats show a similar decrease in safety margin, reportedly due to decreases in resting potential and membrane resistance and an increase in action potential threshold (Frolkis et al., 19761. Conduction block of the presynaptic action potential is more likely to occur in the motor nerve terminals of aged rats (Smith, 1979). It is more difficult to summarize the changes in mEPP frequency with age. It appears to increase from birth to about 6 months, decreases from 6 months to about one year (Kelly, 1978), then, according to another study, subsequently increases again (Smith, 1979). On the other hand, Gutmann, et al. (1971) and Vysko6il and Gutmann (1972) report that mEPP frequency in extremely old (30-33 month) rat diaphragm and soleus is SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 235 much less than in 3 months old muscles. Some of these contradictions might be resolved if genetically homogeneous strains of mice or other animals were used, and different muscles, showing different activity patterns, were studied in the same animals. In spite of the difficulty in interpreting these studies, however, it is clear that age can have profound effects on the structure and function of the neuromuscular system. 3.6. EFFECTS OF DISEASE There a large number of diseases of unknown etiology that affect spinal and/or upper MNs. Most are characterized by extensive M N cell death. With the exception of myasthenia gravis and myotonia congenita, researchers have barely begun to understand the primary defect in any of the human o r animal neuromuscular diseases. Until these fundamental mechanisms are understood, there is a significant possibility that some of the changes seen at the neuromuscular junction are not a direct effect of the disease, but rather a reaction of essentially normal parts of the junction to a lesion elsewhere. Thus, the changes seen in neuromuscular disease might further illustrate features of the regulation of normal junctional properties. We will discuss only those diseases where neuromuscular transmission has been studied in detail and show abnormalities. 3.6.1. M y a s t h e n i a gravis Early electrophysiological observations showed that mEPP amplitude was reduced at neuromuscular junctions in muscle biopsies from myasthenia gravis patients (Dahlb~ick et al.. 1961; Elmqvist et al., 1964). This led to the erroneous conclusion that the defect in this disease was presynaptic, involving a deficiency in the packaging of ACh into quanta (Dahib~ick et al., 1961; Elmqvist et al., 1964) or deficient ACh synthesis (Desmedt, 1966). It is now well established that the primary effect of the disease is postsynaptic, involving an autoimmune response directed against the AChR at the neuromuscular junction (Toyka et al., 1977; Lindstrom, 1977; Engel, et al., 1977; Lindstrom and Lambert, 1978). There are reductions in postsynaptic ACh sensitivity (Albuquerque et al., 1976) and in the number of ct-BTX binding sites (Fambrough, et al., 1973; Engel et al., 1977). However, since we are concerned here with a review of synaptic plasticity, we will focus on recent evidence that there are indeed presynaptic changes in myasthenia gravis, although the changes are not as originally envisioned. Inasmuch as these presynaptic changes are generally of the type that would tend to counteract the postsynaptic deficit, they may illustrate the ability of the junction to regulate synaptic transmission homeostatically. For a complete review of myasthenia gravis, see Drachman (1981). Nerve terminal sprouting and changes in overall terminal size are best detected in whole mount preparations in the light microscope, rather than in the electron microscope. Such light microscopic studies show evidence of terminal sprouting and elongation in myasthenia gravis (CoOrs and Telerman-Toppet, 1976; Brownell, et al., 1972). As previously mentioned, similar findings in cases of motor nerve or postsynaptic block suggest that sprouting may be a common response to decreased muscle activity. On the other hand, electron microscopic studies show a decrease in terminal cross-sectional area, no change in vesicle diameter or density, and reduced postsynaptic folding (Engel and Santa, 1971; Santa, et al., 1972a; see also Johnson and Woolf, 1965; Zacks, et al., 1961, 1962). Although there are no reports of correlated morphology and physiology on the same endplates, an increase in nerve terminal size is consistent with the physiological findings of increased ACh content and release in myasthenia gravis (Cull-Candy, et al., 1978: Cull-Candy, et al., 1980). Total ACh content is twice as high in myasthenia gravis muscles as in controls (Ito et al., 1976; Molenaar et al., 1979). Quantal content at a wide range of Ca 2÷ concentrations is 2-5 times higher (Cull-Candy et al., 1978, 1980; but see Lindstrom and Lambert, 1978) and the initial release of ACh in response to depolarization with high K ÷ is also higher in myasthenia gravis (Molenaar et al., 1979). J,P.N. 174--c 236 ALAN D. GR|NNELL AND ALBERT A. HERRERA 3.6.2. E a t o n - L a m b e r t and other myasthenic syndromes Superficially, Eaton-Lambert syndrome is similar to myasthenia gravis in that there is muscular weakness and easy fatigability, but it differs in that electromyographic recordings show marked potentiation of muscle response with repetitive nerve stimuli, there is little improvement with anticholinesterases, and malignant intrathoracic tumors may be associated (early literature reviewed by Lambert and Elmqvist, 1971). Intracellular recording from biopsy specimens shows that quantal content is reduced to a few percent of normal (Hofmann, et al., 1967; Lambert and Elmqvist, 1971; Lindstrom and Lambert, 1978; Cull-Candy et al., 1980), resulting in subthreshold EPPs at many junctions. The increase in mEPP frequency induced by depolarization with high K ÷ concentration is smaller (Lambert and Eimqvist, 1971; Cull-Candy et al., 1980). However, resting m E P P frequency, mEPP amplitude, and resting potential are unchanged (Lambert and Elmqvist, 1971). Nerve evoked release shows a positive but much reduced dependence on extracellular Ca 2 ÷ concentration (Cull-Candy et al., 1980; Lambert and Elmqvist, 1971, see Fig. 5). There is no obvious morphological basis for low transmitter release. Synaptic vesicle density and nerve terminal cross-sectional area appear normal (Engel and Santa, 1971; Santa, et al., 1972b), as does the overall appearance of the junction seen in the light microscope (Hofmann et al., 1967). In biopsies from one patient, muscle ACh content was normal (Molenaar et al., 1979). Although there is an increase in the size and complexity of postsynaptic folds (Engel and Santa, 1971 ; Santa et al., 1972b), there is no increase in AChR content (Lindstrom and Lambert, 1978). Patients with Eaton-Lambert syndrome commonly have intrathoracic tumors, most of which are small cell carcinomas of the lung (Lambert and Rooke, 1965). These cells secrete a variety of peptide, glycoprotein, and steroid hormones both in situ (Richardson et al., 1978) and in culture (Sorenson et al., 1981), presumably in response to spontaneously generated Ca 2÷ action potentials (McCann et al., 1981). Since some EatonLambert patients improve with successful tumor therapy (Lambert and Rooke, 1965; Norris, et al., 1965), it has been suggested that impaired transmitter release in this disease may be due to some substance produced by the tumor (Rooke et al., 1960: Lambert and Elmqvist, 1971). In this regard, it is interesting that Kim et al. (1980) have reported that bathing a rat nerve-muscle preparation in serum from a patient with Eaton-Lambert syndrome causes an acute reduction in quantal content. 1GO 50 10 N N 5 1 - 0.5 0.1 i 0.1 0.2 i , , Jt,Li 0.5 1.0 , 2.0 Ca+2 Concentration , tt , iltlJ, 5.0 10 (raM) FIG.5. The relationshipbetweenmean endplate potential quantal content and extracellular Ca2 concentration at normal human endplates and at endplates from a patient with Eaton-Lambert syndrome. Regressionline slopes: 3.3 for normal, 1.6 for Eaton-Lambert. Note that the diseased endplates show greatly reduced quantal content and Caz+ dependence.(After Cull-Candy et al., 1980). SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 237 Engel (1980) summarizes reports of 3 other motor endplate diseases which may involve, respectively, (1) a possible defect in ACh resynthesis, (2) lack of endplate ACHE, small nerve terminals, and reduced ACh release, and (3) a possible increase in ACh channel open time. 3.6.3. Murine motor endplate disease Motor endplate disease, first described by Duchen and co-workers (Duchen, et al., 1967; reviewed by Duchen, 1979), is a genetic disease found in a mutant strain of mice. These mice experience rapidly progressive failure of neuromuscular transmission and are usually dead by 3 or 4 weeks of age. Electrophysiological observations on severely affected forelimb muscles (Duchen and Stefani, 1971; Harris and Ward, 1974) show that with increasing age, an increasing proportion of fibers fail to show action potentials or even EPPs in response to nerve stimulation. MEPPs are present, however, and their amplitude is increased due to fiber atrophy. MEPP frequency could be increased by high K ÷ concentration but not by tetanic nerve stimulation, leading Duchen and Stefani (1971) to suggest that the defect was due to failure of the presynaptic action potential to invade the nerve terminal. Postsynaptically, the muscle fibers show the expected effects of functional denervation: atrophy, fibrillation, low resting potentials, reduced action potential rate of rise and overshoot, TTX resistance, and extrajunctional ACh sensitivity (Duchen, 1979). Weinstein (1980) found similar physiological changes in the less severely affected EDL, although in this muscle neuromuscular transmission does not fail completely. Mallart et al., (1980) reported that axonal conduction is altered and both evoked release and safety margin are diminished in intercostal muscles. Light and electron microscopic observations (Duchen and Searle, 1970) show profuse terminal sprouting, again suggesting a regulatory mechanism whereby the terminal increases in size when transmission is ineffective. Sprouting is apparently not a response to partial denervation, since there is no evidence for degeneration of MN somata or peripheral axons and terminal ultrastructure appears normal. 3.6.4. Wobbler mice The wobbler mutant of the mouse is characterized by partial denervation of affected muscles due to degeneration of lower MNs in the brain stem and spinal cord (Duchen, et al., 1966; Duchen, et al., 1968; Duchen, 1979). Electrophysiological findings are characteristic of partial denervation: w_ith denervati0n-_like changes inaffected areas of the muscle and essentially normal properties elsewhere (Harris and Ward, 1974). Intact motor axons sprout to reinnervate the denervated fibers. Since clinical weakness does not appear until at least 1 week after, and denervation atrophy 3 weeks after MN degeneration is first detected, it appears that the sprouting response can temporarily compensate for the partial loss of MNs (Duchen et al., 1968). 3.6.5. Murine muscular dystrophy Muscular dystrophy in the mouse is perhaps the most thoroughly studied potential model for human muscular dystrophy. Since the subject has been recently reviewed (Harris and Ribchester, 1979b), we will limit our dicussion here to a brief treatment of the contribution of recent papers to our general understanding of the disease. McComas and Mrozek (1967) and Law and co-workers (Law, et al., 1967; Law and Atwood, 1972, 1974; Law and Caccia, 1975)proposed that approximately half the fibers in dystrophic mouse muscle are functionally denervated, i.e., they fail to generate a propagated action potential in response to supramaximal nerve stimulation. Law et al. (1967) present evidence that the defects in the dystrophic soleus muscle are both pre- and postsynaptic, viz., 18~o of the fibers show no detectable response at the endplate upon motor nerve stimulation and there are deficiencies in the generation and conduction of muscle action potentials. 238 ALAN D. GRINNELL AND ALBERT A. HERRERA On the other hand, both Carbonetto (1977) and Harris and Ribchester (1978, 1979a,b) find that neuromuscular transmission is normal at endplates in the dystrophic EDL. They find no difference in quantal content or other release parameters even at endplates on fibers which were atrophied (Carbonetto, 1977) or on identified fibers showing gross morphological abnormalities (Harris and Ribchester, 1978). If anything, safety margins for transmission are higher at dystrophic junctions (Harris and Ribchester, 1979b). The variability of mEPP amplitudes at individual endplates, however, does seem to be greater in dystrophic muscles (Carbonetto, 1977; Harris and Ribchester, 1979b). It is somewhat surprising that nerve terminals are able to maintain adequate transmission in view of the motor nerve abnormalities known to occur in mouse dystrophy. The MN soma appears to be little affected, but peripheral nerves show amyelination, loss of axons, electrical "cross-talk', and multiple firing (reviewed by Bradley and Jaros, 1979; Kuno, 1979). Also, mEPP release from dystrophic terminals is more sensitive to the effects of mitochondrial poisons and Ca 2 +-ionophore, suggesting an impaired ability to regulate intraterminal Ca 2 +-concentration (Shalton and Wareham, 1980). Nerve terminals sprout in dystrophic muscles, increasing the area of synaptic contact on hypertrophied fibers (Harris and Ribchester, 1979b; Law and Atwood, 1974). This suggests that the normal regulatory mechanisms which tend to match transmitter output to postsynaptic input impedence (Kuno et al., 1971) may still operate. However, the ability of dystrophic nerves to reinnervate acutely denervated muscles is impaired (Perry and Melenchuk, 198l). Unlike McComas and Law (see above), Harris and co-workers find little evidence for deficient action potential generation in the dystrophic mouse EDL. In a comparison of fibers from dystrophic and normal mice, they found little difference in the incidence of non-overshooting action potentials (Harris and Marshall, 1973) and the relation between the rate of rise of the action potential and membrane potential (Harris and Ribchester, 1979a). They purposely did not examine action potential conduction in muscle fibers because they felt that morphological abnormalities such as fiber splitting and local necrosis, which occurs commonly, would render such observations difficult to interpret (Harris and Ribchester, 1979a). 3.6.6. M u s c u l a r d y s t r o p h y in the chicken Inherited muscular dystrophy of the chicken is another well-studied model for the human disease. Wilson et al., (1979) review the major characteristics of the several genetic lines of dystrophic chickens which have been developed. Physiological studies have primarily used dystrophic lines 304, 413 and 455, together with control animals which, except for the case of line 304, were genetically related. Since the dystrophic changes are most pronounced in fast-twitch muscles, most of the above studies have used the homogenously fast posterior latissimus dorsi (PLD), although according to recent reports (Gunther and Letinsky, 1981) the extensor of the second digit (ED II) may show more promise as an experimentally convenient preparation. For present purposes, we will limit our discussion to recent evidence concerning changes in junctional morphology and physiology that may influence synaptic efficacy. J. S. Gunther and M. S. Letinsky (personal communication) have shown that the morphological development of the ED II of line 413 is abnormal. From 8 weeks e x ovo to 1 year, dystrophic endplates are generally smaller than usual, while from 14 weeks to 1 year frequent examples of denervation, nerve terminal sprouting, and multiple innervation are seen. These morphological abnormalities are rarely seen in similarly aged control muscles. Several important presynaptic changes are seen in chicken dystrophy. MEPP frequency is diminished in the PLD of line 304 (Albuquerque and Warnick, 1971) and the ED II of line 455 (Bryan and Letinsky, 1979; Gunther and Letinsky, 1981), but these parameters are apparently mornal in the PLD of lir/e 413 (Yeagle et al., 1979). MEPPs are present at all visually identified terminals in the ED II (Gunther and Letinsky, 1981). SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 239 Endplate potential quantal content in the PLD of line 304 is normal but in line 413 quantal content both in normal solutions and in Mg 2 +-blocked preparations is elevated (Warnick et al., 1979; but see Warnick and Albuquerque, 1979). J. S. Gunther and M. S. Letinsky (personal communication) found that overall quantal content was normal at most of the individually identified endplates they studied in the ED II of line 413. Transmitter release per unit terminal area, however, was often higher than normal. In spite of the increased quantal content and the reportedly normal levels of transmitter "mobilization" and "available store" (Warnick et al., 1979), neuromuscular transmission is seriously impaired in dystrophic chickens. Nerve-evoked twitches are more easily blocked by curare (Warnick and Albuquerque, 1979) and EPPs show severe depression to subthreshold levels upon repetitive nerve stimulation (Albuquerque and Warnick, 1971; Warnick et al., 1979). As might be expected, muscular dystrophy in the chicken produces a large number of postsynaptic changes as well, but these results are difficult to summarize because they show a high degree of conflict. In some cases, disagreements may stem from the use of different muscles and genetic lines, but clearly much of the problem lies with experimental design. As Gunther and Letinsky (1981) have pointed out, there is a great deal of heterogeneity in the degree to which the disease affects various fibers in any given muscle. Consequently, many of the parameters used to describe dystrophic fibers and endplates show large and differing degrees of variation and may not be normally distributed. Further, these parameters can also vary with age. It has rarely been the case, but is nevertheless essential, that a single preparation, large data samples, non-parametric statistics, and age-specific controls be employed. Since fibers are affected individually, it is likely that the next important advances will come with correlated histological, physiological, and biochemical analyses of individually identified muscle fibers and endplates (see Gunther and Letinsky, 1981). Many of the apparent conflicts in the field of avian muscular dystrophy may be resolved if these mo-e careful techniques are applied. 3.7. SUMMARY It is apparent from this discussion that normal adult neuromuscular junctions are highly modifiable. Long term changes in both structure and function occur seasonally and can result from altered use, aging, hormonal influences, and disease. Thus, in addition to its historical role in the study of basic synaptic physiology, the neuromuscular junction is emerging as a useful model for exploring the mechanisms of synaptic modulation and plasticity. 4. Responses of Adult Motoneurons to Experimental Manipulation When some of the motor axons to a muscle are sectioned or crushed (partial denervation), or when some synapses are inactivated pharmacologically, several long-term adjustments in the function of the system result. The functionally denervated muscle fibers undergo well documented changes in membrane properties, metabolism, and contraction characteristics, which are attributed to cut-off of a neurotrophic substance and/or to altered use (Guth, 1968; Harris, 1974; Gutmann, 1976; Purves, 1976b; Lomo, 1976; Fambrough, 1979; Lmno and Jansen, 1980). For present purposes, however, we are interested instead in the effects of these and other manipulations on MN function. These are seen particularly in three phenomena: (a) intact motor axons sprout new branches to innervate the denervated or inactive fibers, (b) interrupted axons undergo severe metabolic changes before regenerating to reestablish synapses, sometimes displacing the sprouted branches of the intact axons, and (cj transmitter release, and therefore synaptic effectiveness, can be dramatically altered. Some of these phenomena have been reviewed recently (Grafstein and McQuarrie, 1978; Brown et al., 1981), so the following treatment will concentrate mainly on recent work. 240 ALAN D. GRINNELL AND ALBERT A. HERRERA 4.1. SPROUTING Sprouting was first described as a response to partial denervation (Hoffman, 1950; Edds, 1953), but also can be induced by many other forms of manipulation. Motor axons can sprout from intramuscular nodes of Ranvier without previous branches (nodal or collateral sprouts), from the unmyelinated axon proximal to the terminal (preterminal sprouts), or from a portion of an existing terminal (ultraterminal sprouts) (Barker and |p, 1966). Many studies of axonal sprouting have not attempted to distinguish between these, especially between pre- and ultraterminal sprouting. Moreover, nodal and preterminal sprouts may go undetected, since they are often difficult to distinguish from original processes, especially if sufficient time is allowed for myelination to occur. These difficulties are inherent in morphological studies of sprouting. Physiological approaches, such as measurement of the amount of overlap of motor unit fields or quantification of the amount of polyneuronal innervation seen at individual endplates, do not provide information about the type of sprouting. We will distinguish, where possible, mainly between nodal and terminal sprouts. Different kinds of sprouts are seen under different conditions (Hoffman, 1950; Duchen and Strich, 1968; Ironton, et al., 1978; Brown, et al., 1980; Slack, et al., 1979) and to different degrees in different muscles (Brown, Holland and Ironton, 1980); so the distinctions should be kept in mind. Also, there is evidence that motor axons in old animals may sprout (or regenerate) less extensively or quickly than in young adult animals (Pestronk, et al., 1980). 4.1.1. Responses to partial denervation In adult mammalian muscles (in contrast to neonatal muscles, as described above, Section 2.4.4.), the sprouting response can be very strong. Motor units can increase by as much as 4-5 times their normal size when most axons to a muscle are severed (Brown et al., 1976; Jansen et al., 1975; Thompson and Jansen, 1977; Brown and Ironton, 1978). All remaining motor axons appear to enlarge their fields by approximately the same proportion (Brown and Ironton, 1978), although not all terminals in a given motor unit sprout equally, or at all (Brown, et al., 1980). Thus the number of junctions a MN maintains is not fixed, but rather can be regulated somehow by its environment. It is also of interest that the synapses in these enlarged motor units show evidence of being reduced in efficacy. They show increased susceptibility to block by curare, rapid fatigue, and reduced twitch/tetanus ratios, compared with normal motor units (Brown et al., 1976; Thompson and Jansen, 1977; Brown and Ironton, 1978. For similar findings in amphibians, see Bennett and Raftos, 1977; Wigston, 1979a,b, 1980; Haimann et al., 1981a,b). 4.1.2. Response to synaptic block Partial denervation is not the only effective stimulus. It appears that almost any treatment that blocks neuromuscular or muscle activity also causes sprouting in mammals (see Section 3.2, 3.3, 3.6). Even reducing MN activity indirectly by eliminating descending excitatory inputs (spinal isolation) produces terminal sprouting (Brown and Holland, 1979; Brown, et al., 1980). Motor terminals in cat muscles that have been spinally isolated and deafferented for periods of more than a year produce much larger and more variable EPPs, and show extensive collateral sprouting, resulting in polyneuronal innervation at synaptic sites separated by a mean distance about 400/~m (Eldridge et al., 1981). Presynaptic block with botulinum toxin (BoTX), which almost totally abolishes both evoked and spontaneous quantal transmitter release (Spitzer, 1972; Tonge, 1974), possibly by blocking calcium uptake into the nerve terminal (Hirokawa, et al., 1981), causes profuse terminal but not collateral sprouting (Duchen and Strich, 1968; Duchen, 1970; Pestronk and Drachman, 1978; Ironton, et al., 1978; Brown, et al., 1978; Duchen, et al., 1980). Similar, but less extensive terminal sprouting is seen in mice with genetic motor endplate disease (Duchen and Stefani, 1971; see Section 3.6.3) and in muscles paralyzed by tetanus toxin (Duchen and Tonge, 1973). In both these conditions evoked release is blocked, but not spontaneous release. Block of nerve conduction with tetrodo. SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 241 toxin (TTX) (Pestronk and Drachman, 1978; Brown and Ironton, 1977a; Betz et al., 1980a) or local anesthetics (Benoit and Changeux, 1978) similarly induces terminal but not collateral sprouting. Since many of these experiments involve use of nerve cuffs, which can cause damage to axons, the data should be interpreted cautiously. The toxins used may also have other potent effects on nerve or muscle function that are not yet understood. On balance, however, it appears convincing that presynaptic block does induce sprouting. The effects of postsynaptic block are more controversial. Pestronk and Drachman (1978) saw no sprouting in muscles paralyzed by ct-bungarotoxin (~-BTX), even after treatment with BoTX, which would otherwise have induced sprouting. Holland and Brown (1980) on the other hand, did find terminal sprouting when they used frequent injections of a highly purified form of ~t-BTX to block function. They found that ~-BTX increased the sprouting response to BoTX. Postsynaptic block due to loss of ACh receptors in human myasthenia gravis also leads to extensive terminal sprouting (see Section 3.6.1). Thus it seems probable that block of neurouscular activity either pre- or postsynaptically induces enlargement of terminals by sprouting. The absence of degenerating nerves indicates that inactivity per se is somehow an effective stimulus for sprouting. In the presence of denervated muscle fibers or with block due to BoTX these sprouts commonly grow to adjacent fibers and form synapses. In other forms of pharmacological block, however, such as TTX or ~t-BTX poisoning, the sprouts are usually very short and often do not extend to other muscle fibers (Brown and lronton, 1977a; Betz et al., 1980b). Denervation is clearly a more powerful stimulus than inactivity. Using a particularly favorable preparation of rat hind foot muscles, Betz et al., (1980) were able to achieve complete paralysis of one muscle and partial paralysis of others with a TTX cuff. They compared the effects of these levels of paralysis with that caused by partial denervation of the same muscles. Sprouting from active terminals in nerves that were only partially paralyzed was essentially the same as from terminals in totally paralyzed muscles, and in both cases only short terminal sprouts were seen, usually confined to the same muscle fiber. In contrast, more numerous, larger terminal and nodal sprouts were seen in partially denervated muscles. 4.1.3. Effect o f muscle activity on sprouting Since direct muscle stimulation is known to prevent or delay the effects of denervation (Lomo and Rosenthal, 1972; Lomo et al., 1974; Lorno, 1976), experiments have been done to see whether muscle activity can also affect the sprouting response. It was found that chronic direct muscle stimulation in brief tetani caused a dramatic decrease in sprouting both in partially denervated (Brown and Holland, 1979) and BoTX poisoned muscle (Brown et al., 1977). However, direct stimulation affected only terminal sprouts; it did not prevent collateral sprouting (Brown, et al., 1978, 1980). Moreover, stimulating the remaining functional axons in a partially denervated muscle did not affect their terminal sprouting response (Brown and Holland, 1979). The noncontracting denervated muscle fibers were somehow still able to induce sprouting. Based on their observation that slow muscles tend to show more terminal than nodal sprouting, and fast muscles the opposite proportion, Brown, Holland and Ironton (1980) point out that the predominance of terminal sprouting may be correlated with the greater denervation (or inactivity)-induced changes seen in the membrane properties of the slow fibers. Just as direct stimulation of denervated fibers prevents sprouting from intact terminals, functional innervation by a second nerve prevents sprouting by a BoTX-blocked original nerve. In experiments demonstrating this, a foreign nerve is implanted for about two weeks into an innervated muscle. If the original nerve is then sectioned (Fex and Jirmanov& 1969) or BoTX blocked (Fex et al., 1966; Duchen et al., 1975; Duchen and Tonge, 1977) the foreign nerve rapidly forms functional synapses. In the absence of the functional foreign nerve, the BoTX poisoned nerve would sprout; in the presence of the 242 ALAN D. GRINNELL AND ALBERT A. HERRERA foreign innervation it does not sprout. Interestingly, innervation by the foreign nerve, which occurs readily when the original nerve is blocked, can be prevented if the muscle is stimulated directly (Jansen et al., 1973). It should be noted that, in frogs, BoTX is much less effective than denervation in producing an increase in extrajunctional ACh sensitivity, and innervation of a BoTX-poisoned muscle by a foreign nerve takes much longer (Antony and Tonge, 1980). 4.1.4. Terminal and collateral sprouting as responses to different signals Many of the results cited above indicate that terminal and collateral (nodal) sprouting are triggered by different signals. Terminal sprouting is seen either in response to partial denervation or pharmacological synaptic block, while nodal sprouting occurs almost exclusively in cases of partial denervation. The resistance of collateral (nodal) sprouting to inhibition by direct muscle stimulation, and the usual absence of such sprouts except in the case of partial denervation, suggests that this form of sprouting is a separate phenomenon depending on a special factor released by degenerating nerve. Two other experimental results indicate that this is not the whole explanation. Slack, et al. (1979; see also Edds, 1953) carefully examined collateral sprouts in partially denervated muscles and found that they did not originate from randomly located nodes despite the uniform presence of degenerating axons. Instead the new branches were seen preferentially near points where denervated sheaths branched from partially denervated intramuscular nerves. Sprouts first appeared at nodes next to the shortest branches. Moreover, Brown and Holland (1979) cite preliminary experiments in which they prevented terminal sprouting for 6 days by direct stimulation of a partially denervated mouse soleus muscle; upon stopping the stimulation, a good terminal sprouting response was still seen, even though nerve degeneration products should have been removed by that time. The signal to sprout might nevertheless have been communicated to the terminals, which could respond only after the inhibitory effect of muscle stimulation was terminated. It is possible that the distinction between terminal and nodal sprouting may apply principally at short times. Eldridge et al. (1981) have studied the effects of very long-term (2-3 years) spinal isolation and deafferentation on cat neuromuscular junctions. They found extensive collateral sprouting, but relatively little terminal sprouting. It is possible that there was some partial denervation in these muscles, but the primary stimulus was probably inactivity. 4.1.5. Postulated mechanisms for the regulation of sprouting Since nodal and terminal sprouting may be triggered by different signals, it is probable that no single mechanism will explain the M N sprouting response. Partial denervation, which strongly stimulates both types of sprouting, results in the presence of degenerating nerve products in the muscle, as well as denervation changes in the muscle fibers. Muscle inactivity leads to some of the postsynaptic effects of denervation (Lomo, 1976; see above, Section 4.1.3), but no denervation products. 4.1.5.1. Peripheral sprouting factors In the first studies of sprouting, Hoffman (1950; see also Hoffman and Springell, 1951) found that lipids extracted from peripheral nerves caused sprouting in normally innervated muscle and postulated a sprout-inducing substance ("neurocletin") that was released from degenerating axons following partial denervation. Moreover, sectioning of the dorsal root peripheral to the ganglion, which leads to degeneration of sensory nerves in a muscle, causes motor terminal sprouting, albeit less than with damage to the motor system (Brown et al., 1978). Deafferentation without peripheral nerve degeneration, by dorsal root section central to the ganglion, causes no sprouting (Brown and Holland, 1979). Degenerating nerve products thus appear to be an effective stimulus for sprouting. However, if such products are the primary stimulus to sprouting, especially nodal sprouting, why are such sprouts seen only in the intramuscular nerve, and at branch points, SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 243 whereas degenerating axons run the length of the nerve (Slack et al., 1979)? Another regulatory substance, released from denervated or inactive muscle fibers, or a degeneration product unique to the terminals, is indicated. On the other hand, there is growing evidence that terminal sprouting, at least in mammals, is not a long-range response to a diffusible substance, but instead is triggered by the immediate proximity of denervated or inactive muscle fibers. Brown, Holland, Hopkins and Keynes (1981) have recently examined sprouting in a number of partially denervated rat and mouse muscles in which proximity between denervated and intact endplates differed. Sprouting was seen only when innervated and denervated endplates were close together, or after BoTX treatment. Intermingling of innervated and denervated fibers, without endplate proximity, did not result in terminal sprouting. They also confirmed the finding of Weiss and Edds (1946) in that no sprouting was seen between an intact and a denervated hemidiaphragm, or when an intact muscle was surrounded by denervated muscles. If the signal for terminal sprouting is as localized as the results of Brown, Holland, Hopkins, and Keynes (1981) indicate, it may be a non-diffusible integral component of the denervated (or inactive) muscle fiber membrane or surrounding basement membrane. ACh receptors would be an obvious candidate for this signal since they appear in large numbers on denervated fibers. Pestronk and Drachman (1978) proposed that ACh receptors are the sprouting signal, since they found that the extent of sprouting was proportional to ACh sensitivity and sprouting was blocked by ~t-BTX. As was mentioned above, however, Holland and Brown (1980) were unable to block sprouting with ~-BTX. Moreover, there are several experiments that indicate that a sprouting signal can spread for significant distances from denervated to innervated fibers (Betz et al., 1980b; Brown and Holland, 1979). If such signals are not widely diffusible, these results force one to postulate that intact nerve terminals elaborate long processes that probe the environment and grow into stable sprouts only when they contact a denervated fiber (Brown, Holland, Hopkins and Keynes, 1981). Further information on the nature of the signal for terminal sprouting is lacking. It should be noted that sprouting might be either a positive response to a "sprouting factor", or a normal tendency held in check by an antisprouting factor produced by active muscle fibers or the nerve innervating them. Some sprouting occurs normally (Tuffery, 1971; Wernig et al., 1980), but whether this is a response to fluctuations in extrinsic signals, or muscle feedback, is unknown. On the basis of studies of sprouting by intact sensory axons to innervate denervated salamander skin, Diamond and his associates (see reviews by Diamond et al., 1976; Diamond, 1979) have postulated a sprouting substance, released by denervated Markel (touch sensory) cells (Diamond, et al., 1976; Cooper, et al., 1977), that is either removed from the environment by an intact sensory nerve or suppressed by an "anti-sprouting" factor released by the intact nerve. The primary evidence for this is the finding that block of axoplasmic transport by colchicine (Dahlstrom, 1968; Kreutzberg, 1969; Lubinska, 1975), which does not appear to interfere with the sensory function of the nerve, leads to sprouting of adjacent normal axons to innervate the field of the colchicine-blocked nerves (Aguilar et al., 1973; Diamond, 1979). The axons with blocked transport do not sprout, but neither apparently can they prevent sprouting by nearby axons. Complicating the interpretation of these experiments is the fact that colchicine, in concentrations only slightly exceeding those necessary to obtain significant block of transport, kills some axons (Jackson and Diamond, 1977). Also, colchicine causes regression of neurites when applied directly to a muscle (Cangiano and Fried, 1977). However, Jackson and Diamond (1977) found that concentrations which block transport by 30)o cause sprouting but only negligible nerve damage. Indeed, if sprouting was primarily a response to colchicine-induced nerve degeneration (Brown, Holland and Hopkins, 1981), surviving axons of the colchicine-blocked nerve should also have sprouted. These are intriguing experiments, but it seems likely that any experiments using this substance will be subject to question unless one rigorously controls for its toxicity. The observation that spinal isolation or TTX block induce sprouting (Brown 244 ALAN D. GRINNELL AND ALBERT A. HERRERA and Ironton, 1977a; Brown, et al., 1980; Pestronk and Drachman, 1978; Betz et al., 1980a; Eldridge et al., 1981) but do not affect axonal transport, argues against an antisprouting factor. In addition, since direct muscle stimulation acts like functional innervation in preventing sprouting (Brown and Holland, 1979; Brown et al., 1980), it may be adequate to postulate that motor nerves control sprouting only by modulating muscle fiber activity which in turn regulates production of a sprouting substance. Sensory nerve sprouting shows another revealing peculiarity that should be examined more carefully in motor innervation. In the skin of both salamanders (Aguilar et al., 1973; Cooper et al., 1977) and rats (Diamond and Jackson, 1980), there are borders beyond which a particular sensory nerve will not sprout. These "domains" are consistently oriented with respect to underlying body coordinates, whether the skin is in its normal position or experimentally rotated (Diamond et al., 1976). In rat skin, they are confined to single dermatones (Diamond and Jackson, 1980). The only comparable observation in the case of neuromuscular sprouting is that of Slack (personal communication cited by Mark, 1980) that segmental limb nerves in axolotls will sprout throughout an adjacent spinal nerve's territory, but not into the territory of the next spinal nerve beyond that. In most cases, of course, individual muscles are separated from one another by barriers which sprouts cannot traverse. However, there are instances in which different axons from the same spinal segment end selectively in different regions of a given muscle (Burke et al., 1977; Genat and Mark, 1977). It is possible that within single muscles there may be some degree of specificity restricting the area within which the axons of a given nerve will sprout. This possibility must be kept in mind when assessing the results of experiments involving competition between nerves in doubly innervated muscles (see below). It is noteworthy that regenerating axons, whether sensory or motor, are not subject to the same spatial constraints as sprouting axons. We must also call attention to a remarkable dependence of sensory nerve sprouting on activity reported by Nixon et al. (1980). They found that intact nociceptive afferents in rat skin showed little tendency to sprout when surrounded by denervated skin. However, if the intact nerve was stimulated, even for as little as 2 minutes at the time when surrounding skin was denervated, the axons sprouted extensively to innervate an area several times as large as normal, with no regard to dermatone boundaries. 4.1.5.2. Central regulatory mechanisms and contralateral sproutin9 It is also important to note that regulation of sprouting may occur in the spinal cord. Axotomy or peripheral block of function in some MNs leads to changes in their synaptic input and that of surrounding MNs (see next section). Recent experiments on frog muscle imply that these changes may be effective in influencing sprouting. Denervation of one frog cutaneous pectoris (CP) muscle causes a large increase in sprouting and polyneuronal innervation in the contralateral CP muscle (Rotshenker and McMahan, 1976; Rotshenker, 1979; Rotshenker and Reichert, 1980). This contralateral response is seen after complete removal of the CP muscle (Rotshenker, 1978b, 1979) and after colchicine block of the CP nerve (Rotshenker, 1980), but not after denervation of other muscles near the CP or the entire hind limb (Rotshenker, 1978b). These findings suggest that sprouting is a response to injury or block of retrograde transport of some substance to MNs, and that the response is mediated centrally. The role of changes in activity pattern in the intact MNs has not been investigated, but it is noteworthy that in an analogous preparation, the nerve terminals of a frog sartorius muscle show sharply enhanced release of transmitter (without apparent terminal sprouting in the 2-3 month period studied) following denervation of the contralateral sartorius (Herrera and Grinneli, 1981; see Section 4.3.2.). Weakly and Yao (1981) postulated that the morphological substrate for the interaction may be MN dendrites that extend across the midline (Sz6kely, 1976; Liu, 1969; Bregman and Cruce, 1980) and can even make synaptic contact with contralateral MNs (Erulkar and Soller, 1977; Soller and Erulkar, 1978). In mice, no evidence of sprouting has been found in muscles of one leg following section of the contralateral sciatic nerve, or in the soleus muscle after denervation of all SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 245 other muscles in the same leg (Brown et al., 1980). O n t h e other hand, Steinbach (1981), in experiments on the cat hindlimb, did find a slight increase in mean junctional length in 2 fast muscles, but not in the soleus, many months after contralateral ventral root or sciatic nerve section. This change may likely be due to altered use. Further, he found no histological evidence for multiple innervation. The spinal cord's role in sprouting, if there is one, may not be important in mammals. 4.t.6. Summary Sprouting of motor axons appears to parallel the development of denervation-like changes in muscle fibers, whether the changes are caused by actual denervation or by pharmacological block of synapses. Sprouts may originate either from intramuscular nodes or from the axon terminals and can enlarge an axon's field by 4-5 x. Nodal sprouting probably is a response to one or more diffusible substances derived from degenerating axons and denervated muscle. Terminal sprouting appears to result from a more localized signal, perhaps a component of the muscle fiber membrane. Recent evidence for central nervous system (CNS) involvement in the sprouting of contralateral MNs suggests the possibility that spinal cord interactions may be important, and emphasizes how little we really know about sprouting. 4.2. AXOTOMYAND THE DEPENDENCEOF ADULT MOTOR NEURONS ON THE PERIPHERY 4.2.1. Metabolic responses When a vertebrate axon is sectioned (axotomy), the portion distal to the cut degenerates while the proximal portion and the cell body undergo profound changes (Watson, 1968, 1974; Lieberman, 1971; Harris, 1974; Grafstein and McQuarrie, 1978). This phenomenon, which has been investigated mainly in mammalian MNs, provides convincing evidence that peripheral target tissues can influence the metabolic state and spinal connections of neurons which innervate them. The neuron soma shows pronounced changes in metabolism within two to ten days after nerve section or ligation, depending on the distance between the lesion and the cell body, and the preparation being studied. These changes, classically described as chromatolysis, are thought to represent a switching of the synthetic apparatus from routine tasks of maintenance and transmitter production to manufacture of material for axon regeneration (Watson, 1974; Matthews and Raisman, 1972). After an initial burst of metabolic activity and nucleic acid synthesis there is a period of relatively depressed cellular metabolism. If the axon does not regenerate functional connections, protein synthetic rates remain low or may decline further until the cell dies. However, if the axon restores functional connections, there is a second burst of nucleic acid metabolism and synthesis. This was convincingly demonstrated by Watson (1970), who implanted the hypoglossal nerve into normally innervated sternomastoid muscle in mice. The hypoglossal could not form synapses in the presence of the functional original nerve. Under these conditions cell body dry mass and nucleolar nucleic acid content remained low indefinitely. However, a few days after the original nerve (the spinal accessory) was sectioned, the hypoglossal MNs showed a strong metabolic resurgence, coincident with the formation of functional synapses. Clearly, the axon tip plays an important role in determining the metabolic state of the MN. 4.2.2. Changes in motoneuron electrical properties Kuno and his collaborators (Kuno and Llinas, 1970a,b; Kuno et al., 1974a,b; Goldring et al., 1980) carefully studied the electrical properties of mammalian MNs following axotomy. Within a few days after ventral root section there was an increase in soma excitability and the appearance of dendritic action potentials (Kuno and Llinas, 1970a). Significantly, when the axotomy was performed more peripherally, dendritic spikes were rarely seen. MNs innervating fast and slow muscles, which usually differ with regard to 246 ALAN D. GRINNELL AND ALBERT A. HERRERA somatic spike overshoot and after hyperpolarization and axonal conduction velocity tend to lose these differences following axotomy (Kuno et al., 1974a). MNs of slow muscles show more pronounced responses to axotomy than do those of fast muscles (Goldring et al., 1980). Both types of MNs regain their original characteristics upon reinnervation of muscle, even if the nerves have been crossed so that they now innervate the "wrong" kind of muscle (Kuno et al., 1974b). The muscles, on the other hand, acquire contraction characteristics appropriate to the MNs innervating them. Surprisingly, the restoration of normal MN electrical properties was reported to be as successful when the MNs had not established functional endplates as when they had (Kuno et al., 1974b). 4.2.3. Changes in synaptic circuitry in the spinal cord A most fascinating consequence of axotomy is the disruption of synaptic input onto the cell body which in turn can drastically affect neuromuscular activity. In the 1950's, Eccles and his collaborators observed that there was a decrease in the magnitude of monosynaptic excitatory postsynaptic potentials (EPSPs) recorded from axotomized cat MNs, with little or no change in inhibitory postsynaptic potentials (Downman, et al., 1953; Eccles et al., 1958). Intracellular recording from the soma suggested a somewhat selective loss of synapses onto the soma and proximal portion of dendrites (Kuno and Llinas, 1970b; Mendell, et al., 1974), although one might expect that such techniques would have only a limited ability to detect changes in distal dendritic input. As with other effects of axotomy, inputs onto MNs of the soleus were more rapidly and severely affected than inputs onto MNs of fast twitch muscles in rats (Goldring et al., 1980). If regeneration of motor axons back to muscles occurred within about two months, the muscles became well innervated and EPSPs recovered almost completely (Kuno et al., 1974b; Goldring et al., Goldring et al., 1980). If the sectioned nerve was prevented from regnerating into a target muscle for 6-8 months, however, there was little functional innervation by either sensory or motor axons (Goldring et al., 1980). Nevertheless, there was almost complete restoration of normal EPSPs in the medial and lateral gastrocnemius~ the soleus MNs still showed depressed EPSPs (Goidring et al., 1980). This physiological evidence is confirmed by morphological findings. Loss of boutons in the axotomized MNs can be seen both at the ultrastructural level (Blinzinger and Kreutzberg, 1968; Hamberger, et al., 1970; Torvik and Skj6rten, 1971; Sumner and Sutherland, 1973; Mendell, et al., 1976) and with zinc iodide stain and light microscopy (Sumner and Sutherland, 1973; Mendell et al., 1974; Cull, 1974). Since there is no sign of bouton degeneration, the general conclusion is that boutons must be withdrawn or gradually resorbed (Sumner and Sutherland, 1973). This withdrawal (an interesting form of synapse elimination) is preceded by the loss of post-synaptic thickenings and changes in EPSP amplitude and rise time (Sumner, 1975). Consistent with the electrophysiological data, boutons tend to disappear earliest and in greatest numbers from the soma, later from the dendrites. It is also reported that there is preferential loss of boutons with spherical vesicles, leaving mainly endings with flat or mixed populations of vesicles (Sumner, 1975). While synapses are being lost there is a decrease in size of dendrites (Sumner and Sutherland, 1973). Most of these changes are reversible after regeneration, hut only if functional reinnervation of muscle takes place (Watson, 1974; Sumner, 1975; but see Goldring et al., 1980). Whether the restored boutons are from the original afferents is now known, but Mendell and Scott (1975) observed that, even after peripheral nerve cross (and contrary to earlier evidence by Eccles et al., 1962), there is restoration of the original numbers and types of EPSPs, suggesting accurate reinnervation. Likewise, in the guinea pig superior cervical ganglion (SCG), where axotomy causes similar chromatolytic changes and loss of synaptic inputs (Matthews and Raisman, 1972; Matthews and Nelson, 1975; Purves, 1975), the normal pattern and organization of preganglionic inputs return with peripheral reinnervation of target tissue even when the wrong peripheral connections are made (Purves and Thompson, 1979). SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 247 4.2.4. Effects on neuro#lia In addition to the effects of axotomy on MNs and their afferent inputs, the gila in the anterior horn are also profoundly affected. Although the role of the neuroglia in the CNS is poorly understood, there is increasing appreciation of their possible importance in maintaining an appropriate extracellular environment and in trophic interaction with neurons (Orkand, 1980). These interactions may prove important in regulation of MN health and activity. At about the time when boutons begin to disappear, the microglia proliferate in large numbers (Watson, 1965), only to disappear 3-4 weeks later, whether or not there is peripheral reinnervation of muscle (Sumner, 1975). If an axotomized nerve regenerates and makes functional connections, synaptic boutons reappear and a second nerve section causes the same chromatolytic response, loss of boutons, and proliferation of microglia as the initial axotomy. If the nerve has not made functional connections, however, boutons are not restored and, although there is a chromatolytic response to the second section, there is no glial response (Watson, 1974; Sumner, 1975). The role of the microglia is not well understood. They proliferate only when boutons are being lost, but there is no evidence for phagocytosis of boutons or axonal processes (Sumner, 1975). As the boutons detach, the space between the terminal and the postsynaptic MN is occupied by microglial processes (Blinzinger and Kreutzberg, 1968; Kerns and Hinsman, 1973). Whether the glia are in any way responsible for the synaptic detachment is not clear. Astroglia showed two periods of response to nerve section. At the time of synapse detachment, astrocytes proliferate, increase their metabolic activity, elaborate new processes, and eventually replace the microglia covering the old synaptic surface. Subsequently, astrocytes show another metabolic response at the time of reconnection of synaptic boutons (Sumner and Sutherland, 1973; Watson, 1972). (It is known that astrocyte metabolism is stimulated by an increase in extracellular K ÷ and may be involved in regulation of extracellular concentrations of this ion IOrkand, et al., 1966; Orkand, 1980; but see Tang et al., 1980]). A slight but rapid increase in astroglial numbers is seen on the side of the cord contralateral to axotomy, perhaps as a result of changes in activity (Watson, 1968, 1972). Finally, the oligodendroglia show a metabolic reaction only at the time of synaptic reconnection, possibly because of the need for more myelin to wrap new axonal branches at that time (Watson, 1972, 1974). These changes in glial number and metabolism may only be responses to changes in local conditions within the spinal cord, but they are eloquent testimony to the dynamic state of intercellular interactions in the cord, and the susceptibility of these interactions to peripheral damage. 4.2.5. Role of sensory axons in the axotomy response It is important to point out that most of the experiments described above involved not only axotomy of motor axons, but section of peripheral sensory axons as well. The observations that the changes in synaptic input occur only when the ventral root is sectioned (Eccles et al., 1958; Kuno and Minas, 1970b; Farel, 1978), and that similar changes are seen, at greater delay, with peripheral nerve section (Mendell, et al., 1974; Mendell and Scott, 1975), imply that axotomy of motor axons is adequate to evoke the changes. Moreover, Cull (1975) found that section of the lingual nerve had no effect on the afferent innervation of hypoglossal MNs. However, papers by Kuno and his colleagues strongly suggest that cutting the peripheral process of a sensory axon affects the function of its central processes, and that these changes are responsible for at least some of the affects of axotomy. Gallego et al. (1979b) and Goldring et al. (1980) found that section of the medial gastrocnemius nerve caused a decrease in EPSPs not only in the axotomized MNs but in all MNs postsynaptic to the sensory axons in the sectioned nerve, i.e. onto the MNs of the lateral gastrocnemius and soleus as well. In contrast, synaptic inputs from sensory afferents in the intact lateral gastrocnemius and soleus nerves had no diminution of effect, even on medial gastrocnemius MNs. Since the operative procedure necessary for achieving ventral root section often results in dorsal root 248 ALAN D. GRINNELL AND ALBERT A. HERRERA damage (L. Eldridge, personal communication), it is important to consider the possibility that many of the effects of axotomy on synaptic boutons, and perhaps even neuroglia, might be due to sensory nerve damage. 4.2.6. Altered neuromuscular activity and peripheral regulatory substances Both physical damage to the MNs themselves, and disruption of their sensory input, may contribute to the changes seen in MNs following axotomy. In addition, however, there is convincing evidence for a regulatory substance supplied by the periphery. The fact that chromatolytic and synaptic changes occur at greater delay when the site of nerve section is more peripheral (Watson, 1974; Mendell, et al., 1974; Mendell and Scott, 1975) might be best interpreted as the exhaustion of a substance that is transported from the periphery to the cell body. Moreover, some of the effects of axotomy can be mimicked by intramuscular injection of BoTX (Watson, 1969; Rong and Xing, 1981). This is an interesting but potentially complicated phenomenon, for the mechanism of action of BoTX is not well understood. The toxin has this effect only if the MNs have formed functional synapses; a supernumerary nerve in already-innervated muscle does not respond to the BoTX directly, but does show a delayed metabolic response associated with synapse formation after the synapses of the original nerve are blocked (Watson, 1969). One might postulate that normally innervated muscle produces a substance that is picked up by functional synapses and transported to the soma where it maintains metabolic activity appropriate to that state. If BoTX acts by interfering with such a process, it is probably by block of uptake, rather than of transport, since it has been shown that the toxin does not block axoplasmic transport (J. W. Griffin, personal communication). As we saw above (Section 4.1.2.), BoTX, which blocks functional synapses, simulates denervation and has the expected effect of inducing sprouting and synapse formation by a supernumerary nerve. However, this synapse formation occurs at considerably greater delay than would be observed if the original nerve was simply sectioned (Watson, 1970), implying that BoTX may delay the response of the supernumerary nerve to the denervation signal, or may delay the production of that signal. More straightforward evidence for production of_a~u'~n-maintaining substances by target tissues has been found in other systems. Colchicine, which blocks axoplasmic transport in the postganglionic nerve, mimics the effects of axotomy in both the chick ciliary ganglion (Pilar and Landmesser, 1972) and the guinea pig SCG (Purves, 1976a). In the case of the sympathetic ganglion, a good candidate for the substance has been identified. Nerve growth factor, which is produced by the normal target tissues of these ganglion cells, has been shown to prevent or reverse the changes that normally occur with axotomy: dendritic spikes, metabolic changes, loss of boutons, and decrease in EPSP amplitude (Hendry, 1975; Purves and Nja, 1976; see also Section 2.3.2. for discussion of evidence for factors influencing 16arasympathetic ganglia/. Simple changes in the activity level of neuromuscular synapses do not mimic most of the effects of axotomy but can have profound effects on synaptic inputs to MNs. For example, early studies showed that tenotomy caused an increase in EPSP amplitudes in MNs innervating the tenotomized, and presumably less active, muscle (Kozak and Westerman, 1961}. Unfortunately, the effects of tenotomy on spinal cord activity are not well known. Recently Kuno and his associates have studied the effects of altered activity more systematically. They find that changes in soleus MN membrane properties can be brought about by several different manipulations, all of which produce severe muscle inactivity, e.g. spinal cord transection or a TTX cuff on the nerve (Cz~h et al., 1978~. Severe changes are seen even in MNs that continue to function normally but innervate partially denervated muscle (Huizer et al., 1977t..When nerves are blocked with TTX. these changes can be reversed by daily nerve stimulation distal to the cuff, which prevents many of the denervation-like alterations in the muscle, but not by stimulation central to the cuff, which activates the MNs but not the muscle {Cz6h et al., 1978). This result is reassuring evidence that the changes in MN properties are not due simply to axonal SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 249 damage caused by the cuff itself, always a potential problem that must be considered (Cangiano and Fried, 1977). In similar experiments a TTX cuff was used to block conduction in the medial gastrocnemius nerve (Gallego et al., 1979b). This was found to cause an increase in EPSP amplitude for sensory inputs from the medial gastrocnemius muscle to medial or lateral gastrocnemius MNs but not for afferents from other nerves onto medial gastrocnemius MNs. Thus the increase in EPSP size is seen in pathways where sensory input has been reduced but not in other pathways. If MNs reflect the physiological state of the muscles they innervate, as implied by the results above, then other procedures that drastically alter muscle physiology might be expected to have effects on the MNs. It has long been known that muscles that are chronically immobilized in a shortened position undergo atrophy, whereas immobilization in a lengthened position prevents or delays even the effects of severe disuse such as that caused by spinal cord transection (Thomsen and Luco, 1944; Tabery et al., 1972; Goldspink, 1977). Gallego et al. (1979a) found that deafferented soleus MNs isolated by spinal cord transection maintained several of their normal properties when the muscle was immobilized in a lengthened position, but showed changes when the muscle was immobilized in a shortened position. Thus the normal electrophysiological properties of these MNs appear to depend on the metabolic state of the muscle, rather than on contractile activity per se. It is probable that this influence is exerted by retrograde transport of a trophic substance. These findings in mammals become all the more interesting in view of the strong enhancement of transmitter release from frog motor nerve terminals in stretched muscles (see above, Section 3.1.). 4.2.7. Summary Sectioning a motor axon, or blocking axonal transport or its ability to drive postsynaptic fibers, causes dramatic changes in a motoneuron's (MN's) metabolism, its synaptic inputs, and the surrounding neuroglia. Adjacent intact MNs also undergo changes in circuitry and may be affected by glial changes. Thus at the same time that there are changes in the muscle fibers, and sprouting responses from intact MNs, there are important changes occurring in the spinal cord. Some of the axotomy-induced changes in MN metabolism and connections are summarized in Fig. 6. It is probable that part of the regulation of neuronal health and electrical properties is due to retrograde transport of one or more substances supplied by the muscle fibers, perhaps analogous to N G F in target cells of sympathetic neurons. The role of sensory nerve damage in the MN response needs careful further assessment. 4.3. EXPERIMENTAL ALTERATION OF SYNAPTIC EFFECTIVENESS It has recently become clear that synaptic effectiveness can be semi-permanently altered by various forms of experimental manipulation. These studies imply a close relationship between synaptic effectiveness and motor unit size, and suggest that interactions between cells in the spinal cord may be involved in the long term regulation of synaptic effectiveness. 4.3.1. The dependence of synaptic effectiveness on motor unit size Several studies of muscle reinnervation have suggested that a neuron's success in competitive synaptic interaction depends on the size of its peripheral field. MNs that have enlarged their peripheral fields by sprouting to innervated denervated muscle appear to suffer a competitive disadvantage (Brown et al., 1976; Thompson and Jansen, 1977; Thompson, 1978; Brown and Ironton, 1978; Wigston, 1979b, 1980; Haimann et al., 1981a, b; see also Purves, 1976c; Sargent and Dennis, 1981). Strong support for the regulation of synaptic effectiveness as a function of axonal arborization size comes from experiments in which reducing the size of frog sartorius motor units caused a marked increase in transmitter release and safety margin (Herrera and Grinnell, 1980b, see Fig. 7). 250 ALAN D. GRINNELL AND ALBERT A. HERRERA N ICRE ~/$ED ~ FIG. 6. Schematic diagram summarizing many of the effects of partial denervation on the axotomized motoneurons and adjacent and contralateral cells. As the distal stumps of the cut axons degenerate, the denervated muscle cells undergo a number of changes (see Section 4). These, coupled with the presence of degenerating nerve products and perhaps events in the spinal cord, lead to sprouting by adjacent intact axons to innervate the old endplate sites (Section 4.1.1, 4.1.4, 4.1.5). Meanwhile. the axotomized motoneurons undergo metabolic [Section 4.2.11 and electrophysiological (Section 4.2.2) changes. Synaptic boutons are withdrawn from the injured motoneurons ISection 4.2.3) and from intact motoneurons of the same muscle if sensory axons have been damaged [Section 4.2.5). Glial cells proliferate to fill the old synaptic surfaces ISection 4.2.41. The effects of denervation are seen even contralaterally, where, in some instances, terminals may sprout or show increased transmitter release (Section 4.1.5.2, 4.3.2). As mentioned above (see Section 3.1), the effectiveness of synaptic transmission is dependent on m o t o r unit size even in normally innervated frog muscle. The observation that in normal animals, the largest m o t o r units have the most effective synapses is consistent with the hypothesis that M N s with large peripheral fields supply more of an essential material to each terminal, resulting in higher safety margins. O n e can speculate that production and/or transport may be fixed during development so that reducing the peripheral field size (Herrera and Grinnell, 1980bi would cause the original a m o u n t of this material to be distributed to fewer terminals, resulting in enhanced transmitter release. 4.3.2. Synaptic effecticeness can be altered by contralateral denerl'ation Several old and recent reports indicate that unilateral denervation of skeletal muscles can cause changes in contralateral innervation. At least four contralateral effects have been described: (1)loss of m o t o r axons, (2t sprouting, (3) increased polyneuronal innervation, and (4) increased synaptic effectiveness. D u n n (1909) provided the first description of contralateral effects of denervation. She found a loss of contralateral m o t o r axons following unilateral nerve section in frogs. Similar findings were reported in rats (Greenman, 1913; Tamaki, 1933, 1936), but m o d e r n techniques indicate that this does not occur in the mouse (Brown et al., 1980). C o n t r a lateral sprouting, and a consequent increase in focal polyneuronal innervation, has been described by Rotshenker and colleagues, who used the frog cutaneous pectoris (CP) preparation (Rotshenker and M c M a h a n , 1976; Rotshenker, 1978a,b, 1979; Reichert and Rotshenker, 1979: Rotshenker and Reichert, 1980}. These and related findings are discussed above, under Sprouting (see Section 4.1.5.2.). However, histological studies have shown that unilateral denervation does not cause wide.spread contralateral nerve terminal sprouting in the frog sartorius preparation (Herrera and Grinnell, 1980c, 1981) or mouse hindlimb muscles (Brown et al., 1980; but see Steinbach, 1981). Indirect evidence SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 251 2.5 I 2.0 _2 g 1.5 I E E 8 1.0 8 0.5 C reinnerv, D E normal re=nnerv. contralat, cutaneous sarI. sarI. 1/2 sarI. sart. peclor~s (31) (16) (2 t) (26) (42) FIG. 7. Naturally occurring (A,E) and experimentally induced (B-D) differences in synaptic effectiveness at frog neuromuscular junctions. Effectiveness is expressed as mean number of quanta ( + s.e.m.) released per 100 pm nerve terminal length at individually identified junctions. Number of junctions examined in parentheses. (A) Normal sartorius. (B) Reinnervated sartorius 8-12 weeks after nerve crush. (C) Reinvervated sartorius 8 12 weeks after crushing nerve and excising half the muscle fibers. Motor unit size is reduced in these preparations. (D) Sartorius muscles 5-16 weeks after denervation of contralateral sartorius. (El Normal cutaneous pectoris. (After Grinnell and Herrera, 1980a; Herrera and Grinnell, 1980b, 1981). has led others to conclude there is no contralateral sprouting response in the case of the frog sartorius (Weakly and Yao, 1981) and Xenopus pectoral muscles (Haimann et al., 1981b). Unilateral denervation can also change contralateral synaptic effectiveness. We recently reported that denervation of one frog sartorius muscle causes a large and persistent increase in safety margin in the contralateral sartorius (Herrera and Grinnell, 1981). Measurements of nerve terminal length show that the increased effectiveness is not due to an increase in synaptic size. Rather, it appears that the inherent release properties of the terminal are altered, since quantal content measurements in low [Ca 2+] show a 3-8 fold increase in transmitter release per unit nerve terminal length (Fig. 7). In contrast to the results of Rotshenker (Reichert and Rotshenker, 1979; Rotshenker and Reichert, 1980), we find no morphological evidence for contralateral nerve terminal sprouting. However, unlikely Weakly and Yao (1981), we do see an apparent contralateral increase in focal polyneuronal innervation. It is possible that this increase may not be due to sprouting, but rather to the increase in detectability of normally present but undetectably weak polyneuronal inputs (see Herrera and Grinnell, 1981). It has been proposed that the sprouting signal passes transneuronally across the thoracic spinal cord (Rotshenker, 1979) to the contralaterai CP muscle. This may involve the medial dendrites of MNs (Bregman and Cruce, 1980) which, in the frog, cross the midline of the cord. Weakly and Yao (1981) suggest that their observation of no increase in polyneuronal innervation in the contralateral sartorius may be expected if MNs with medially projecting dendrites are much less common in lumbar segments of the cord I.P.N. 17/4--0 252 ALAN D. GRINNELL AND ALBERT A. HERRERA than in thoracic segments. However, such dendrites have been reported in the lumbar cord of frogs (Liu, 1969: Sz6kely. 1976). In addition, Erulkar has clearly shown that MNs on the two sides of the frog lumbar spinal cord make direct electrical and chemical synaptic contact with each other via contralaterally projecting dendrites (Erulkar and Soller, 1977; Soller and Erulkar, 1978; see also Grinnell, 1966). These synapses are sufficiently effective that, under some conditions, antidromic stimulation of one lumbar ventral root can cause the appearance of action potentials in the contralateral root. Unilateral denervation is likely to alter the work load of the contralateral intact muscle, especially in the case of the CP. There is no unequivocal evidence (but see Rotshenker, 1979) whether some or all of these contralateral effects might be due to changes in muscle activity in response to the demands of increased work. Watson has shown that unilateral section of the hypoglossal nerve in the rat can cause changes in nucleolar function in neurons (Watson, 1968) and astroglia (Watson, 1972) in the contralateral hypoglossal nucleus. He proposes that these changes may be due to work hypertrophy since, after the operation, all tongue movement is controlled by the single intact nerve. 4.3.3. Summary These studies clearly establish that synaptic effectiveness at the neuromuscular junction is subject to long term regulation. Correlations between synaptic effectiveness and motor unit size suggest that the commitment of each motoneuron to support transmitter release is finely tuned to the size and activity level of its peripheral field. The effects of contralateral denervation on synaptic effectiveness indicate that spinal mechanisms may exist that coordinate regulation of bilaterally symmetrical muscles. Perhaps most exciting is the possibility that further study of this well-characterized synapse will reveal the intraceilular mechanism by which transmitter release is enhanced, providing a specific model for long term physiological plasticity in the central nervous system. 5. Regulation of Motoneuron Function During Regeneration 5.1. SELECTIVE REINNERVATION OF OLD ENDPLATE SITES Before developing muscle fibers are innervated, there are no detectable specializations that mark the future site of synapse formation. However, once a functional synapse has formed, the synaptic site is permanently altered. Even though there is polyneuronal innervation during development, almost all the axons innervating a given fast fiber end at the same spots and all but one of these inputs are lost during the period of synapse elimination (see Section 2.3). Normally. an innervated adult muscle is refractory to further innervation (Elsberg, 1917: Aitken. 1950; Guth and Zalewski, 1963: Gwynn and Aitken, 1966). If a foreign motor nerve is implanted into already innervated muscle, processes ramify near the point of implantation but no new junctions are formed as long as the muscle has not been damaged. An interesting exception to this general rule is the report by Bixby and Van Essen (1979b) that foreign motor axons transplanted onto the rat soleus in the region of the intact normal nerve can occasionally form synapses. Foreign synapses are always formed at the site of the original endplate, confirming the "'specialness" of that location. This result suggests that processes of the foreign nerve can somehow force the original nerve out of synaptic gutters or occupy gutters exposed by the normal remodelling process (Tuffery. 1971). The fact that some endplates appear to be taken over entirely by a foreign nerve suggests that there is some element of competition involved. Upon denervation of a muscle, the barrier to innervation is rapidly lost. In mammals. reinnervation is essentially complete 2-4 weeks after nerve section (Saito and Zacks, 1969; Fex and Jirmanov& 1969). Most of this time is presumably taken up by axonal regrowth, since Fex and Thesleff (1967) demonstratecl that synapse formation itself can occur quite rapidly. They implanted the deep peroneal nerve into the rat gastrocnemius SPECIFICITY AND PLASTICITYoF NEUROMUSCULARCONNECTIONS 253 muscle 70 days before sectioning the native (tibial) nerve. Since the foreign nerve had already recovered from axotomy and grown into the muscle, it was able to form synapses within 2-3 days after muscle denervation. Many subsequent studies have confirmed this finding, and the technique is commonly used to achieve rapid foreign innervation (Fex and Jirmanovfi, 1969; Watson, 1970; Jansen et al., 1973; Frank et al., 1975; Kuifler et al., 1977, 1979; Dennis and Yip, 1978; L ~ n o and Slater, 1978, 1980a,b; Weinberg et al., 1981; and others). The preimplanted foreign nerve can form functional junctions almost as soon as postsynaptic effects of denervation appear, such as increased ACh sensitivity and TTX-resistant spikes (see reviews by Gutmann, 1976; Harris, 1974; Purves, 1976b; Fambrough, 1979; Lomo and Jansen, 1980). Several treatments that simulate denervation in causing neuromuscular inactivity also remove the barrier to innervation. Fibers accept innervation at sites of damage (Miledi, 1963) or when muscle activity is stopped by local anesthetic block of nerve conduction (Jansen et al., 1973), block of release by BoTX (Hoffman, et al., 1964; Fex et al., 1966), or Naja-toxin postsynaptic block (Duchen et al., 1975). The ability of a muscle fiber to be innervated can be generally correlated with the presence of high levels of ACh sensitivity on the membrane. Denervated or damaged muscle develops extrajunctional ACh sensitivity, and can be innervated ectopically. Moreover, direct muscle stimulation, which eliminates extrajunctional ACh sensitivity, also blocks reinnervation except at the old endplate sites (Jansen et al., 1973; Frank et al., 1975). These observations are consistent with the hypothesis that synapses form selectively where there are high concentrations of ACh receptors (Katz and Miledi, 1964; Fex et al., 1966). This possibility has not been eliminated; but the portion of the receptor molecule that binds ~-BTX does not appear to be involved, since the toxin does not block reinnervation in the diaphragm of chronically anesthetized rats (Jansen and Van Essen, 1975). Although denervated fibers develop many of the properties of uninnervated embryonic fibers, they continue to differ in at least one important respect: they retain the old postsynaptic site, with junctional folds and concentrations of ACh receptor and acetylcholine esterase (ACHE) that persist for several months (Miledi, 1960; Miledi and Slater, 1968; McArdle and Albuquerque, 1973; Frank et al., 1975). (Newly developed endplate sites in young animals show much greater lability than do adult endplates [Brown et al., 1976; Jansen et al., 1976]). This appears to be a significant difference, for regenerating fibers of either the native or a foreign nerve show a strong tendency to reinnervate old endplate sites, rather than form new synapses ectopically (in mammals: Gutmann and Young, 1944; Saito and Zachs, 1969; Watson, 1970; Bennett, McLachlan and Taylor, 1973a; Bennett and Pettigrew, 1974b; in birds: Hnik et al., 1967; Bennett, Pettigrew and Taylor, 1973; in amphibians: Miledi, 1960; Letinsky et al., 1976). Old endplates can attract regenerating nerves for several months, even when the muscle fibers are kept active by direct muscle stimulation (Frank et al., 1975; Brown and Ironton, 1978). The tendency for motor axons to reinnervate old endplate sites appears to be greater in fast muscles than in muscles having a significant fraction of slow fibers (Harris, 1974). Very few ectopic synapses form in the diaphragm and sternomastoid, for example, (Shukla and Aitken, 1963; Bennett, McLachlan and Taylor, 1973; Watson, 1970), while ectopic synapses can easily be induced in the soleus and tibialis anterior (Fex and Jirmanova, 1969; Gwynn and Aitken, 1966; Saito and Zachs, 1969). Ectopic synapse formation is more likely when the sectioned original nerve is transplanted at some distance from the old nerve sheath and endplate area (Saito and Zachs, 1969), or when a foreign nerve is implanted far from the endplate zone (Jansen et al., 1973; Frank et al., 1975), although many axons still find their way back to old endplates (Bennett, McLachlan and Taylor, 1973a). In amphibians, the selectivity of reinnervation of old endplate sites appears to be especially strong. If a frog CP nerve is sectioned and transplanted to a point in the same muscle far from the distal nerve stump and endplate zone, virtually all of the axons still find their way back selectively to the old endplates (A. Jitsumyo and Grinnell, unpublished). The rare ectopic synapses found are located at the site of transplantation where it is likely that muscle fibers were damaged. 254 ALAN D. GRINNELL AND ALBERT A. HERRERA The remarkable precision of reinnervation of old endplate sites by regenerating nerves in frogs was well documented by Letinsky et al. (1976), in a correlated light and electron microscopic analysis. Subsequently, in a fascinating series of studies, McMahan+ Sanes, and their collaborators have been analyzing the factors responsible for this selective reinnervation of old sites. In these experiments they carefully damaged and denervated the frog CP muscle, causing degeneration and phagocytosis of muscle fibers. The muscle was then X-irradiated to prevent muscle fiber regeneration. Under these conditions the basement membrane that previously surrounded each muscle fiber remained intact even when the muscle fibers were gone. Moreover, the basal lamina which normally extends into the synaptic clefts still retained the conformation of junctional folds. Most interestingly, regenerating axons selectively reinnervated these basement membrane sheaths at the original synaptic sites, and differentiated into morphologically identifiable nerve terminals (Marshall, et al., 1977; Sanes, et al., 1978). Thus, in the absence of postsynaptic cells, cues in the basement membrane are sufficient to specify the site of reinnervation. Although AChE is a major component of the basal lamina at the site of junctional folds (McMahan, et al., 1978), irreversible block of AChE activity with 10 mM diisopropylfluorophosphate (DFP) did not affect the selectivity of reinnervation (Marshall, et al., 1977; see also Filogamo and Gabella, 1966). (This of course does not rule out the possibility that a different part of the molecule than that affected by DFP is the critical component.) Moreover, it has recently been discovered that the basement membrane at the site of the synapse contains a unique antigenic site not associated with AChE (Sanes and Hall, 1979). In the same preparation, it was also recently shown that if muscle fibers are permitted to regenerate while the nerve is prevented from reinnervating, newly formed muscle fibers accumulate ACh receptors precisely in register with the basal lamina junctional folds surviving from the original synaptic sites (Burden, et al., 19791. However, the basement membrane alone may not be entirely sufficient for selective "reinnervation" of old gutters. If the Schwann cells are killed by freezing and regeneration of muscle fibers is prevented, very few regenerating axons find their way back to old sites (D. P. Kuffler, D. Edington, U. J. McMahan, personal communication). Thus some living cellular element, either muscle or Schwann cell, appears to be necessary in addition to the basal lamina to assure that regenerating axons selectively reinnervate the sites of old endplates. 5.2. COMPETITIVE INTERACTIONS AND SPECIFICITY OF REINNERVATION The ability of damaged nerves to specifically regenerate their appropriate connections is a question of great medical interest and has consequently been the subject of much study. There is a general conception that specific reinnervation does not ocur in adult mammals, since if a nerve trunk is sectioned and resutured, regenerating motor axons seem to randomly reinnervate both appropriate and inappropriate muscles (see Sperry, 1945; Guth, 1956; Bernstein and Guth, 1961; Miledi and Stefani, 1969; Fambrough, 1976; Purves, 1976b,c; Tada et al., 1979; Brushart and Mesulam, 1980). When the original nerve and a foreign nerve regenerate simultaneously into a muscle, each appears to be equally successful in forming functional synapses (Weiss and Hoag, 1946; but see Hoh, 1975 below). In contrast, when nerves to a frog or salamander limb are cut, regeneration restores normal function (Sperry, 1945). In early experiments on the reinnervation of normal and supernumerary limbs in lower vertebrates, Weiss (see reviews 1936, 1941, 1950) concluded that regenerating nerves innervated muscles randomly, but that each muscle then somehow influenced the MNs innervating it to reorganize their central connections to the configuration appropriate for that muscle tmyotypic specificationl. Subsequent work, initially in Sperry's lab, indicated that this was not the case. Normal function was restored by the regeneration of motor axons to their appropriate muscles--a true form of specificity (see belowl. Although motor axons can regenerate specifically ]n some systems, it is clear that any implanted cholinergic nerve can innervate any muscle if the nerve has no alternative target SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 255 and the muscle has no alternative source of innervation. For example, nerves can be successfully crossed between antagonists or muscles of different fiber type. Since the time of Langley it has been known (Langley and Anderson, 1904) that even preganglionic autonomic nerves can innervate skeletal muscle (see also Guth and Frank, 1959; Landmesser, 1971, 1972; Bennett, McLachlan, and Taylor, 1973b; Grinnell and Rheuben, 1979; Gordon, et al., 1980; Breitschmid and Brenner, 1981). However, in most situations, regenerating axons have the opportunity to innervate any of several muscles, including the correct one, and muscles are in a position to receive inputs from several different regenerating muscle nerves, one of which may be appropriate. In these cases, both axons and muscle fibers could in principle participate in the choice of which connections are formed and maintained. Today, with improved techniques of visualizing terminals, tracing pathways, and analyzing synaptic events, the study of adult muscle reinnervation continues to be a lively and fascinating field. 5.2.1. Specific reinnervation of different muscle fiber types 5.2.1.1. Mammals Miledi and Stefani (1969) saw no evidence for specific reinnervation when they sectioned a rat sciatic nerve and allowed it to regenerate. The soleus, a slow-twitch muscle, was reinnervated by axons that had previously innervated fast-twitch muscle, presumably because these axons regenerated more rapidly. Under the influence of these axons (or the activity pattern they imposed) the soleus acquired fast-twitch characteristics. The fast innervation remained indefinitely. However, in a more direct test for specific reinnervation of fast and slow twitch muscles in rats, Hoh (1975) sutured both fast and slow nerves (from the extensor digitorium longus I-EDL] and the soleus) into the distal nerve stump of one or the other of these muscles. Under these conditions the EDL became innervated selectively by the EDL nerve, the soleus almost equally well by both nerves. Even when the soleus nerve alone was implanted into the EDL, it formed junctions slowly and with incomplete success (Hoh, 1975; see also Fex and Jirmanovfi, 1969). Thus there may be some specificity in the reinnervation of different fiber types. Moreover, Brown and Butler (1976) reported selective reinnervation of intrafusal muscle fibers by the correct fusimotor axons in adult cats (but see Takano, 1976; Brushart and Mesulam, 1980 for evidence of poor reinnervation by gamma fibers when they were injured far from their points of termination). 5.2.1.2. Birds In birds, selectivity in the reinnervation of fast and slow-twitch muscle is also seen, and has been analyzed in great detail. Muscle fibers in the fast posterior latissimus dorsi (PLD) muscle of the chicken are innervated at one site while fibers of the slower anterior latissimus dorsi (ALD) are diffusely innervated. When their nerves were crossed, each innervated the endplate sites of the inappropriate muscle, but with unequal success (Feng et al., 1965). The PLD nerve caused good contraction of the ALD muscle, but not vice versa. Further examination of this preparation showed that the PLD axons formed terminals at each of the distributed old endplate sites on the ALD fibers, while the ALD nerve was restricted to the single endplate site of the orignal axon (Hnik et al., 1967; Zelen~i et al., 1967; Bennett, Pettigrew and Taylor, 1973). The ineffectiveness of the ALD nerve in driving PLD fibers can be attributed to the low quantal content of release at the single endplate (Vysko6il, Vyklik~, and Huston, 1971). 5.2.1.3. Lower vertebrates In lower vertebrates, many examples of selective reinnervation of muscles of similar fiber type have been reported, especially in urodele amphibians (see below). Specificity in the reinnervation of frog slow muscle (e.g. the iliofibularis or pyriformis) is striking, and 256 ALAN D. GRINNELL AND ALBERT A. HERRERA has been extensively investigated. Frog slow muscle fibers differ from fast fibers in contraction speed, pattern of innervation (diffuse), inability to produce action potentials, graded contractile reponse to depolarization, and prolonged contractures when exposed to high concentrations of K ÷ or ACh (Kuffler and Vaughn-Williams, 1953). When slow fibers are denervated, they lose their K+-contracture response (Elul, et al., 1970) and develop TTX-resistant spikes (Miledi, et al., 1971). After crushing the nerve to a muscle containing both fast and slow fibers all fibers are initially reinnervated by fast axons, due to their faster rate of regeneration. The slow fibers retain the spikes and altered contraction properties they acquired when denervated. Slow motor axons begin to reinnervate the slow fibers about a month later. Even before slow axon-elicited contraction is seen, slow fibers begin to lose their spikes and regain K ÷ contractures (Elul et al., 1970; Schmidt and Stefani, 1976, 1977). Most interestingly, when the slow motor axons reinnervated a mixed muscle (the pyriformis), they selectively reinnervated slow fibers. Few if any slow nerve synapses were found on fast-twitch muscle fibers. Initially, some slow fibers were innervated by both fast and slow axons but the fast axons subsequently lost their e~ctiveness (SchmidtandStefani, 19~/6). Although the mechanism is not known, this displacement appears to represent evidence for specific reinnervation by slow axons, and successful competitive removal of inappropriate fast axons. Vrbov~i et al. (1978) have postulated that the fast nerves may form so many terminals in attempting to innervate all of the distributed endplate sites on the slow fibers that each terminal is relatively weak and unable to compete with a regenerating slow axon process (see also Section 5.2.2.3.). How this would explain the total displacement of the fast axons is not clear. It would be of interest to determine whether such displacement occurred if the normal targets for the fast axons (fast twitch fibers) were removed prior to regeneration. 5.2.2. Reinnervation of muscles of similar fiber type Truly specific reinnervation of fast-twitch muscle has not been shown in mammals or birds, although the axons of different nerves can compete for effectiveness (see below). However, in fish and amphibians, particularly urodele amphibians, there have been several reports that regenerating fast axons can selectively reestablish connections with appropriate muscles. These experiments, their attempted verification, and determinations of the mechanisms involved have generated a literature filled with lively controversy. It has become obvious that one must distinguish between cases where regenerating axons have an equal chance to reinnervate a given target and cases where foreign and original nerves do not compete under equal conditions, e.g. where sprouted collaterals are displaced by a regenerating nerve. In the latter situation nonspecific competitive interactions can give the appearance of specificity. 5.2.2.1. Where correct and foreign nerves have equal opportunity Beginning with the work of Sperry and his colleagues in the mid 1960's, it became clear that myotypic specification was not an adequate explanation for the recovery of normal function following regeneration; it could be explained better as specific reinnervation of muscles by their appropriate nerves. Sperry and Arora (1965) introduced this interpretation when they showed that the main oculomotor nerve trunk of cichlid fish would regenerate to restore coordinated eye movements, but if the nerves to different extraocular muscles were surgically crossed, weak misdirected reflexes resulted. Interestingly, in a few cases, misdirected reflexes gave way in time to correct reflexes. In these cases, some axons were found to have grown past the incorrect muscle to their normal target muscle. This, by itself, is not proof of specificity, since both muscles were denervated and regenerating axons might have accidentally encountered the correct muscle. Specificity would be indicated if it could be shown that correct axons have a selective advantage in reinnervating or maintaining synapses in their appropriate muscle. Mark and his colleagues, in further study of the same preparation, purported to show this, although their findings were somewhat controversial (Marotte and Mark, 1970a, b; Mark and SPECIFICITY AND PLASTICITYOF NEUROMUSCULAR CONNECTIONS 257 Marotte, 1972; Mark, et al., 1972). They denervated the superior oblique (SO) muscle by sectioning its nerve near the spinal cord, and removed the inferior oblique (I0) muscle. At first, the regenerating I0 nerve reinnervated the SO muscle, producing reversed eye rotation reflexes. Later, when the SO nerve regenerated to the periphery, normal reflexes returned abruptly (over a period of about two days). This implies that the appropriate (SO) nerve was able to reinnervate the SO muscle and rapidly suppress the effectiveness of the I0 nerve. The foreign (I0) nerve continued to show electrical activity, but, according to Mark and Marotte (1972), was not effective in driving a significant number of muscle fibers above threshold. Moreover, 5-7 days after sectioning the appropriate SO nerve in such a doubly innervated muscle, synapses with normal ultrastructure were still present. These presumably belonged to the ineffective I0 nerve (Marotte and Mark, 1970b; Mark, et al., 1972). These authors proposed, therefore, that inappropriate axons could be "repressed" by appropriate ones, while remaining morphologically unchanged. Not surprisingly, this remarkable hypothesis was soon subjected to other tests. Scott (1975, 1977) used the fish extraocular muscle preparation, but studied it with intracellular electrodes. She found that individual muscle fibers could be doubly innervated by foreign and correct axons and that both remained functional indefinitely. She concluded that the correct nerve did not repress or displace the foreign nerve. However, endplate potentials (EPPs) at either foreign or native nerve synapses were not as large as those seen when either innervated the muscle alone. Hence, these experiments did, in fact, support the existence of some competitive interaction between the two nerves, but not specificity. Moreover, muscle tissue sometimes regenerated when an extraocular muscle was removed (Scott, 1977). If this regenerated muscle became innervated by branches of its appropriate axons, it would produce contractions that, in behavioral tests, would suggest return of normal reflexes. Another report of specific reinnervation of fish muscle is deserving of further study. Mark (1965) found that crossing the retractor and protractor nerves in a fish fin led to permanently reversed fin movement. However, if both nerves were sectioned near the spinal cord, regeneration restored normal coordinated movement, implying specific regrowth of axons to their appropriate target muscles. Frank and Jansen (1976), on the other hand, reported that fish gill muscle innervated by a foreign fin nerve could also be reinnervated by the native vagus nerve and that dual innervation was maintained indefinitely. Since the classic work of Detwiler (1933) and Weiss (1936, 1950), urodele amphibians have been favorite subjects for studying the specificity of reinnervation by motor nerves. The phenomenon is clearly illustrated by the experiments of Grimm (1971), who crossed flexor and extensor nerves in the axolotl forelimb. The crossed nerves grew into the muscles toward which they were directed, and crossed reflexes resulted initially. However, with time this changed, and the muscles contracted as if they had been reinnervated by the correct nerve. Anatomical examination, electromyographic recording, and nerve stimulation revealed that the nerves sprouted tiny branches which reinnervated their normal target muscles. Within each muscle, it was these few axons that were effective rather than the majority of axons which were from the inappropriate nerve. Cass and Mark (1975), in a similar study of the axolotl hind limb, crossed flexor and extensor nerve trunks and found correctly coordinated movement and accurate reinnervation of different muscle groups by each segemental nerve. Perhaps the best evidence for specificity in the competition between native and foreign nerves in urodeles comes comes from the experiments of Dennis and Yip (1978; see also Yip and Dennis, 1976). They excised a newt forelimb flexor muscle and implanted its nerve onto the surface of an extensor, followed two weeks later by denervation of the extensor. Foreign nerve synapses appeared about two weeks after section of the normal nerve. Within one month after native nerve section, the foreign axons had innervated virtually all muscle fibers as effectively as in the unoperated muscle. Nevertheless, the native nerve subsequently reinnervated the muscle and formed synapses that returned rapidly to normal suprathreshold strength. The quantal content of foreign nerve synapses 258 ALAN D. GRINNELL AND ALBERT A. HERRERA simultaneously decreased until transmission failed at nearly all synapses. Dennis and Yip (1978) used the horseradish peroxidase (HRP) uptake technique (Heuser and Reese, 1973) to show that in muscles where foreign terminals had become ineffective, stimulation of the correct nerve labeled vesicles in the vast majority (94~o) of terminals. They concluded that the foreign terminals were physically displaced. While it is possible that the excised flexor muscle may have partially regenerated, and become reinnervated by some of its displaced axons, these experiments provide strong evidence for true specificity of neuromuscular connections. The correct nerve terminals may have a competitive advantage peripherally, or neuronal interactions in the spinal cord may decrease the ability of the foreign MNs to maintain synapses in the inappropriate muscle. Bennett, et al. (1979) reported very similar findings with essentially the same preparation, except the flexor mucle was not excised and the flexor nerve was not preimplanted. Under these conditions, foreign nerve terminals required 8 weeks to reach the same size and level of transmitter release as normal extensor nerve terminals. An unfortunate disadvantage of these salamander preparations is that both native and foreign nerves form distributed junctions in close proximity along the length of muscle fibers, making it difficult to correlate structure and function for any given terminal. Another approach to assessing the specificity of reinnervation was employed by Stephenson (1979), who simultaneously grafted a supernumerary forelimb into the shoulder region of young axolotls and cut the 4th or 5th spinal nerve. Many of the cut axons innervated the extra limb, which then moved homologously with the normal limb. The rare axons which branched and innervated muscles in both limbs could be demonstrated by cutting all dorsal roots and stimulating individual muscle nerves in one limb while looking for muscle contraction in the other limb. The majority of such axon reflexes encountered (20/26) were between the same or synergistic muscles in the two limbs, while only 5 out of 26 were found in muscles of clearly unrelated function. Selectivity was not perfect, but some specificity is strongly indicated. 5.2.2.2. Competition between two foreign nerves in frogs Competitive suppression of synaptic effectiveness can also occur in case where neither nerve is appropriate. Grinnell et al. (1977, 1979) transplanted the frog sartorius muscle to the lymph sac of the back and implanted one or two foreign nerves (spinal nerves 4 and 5) into it. Motor axons in either nerve alone could innervate virtually the whole muscle, with mostly suprathreshold EPPs. In doubly innervated muscles, each nerve innervated most muscle fibers, but on any given fiber usually only one input was suprathreshold while the other showed very low quantal content EPPs and low frequency mEPPs. Even if one nerve was implanted 60-90 days earlier than the other, a second nerve was still able to innervate most fibers, acquire "control" of many of them, and reduce the effectivetiveness of many of the earlier nerve's synapses below threshold. Although neither nerve appeared to be innately preferred by the muscle, either spinal nerve had a clear-cut competitive advantage over a symapthetic preganglionic nerve (Grinnell and Rheuben, 1979; Grinnell, et al., 1979). The preganglionic splanchnic nerve formed almost no suprathreshold junctions when implanted at the same time as a spinal nerve. However, when implanted 60-90 days before a spinal nerve, it innervated much of the muscle, only to be totally displaced as the somatic motor axons grew in. 5.2.2.3. Displacement of sprouted collaterals A number of other studies, involving displacement of foreign nerve collateral sprouts by regenerating correct nerves, were originally viewed as demonstrating specific reinnervation. Recently, an alternative explanation has been advanced. Most of these findings could be explained if sprouts of motor axons which maintained full peripheral fields elsewhere were at a competitive disadvantage compared to regenerating axons which initially have fewer terminals to support (Brown et al., 1976; Thompson and Jansen, 1977; Thompson, 1978; Jansen, et al., 1978; Purves and Lichtman, 1978; Wigston, 1979b, SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 259 1980; Haimann et al., 1981a,b). One of the most dramatic of these studies was that of Cass, et al. (1973) who observed that when one of the spinal nerve innervating the axolotl hindlimb was sectioned, adjacent segmental nerves sprouted collaterals that innervated the entire denervated territory within approximately one month. As the sectioned nerve regenerated, it reestablished functional synapses throughout its original territory, and the sprouted collaterals lost their influence. Most interestingly, if the correct nerve was resectioned several months later, the adjacent nerves reportedly established strong synapses throughout the denervated area within 3-4 days. Cass et al. (1973) speculated that the sprouted terminals remained in place but were made silent by some mechanism of "repression". These exciting findings have been confirmed in some respects, but not in others, by studies using the same axolotl hind limb preparation (Bennett and Raftos, 1977; Slack, 1978), or a doubly innervated salamander shoulder muscle, the supracoracoideus (Genat and Mark, 1977; Harris et al., 1977; Wigston, 1979a,b, 1980). In most of these studies the size and strength of the synapses formed by sprouted collaterals were not analyzed, although in one case they reached a mean maximum value of about 70% of the normal synaptic strength (Bennett and Raftos, 1977). In the supracoracoideus muscle preparation, sprouts succeeded in innervating only a portion of the denervated muscle (Genat and Mark, 1977; Wigston, 1980). Further evidence that these experiments illustrate the competitive disadvantage of sprouted terminals, and not specific recognition, was provided by Wigston (1980). He showed that a foreign nerve implanted into a muscle partly innervated by sprouted collaterals could displace the sprouts as effectively as a regenerating original nerve. The properties and stability of sprouted terminals have been further examined in two Xenopus preparations. Fangbonner and Vanable (1974) found that the denervated superior oblique muscle was reinnervated by collateral sprouts from the oculomotor nerve. When the correct nerve (the trochlear) regenerated, the sprouts were partly, but usually not entirely, displaced. If regeneration of the trochlear nerve was delayed, so was the displacement of the foreign nerve sprouts, and the degree of displacement was correlated with the number of correct axons reinnervating the muscle. More directly comparable to the axolotl work are studies of the doubly innervated Xenopus pectoral muscle by Mallart and his colleagues (Haimann et al., 1976, 1981a,b). Approximately one half of this muscle is innervated by an anterior nerve, the other half by a posterior nerve, with an area of overlap. If the posterior nerve is cut, anterior axons sprout to innervate much, but not all, of the denervated territory. The innervation of this territory by anterior nerve sprouts was less rapid and less complete than innervation by regenerating posterior nerve axons (Haimann et al., 1981b). Bennett and Raftos (1977) also observed that synapse formation by sprouts is slower than that by a regenerating nerve; but both occur more quickly in salamanders than in Xenopus. In Xenopus, as in salamanders, the EPPs produced by the sprouted terminals had lower quantal content than both normal anterior nerve junctions and junctions formed by the posterior nerve. It is noteworthy that the regenerating posterior nerve reinnervated only its original area, apparently not invading the territory of the anterior nerve. Thus sprouted terminals are not only easily displaced, but also appear to be less effective than the original terminals in excluding synapse formation by the regenerating nerve. It would be of interest to determine more exactly the differences between normal and sprouted synapses formed by a given axon: do sprouts only partially fill synaptic gutters, allowing a regenerating axon access to the synaptic site? Do sprouted terminals release less transmitter per unit area of synaptic contact? Does sprouting change the properties of an axon's intact original terminals? 5.2.2.4. Mechanisms of competitive displacement in lower vertebrates Mark, in his early descriptions of synaptic repression, concluded that foreign nerve terminals that had been functionally replaced by regenerating nerve terminals retained their normal ultrastructure (Marotte and Mark, 1970b; Mark, et al., 1972) and could be 260 ALAN D. GRINNELL AND ALBERT A. HERRERA reactivated (derepressed)within 3-7 days after recutting the normal nerve (Cass et al., 1973; Genat and Mark, 1977). The morphological argument was based on the presence of apparently normal synapses in extraocular muscles several days after cutting the correct nerve a second time. However, since Scott (1975, 1977) found that foreign nerve synapses were not repressed in this preparation, and since Mark and his coworkers could not have detected subthreshold synapses with their behavioral tests, it is probable that Mark's foreign synapses were at least partly functional and not completely repressed. Moreover, Dennis and Yip (1978) found no evidence for rapid depression (< 20 days), and Wigston (1979a), in a more extensive study of the same preparation used by Genat and Mark (1977), found that only a few foreign nerve synapses had reestablished function 10 days after cutting the correct nerve. In addition, HRP labeling experiments have failed to demonstrate any foreign nerve terminals at endplates in muscles where the regenerating correct nerve had completely suppressed EPPs from the foreign nerve (Dennis and Yip, 1978; Bennett, et al., 1979). One study of "repressed" terminals in axolotl muscle has suggested that the mechanism of transmitter release is affected, independent of terminal size (Harris, et al., 1977). Using the supracoracoideus muscle of Ambystoma, these authors examined mEPPs which were presumed to arise at sprouted foreign nerve synapses which were repressed, since stimulation of the foreign nerve evoked no EPPs. These mEPPs differed from those at normal, developing, or degenerating terminals in that they showed no increase in frequency in the presence of La 3+ or increased K +. They differed from Schwann cell mEPPs (Birks, et al., 1960) by showing increased frequency in hypertonic solutions, and disappearing on section of the posterior nerve, These results led Harris et al. (1977) to conclude that repressed terminals lost their Ca 2 ÷-dependent release mechanism. Unfortunately, these interesting results were not confirmed by Wigston (1979b), who continued studying the same preparation. He found that the foreign nerve terminals were sensitive to increased K + concentration, even when they were supposedly repressed. Moreover, he suggested that the absence of evoked release in the earlier series of experiments was probably due to nerve damage during dissection. These observations, coupled with Wigston's (1979b) documentation of normal Ca 2+ dependent release in the sprouted "repressed" terminals, leads us to conclude there is not yet any convincing evidence that repression results in "silent" synapses with normal ultrastructure. Present evidence favors the view that foreign nerve terminals are physically removed from the junctional area in lower vertebrates. This would explain the findings that, upon regeneration of the original nerve, foreign nerve terminals show a drop in EPP quantal content and mEPP frequency, with no obvious change in quantal size or postsynaptic characteristics (Yip and Dennis, 1976; Bennett and Raftos, 1977; Harris et al., 1977; Dennis and Yip, 1978; Grinnell et al., 1977, 1979; Wigston, 1980). On the other hand, the finding that transmitter release from frog motor nerve terminals can be altered by up to 5-8 fold by experimental manipulations such as change in motor unit size or contralateral denervation, without apparent change in terminal size (Herrera and Grinnell, 1980b, 1981), suggests that physiological mechanisms of synaptic suppression may exist and be important. Of course, proposing that repression occurs by withdrawal of foreign nerve terminals does not explain why the terminals retract. The mechanism of competitive interaction is still quite unknown. It seems necessary to postulate that the size and effectiveness of a terminal is dependent on a balance of at least three factors: (1) the amounts of critical materials supplied by the soma, levels of which will depend on the physiological state of the neuron, its axonal transport, and the number of terminals it is maintaining. This self-evident view is supported by inherent differences in the efficacy of terminals of different motor units in the same muscle (Grinneli and Herrera, 1980b), observations of the low efficacy of sprouts (Thompson and Jansen, 1977; Bennett and Raftos, 1977; Haimann et al., 1981a,b), and changes in transmitter release efficacy brought about by changes in motor unit size (Herrera and Grinnell, 1980b); (2) One or more growth (or sprout) promoting substances, produced by inactive muscle and/or denervating nerve SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 261 products. This is needed to explain sprouting in paralyzed and partially denervated muscles (see Section 4.1.5.1.); and (3) an inhibitor)' substance produced by a muscle fiber within a significant distance of an active endplate, which mediates the competition observed between terminals on the same fiber. In lower vertebrates, some additional mechanism must exist to permit selective removal of inappropriate terminals and maintenance of correct terminals. Vrbovfi and her colleagues (1978) have suggested that both synapse elimination and competitive displacement are the result of synaptic activity-induced proteolytic digestion of weak terminals near strong ones, and that sprouting takes place normally in the absence of synaptic activity. However, for the reasons outlined above (Section 2.4.6.), it seems unlikely that this can be a complete explanation of either synapse elimination or competitive displacement. One potential source of variability in the results of these competition experiments may be the length of time allowed for the foreign nerve to consolidate its innervation before facing competition. With time, foreign nerve terminals seem to become more resistant to displacement. Bennett et al. (1979) found that foreign innervation of a salamander muscle was completely eliminated by the original nerve if the original nerve regenerated within 6 weeks, but not if reinnervation was delayed for 10 weeks or more. Similarly, Slack (1978) showed that terminals formed by collateral sprouts of foreign nerves in axolotl limbs were displaced by the regenerating normal nerve if it returned within 2 weeks, but not if regeneration was delayed 9-10 weeks by multiple crushes and maintenance at 10°C. It is not known why foreign nerve terminals are more resistant to displacement after longer times. The ability to withstand displacement may depend on filling available synaptic gutters completely and thus depriving the regenerating nerve of synaptic space. Wigston (1979b) found that at a time when the normal nerve was able to successfully suppress foreign synapses, there were many vacant synaptic gutters at foreign-innervated endplates. However, stable long-term foreign synapses did not prevent successful reinnervation by the correct nerve at the same or nearby sites in the experiments of Bennett et al. (1979). Since there is a strong tendency for reinnervating cholinergic nerves to terminate at the sites of old junctions (see above), it is likely that competing terminals will often find themselves in the same endplate region. Do foreign and correct nerves need to be in close proximity to compete, perhaps even within the same synaptic gutter? This question is difficult to answer using the salamander preparation, since endplates are distributed along the entire length of each muscle fiber, making it difficult to distinguish which axon supplies which terminal. Dennis and Yip (1978) frequently saw, in the electron microscope, what were probably correct and foreign terminals in close apposition in the same cross-section, but did not venture a positive identification. Bennett, et al. (1979), on the basis of EPP rise times, concluded that weak foreign nerve junctions were located very close ( < 1 mm) to sites of native nerve terminals, while strong foreign junctions were not close to competing terminals. In other preparations, it is clear that competitive interaction can occur between well separated terminals. Grinnell et al. (1977, 1979) observed that mutual suppression between foreign nerves in transplanted frog sartorius muscles could occur over distances of several mm. Wigston (1979a) found that 92~o of the foreign nerve terminals he studied during the period of sprout suppression were more than 120/~m from regenerated native terminals. Similar findings were reported by Haimann et al. (1981b). These findings suggest that competitive interaction may be mediated via the muscle fiber, a conclusion supported by similar findings in mammalian muscle (see below). Some such interaction might explain the finding that in muscles whose fibers are normally multiply innervated, endplates tend to be distributed along the length of each fiber, with a minimum distance between junctions, characteristic of each muscle (Bennett and Pettigrew, 1974b). In muscles that are normally multiply innervated, such as the axolotl supracoracoideus or the X e n o p u s pectoralis, ongoing competition may result in synaptic remodelling throughout the life of the animal. Haimann et al. (1981a) found that in the region of the X e n o p u s pectoralis muscle where anterior and posterior nerve territories overlapped (about 20~o of the muscle), many fibers received inputs from both nerves. When both 262 ALAN D. GRINNELL AND ALBERT A. HERRERA axons innervated the same synaptic site, or even distant sites, the summed EPPs from both inputs were smaller than those of synapses on singly innervated fibers. Silver-stains revealed signs of sprouting and regression, supporting the view that the innervation of the region of overlap is in a state of dynamic equilibrium. Transmitter release at nerve terminals in the multiply innervated sartorius muscle of Rana pipiens is lower than release from terminals in the singly innervated cutaneous pectoris muscle (Grinnell and Herrera, 1980a). This reduction in synaptic effectiveness may be due to competition between the multiple junctions, or due to the larger numbers of terminals each sartorius axon must maintain, with present evidence suggesting that the latter effect is predominant (Herrera and Grinnell, 1980a). 5.2.2.5. Competitive interaction between synapses on adult mammalian muscle fibers Normally, adult mammalian muscle will not accept further innervation (see above, Section 5.2.1.1.). However, under special circumstances mammalian muscle fibers can become multiply innervated. In many cases competitive suppression can eliminate the extra inputs. It remains to be determined whether this suppression involves a process similar to that whereby polyneuronal innervation is eliminated in neonatal muscle (see Section 2.4.). No specificity is indicated in most cases, but the competitive interaction may be very similar to that seen in lower vertebrates. Following nerve crush or section, for example, many endplates become polyneuronaily innervated by regenerating axons. All but one of the multiple inputs is lost by a process of synapse elimination comparable to that seen in neonatal animals, but which requires a much longer time (McArdle, 1975; Jansen and Van Essen, 1975; Benoit and Changeux, 1978). Blocking ACh receptors with ct-BuTX during the process of regeneration does not prevent synapse formation or polyneuronal innervation (Jansen and Van Essen, 1975: but see Pestronk and Drachman, 1978). As in neonatal muscles, blocking neuromuscular activity by applying local anesthetic to the sciatic nerve may result in an increased amount of polyneuronal innervation, followed by gradual synapse elimination (Benoit and Changeux, 1978). However, Betz et al. (1980b) found no evidence for polyneuronal innervation following TTX block in a rat foot muscle. As in amphibians, if a mammalian muscle is partially denervated, remaining intact axons sprout to innervate most or all of the denervated fibers (Hoffman, 1951; Edds, 1953; Guth, 1962; Thompson and Jansen, 1977; see also Section 2.4.4.). Interestingly, sprouts form synapses more rapidly and on a higher proportion of fibers in smaller muscles or muscles with smaller average motor unit size (Brown and Ironton, 1978). Both sprouts and regenerating axons appear to innervate previous synaptic sites selectively. In some cases, double innervation persists. However, in many others, sprouted terminals lose their effectiveness (Brown and Ironton, 1978; Thompson, 1978; Hopkins, et al., 1980; Betz, personal communication in Brown and Ironton, 1978), showing EPPs of progressively smaller mean quantal content (Hopkins et al., 1980). As before, it is not clear whether the selective loss of sprouts indicates that the regenerating axons are specifically preferred, or simply that terminals in enlarged motor units are more vulnerable. As in the case of lower vertebrates (c.f. Section 5.2.2.3.), sprouted terminals become more resistant to displacement with time (Thompson, 1978; see also Brown and Ironton, 1978). The most economical explanation for these results is that regenerating axons can reinnervate old endplate sites whenever vacant synaptic space is available and that in a high percentage of cases, regenerating terminals can outcompete overextended sprouts. Other experiments, however, show that competitive displacement of synapses can occur well beyond the confines of a given endplate. When a foreign nerve is carefully implanted onto a muscle which is subsequently functionally denervated by botulinum toxin (Fex et al., 1966), nerve block with local anesthetics (Jansen et aL, 1973), postsynaptic block with Naja-neurotoxin (Duchen et al., 1975), or native nerve section (Fex and Thesleff, 1967; Gutmann and Hanzlikov~i, 1967; Fex and Jirmanov~i, 1969; Frank et al., 1975; Kuttter et al., 1979), the foreign axons form synapses ectopically. In most cases reported, only a fraction of the muscle became innervated by the foreign nerve. If the SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 263 original nerve was crushed so that regeneration was rapid, it reinnervated essentially all the old endplates, with many fibers becoming dually innervated. If the original nerve was sectioned so that regeneration was delayed, it could reinnervate few of the fibers innervated by the foreign nerve (Frank et al., 1975). In these cases, there is apparently some change in the muscle fiber making it refractory to innervation, even when the original postsynaptic site survived, with high ACh sensitivity, at significant distance. (The fact that these old sites survive for several months, but do not accept reinnervation by the regenerating original nerve, provides additional evidence that sites of high ACh sensitivity are not a sufficient condition for acceptance of innervation [see Section 5.1.1). When foreign nerve synapses were located far from the original endplate region, they appeared to survive indefinitely, even after the original nerve had reinnervated the same fiber (Gutmann and Hanzlikov~, 1967; Frank et al., 1975; Kuffler et al., 1977, 1980). However, if the foreign nerve was implanted near the original endplates (Kuffler et al., 1977), or two foreign nerves were implanted together (Kuffler et al., 1980), ectopic endplates were found close to the old endplates, and to each other. Under these circumstances the period of polyneuronal innervation was transient. Each fiber was eventually innervated at only one site, leaving one or more nearby vacant endplate sites on many fibers. This competitive displacement can apparently operate over a distance as great as 3-4 mm (Kuffler et al., 1980). When two endplates remain innervated, they are often from the same axon, implying some protection of terminals from competition by other terminals of the same axon (Kuffler et al., 1980). 5.2.3. Summary Denervated vertebrate skeletal muscle can be reinnervated by any cholinergic axon, and all show a strong tendency to terminate selectively at the old endplate sites. In frog muscle, some of the cues for recognition of the old endplate site are known to be built into the basal lamina. Whether the same is true of other vertebrates has not been determined. In all vertebrates studied, there is at least some evidence for specificity in the reinnervation of different muscle fiber types, i.e. "slow" axons are usually more successful in innervating slow than fast muscle fibers. "Fast" axons readily innervate slow fibers, but, in frogs at least, can be displaced by regenerating "slow" axons. Truly specific reinnervation of fast-twitch muscle has been claimed only for amphibians. Many of the experiments purporting to show this have demonstrated the ability of an appropriate nerve to regenerate into a muscle, displacing the terminals of surrounding axons that have sprouted to innervate the denervated territory. Such examples cannot be taken to be proof of specificity, since a regenerating foreign nerve is equally successful in displacing the sprouts. It is probable that the sprouts, being part of an expanded axonal arborization, are unable to compete with regenerating axons. On the other hand, there are a few well documented cases in which a foreign nerve and a correct nerve are given equal opportunity to innervate the muscle and the correct nerve clearly prevails. Similarly, axon reflexes between a normal and supernumerary limb show that branches of the same axon tend to innervate synonymous or synergistic muscles in the same part of both limbs. Thus specificity in reinnervation does exist. Its mechanisms are unknown. Competitive displacement or suppression, however, is known to be a presynaptic phenomenon, marked by decrease in the amount of transmitter released from the sprouts or foreign nerve terminals, and their eventual physical disappearance. This interaction may be strongest when competing terminals are close together, but can occur over distances of at least a few mm. On the other hand, as sprouts, and perhaps foreign terminals as well, mature and fully occupy pre-existent synaptic gutters, they lose their competitive disadvantage. There is no evidence for specificity in reinnervation of mammalian muscles, but a form of competition is seen, comparable to that in lower vertebrates. Normally innervated and active fibers cannot be hyperinnervated. A denervated fiber can accept ectopic innervation, but will not do so if the muscle is kept active by direct stimulation. On the other 264 ALAN D. GRINNELL AND ALBERT A. HERRERA hand, the original nerve will regenerate to reinnervate its old endplates even when the muscle remains active. Following denervation, if an ectopic junction is formed on a muscle fiber, if gradually acquires the ability to prevent reinnervation of the old endplate by the regenerating nerve, even when the original endplate is far away. If the fiber becomes dually innervated, both synapses survive if they are far apart on the fiber, but one (usually the new ectopic endplate) is eliminated if they are within 3-4 mm of each other. Most interestingly, if the foreign nerve is cut or crushed in a muscle that has been hyperinnervated for several months, the foreign nerve will regenerate to reinnervate its previously established foreign endplates, despite the continued functional presence of the original nerve synapses (Frank et al., 1975). 6. Conclusions Neuromuscular synapses are not fixed, unchanging, all-or-none connections. Motoneurons (MNs) and their peripheral processes are influenced throughout development and adult life by a large variety of regulatory processes. Research has concentrated mainly on factors regulating the peripheral processes of MNs, largely because of the relative complexity and inaccessibility of the soma, dendrites, and their connections. Undoubtedly, factors affecting the central aspects of MNs are at least equally numerous. Although none of the mechanisms of regulation have been elucidated, many of the phenomena have been well described. A recurring theme is competition between MNs or their terminals for survival, synaptic space, and influence. In some cases the goal of the competition may be acquisition of one or more trophic substances analogous to nerve growth factor in the sympathetic nervous system. In fact, it seems unlikely that any one trophic substance could explain all of the regulatory phenomena observed. Success in this competition no doubt depends not only on trophic substances, but on a variety of other factors, such as the arrival time of a terminal at its target fiber, the site of contact, the number of other terminals each soma is trying to maintain, their relative activity level, and mysterious factors governing what is loosely termed specificity. Most aspects of MN behavior seem to involve as yet ill-defined interactions between MNs and their peripheral targets. It is perhaps instructive to list some of the aspects of MN behavior that appear to involve regulation by different features of their environment or the muscle fibers they innervate: (1) Guidance of axonal growth to and along appropriate pathways. (2) Induction of synapse formation and cessation of further growth upon contact with appropriate targets. (3) Maintenance of some MNs, not others, during the period of cell death. (4) Determination of the timing and rate of synapse elimination, as a function of activity levels and other factors. (5) Production of sprout-inducing factors upon denervation, partial or complete block of activity. (6) Rejection of hyperinnervation when already innervated. (7) Rejection of innervation by the inappropriate MN type, e.g. innervation of fast muscle fibers by slow axons. (8) Restriction of synapse formation during regeneration to old endplate sites (known to involve basal lamina components). (9) Regulation of MN electrical properties (overshoot, after-hyperpolarization). (10) Regulation of MN metabolism, synaptic input, and glial environment. (11) Maintenance of the dynamic equilibrium between terminal size, synaptic effectiveness, and muscle fiber size. (12) Mediation of competitive interaction when one junction suppresses or displaces another at a distance. The mechanisms of interaction that govern these phenomena are mostly unknown, but they probably involve response of a nerve terminal to contact with specific components SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 265 of the muscle fiber membrane or basal lamina, and to a variety of diffusible substances released by muscle fibers or Schwann cells and effective primarily in the immediate vicinity of their release. Signals may trigger a local response by a terminal, or the effect may involve retrograde transport of the substances to the soma where they govern metabolism and/or gene expression. It is possible that a given environmental signal or substance can act in different ways, but almost certainly many different factors are necessary for the whole spectrum of effects. It must be pointed out, however, that no regulatory substances governing M N properties have been identified to date, or even proven to exist. There are, in addition, other ways in which neuromuscular junctions can be regulated. For example, there is evidence from the effects of hormones, seasonal changes, and certain disease states that membranes at the neuromuscular junction, both pre- and post-synaptic, have receptors for chemical messengers other than ACh. A word of caution is needed. Many of the aspects of MN regulation and plasticity that we have reviewed were demonstrated by extreme surgical or pharmacological manipulations. Their importance in the regulation of normal MN function has yet to be established, since in normal animals, changes and responses to changes are likely to be more subtle. In the study of normal regulatory processes and plasticity the most rapid progress is likely to come from studies which simultaneously apply and correlate findings from several different disciplines (physiology, morphology, biochemistry, etc.) on single identified units. In this approach may lie our best chance of understanding how MN physiology, morphology, and connectivity are regulated, and how such plasticity enables the motor system to remain flexible and adaptive. Herein also lies the major advantage of using the neuromuscular junction as a model for plasticity and regulation of synaptic function in the central nervous system. Acknowledgements We gratefully acknowledge the help of Frances Knight, Robyn Cheeseman, Gretchen Wooden, and Dev Mishra in preparation of the manuscript, and the permission by several colleagues in the field to use their figures or unpublished findings. References AGUILAR, C. E., BISBY,M. A., COOPER E. and DIAMOND,J. (1973) Evidence that axoplasmic transport of trophic factors is involved in the regulation of peripheral nerve fields in salamanders. J. Physiol. 234, 449-464. AITKEN, J. T. (1950). Growth of nerve implants in voluntary muscle. J. Anat. 84, 38-49. ALBUQUERQUE, E. X.. DESHPANDE, S. S. and GUTH, L. (1978). Physiological properties of the innervated and denervated neuromuscular junction of hibernating and non-hibernating ground squirrels. Exp. Neurol. 62, 347-373. ALBUQUERQUE, E. X., RASH, J. E., MAYER R. F. and SATTERFIELD, J. R. (1976). An electrophysiological and morphological study of the neuromuscular junction in patients with myasthenia gravis. Exp. Neurol. 51, 536-563. ALBUQUERQUE, E. X. and WARNICK, J. E. (1971). Electrophysiological observations in normal and dystrophic chicken muscles. Science, 172, 1260-1263. ANGAUT-PETIT, D. and MALLART, A. (1979). Dual innervation of endplate sites and its consequences for neuromuscular transmission in muscles of adult Xenopus laeris. J. Physiol. 289, 203-218. ANTONY. M. T. and TONGE, D. A. (1980). Effects of denervation and botulinum toxin on muscle sensitivity to acetylcholine and acceptance of foreign innervation in the frog. J. Physiol. 303, 23-31. ANZENBACHER, H. and ZENKER, W. (1963). Uber die Gr6ssenbeziehung der Muskelfasern zu ihren motorischen Endplatten und Nerven. Z. Zellforsch. Mikroskop. Anat. 60, 860--871. ARNOLD, A. P. (1981). Logical levels of steroid hormone action in the control of vertebrate behavior. Amer. Zool. 21,233-242. AUERBACH, A. and BETZ, W. (1971). Does curare affect transmitter release? J. Physiol. 213, 691-705. BAGUST, J., LEWIS, D. M. and WESTERMAN. R. A. (1973). Polyneuronal innervation of kitten skeletal muscle. J. Physiol. 229, 241-255. BARKER, D. and IP, M. C. (1966). Sprouting and degeneration of mammalian motor axons in normal and de-afferented skeletal muscle. Proc. R. Soc. Lond. B 163, 538--554. BARNARD, E. A., WIECKOWSKI, J. and CHIU, T. H. (1971). Cholinergic receptor molecules and cholinesterase molecules at mouse skeletal muscle junctions. Nature, 234, 207-209. BARONDES, S. H. (Ed.) (1976). Neuronal Recognition, Plenum Press, New York, 266 ALAN D. GRINNELL AND ALBERT A. HERRERA BARSTAD, J. A. B. (1962). Presynaptic effect of the neuromuscular transmitter. Experientia, lg, 579-581. BAUER, H. (1971). Die Freisetzung von Acetylcholin an der motorischen Nervenendigung unter dem Einfluss von d-Tubocurarin. Pflii9 Arch. 326, 162-183. BEANI, L.. BIANCHI. C. and LEDDA, F. (19641. The effect of tubocurarine on acetylcboline release from motor nerve terminals. J. Physiol. 174, 172-183. BEAUDOIN, A. R. (1956L The development of lateral motor column cells in the lumbosacral cord in Rana pipiens. II. Development under the influence of thyroxin. Anac Record 125, 247 259. BEKOFF, A. (1976). Ontogeny of leg motor output in the chick embryo: a neural analysis. Brain Res. 106, 271- 291. BEKOFE, A., STEIN, P. S. G. and HAMBURGER, V. (1975), Coordinated motor output in the hindlimb of the 7 day chick embryo. Proc. Nat. Acad. Sci. 72, 1245-1248. BENNETT, M. R., DAVEV, D. and U-~.BEL,K. (1980). The growth of segmental nerves from the brachial myotomes into the proximal muscles of the chick forelimb during development. J. Comp. Neurol. 189, 335-357. BENNETT. M. R. and LAVtDlS, N. A. (1979). The effect of calcium ions on the secretion of quanta evoked by an impulse at nerve terminal release sites. J. Gen. Physiol. 74, 429-456. BENNETT M. R.. MCGRATH. P. A. and DAVEY D. F. (19791. The regression of synapses formed by a foreign nerve in a mature axolotl striated muscle. Brain Re.s. 173, 451 469. BENNETT M. R,, MCLACHLAN, E. M. and TAYLOR, R. S. (1973a). The formation of synapses in reinnervated m a m m a l i a n striated muscle. J. Physiol. 233, 481-500. BENNETT, M. R., McLACHLAN. E. M. and TAYLOR, R. S. (1973b). The formation of synapses in m a m m a l i a n striated muscle reinnervated with autonomic preganglionic nerves. J. Physiol. 233, 501 518. BENNETT M. R. and NURCOMBE, V. (1979L The survival and development of cholinergic neurons in skeletal muscle conditioned media. Brain Res. 173, 543-548. BENNETT M. R. and PETTIGREW. A. G. (1974ai. The formation of synapses in striated muscle during development. J. Physiol. 241,515-545. BENNETT, M. R, and PETTIGREW, A. G. (1974bL The formation of synapses in reinnervated and cross-reinnervated striated muscle during development. J. Physiol. 241,547-573, BENNETT M. R and PETTIGREW, A. G. (1975). The formation of synapses in amphibian striated muscle during development. J. Physiol. 252, 203 239. BENNETT M. R., PETTIGREW, A. G. and TAYLOR, R. S. 11973). The formation of synapses in reinnervated and cross-reinnervated adult avian muscle. J. Physiol. 230, 331-357. BENNETT M. R. and RAFTOS. J. (1977L The formation and regression of synapses during the reinnervation of axolotl striated muscles. J. Physiol. 265, 261 295. BENOIT, P. and CHANGEL'X. J.-P. (1975). Consequences of tenotomy on the evolution of multi-innervation in developing rat soleus muscle. Brain Res. 99, 354 358. BENOIT. P. and CHANGEt:X. J.-P. (1978). Consequences of blocking the nerve with a local anesthetic on the evolution of multi-innervation at the regenerating neuromuscular junction of the rat. Brain Res. 149, 89 96. BERANEK, R. and VYSKO~'tL, F. (1967). The action of tubocurarine and atropine on the normal and denervated rat diaphragm. J. Physiol. 188, 53-66. BERG, D. and HALL, Z. (1975). Increased extrajunctional acetylcholine sensitivity produced by chronic postsynaptic neuromuscular blockade. J. Physiol. 244, 659-676. BERNSTHN, J. J. and Gt~TH. L. 119611. Nonselectivity in establishment of neuromuscular connections following nerve regeneration in the rat. Exp~ Neurol. 4, 262--275. BETZ. W. J., CALDWELL. J. H. and RtBCHESTER, R. R. (1979). The size of motor units during post-natal development of rat lumbrical muscle. J. Physiol. 297, 463 478. BETZ, W. J., CALDWELL, J. H. and RtBCH~TER, R. R. (1980a). The effects of partial denervation at birth on the development of muscle fibers and motor units in rat lumbrical muscle. J. Physiol. 303, 265-279. BETZ. W. J., CALDWELL, J. H. and RtBCHESTER, R. R. (1980bt. Sprouting of active nerve terminals in partially inactive muscles of the rat. J. Physiol. 303, 281-297. BtRKS, R., KATZ, B. and MtLED1, R. (19601. Physiological and structural changes at the amphibian myoneural junction, in the course of nerve degeneration. J. Physiol. 150, 145 168. BtXBY, J. L. (1981). Ultrastructural observations on synapse elimination in neonatal rabbit skeletal muscle. ,/. Neurocytol. 10, 81-100. BtxBv, J. L. and VAN ESSEN, D. C. 11979ai. Regional differences in the timing of synapse elimination in skeletal muscles of the neo-natal rabbit. Brain Res. 169, 275-286. BIXBV. J, L. and VAN ESSEN, D. C. (1979bt. Competition between foreign and original nerves in adult mammalian skeletal muscle. Nature, 282, 726 728. BLABER. L. C. (1970L The effect of facilitatory concentrations of decamethonium on the storage and release of transmitter at the neuromuscular junction of the cat. J. Pharmacol. Exp. Ther. 175, 664-672. BLABER. L. C. (19731. The prejunctional actions of some non-depolarizing blocking drugs. Br. J. Pharmacol. 47, 109~ 116. BLABER, L. C. and CHRIST. D. D. (19671. The action of facilitatory drugs on the isolated tenuissimus muscle of the cat. Int. J. Neuropharmucol. 6, 473 4.84. BLACK, I. B., HENDRV, I. A, and IVERSON, L. L. (1971). Transynaptic regulation of growth and development of adrenergic neurones in mouse sympathetic ganglion. Brain Res. 34, 229--240. BLACK. I. B., HENDRV, I. A. and IVERSON, L. L. (19721. The role of postsynaptic neurones in the biochemical maturation of presynaptic cholinergic nerve terminals in a mouse sympathetic ganglion. 3. Physiol. 221, 149. 159. BLACKSHAW.S. E. and WARNER. A. E. (19761. LOW resistance junctions between mesoderm cells during development of trunk musculature. J. Physiol. 225, 209--230. BLASDEL. G. G. and PETTIGREW, J. D. (1978). Effects of prior visual experience on cortical recovery from the effects of unilateral eyelid suture in kittens. J. Physiol. 274, 601 619. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 267 BLINZINGER, K. and KREUTZBERG, G. (1968). Displacement of synaptic terminals from regenerating motoneurons by microglial cells. Z. Zellforsch. Microsk. Anat. 85, 145-157. BOWEN, J. M. and MERRY, E. H. (19691. Influence of d-tubocurarine, decamethonium and succinylcholine on repetitively evoked end-plate potentials. J. Pharmac. Exp. Ther. 167, 334-343. BOYD, I. A. and MARTIN, A. R. (1956). The endplate potential in mammalian muscle. J. Physiol. 132, 74-91. BRADLEY, W. G. and JAROS, E. (1979). Involvement of peripheral and central nerves in murine dystrophy. Ann. N.Y. Acad. Sci. 317, 132-142. BRAUN, M., SCHMIDT, R. F. and ZIMMERMANN,M. (1966). Facilitation at the frog neuromuscular junction during and after repetitive stimulation. Arch. Ges. Physiol. 287, 41-55. BRAY, J. J., HUaaARD, J. I. and MILLS, R. G. (1979). The trophic influence of tetrodotoxin-inactive nerves on normal and reinnervated rat skeletal muscles. J. Physiol. 297, 479-491. BREEDLOVE,S. M. and ARNOLD,A. P. (1980). Hormone accumulation in a sexually dimorphic motor nucleus of the rat spinal cord. Science, 210, 564-566. BREGMAN, B. S. and CRUCE, W. L. R. (1980). Normal dendritic morphology of frog spinal motoneurons: a golgi study. J. Comp. Neurol. 193, 1035-1045. BREITSCHM1D, P. and BENNER,H. R. (1981L Channel gating at frog neuromuscular junctions formed by different cholinergic neurones. J. Physiol. 312, 237-252. BROWN, G. L. and HARVEY, A. M. (1938). Neuromuscular conduction in the fowl. J. Physiol. 93, 285-300. BROWN, M. C. and BUTLER, R. G. (1976). Regeneration of afferent and efferent fibres to muscle spindles after nerve injury in adult cats. J. Physiol. 260, 253-266. BROWN, M. C., GOODWlN, G. M. and 1RONTON, R. (1977). Prevention of motor nerve sprouting in botulinum toxin poisoned mouse soleus muscles by direct stimulation of the muscle. J. Physiol. 267, 42-43P. BROWN, M. C. and HOLLAND. R. L. (1979). A central role for denervated tissues in causing nerve sprouting. Nature, 282, 724-726. BROWN, M. C,, HOLLAND, R. L. and HOPKINS, W. G. (1981). Motor nerve sprouting. Ann. Rev. Neurosci. 4, 17-42. BROWN, M. C., HOLLAND, R. L., HOPKINS, W. G. and KEYNES,R. J. (1981). An assessment of the spread of the signal for terminal sprouting within and between muscles. Brain Res. 210, 145-151. BROWN, M. C., HOLLAND, R. L. and IRONTON, R. (1978). Degenerating nerve products affect innervated muscle fibres. Nature. 275, 652-654. BROWN, M. C., HOLLAND,R. L. and IRONTON, R. (1979). Evidence against an intraspinal signal for motoneurone sprouting in mice. ,L Physiol. 291, 35P-36P. BROWN, M. C.. HOLLAND, R. L. and IRONTON, R. 11980). Nodal and terminal sprouting from motor nerves in fast and slow muscles of the mouse, d. Physiol. 306, 493-510. BROWN, M. C. and IRONTON, R. (1977). Motoneurone sprouting induced by prolonged tetrodotoxin block of nerve action potentials. Nat,re, 265, 459-461. BROWN, M. C. and IRONTON, R. (1978). Sprouting and regression of neuromuscular synapses in partially denervated mammalian muscles. J. Physiol. 278, 325-348. BROWN, M. C., JANSEN, J. K. S. and VAN ESSEN, D. (1976). Polyneuronal innervation of skeletal muscle in new-born rats and its elimination during maturation. J. Physiol. 261,387-422. BROWNELL. B., OPPENHEIMER, D. R. and SPALDING,J. M. R. (1972). Neurogenic muscle atrophy in myasthenia gravis. J. Neurol. Neurosury. Psychiat. 35, 311-322. BRUSHART, T. M. and MESULAM,M.-M. (19801. Alteration in connections between muscle and anterior horn motoneurons after peripheral nerve repair. Science, 208, 603-605. BRYAN, J. S. and LETINSKY,M. S. (1979). Morphology and electrophysiology of dystrophic chicken muscle. Soc. NetJrosci. Ahstr. 5, no. 1611. BORDEN, S. J., SARGENT, P. B. and MCMAriAN, U. J. (1979). Acetylcholine receptors in regenerating muscle accumulate at original synaptic sites in the absence of the nerve. J. Cell Biol. 82, 412--425. BURKE, R. S., STRICK, P. L., KANDA, I. K., KIM, C. C. and WALmLEY, B. (1977). Anatomy of medial gastrocnemius and soleus nuclei in cat spinal cord. J. Neurophysiol. 40, 66%680. CAMPENOT, R. B. (1977). Local control of neurite development by nerve growth factor. Proc. Nat. Acad. Sci. 74, 4516-4519. CANGIANO, A. and FRIED, J. A. {1977). The production of denervation-like changes in rat muscle by colchicine without interference with axonal transport or muscle activity. J. Physiol. 265, 63-84, CARBONEITO, S. (19771. Neuromuscular transmission in dystrophic mice. J. Neurophysiol. 40(41, 836-843. CARDASIS, C. A. and PADYKULA,H. A. (19791. Ultrastructural evidence of reorganization at the neuromuscular junction in the soleus and gastrocnemius muscles in the normal adult rat. Anat. Rec. 193, 497. CASS, D. T. and MARK. R. F. (1975L Re-innervation of axolotl limbs. I. Motor nerves. Proe. R. Soc. Lon. B 190, 45-58. CASS, D. T., SUTTON,T. J. and MARK, R. F. (1973). Competition between nerves for functional connexions with axolotl muscles. Nature, 243, 201-203. CASTRO, G. DE O. (1963L Effects of reduction of nerve centers on the development of residual ganglia and on nerve patterns in the wing of the chick embryo. J. Exp. Zool. 152, 279-295. CHANG. C. C., CBENG, H. C. and CHEN, T. F. (19671. Does d-tubocurarine inhibit the release of acetylcholine from motor nerve endings': Jap. J. Physiol. 17, 505-515. CHANG. C. C., CHUANG, S. and HUANG, M. C. (19751. Effects of chronic treatment with various neuromuscular blocking agents on the number and distribution of acetylcholine receptors in the rat diaphragm. J. Physiol. 250, 161-173. CHANG. C. C. and LEE, C. Y. (19661. Electrophysiological study of neuromuscular blocking action of cobra neurotoxin. Br. J. Pharmacol. Chemother. 28, 172-181. CHIAKULAS,J. J. and PAULY,J. E. {1965}. A study of postnatal growth of skeletal muscle in the rat. Anat. Rec. 152, 55-61. CHU-WANG, 1.-W. and OPPENHEIM, R. W. (1978al. Cell death of motoneurones in the chick embryo spinal cord. J,P.N. 174--E 268 ALAN D. GRtNNELL AND ALBERT A. HERRERA I. A light and electron microscopic study of naturally occurring and induced cell loss during development. J. Comp. Ne,rol. 177, 33-58. CHU-WANG, I.-W. and OPPENHEtM, R. W. (1978b). Cell death of motoneurones in the chick embryo spinal cord. II. A quantitative and qualitative analysis of degeneration in the ventral root. including evidence for axon outgrowth and limb innervation prior to cell death. J. Comp. Nellrol. 177, 59-86. CIANI, S. and EDWARDS. C. (1963). The effect of acetylcholine on neuromuscular transmission in the frog. J. Pharmacol. Exp. Ther. 142, 21-23. CLARKE. P. G. H. and COWAN. W. M. (1976). The development of the isthmo-optic tract in the chick, with special reference to the occurrence and correction of developmental errors in the location and connections of isthmo-optic nerves. J. Comp. Neurol. 167, 143-164. CLOSE, R. (19671. Properties of motor units in fast and slow skeletal muscles of the rat. d. Physiol. 193, 45 55. ColORs, C. (1955). Les variations structurelles normales et pathologiques de la jonction neuromusculaire. Actu Neurol. Psychiat. Belq. 55, 741-866. COORS, C. and TELERMAN-Tor'PEt, N. (19761. Morphological and histochemical changes of motor units in myasthenia. Ann. N.Y. Acad. Sci. 274, 6~19. Co~as, C. and WOOLF, A. L. (1959}. The lnnerration ofMttscle. A Biopsy Study. Blackwell Scientific, Oxford. COr~RADt, S. and RONNEVI, L.-O. (1977) Ultrastructure and synaptology of the initial segment of cat spinal motoneurons during early postnatal development. J. Neurocytol. 6, 195-210. CONRADI, S. and SKOGLUND,S. (1969). Observations on the ultrastructure of the initial motor axon segment and dorsal root boutons on the motoneurons in the lumbrosacral spinal cord during postnatal development. .4eta Physiol. [Sound. Suppl). 333, 53 76. COOI,ER. E., DIAMOND.J. and TURNER. C. (1977). The effects of nerve section and of colchicine treatment on the density of mechano sensory nerve endings in salamander skin. J. Physiol. 264, 725-749. COTMAN, C. W. (Ed.) (1978). Neuronal Plasticity. Raven Press, New York. COWAN. W. M. (1973). Neuronal death as a regulative mechanism in the control of cell number in the nervous system. In: Det'elopment uml A qin~ in the Nerrous System. pp. 19 41. Acad. Press, New York. COWAN, W. M. (1978). Aspects of neural development. In: International Review of Physiology, Neurophysiolooy I l L A. C. GUVTON. Ed. University Park Press, Baltimore, 17, 149-191. CRAGG, B. G. {1975~. The development of synapses in the visual system of the cat. J. Comp. Neto'ol. 160, 147 166. CRAIN. B.. COTMAN, C.. TAYLOR. D. and LYNCH, G. (1973). A quantitative electron microscopic study of synaptogenesis in the dentate gyra of the rat. Brain Res. 63, 194-204. CrEAZZO. T. L. and SOHAL. G. S. (19781. Effects of alpha and beta bungarotoxin on the development of trochlear nucleus and superior oblique muscle. Soc. Nettrosci. Abstr. 4, 110. CREPEL F.. MartANt. J. and DELHAVE-BoUCHAL'D,N. (19761. Evidence for multiple innervation of Purkinje cells by climbing fibers in the immature rat cerebellum. J. Neurobiol. 7, 567-578. CRUCE. W. L. R. (1974). The anatomical organization of hindlimb motoneurons in the lumbar spinal cord of the frog. Runa catesbeiana. J. Comp. Ne,rol. 153, 59-76. CULL. R. E. 0974). Role of nerve-muscle contact in maintaining synaptic connections. Exp. Brain Res. 20, 307 310. CULL. R. E. (19751. Effect of sensory nerve division on the afferent synapses of axotomized motor neurones. Exp. Brain Res. 22, 421--426. CULt:CANDY, S. G.. MILEDI. R. and TRAUTMANN. A. 119781. Acetylcholine-induced channels and transmitter release at human endplates. Ntm,'e. 271, 74-75. CULL-CANDY, S. G.. MILEDI, R.. TRAUTMANN, A. and UCHITEL. O. D. 11980}. On the release of transmitter at normal, myasthenia gravis and myasthenic syndrome affected human end-plates. J. Physiol. 299, 621-638. CzI~tt. G.. GALLEGO, R., KUEX), N. and Kt:NO. M. (1978J. Evidence for the maintenance of motoneurone properties by muscle activity, d. Physiol. 281,239-252. DAHLB,~CK, O.. ELMQVlST, D.. JOHNS. T. R.. RADNeR. S. and THESLEW. S. 11961 ~. An electrophysiologic study of the neuromuscular junction in myasthenia gravis. ,/. Physiol. 156, 336-343. DAHLS'rR6M (19681. Effect of colchicine on transport of amine storage granules in sympathetic nerves of rat. Eur. J. Pharmacol. 5, 111-113. DANIELS. M. P. andd VOGEL. Z. 119751. lmmnnoperoxidase staining of :t-bungarotoxin sites in muscle endplates show distribution of acetylcholine receptors. Nat,re. 254, 339 341. DAvis. R. and KOELLE, G. B. ~1967). Electron microscopic localization of acetylcholinesterase and nonspecific cholinesterase at the neuromuscular junction by the gold-thiocholine and gold-thiolacetic acid methods. J. Cell Biol. 34, 157-17t. DECINo. P. 119811. Transmitter release properties along regenerated nerve processes at the frog neuromuscular ,junction. J. Ne,rosci. !, 308 317. D~NNts. M. J. (19751. Physiological properties of junctions between nerve and muscle de',eloping during salamander limb regeneration. J. Physiol. 244, 683-702. De~lS. M. J. (1981~. Development of the neuromuscular junction: inductive interactions between cells. Ann. oe~. .Ve,ro~ci. 4, 43 68. DEN~I~. M. J. and HARRIS. A. J. t1979). Elimination of inappropriate nerve-muscle connections during development of rat embryos. Bruin Res. 49, 359- 364. DEN~lS. M. J. and HARRIS. A. J. t1980L Transient inability of neonatal rat motoneurons to reinnervatc muscle. Dec. Biol. 74, 173 183. DENNXS, M. J. and YIP, J. W. (1978). Formation and elimination of foreign synapses on adult salamander muscle. J. Physiol. 274, 299-310. DENNIS, M. J.. ZtSKIND-CO~HAtm. L. and HARRtS. A. J. (19811. Development of neuromuscular junctions in rat embryos. Def. Biol. 81,266 279. DESHI'ANDE, S. S.. ALBt QL ERQUE. E. X. and GtTH, L. (19761. Neurotrophic control of skeletal muscles in normal and hibernating ground squirrels. Soc. Neurosci. Ahm'. 2, 1035. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 269 DESMEDT. J. E. (19661. Presynaptic mechanisms in myasthenia gravis. Ann. N.Y Acad. Sci. 135, 209-246. DETWILER, S. R. (19331. Experimental studies upon the development of the amphibian nervous system. Biol. Rel'. 8, 269-310. DIAMOND, J. (19791. The regulation of nerve sprouting by extrinsic influences. In: The Neurosciences: Fourth Study Program, pp. 937-955. Eds. F. O. SCHMIYr and F. G. WORDEN. M.I.T. Press. DIAMOND,J., COOPER,D., TURNER, C. and MACISTYRE,L. (1976). Trophic regulation of nerve sprouting. Science 193, 371-377. DIAMOND, J. and JACKSON, P. C. (19801. Regneration and collateral sprouting of peripheral nerves. In: Nert'e Repair and Regeneration, pp. 115-129. Eds. D. C. JEWETTand H. R. MCCARROLL,C. V. Mosby, St. Louis. DODGE, F. A. and RAHAMINOFF, R. (1967). Co-operative action of calcium ions in transmitter release at the neuromuscular junction. J. Physiol. 193, 419-432. DOWNMAN. C. B. B., ECCLES, J. C. and MCINTYRE, A. K. (1953). Functional changes in chromatolysed motoneurones. J. Comp. Neurol. 98, 9-36. DRACHMAN, D. B. (1981). The biology of myasthenia gravis. Ann. Rer. Ne,rosci. 4, 195-225. DUCHEN. L. W. (1970}. Changes in motor innervation and cholinesterase localization induced by botulinum toxin in skeletal muscle of the mouse: differences between fast and slow muscle. J. Neurol. Neurosurg. Psychiat. 33, 40--54. DUCHEN. L. W. (1979}. Hereditary disorders of motor and sensory neurons in the mouse. Ann. N.E Acad. Sci. 317, 506-511. DUCHEN, L. W., FALCONER,D. S. and STRICH, S. J. (19661. Hereditary progressive neurogenic muscular atrophy in the mouse, d. Physiol. 183, 53P-55P. DUCHEN. L. W., GOMEZ, S. and QUEIROZ, L. S. (19801. Axonal growth and regeneration at the neuromuscular junction. In: Ontoyenesis and Functional Mechanisms of Peripheral Synapses, pp. 265-270. Ed. J. TAXI. I.NS.E.RM. Symp. No. 13, Elsevier Publ., Amsterdam, North Holland. DUCHEN, L. W., ROGERS, M., STOLKIN, C. and TONGE, D. A. (19751. Suppression of botulinum toxin induced axonal sprouting in skeletal muscle by implantation of an extra nerve. J. Physiol. 248, IP-2P. DUCHEN, L. W. and SEARLE,A. G. (19701. Hereditary motor end-plate disease in the mouse: light and electron microscopic studies. J. Neurol. Neurosury. Psychiat. 33, 238-250. DUCHEN. L. W., SEARLE,A. G. and STRICH. S. J. (19671. An hereditary motor end-plate disease in the mouse. J. Physiol. 189, 4P-6P. DUCHEN. L. W. and STEEANI, E. (19711. Electrophysiological studies of neuromuscular transmission in hereditary "motor end-plate disease' of the mouse. J. Physiol. 212, 535-548. DUCHEN, L. W. and STRIC8. S. J. (1968}. The effects of botulinum toxin on the pattern of innervation of skeletal muscle in the mouse. Quant. J. Exp. PhysioL 53, 84-89. DUCHEN. L. W., STRICH. S. J. and FALCONER, D. S. (1968}. An hereditary motor neurone disease with progressive denervation of muscle in the mouse: the mutant 'wobbler'. J. Neurol. Neurosurg. Psychiat. 31, 535542. DUCHEN. L. W. and TONGE. D. A. (1973}. The effects of tetanus toxin on neuromuscular transmission and on the morphology of motor endplates in slow and fast skeletal muscle of the mouse. J. Physiol. 228, 157-172. DUCHEN. L. W. and TONGE. D. A. (1977}. The effects of implantation of an extra nerve on axonal sprouting usually induced by botutinum toxin in skeletal muscles of the mouse. J. Anat. 124, 205-215. DUNCAN, C. J. and PUBLICOVER,S. J. (19791. Inhibitory effects of cholinergic agents on the release of transmitter at the frog neuromusclar junction. J. Physiol. 294, 91-103. DUNLAP, D. G. (19661. The development of the musculature in the hindlimb in the frog, Rana pipiens. J. Morphol. 119, 241-258. DUNN. E. H. {19091. A statistical study of the medullated nerve fibers innervating the legs of the leopard frog, Rana pipiens, after unilateral section of the ventral roots. J. Comp. Neurol. 14, 684-719. EBENDAL, T., OLSON. L., SEIGER, ~. and HEDLUND, K.-O. {1980}. Nerve growth factors in the rat iris. Nat,re, 286, 25-28. ECCLES. J. C. (19641. The Ph.vsiolo(ty of Synapses, pp. 127-131. Springer-Verlag, New York. ECCLES, R. M., KOZAK,W. and WESTERMAN,R. A. (19621. Enhancement of spinal monosynaptic reflex responses after denervation of synergic hind-limb muscles. Exp. Ne,rol. 6, 451-464. ECCLES, J. C., LinEr. B. and YOUNG, R. R. (19581. The behavior of chromatolysed motoneurones studied by intracellular recording. J. Physiol. 143, 1i-40. EDDS. MAC.V. (1953}. Collateral nerve regeneration. Quart. Rer. Biol. 28, 260-276. ELDRIDGE, L., LIEBHOLD, M. and STEINaACn, J. (19811. Alterations in cat skeletal neuromuscular junctions ................ following prolonged inactivity. J. Physiol. 313, 529-545. ELMQVIST, D., HOFMANN, W. W., KUGELnERG, J. and QUAS~L, D. M. J. (19641. An electrophysiological investigation of neuromuscular transmission in myasthenia gravis. J. Physiol. 174, 417-434. ELSBERG. C. A. (19171. Experiments on motor nerve regeneration and the direct neurotization of paralyzed muscles by their own and by foreign nerves. Science, 45, 318-320. ELUL. R.. MILEDI, R. and STEFANL E. (19701. Neural control of contracture in slow muscle fibres of the frog. Acta Physiol. Lat. Am. 10, 194-226. ENGEL, A. G. (19801. Cytochemical and immunoelectron microscopic approaches to altered functions of the motor end-plate. In: Ontogenesis and Functional Mechanisms of Peripheral Synapses, pp. 271-285. Ed. J. TAXI. I.N.S.E.R.M. Symp. No. 13, Elsevier Publ., Amsterdam, North Holland. ENGEL, A. G.. LINDSTROM, J. M.. LAMBERT,E. H. and LENNON, V. A. (19771. Ultrastructural localization of the acetylcholine receptor in myasthenia gravis and in its experimental autoimmune model. Neurolo~ty, 27, 307-315. ENGEL. A. G. and SANTA,T. (19711. Histometric analysis of the ultrastructure of the neuromuscular junction in myasthenia gravis and myasthenic syndrome. Ann. N. Y Acad. Sci. 183, 46-63. ERt:LKAR, S. D. and SOLLER, R. W. (19771. Activation of frog motoneurons by stimulation of contralateral ventral roots. Soc. for Neurosci. Ahstr. 3, 514. no. 1642. 270 ALAN D. GRINNELL AND ALBERT A. HERRERA FAMBROUGH. D. M. (19761. Specificity of nerve muscle interactions. In: Neuronal Reeo~lnition. pp. 25 67. Eds. S. BARONOESand F. E. BLOOM. Plenum, New York. FAmBROt'GH. D. M. (1979). Control of acetylcholine receptors in skeletal muscle. Phlsiol. Rev. 59, 165 227. FAMBROrGH. D. M.. DRACHMAN, D. B. and SATYAmURIL S. 11973). Neuromuscular junction in myasthenia gravis: Decreased acetylcholine receptors. S~'ienee, 182, 293 295. FANGBONER, R. F. and VANABLE, JR.. J. W. (1974). Formation and regression of inappropriate nerve sprouts during trochlear nerve regeneration in Xem)pus hlevis. J. Comp. Neurol. 157, 391 406. FAREL, P. B. (1978). Reflex activity of regenerating frog spinal motoneurons. Bruin Res. 158, 331 341. FATT. P. and KATZ. B. (1951). An analysis of the end-plate potential recorded with an intra-cellular electrode. J. Physiol. 115, 320 370. FAtT. P. and KA'rZ. B. 11952). Spontaneous subthreshold activity at motor nerve endings. J. Physiol. I1% 109 128. F t u l z . A. and MALLARt. A. 11971). An analysis of aeetylcholine responses of junctional and extrajunctional receptors of frog muscle fibres. J. Physiol. 218, 85 100. FEYG, T. P.. Wu, W. Y. and YANG. F. Y. 11965). Selective reinnervation of a "slow" or "'fast" muscle by its original motor supply during regeneration of mixed nerve. Sci. Sin. 16, 1717 1720. Fix. S. and JtRMANOV~,. I. (1969). Innervation by nerve implants of"fast'" and "'stow" skeletal muscles of the rat. Acta Phrsiol. Sound. 76, 257-269. FEX. S.. SONES~)N. B., THESLEEF, S. and ZELEN~,. J. 11966). Nerve implants in botulinum poisoned m a m m a l i a n muscles. J. Physiol. 184, 872 882. FEX, S. and THESLEEF. S. 11967). The time required for innervation of denervated muscle by ncr,,e implants. Li[? Sci. 6, 635 639. FILOGAMO. G. and GABELLA. G. 11966). Cholinesterase behavior in denervated and reinnervated muscles. Aeta Anat. 63, 199 214. FISCHBACH. G. n. and ROBBINS. N. (1971L Effect of chronic disuse of rat soleus neuromt, scular junctions on postsynaptic membrane. J. Neurophysiol. 34, 562-569. FLETCHER, P. and FORRESTER, T. {1970). The measurement of acetylcholine released from m a m m a l i a n skeletal muscle in the presence of curare. J. Physiol. 211, 39P. FLETCHER. P. and FORRESTER. T. {1975). The effect of curare on the release of acetylcholine from mammalian motor nerve terminals and an estimate of q u a n t u m content. J. Phvsiol. 251, 131 144. FORm:S, A., RAY. L. H. and HOPKINS, A.MCH. (19231. The effect of tension on the action current of skeletal muscle. Am. J. Phl'siol. 65, 300 311. FORRESTt R, T. and SCHMtDT. H. {1970). An electrophysiological investigation of the slow muscle fiber system in the frog rectus abdominis muscle. J. Physiol. 207, 477 4.91. FRANK, E. and JANSEN. J. K. S. 11976). Interaction between foreign and original nerxes innervating gill muscles in fish. J. Neurophysiol. 39, 84-90. FRANK, E.. JANSEN. J. K. S.. L()MO, T. and WESI¢iAARD. R. H. 1t975). The interaction between foreign and original motor nerves innervating the soleus muscle of rats. J, Physiol. 24% 725 -743. FREEDMAN. S. D. and LEN1-Z, T. L. 11980). Binding of horseradish peroxidase-z~-bungarotoxin to axonal membranes at the node of Ravier. J. Comp. Neurol. 193, 179-185. FROLKIS.. V. V., MARTYNI'NKO, O. A. and ZAMOSIYAN, V. P. 11976). Aging of the neuromuscular apparatus. Gerontoh),qv. 22, 244 279. F'Lt TON. J. F. (1925t. The influence of tension upon the electrical response to repetitive stimuli. Proc. R. So~. Lond. B 97, 406 423. GALINDO, A. (1971a). Depolarizing neuromuscular block. J. Pharmaeol. Exp. Ther. 178, 339 349. GALINDO, A. (1971b). Prgiunctional effect of curare: its relative importance. J. Ne,roph)'siol. 34, 289 301. GALLEGO, R., KUNO, M., NOIqEZ. R. and SNIDER, W. D. (1979a) Dependence of motoneurone properties on the length of immobilized muscle. J. Physiol. 291, 179-189. GALLi~(;O. R.. KU:NO, M., N(:~qeZ, R. and SNtDt~R. W. D. 11979b). Disuse enhances synaptic efficacy in spinal motoneurones. J. Ph.l'siol. 291, 191 205. GANC;t'LV. D. K. and DAS, M. (1979L Effects of oxotremorine demonstrate presynaptic muscarinic and dopaminergic receptors on motor nerve terminals. Nature, 278, 645-646. GENAT, B. R. and MARK, R. F, ~1977i. Electrophysiological experiments on the mechanism and accuracy of neuromuscular specificity in the axolotl. Phil. Trans. R. Soc. Lond. B 278, 335 347. GIRGIS. S. D.. DREICHEN, K. L., SOKOLL. M. D. and LONG, J. P. (1971t, The effect of neuromuscular blocking agents on acetylcholine release. Proc. Soc. Exp. Biol. Med. 138, 693 695. GIRVLER. R. A. and ROBBINS, N. (19781. Differences in neuromuscular transmission in red and white muscles. Brain Res. 142, 160 164. GIACOBINI, G.. FILOGAMO, G.. WEBER. M.. BootEl. P. and CHANGEt'X, J. P. 11973). Effects of a snake :~-neurotoxin on the development of innervated skeletal muscles in chick embryo. Proc. Nut..4cad. Set. 70, 1700 1712. GIA(OBINI-ROBECCHI. M. G.. GIACOBINI, G.. FILOGAMO. G. and CHANGEUX. J. P. 11975). Effects of the type A toxin from Clostridium botulinunl on the development of skeletal muscles and on their innervation in the chick embryo. Brain Res. 83, 107 121. GIBSON. C. T. and SXnIH. D. O. (1980L Reduced levels of acetylcholine in the motor ner',e terminals of aged rats. Soc..lbr Neurosci. Ahstr. 6, no. 55.1. GLAV|NOVI('. M. 1. (1979~. Presynaptic action of curare. J. Physiol. 290, 499 506. GOLD, M. R. and MARTIN. A. R. [1980). V.I.P. increases quantal content at the neuromuscular junction. So~. fi)r Neurosci. Ahstr. 6, 842 no. 283.1. GOLDRINr3, J. M., KtNO. M.. N('~EZ. R. and SNtDER. W. D. 11980). Reaction of synapses on motoneurones to section and restoration of peripheral sensory connexions in the cat. J. PhvsioL 309, 185 198. GOLDSPINK. D. F. {1977}. The influence of immobilization and stretch on protein turnover of rat skeletal muscle. J. Physiol. 264, 267 282. SPECIFICITY AND PLASTICITY OF NEUROMUSCULAR CONNECTIONS 271 GORDON, T., NIVEN-JENKINS,N. and VRBOV/,,G. (1980). Observations on neuromuscular connections between the vagus nerve and skeletal muscle. Neurosci. 5, 597-610. GORDON, H. and VAN ESSEN, D. C. 11981). Motor units diversify in size as synapse elimination proceeds in the neonatal rabbit soleus muscle. Soc.for Neurosci. Abstr. 7, 179. GOTTLIEB, D. I. and GLASER, L. (1980). Cellular recognition during neural development. Ann. Rer. Nem'osci. 3, 303-318. GRAFSTEIN, B. and McQUARRIE, I. G. 11978). Role of the nerve cell body in axonal regeneration. In: Neuronul Plasticity, pp. 155-195. Ed. C. W. COTMAN. Raven. New York. GRANaACEIER,N. (1971). Uber die Grtissenbeziehungen der Muskelfasern zu ihren motorischen Endplatten und Nerven bei Hypertrophie and Atrophie. Z. Anat. EntwickI-Gesch. 135, 76 87. GREENMAN, M. J. (1913j. Studies on the regeneration of the peroneal nerve of the albino rat: Number and sectional areas of fibres: Area relation of axis to sheath. J. Comp. Neurol. 23, 479-514. GRIMM. L. M. (1971). An evaluation of myotypic respecification in axolotls. J. Exp. Zool. 178, 479-496. GRINNELL, A. D. (1966). A study of the interaction between motoneurones in the frog spinal cord. J. Physiol. 182, 612-648. GRINNELL, A. D. and HERRERA,A. A. (1980a). Physiological regulation of synaptic effectiveness at frog neuromuscular junctions. J. Physiol. 307, 301-317. GRINNELL, A. D. and HERRERA, A. A. (1980bL Synaptic efficacy depends on motor unit size in normal frog neuromuscular junctions. Soc. for Neurosci. Ahstr. 6, no. 192.11. GRINNELL. A. D.. LETINSKY. M. S. and RHEUBEN,M. B. (19791. Competitive interaction between foreign nerves innervating frog skeletal muscle, d. Physiol. 289, 241 262. GRINNELL. A. D. and R,EUaEN, M. B. (1979). The physiology, pharmacology and tropic effectiveness of synapses formed by autonomic preganglionic nerves on frog skeletal muscles, d. Physiol. 289, 219- 240. GRINNELL, A. D., RHEUSEN,M. B. and LETINSKV,M. S. (1977). Mutual repression of synaptic efficacy by pairs of foreign nerves innervating frog skeletal muscle. Nat,re, 265, 368-370. GUNOERSON, R. W. and BARRETT,J. N. (1979). Neuronal chemotoxis: Chick dorsal-root axons turn toward high concentrations of nerve grrowth factor. Science, 206, 1079-1080. GUNTHER, J. S. and LETINSKY, M. S. (19811. A preparation for studying dystrophic avian muscle and neuromuscular junctions. Muscle and Nerre (in press~. GUTH, L. (1956~. Regeneration in the mammalian peripheral nervous system. Physiol. Ret'. 36, 441-478. GUT,, L. (1962). Neuromuscular function after regeneration of interrupted nerve fibers into partially denerrated muscle. Exp. Neurol. 6, 129-141. GUTH. L. (1968). "'Trophic'" influences of nerve on muscle. Phy.~iol. Re~. 48, 645 687. GUTH, L. and FRANK, K. (1959). Restoration of diaphragmatic function following vagophrenic anastomosis in the rat. Exp. Neurol. I, 1-12. GUTH. L. and ZALEWSKL A. A. (1963). Disposition of cholinesterase following implantation of nerve into innervated and denervated muscle. Exp. Nem'ol. 7, 316-326. GUTMANN, E. (1976). Neurotrophic relations. Ann. Rer. Physiol. 38, 177 216. GUTMANN, E. and HANZLiKOVA,V. (1965). Age changes of motor endplates in muscle fibers of the rat. Gerontoloqia, I1, 12-24. GUTMANN, E. and HANZLiKOVA,V. (1966). Motor unit in old age. Nature. 209, 921 922. GUTMANN, E. and HANZLiKOV/,,V. (1967). Effects of accessory nerve supply to muscle achieved by implantation into muscle during regeneration of its nerve. Physiol. Bohemoslar. 16, 244-250. GUTMANN. E. and HANZLiKOVA,V. (1972). A.qe Changes in the Neuromuscular System. Scientechnica, Bristol. GUTMANN, E., HANZLiKOV/~,V. and JAKOUBEK,B. (1968). Changes in the neuromuscular system during old age. Exp. Gerontol. 3, 141-146. GUTMANN, E.. HANZLiKOV/,, V. and VYSOKOOL, F. (1971). Age changes in cross striated muscle of the rat. d. Physiol. 216, 331-343. GUTMANN, E.. TU6EK, S. and HANZLiKOVA,V. 11969). Changes in cholineacetyl-transferase and cholinesterase activities in the levator ani muscle of rats following castration. Physiol. Bohemoslot'. 18, 195 203. GUTMANN, E. and YOUNG. J. Z. (1944). The re-innervation of muscle after various periods of atrophy. J. Amtt. 78, 15--43. GWYNN, D, G. and AITKEN,J. T. (1966~. The formation of new motor endplates in mammalian skeletal muscle. d. Anat. I00, 111-126. GWVNN, D. G. and VRSOVA, G. (19661. Changes in the histochemical appearances of cholinesterase at the neuromuscular junction in atrophic muscles. J. Physiol. 186, 7P-8P. HAIMANN. C., MALLART.A., TOMASI FERRL J., ZILBER GACHELIN,N. F. (1981al. Patterns of motor innervation in the pectoral muscle of adult Xenopus lueris: evidence for possible synaptic remodelling, d. Physiol. 310, 241 256. HAIMANN, C., MALLART,A., TOMAS I FERRL J., ZILBER GACHELIN, N. F. ~1981b~. interaction between motor axons from two different nerves reinnervating the pectoral muscle of Xenopus hteris. J. Physiol. 310, 257-272. HAIMANN, C., MALLART, A. and ZILaER-GACHELIN, N. F. (1976). Competition between motor nerves in the establishment of neuromuscular junctions in striated muscles of Xenopus laevis. Neurosci. Lett. 3, 15-20. HAMBERGER. A., HANSSON, H. and SJ6STRANO. J. (19701. Surface structure of isolated neurons. Detachment of nerve terminals during axon regeneration. J. Cell Biol. 47, 319-331. HAMnURGER, V. (1939). Motor and sensory hyperplasia following limb bud transplantations in chick embryos. Physiol. Zool. 12, 268-284. HAMaURGER, V. (1946). Isolation of the branchial segments of the spinal cord of the chick embryo by means of tantalum foil blocks. J. Exp. Zool. 103, 113-142. HAMBURGER,V. (19581. Regression versus peripheral control of differentation in motor hypoplasia. Ant. d. Anat. 102, 365 410. 272 ALAN D. GRINNELL AND ALBERTA. HERRERA HAMBURGER, V. (1975a). Cell death in the development of the lateral motor column of the chick embryo, d. Comp. Neurol. 160, 535-546. HAMBURGER, V. (1975b). Changing concepts in developmental neurobiology. Perspectit'es Biol. Med. 18, 162-178. HAMBURGER, V. (1977). The developmental history of the motor neuron. Neurosci. Res. Pro.qram Bull. S,ppl. I l l , 1-37. HAMBURGER, V. and LEvI-MONTALCIN1, R. (1949). Proliferation, differentiation, and degeneration in the spinal ganglia of the chick embryo under normal and experimental conditions. J. Exp. Zool. 111, 4 5 7 502. HANZLiKOV,~, V. and GUTMANN, E. (1972). Effect of foreign innervation on the androgen sensitive levator ani muscle of the rat. Z. Zellforsch. Mikrosk. Anat. 135, 165 174. HANZLiKOV,~, V. and GUTMANN, E. (1974). The absence of androgen-sensitivity in the grafted soleus muscle innervated by the pudendal nerve. Cell Tissue Rex. 154, 121-129. HANZLiKOV.~, V. and GUTMANN, E. 0978). Effect of castration and testosterone administration on the neuromuscular junction in the levator ani muscle of the rat. Cell Tissue Res. 189, 155-166. HARRIMAN, D. G. F., TAVERNER,D. and WOOLF, A. L. (1970). Ekbom's syndrome and burning paresthesiae. A biopsy study of vital staining and electron microscopy of the intramuscular innervation with a note on age changes in motor nerve endings in distal muscles. Brain, 93, 393-406. HARRIS, A. J. (1974). Inductive functions of the nervous system. Ann. Rez'. Phrsiol. 36, 251 305. HARRIS, A. J., ZISKIND, L. and WIGSTON, D. (1977). Spontaneous release of transmitter from "repressed" nerve terminals in axolotl muscle. Nature, 268, 265 267. HARRIS, C. (1954). The morphology of the myoneural junction as influenced by neurotoxic drugs. Amer. d. Pathol. 30, 501-519. HARRIS, J. B. and MARSHALL,M. W. (1973). A study of action potential generation in murine dystrophy with reference to "functional denervation". Exp. Ne,rol. 41,331 344. HARRIS, J. B. and RIBCrtESTER, R. R. (1978). Neuromuscular transmission is adequate in identified abnormal dystrophic muscle fibers. Nature, 271,362-364. HARRIS, J. B. and RIBCHESTER, R. R. (1979a). The relationship between endplate size and transmitter release in normal and dystrophic muscles of the mouse. J. Physiol. 296, 245-265. HARRIS, J. B. and RIBCHESTER, R. R. (1979b). Muscular dystrophy in the mouse: neuromuscular transmission and the concept of functional denervation. Ann. N.Y. Acad. Sci. 317, 152 170. HARRIS, J. B. and WARD. M. R. (1974). A comparative study of 'denervation" in muscles from mice with inherited progressive neuromuscular disorders. Exp. Neurol. 42, 169-180. HARRIS, W. A. (1981}. Neural activity and development. Ann. Ret'. Physiol. 43, 689 710. HARRIS-FLANAGAN, A. W. (1969). Differentiation and degeneration in the motor horn of the foetal mouse. J. Morph. 129, 281-306. HEATON, M. B. (1977). Retrograde axonal transport in lateral motor neurons of the chick embryo prior to limb bud innervation. Det'. Biol. 58, 421-427. HENDRY, I. A. (1975). Response of adrenergic neurones to axotomy and nerve growth factor. Brain Res. 94, 87-97. HERRERA, A. A. and GRINNEI,L, A. D. (1980a). Differences in synaptic effectiveness at frog neuromuscular junctions: evidence for long term physiological regulation. Brain Res. 194, 228-231. HERRERA,A. A. and GRINNELL,A. D. (1980b). Transmitter release from frog motor nerve terminals depends on motor unit size. Nature, 287, 649-651. HERRERA, A. A. and GRINNELL, A. D. (1981). Contralateral denervation causes enhanced transmitter release from frog motor nerve terminals. Nature, 291,495-497. HEUSER, J. E. and REESE,T. S. (1973). Evidence for recycling of synaptic vesicle membrane during transmitter release at the frog neuromuscular junction. J. Cell. Biol. 57, 315-344. HIDAKA,T. and KURIYAMA,H. (1969). Effects of catecholamines on the cholinergic neuromuscular transmission in fish red muscle. J. Physiol. 201, 61 71. HmOKAWA, N., HEUSER,J. E. and EVANS, L. (1981). Structural evidence that botulinum toxin blocks neuromuscular transmission by impairing the calcium influx that normally accompanies nerve depolarization. J. Cell Biol. 88, 160-171. HNIK, P., JIRMANOV.~,I., VYKLI~K~'.L. and ZELEN,~,J. (1967). Fast and slow muscles of the chick after nerve cross-union. J. Physiol. 193, 309-325. HOFFMAN, H. (1950). Local reinnervation in partially denervated muscle: a histophysiological study. A,,~t. J. Exp. Biol. Med. Sci. 28, 383-397. HOFFMAN, H. (1951). A study of the factors influencing innervation of muscle by impanted nerves. Au.st. J. Exp. Biol. Med. Sei. 29, 280-307. HOFFMAN, H. and SPRINGELL, P. H. (1951). An attempt at the chemical identification of "neurocletin' (the substance evoking axon-sprouting). Aust. J. Exp. Biol. Med. Sci. 29, 417-424. HOFMANN, W. W., KUNDIN, J. E. and FARRELL, D. F. (1967). The pseudomyasthenic syndrome of Eaton and Lambert: An electrophysiological study. Electroenceph. Clin. Neurophysiol. 23, 214-224. HOFMANN, W. W., THESLEFF,S. and ZELEN~., J. (1964). Innervation of botulinum poisoned skeletal muscles by accessory nerves. J. Physiol. 171, 27P-28P. HOH, J. F. Y. (1975). Selective and non-selective reinnervation of fast-twitch and slow-twitch rat skeletal muscle. J. Physiol. 251, 791-801. HOLLAND, R. L. and BROWN, M. C. (1980). Postsynaptic transmission block can cause terminal sprouting of a motor nerve. Science, 207, 649-651. HOLLYDAY, M. and HAMBURGER,V. (1976). Reduction of the naturally occurring motor neuron loss by enlargement of the periphery. J. Comp. Neurol. 170, 311-320. HOLLYDAY, M. and HAMBURGER,V. (1977). An autoradiographic study of the formation of the lateral motor column in the chick embryo. Brain Res. 132, 197-208. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 273 HOLLYDAY, M., HAMBURGER, V. and FARRIS, J. M. G. (1977). Localization of motor neuron pools supplying identified muscles in normal and supernumerary legs of chick embryo. Proc. Nat. Acad. Sci. 74, 3582-3586. HOLT, R. K. and SOHAL, G. S. (19781. Elimination of multiple innervation in the developing avian superior oblique muscle. Am. J. Anat. 151,313-318. HOPKINS, W. G., SINGH, A. D. and SLACK. J. R. (19801. Elimination of polyaxonal innervation in adult mammalian muscle. J. Physiol. 299, 67P. HORDER. T. J. (1978). Functional adaptability and morphogenetic opportunism the only rules for limb development'? Zoon, 6, 181-192. HUBBARD, J. 1., SCHMIDT, R. F. and YOKOTA, T. (1965). The effect of acetylcholine upon mammalian motor nerve terminals. J. Physiol. 181, 810-829. HUBBARD, J. I. and WILSON. D. F. (1973). Neuromuscular transmission in a mammalian preparation in the absence of blocking drugs and the effect of D-tubocurarine. J. Phl'siol. 228, 307 325. HUBBARD. J. 1., WILSON, D. F. and MIYAMOTO, M. (19691. Reduction of transmitter release by D-tubocurarine. Nature, 223, 531-533. HUBEL O. H. (1978). Effects of deprivation on the visual cortex of cat and monkey. In: The Harcey Lectures, Series 72, pp. 1-51. Academic Press, New York. HUBEL, O. H. and WIESEL, T. N. (1970). The period of susceptibility to the physiologial effects of unilateral eye closure in kittens. J. Physiol. 206, 419-436. HUBEL, D. H., WIESEL. T. N. and LEVAY, S. (1977). Plasticity of ocular dominance columns in monkey striate cortex. Proc. R. Soc. Lond. B 278, 377~,09. HUGHES, A. (1968). Aspects of Neural Ontogeny. Logos, London. HUGHES, A. and TSCHUMI, P. A. (1958). The factors controlling the development of the dorsal root ganglia and ventral horn in Xenopus laecis (Daud). J. Anat. 92, 498-527. HUIZAR, P., KUNO, M., KUDO, N. and MIYATA, Y. (1977). Reaction of intact spinal motor neurones to partial denervation of the muscle. J. Physiol. 265, 175-191. HUTTER, O. F. and LOEWENSTEIN, W. R. (1955). Nature of neuromuscular facilitation by sympathetic stimulation in the frog. J. Physiol. 130, 559-571. HUTTER, O. F. and TRAUTWEIN, W. (1956). Neuromuscular facilitation by stretch of motor nerve-endings. J. Physiol. 133, 610-625. INNOCENTI, G. M., FIORE, L. and CAMINITI, R. (1977). Exuberant projection into the corpus callosum from the visual cortex of newborn cats. Neurosci. Lett. 4, 237-242. IRONTON, R., BROWN, M. C. and HOLLAND, R. L. (1978). Stimuli to intramuscular nerve growth. Brain Res. 156, 351-354. ITO, Y., MILEDI, R., MOLENAAR, P. C., VINCENT, A., POLAK, R. L., VAN GELDER, M. and NEWSOM-DAVIS. J. (1976). Acetylcholine in human muscle. Proc. R. Soc. B. 192, 475-480. JACKSON, P. and DIAMOND, J. (1977). Colchicine block of cholinesterase transport in rabbit sensory nerves without interference with long-term viability of the axons. Brain Res. 130, 579-584. JAcons, R. S. and BLABER,L. C. (1971). The anti-curare action of sodium fluoride at the neuromuscular junction of the cat. Neuropharmacol. 10, 607-612. JANSEN,J., LOMO,T., NICOLAYSEN,K. and WESTOAARD,R. (1973). Hypcrinnervation of skeletal muscle fibers: its dependence upon muscle activity. Science, 181,559-561. JANSEN, J. K. S., THOMPSON, W. and KUFFLER, D. P. (1978). The formation and maintenance of synaptic connections as illustrated by studies of the neuromuscular junction. Prog. Brain Res. 48, 3-18. JANSEN, J. K. S. and VAN ESSEN, D. C. (1975). Re-innervation of rat skeletal muscle in the presence of ct-bungarotoxin. J. Physiol. 250, 651-667. JANSEN, J. K. S., VAN ESSEN, D. C. and BROWN, M. C. (1975). Formation and elimination of synapses in skeletal muscles of rat. Cold Spring Harbor Syrup. Quant. Biol. 40, 425-434. JEDREZEJCZYK, J., WIECKOWSKI, J., RYMASZEWSKA,T. and BARNARD, E. A. (1973). Dystrophic chicken muscle: altered synaptic acetylcholinesterase. Science 180, 406-408. JENKINSON,D. H., STAMENOVICB. A. and WHITAKER,B. D. L. (1968). The effect of noradrenaline on the endplate potential in twitch fibers of the frog. J. Physiol. 195, 743-754. JOHNS, T. R. and THESLEFF, S. (1961). Effects of motor inactivation on the chemical sensitivity of skeletal muscle. Acta Physiol. Scand. 51, 136-141. JOHNSON, A. G. and WOOLF, A. L. (1965). Replacement at the neuromuscular junction of the terminal axonic expansion by the Schwann cell. Acta Neuropath. 4, 436-441. JO~ES, S. W. and SALPETER,M. M. (1981). Do presynaptic acetylcholine receptors exist at the frog neuromuscular junction? Fed. Proc. 40(3), 261. KANDEL, E. R. (1976). Cellular Basis of Behavior: An Introduction to Behavioral Neurobiology. Freeman, San Francisco. KATZ, B. (1969). Repetitive "back-firing" in mammalian nerve-muscle preparations. In: The Release of Neural Transmitter Substances, III, pp. 45-56. Ed. C. C. THOMAS. Springfield. KATZ, B. and MILEDI, R. (1964). The development of acetylcholine sensitivity in nerve-free segments of skeletal muscle. J. Physiol. 170, 389-396. KATZ, B. and MILEDL R. (1972). The statistical nature of the acetylcholine potential and its molecular components. J. Physiol. 224, 665-699. KATZ, B. and MILEDI, R. 0973). The effect of atropine on acetylcholine action at the neuromuscular junction. Proc. R. Soc. Lond. B. 184, 221-226. KATZ, B. and MILEDI, R. (1977). Transmitter leakage from motor nerve endings. Proc. R. Soc. Lond. B. 196, 59-72. KELLEY, D. B. (1980). Auditory and vocal nuclei of frog brain concentrate sex hormone. Science, 207, 553-555. KELLEY, D. B., MORRELL, J. I. and PFAFF, D. W. (1975). Autoradiographic localization of hormone-concentrating cells in the brain of an amphibian Xenopus laevis. I. Testosterone. J. Comp. Neurol. 164, 47-61. KELLY, A. M. and ZACKS, S. I. (1969). The histogenesis of rat intercostal muscle. J. Cell Biol. 42, 135-153. 274 ALAN D. GR1NNELLAND ALBERT A. HERRERA KELLY, S. S. (1978). The effect of age on neuromuscular transmission. J. Physiol. 274, 51-62. KERNS, J. M. and HINSMAN,E. J. (19731. Neuroglial response to sciatic neurectomy. I1. Electron microscopy. J. Comp. Neurol. 151,255-280. KIDOKORO, Y., HEINEMANN,S., SCHUBERT, D., BRANDT, B. L. and KLIER, F. G. (19751. Synapse formation and neurotrophic effects on muscle cell lines. Cold Spring Harbor Syrup. Quant. Biol. 40, 373-388. KIM, Y. 1., SANDERS,O. B. and JOHNS, T. R. (1980). Eaton-Lambert syndrome: decreased quantum content of neuromuscular transmitter release produced by serum factors. Soc.for Neurosci. Ahstr. 6, no. 137.19. KIRSCHEN, t . A. and LaVelle, A. (1977). Leucine incorporation by immature and mature axotomized facial neurons. Soc. for Neurosci. Abstr. 3, 328. KOELLE, G. B. (1961). A proposed dual neurohumoral role of acetylcholine: its functions at the pre- and postsynaptic sites. Nature, 190, 208 211. KOELLE, G. n. (1962). A new general concept of the neurohumoral functions of acetylcholine and acetylcholinesterase. J. Pharm. Pharmae. 14, 65-90. KOELLE, G. B. (1971). Current concepts of synaptic structure and function. Ann. N . Y Aead. Sci. 183, 5 20. KORDA~, M., BRZIN, M. and MAJCEN, Z. 0975). A comparison of the effect of cholinesterase inhibitors on end-plate current and on cholinesterase activity in frog muscle. Neuropharmacol. 14, 791 800. KORNELIUSSEN, H. and JANSEN, J. K. S. (1976). Morphological aspects of the elimination of polyneuronal innervation of skeletal muscle fibers in newborn rats. J. Neuroeytol. 5, 591-604. KORNELIUSSEN, H. and WAERHAUG,O. (1973). Three morphological types of motor nerve terminals in the rat diaphragm and their possible innervation of different muscle fiber types. Z. Anat. Entwiek. 140, 73-84. KOZAK, W. and WESTERMAN,R. A. (1961). Plastic changes of spinal monosynaptic responses from tenotomized muscle in cats. Nature, 189, 753-755. KREUTZBERG, G. W. (1969). Neuronal dynamics and axonal flow. IV. Blockage of intra-axonal enzyme transport by colchicine. Proc. Nat. Acad. Sci. 62, 722-728. KRNJEVI~', K. and MILEDI, R. (1958). Some effects produced by adrenaline upon neuromuscular propagation in rats. J. Physiol. 141,291-304. KRNJEVI(~, K. and MITCHELL,J. F. (1961). The release of acetylcholine in the isolated rat diaphragm. J. Physiol. 155, 246-2.62. KUBA, K. (1970). Effects of catecholamines on the neuromuscular junction in the rat diaphragm. J. Physiol. 21 l, 551 570. KUBA, K. and TOMITA,T. (1971). Noradrenaline action on nerve terminals in the rat diaphragm. J. Physiol. 217, 19-31. KUBA, K. and TOMITA, T. (1972). Effects of noradrenaline on miniature end-plate potentials and on the end-plate potential. J. Theor. Biol. 36, 81-88. KUEFLER, D., THOMPSON. W. and JANSEN, J. K. S. (19771. The elimination of synapses in multiply-innervated skeletal muscle fibres of the rat: dependence on distance between end-plates. Brain Res. 138, 353-358. KU~FLER, D. P., THOMPSON,W. and JANSEN,J. K. S. (19801. The fate of foreign endplates in cross-innervated rat soleus muscle. Proe. R. Soc. Lond. B. 2tl8, 189-222. KUFFLER, S. W. (1952). Incomplete neuromuscular transmission in twitch system of frog's submentalis muscles. Fed. Proc. 11, 87. KUEFLER, S. W. and VAUGHAN-WILHAMS,E. M. (1953). Small-nerve junctional potentials. The distribution of small motor nerves to frog skeletal muscle, and the membrane characteristics of the fibres they innervate. J. Physiol. 121,289-317. KULLBERG, R. W., LENTZ, T. L. and COHEN, M. W. 119771. Development of the myotomal neuromuscular junction in Xenopus laet'is: An electrophysiological and fine-structural study. Del. Biol. 60, 101-129. KUNO, M. 119791. Physiologic consequences of neural abnormalities in murine dystrophy. Ann. N.Y. Acad. Sci. 317, 143 151. KUNO, M. and LLIN,~S, R. (1970a). Enhancement of synaptic transmission by dendritic potentials in chromatolysed motoneurones of the cat. J. Physiol. 210, 807-821. KUNO, M. and LLIN,~S, R. (1970b). Alterations of synaptic action in chromatolysed motoneurones of the cat. J. Physiol. 210, 823-838. KUNO, M., MIYATA, Y. and MUf~OZ-MARTINi~Z, E. J. {1974a). Differential reaction of fast and slow alpha motoneurones to axotomy. J. Physiol. 240, 725-739. Ku~o. M., MIVATA,Y. and MU~OZ-MART1Ni~Z, E. J. (1974b}. Properties of fast and slow alpha motoneurones following motor reinnervation. J. Physiol. 242, 273-288. KUNO, M., TURKANIS,S. A. and WEAKLY,J. N. (1971). Correlation between nerve terminal size and transmitter release at the neuromuscular junction of the frog. J. Physiol. 213, 545-556. LAING, N. G. and PRESTIGE, M. C. (19781. Prevention of spontaneous motoneurone death in chick embryos. J. Physiol. 282, 33P-34P. LAMANSr-Y, S. i1870). Ueber die negative Stromesschwankung des arbeitenden Muskels. Pfliiyers Arch. Bd. 3, 193-204. LAMB,A. H. (1976). The projection patterns of the ventral horn to the hind limb during development. Del'. Biol. 54, 82-99. LAMB, A. H. (1977). Neuronal death in the development of the somatotopic projections of the ventral horn in Xenopus. Brain Res. 134, 145-150. LAMB, A. H. (1979). Evidence that some developing limb motoneurons die for reasons other than peripheral competition. Det'. Biol. 71, 8-21. LAMB, A. H. (1980). Motoneurone counts in Xenopus frogs reared with one bilaterally-innervated hind limb. Nature, 284, 347-350. LAMBERT, E. H. and ELMQVlS'r, D. (1971). Quantal components of endplate potentials in the myasthenic syndrome. Ann. N . Y Acud. Sci. 183, 183-199. LAMBERT, E. H. and ROOKE, E. D. (1965). Myasthenic state and lung cancer. In: The Remote Effects of Cancer on the Nert~ous System. pp. 67-80. Eds. LORD BRAIN and F. H. NORRIS. Grune and Stratton, New York. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 275 LANCE-JONES, C. and LANDMESSER, L. (1980a). Motoneuron projection patterns in embryonic chick limbs following partial deletions of the spinal cord. J. Physiol. 302, 55%580. LANCE-JONES,C. and LANDMESSER,L. (1980b). Motoneuron projection patterns in the chick hind-limb following early partial reversals of the spinal cord. J. Physiol. 302, 581-602. LANCE-JONES, C. and LANDMESSER,L. T. (198la). Pathway selection by chick lumbosacral motoneurons during normal development. Proc. R. Soc. Lond. B (in press). LANCE-JONES~ C. and LANDMESSER,L. T. (1981b). Pathway selection by embryonic chick motoneurons in an experimentally altered environment. Proc. R. Soc. Lond. B (in press). LANDMESSER, L. (1971). Contractile and electrical responses of vagus-innervated frog sartorius muscle. J. Physiol. 213, 707-725. LANDMESSER,L. (1972). Pharmacological properties, cholinesterase activity, and anatomy of nerve-muscle junctions in vagus-innervated frog sartorius. J. Physiol. 220, 243-256. LANDMESSER, L. T. (1978a). The distribution of motoneurones supplying chick hind limb muscles. J. Physiol. 284, 371-389. LANDMESSER, L. T. (1978b). The development of motor projection patterns in the chick hind limb. J. Physiol. 284, 391~,14. LANDMESSER, L. T. (1979). Competitive interactions between developing cholinergic neurones. Pros. Brain Rex. 49, 373-384. LANDMESSER,L. T. (1980). The generation of neuromuscular specificity. Ann. Rer. Neurosei. 3, 279-302. LANDMESSER,L. T. and MORRIS, D. G. (1975). The development of functional innervation in the hind limb of the chick embryo. J. Physiol. 249, 301-326. LANGLEY, J. N. and ANDERSON, H. K. (1904). The union of different kinds of nerve fibers. J. Physiol. 31, 365-391. LAVOIE. P.-A., COLLIER,B. and TENENHOUSE,A. (1976). Comparison of alpha-bungarotoxin binding to skeletal muscles after inactivity or denervation. Nature. 260, 349-350. LAW, P. K. and ATWOOD, H. L. (1972). Nonequivalence of surgical and natural denervation in dystrophic mouse muscles. Exp. Neurol. 34, 200-209. LAW, P. K. and ATWOOD, H. L. (1974). Does axonal sprouting occur in dystrophic mouse muscles? Experentia, 30, 155-156. LAW, P. K., ATWOOD, H. L. and MCCOMAS, A. J. (1967). Functional denervation in the soleus muscle of dystrophic mice. Expl. Neurol. 51,434-443. LAW, P. K. and CACCIA,M. R. (1975). Physiological estimates of the sizes and numbers of motor units in soleus muscles of dystrophic mice. J. Neurol. Sei. 24, 251-256. LENTZ, T. L., MAZURKIEWICZ, J. E. and ROSENTHAL, J.. (1977). Cytochemical localization of acetylcholine receptors at the neuromuscular junction by means of horseradish peroxidase-labeled ~-bungarotoxin. Brain Res. 132, 423~t42. LETINSKY, M. (1974a). The development of nerve-muscle junctions in Rana catesheiana tadpoles. /)el'. Biol. 40, 129-153. LETINSKY, M. (1974b). Physiological properties of developing frog tadpole nerve-muscle junctions during repetitive stimulation. Def. Biol. 40, 154-161. LETINSKY, M. S., FISCHBACH,G. D. and McMAHAN, U. J. (1976}. Precision of reinnervation of original postsynaptic sites in muscle after a nerve crush. J. Neuroeytol. 5, 691-718. LETINSKY, M. S. and MORRISON-GRAHAM,K. (1980). Structure of developing frog neuromuscular junctions. J. Neurocytol. 9, 321-342. LETOURNEAU, P. C. (1978). Chemostatic response of nerve fiber elongation to nerve growth factor. Del. Biol. 66, 183-196. LEVAY. S.. STRYKER.M. P. and SHATZ, C. J. (1978). Ocular dominance columns and their development in layer IV of the cat's visual cortex. A quantitative study. J. Comp. Neurol. 179, 223-244. LEvI-MONTALCINI, R. (1949). The development of the acoustico-vestibular centers in the chick embryo in the absence of the afferent root fibers and of descending fiber tracts. J. Comp. Neurol. 91,209-242. LIBET, B., RALSTON,H. J. and FEINSTEIN,B. (1951). The effect of stretch on action potentials in muscle. Biol. Bull. 101, 194. LIaET, B. and WRIGHT, E. W. (1952). Facilitation at neuromuscular junctions by stretch of muscle. Fed. Proc. ! I, 94. LICHTMAN, J. W. (1977). The reorganization of synaptic connexions in the rat submandibular ganglion during post-natal development. J. Physiol. 273, 155 177. L1CHTMAN,J. W. (1980). On the predominantly single innervation of submandibular ganglion cells in the rat. J. Physiol. 302, 121-130. LICHTMAN, J. W. and PURVES,D. (1980). The elimination of redundant preganglionic innervation to hamster sympathetic ganglion cells in early postnatal life. J. Physiol. 301,213-228. LICHTMAN, J. W., PURVES, D. and YIP, J. W. (1980). Innervation of sympathetic neurons in the guinea pig thoracic chain. J. Physiol. 298, 285-299. LIEBERMAN. A. R. (1971). The axon reaction: a review of the principal features of peripheral responses to axon injury. Int. Ret'. Neurohiol. 14, 99-124. LILLEHEIL, G. and NAESS,K. 0961). A presynaptic effect of d-tubocurarine in the neuromuscular junction. Aeta Physiol. Scand. 52, 120-136. LINDSTROM, J. (1977). An assay for antibodies to human acetylcholine receptor in serum from patients with myasthenia gravis. Clin. Immunol. Immunopath. 7, 3~43. LINDSTROM, J. M. and LAMaERT, E. H. (1978). Content of acetylcholine receptor and antibodies bound to receptor in myasthenia gravis, experimental autoimmune myasthenia gravis, and Eaton-Lambert syndrome. Neurol.. Minneap. 28, 130-138. LIu, C. N. (1969). Comment following M. E. and A. B. Scheibel (1969). A structural analysis of spinal interneurons and Renshaw cells. In: The Interneuron, pp. 193-204. Univ. of Calif. Press, Berkeley. 276 ALAN D. GRINNELL AND ALBERT A. HERRERA LOMO, T. (1976). The role of activity in the control of membrane and contractile properties of skeletal muscle. In: Motor lnnervation of Muscle pp. 289-321. Ed. S. THESLEFF. Academic Press, New York. LOMO, T. and JANSEN, J. K. S. (1980). Requirements for the formation and maintenance of neuromuscular connections. Current Topics in Developmental Biology 16, 253 281. LOMO, T. and ROSENTHAL, J. (1972). Control of ACh sensitivity by muscle activity in the rat. J. Physiol. 221, 493-513. LOMO, T. and SLATER, C. R. (1978). Control of acetylcholine sensitivity and synapse formation by muscle activity, d. Physiol. 275, 391-402. LOMO, T. and SLATER, C. R. (1980a). Acetylcholine sensitivity of developing ectopic nerve-muscle junctions in adult rat soleus muscles. J. Physiol. 303, 173-189• LOMO, T. and SLATER, C. R. (1980b). Control of junctional acetylcholinesterase by neural and muscular influences in the rat. d. Physiol. 303, 191-202. LOMO, T. and WESTGAARD,R. H. (1975). Further studies on the control of ACh sensitivity by muscle activity in the rat. d. Physiol. 252, 603-626. LOMO, T., WESTGAARD, R. H. and DAHL, H. A. (1974). Contractile properties of muscle: control by pattern of muscle activity in the rat. Proc. R. Soc. Lond. B 187, 99-103• LUmNSRA, L. (1975). On axoplasmic flow. Int. Rev. Neurohiol. 17, 241-296. LUFF, A. R. and PROSKE, U. (1976). Properties of moior units of the frog sartorius muscle• J. Physiol. 258, 673-685. LUINE, V., NOTTEBOHM, F., HARDING, C., McEwEN, B. S. (19801. Androgen affects cholinerrgic enzymes in syringeal motoneurons and muscle. Brain Res. 192, 89-107. LUND, R. D. (1978). Development and Plasticity of the Brain, Oxford Univ. Press, New York. LEMAN, C. P. and O'BRIEN, R. C. (1969). Hyperresponsiveness in hibernation• Syrup. Soc. Exp. Biol. 23, 489-509. MAENO, T. (1969). Analysis of mobilization and demobilization process of neuromuscular transmission in the frog. J. Neurophysiol. 32, 793-800. MAGLEBY, K. L., PALLOTTA, B. S. and TERRAR D. A. (19.81). The effect of (+)-tubocurarine on neuromuscular transmission during repetitive stimulation in the rat, mouse, and frog. J. Physiol. 312, 97-113. MALLART, A., ANGAUT-PETIT,D., ZILBER-GACHELIN, N. F., TOM/,S I FERRL J. and HA1MANN, C. (1980). Synaptic efficacy and turnover of endings in pauci-innervated muscle fibres of Xenopus laevis. In: Ontogenesis and Functional Mechanisms of Peripheral Synapses, pp. 213-223. Ed. J. TAXI. I.N.S.E.R.M. Symp. No. 13. Elsevier Press, Amsterdam, North Holland. MARK, R. F. (1965). Fin movement after regeneration of neuromuscular connections: an investigation of myotypic specificity. Exp. Neurol. 12, 292-302. MARK, R, F. (1980). Synaptic repression at neuromuscular junctions. Physiol. Rev. 60, 355 395. MARK, R. F. and MAROT'rE, L. R. (t972). The mechanism o f selective reinnervation of fish eye muscles. III. Functional, electrophysiological and anatomical analysis of recovery from section of the lllrd and IVth nerves. Brain Res. 46, 131-148. MARK, R. F., M^ROTrE,. L. R. and MART, P. E. (1972). The mechanism of selective reinnervation of fish eye muscles. IV. Identification of repressed synapses. Brain Res. 46, 149-157• MAROTTE, L. and MARK, R. F. (1970a). The mechanism of selective reinnervation of fish eye muscle. I. Evidence from muscle function during recovery. Brain Res. 19, 41-51. MAROTTE, L. R. and MARK, R. F. (1970b). The mechanism of selective reinnervation of fish eye muscles. II. Evidence from electron microscopy of nerve endings. Brain Res. 19, 53-69. MARSHALL, L. M., SANES,J. R. and MCMAHAN, U. J. (1977). Reinnervation of original synaptic sites on muscle fiber basement membrane after disruption of the muscle cells. Proc. Nat. Acad. Sci. 74, 3073-3077. MARTIN, A. R. (1955). A further study of the statistical components of the end-plate potential, d. Physiol. 130, 114-122. MATTHEWS, M. R. and NELSON, V. (1975). Detachment of structurally intact nerve endings from chromatolytic neurons of rat superior cervical ganglion during the depression of synaptic transmission induced by post-ganglionic axotomy. J. Physiol. 245, 91-135. MATTHEWS, M. R. and RAISMAN,G. (1972). A light and electron microscopic study of the cellular response to axonal injury in the superior cervical ganglion of the rat. Proc. Roy. Soc. Lond. B 181, 43-79. MATTHEWS-BELLINGER, J. and SALPETER, M. M. (1978). Distribution of acetylcholine receptors at frog neuromuscular junctions with a discussion of some physiological implications. J. Physiol. 279, 197-213. MCARDLE, J. J. (1975). Complex endplate potentials at regenerating neuromuscular junctions of the rat. Exp. Neurol. 49, 629-638. MCARDLE, J. J. and ALBUQUERQUE, E. X. (1973). A study of the reinnervation of fast and slow mammalian muscles. J. Gen. Physiol. 61, 1-23. MCCANN, F. V., PETTENGILL, O. S., COLE, J. J., RUSSELL, J. A. G. and SORENSON, G. D. (1981). Calcium spike electrogenesis and other electrical activity in continuously cultured small cell carcinoma of the lung. Science, 212, 1155-1157. MCCLENNAN, I. S. and HENDRY, I. A. (1980). Parasympathetic neuronal survival induced by factors from muscle. Neurosci. Lett. 10, 269-273. MCCOMAS, A. J. and MROZr~K. K. (1967). Denervated muscle fibres in hereditary mouse dystrophy. J. Neurol. Neurosurg. Psychiat. 30, 526-530. MCGRATH, P. A. and BENNETT, M. R. (1979). The development of synaptic connections between different segmental motoneurones and striated muscles in an axolotl limb. DeE. Biol. 69, 133-145. MCKAV, R., RAFF, M. C., REICHARDT, L. F. (Eds.) (1981). Monoclonal antibodies to Neural Antigens. Cold Spring Harbor Reports in the Neurosciences, Vol. 2., Cold Spring Harbor Laboratory, New York. MCLACHLAN, E. M. and MARTIN, A. R. (1981).Non-linear summation of end-plate potentials in the frog and mouse. J. Physiol. 311,307-324. MCMAHAN, U. J., S^NES, J. R. and MArtSnALL, L. M. (1978). Cholinesterase is associated with the basal lamina at the neuromuscular junction. Nature, 271, 172-174. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 277 MELICHAR, I., BROT,EK, G., JANSK~', L. and VYSKO~'IL. F. (1973). Effect of hibernation and noradrenaline on acetylcholine release and action at neuromuscular junctions of the golden hamster (Mesocricetus auratus). Pfliigers Arch. 345, 107-122. MELICHNA, J. GUTMANN, E., HERBRYCHOVA,A. and STICHOVA, J. (1972). Sexual dimorphism in contraction properties and fibre pattern of the flexor carpi radialis muscle of the frog (Rana temporaria). Experientia, 28, 89-91. MENDELL, L. M., MUNSON, J. B. and SCOTT, J. G. (1974). Connectivity changes of la afferents on axotomized motoneurons. Brain Res. 73, 338-342. MENDELL, L. M., MUNSON,J. B. and SCOTT. J. G. (1976). Alterations of synapses on axotomized motoneurones. d. Physiol. 255, 67-69. MENDEEE, L. M. and SCOTT, J. G. (1975). The effect of peripheral nerve cross-union on connections of single la fibers to motoneurons. Exp. Brain Res. 22, 221-234. MILEDL R. (19601. The acetylcholine sensitivity of frog muscle fibres after complete or partial denervation. J. Physiol. ISI, 1-23. MILEDI, R. (1963L Formation of extra nerve-muscle junctions in innervated muscle. Nature 199, 1191-1192. MILEDI, R., MOLENAAR,P. and POEAK, R. (1978}. :t-Bungarotoxin enhances transmitter "released" at the neuromuscular junction. Nature, 272, 641-642. MIEED1, R. and SEATER,C. R. 11968). Electrophysiology and electron microscopy of rat neuromuscular junctions after nerve degeneration. Proc. R. Soc. Lond. B 169, 289 306. MIEEDI, R. and STEFANI,E. (1969). Nonselective re-innervation of slow and fast muscle fibres in the rat. Nature, 222, 569-571. M1LEDI, R., STEFANI, E. and SrEINBACH, A. B. (1971). Induction of action potential mechanism in slow muscle fibres of the frog. J. Physiol. 217, 737-754. MIYAMOTO, M. D. (19781. The actions of cholinergic drugs on motor nerve terminals. Pharmacol. Rev. 29, 221-247. MIYAMOTO, M. D. and BRECKENRIDGE,B. M. (1974). A cyclic adenosine monophosphate link in the catecholamine enhancement of transmitter release at the neuromuscular junction. J. Gen. Physiol. 63, 61)9-624. MIYATA, Y. and YOSHIOKA, K. (1980}. Selective elimination of motor nerve terminals in the rat soleus muscle during development. J. Physiol. 309, 631-646. MOLENAAR, P. C., POLAK, R. L., MILED1, R., ALEMA,S., VINCENT.A. and NEWSOM-DAVlS,J. (1979). Acetylcholine in intercostal muscle from myasthenia gravis patients and in rat diaphragm after blockade of acetylchofine receptors. Pro O. Brain Res. 49, 449~458. MORAVEC, J., MELICHAR, i.. JANSK~', L., VYSKO~'IL,F. (1973). Effect of hibernation and noradrenaline on the resting state of neuromuscular junction of golden hamster (Mesocricetus auratusL Pfliigers Arch. 345, 93-106. MORAVEC. J. and VYSKO(~IL,F. (1976). Neuromuscular transmission in hibernators. In: Regulation qfDepre.s,wd Metabolism and Thermogenesis. pp. 81-92. Ed. L. JANSKY and J. MUSACCHIAC. C. Thomas Publishers, Springfield. MORRIS, D. G. (1978L Development of functional motor innervation in supernumerary hindlimbs of the chick embryo. J. Neurophysiol. 41, 1450-1465. MORRISON-GRAHAM, K. (1981). Synapse elimination at the developing frog neuromuscular junction. Ph.D. Thesis, Neurosciences Program, UCLA Brain Research Institute. MOSCONA, A. A. (Ed.) (1974). The Cell Surface in Development. Wiley, New York. MOVSHON, J. A. (1976). Reversal of the physiological effects of monocular deprivation in the kitten's visual cortex. J. Physiol. 261, 125-174. MULLER, E. R. A., GALAVAZl.G., SZIRMAI,J. A. (1969). Effect of castration and testosterone treatment on fiber width of the flexor carpi radialis muscle in the male frog (Rana temporaria). Gen. Comp. Emlocrin. 13, 275-284. NACriMANSOriN, D. (1976). Transduction of chemical into electrical energy. Proc. Nat. Acad. Sci. 73, 82 85. NARAVANAN, C. H. and NARAVANAN,Y. 11978}. Neuronal adjustments in developing nuclear centres of the chick embryo following transplantation of an additional optic primordium. J. Embryol. Exp. Morph. 44, 53-70. NASTUK, W. L. (1971). Mechanisms of neuromuscular blockade. Ann. N.Y. Acad. Sci. 183, 171-182. NELSON, P. G. (1969). Functional consequences of tenotomy in hind limb muscles of the cat. J. Physiol. 201, 321-333. NELSON, P. G. (1975). Nerve and muscle cells in culture. Physiol. Rev. 55, 1-61. NiSrll, R. and BERG, D. K. (1977). Dissociated ciliary ganglion processes in vitro: Survival and synapse formation. Proc. Nat. Acad. Sci. 74, 5171-5175. NIsrtL R. and BERG, D. K. (1979}. Survival and development of ciliary ganglion neurones grown alone in cell culture. Nature, 277, 232-234. NISHI, R. and BERG, D. K. (1981). Two components from eye tissue that differentially stimulate the growth and development of ciliary ganglion neurons in cell culture. J. Neurosci. I, 505-513. NIXON, B., JACKSON, P., DIAMOND, A., FOERSTER,A. and DIAMOND, J. (1980). Impulse activity evokes collateral sprouting of intact nerves into available target tissue. Soc. for Neurosci. A hstr. 6, 171. NORRIS, F. H., Izzo, A. J. and GARVEY, P. H. (1965). Brief report: Tumor size and Lambert-Eaton syndrome. In: The Remote Effects of Cancer on the Nervous system pp. 81-82. Eds. LORD BRAIN and F. H. NORRIS. Grune and Stratton, New York. NUDELL, B. (1981L Inverse relationship between terminal length and release per unit length in frog neuromuscular junctions on fibers of uniform input resistance. Soc. Neurosci. Abst. 7, 710. NYSTR6M. B. ~1968). Postnatal development of motor nerve terminals in "slow red" and "fast white" cat muscles. Acta Neurol. Scand. 44, 363-383. OBATA, K. and TANAKA. H. (1980). Conditioned medium promotes minute outgrowth from both central and peripheral neurones. Neurosei. Lett. 16, 27-33. 278 ALAN D. GRINNELL AND ALBERT A. HERRERA O'BRIEN, R. A. D., ()STBERG, A. J. C. and VRBOV.~,G. (1978). Observations on the elimination of polyneuronal innervation in developing mammalian skeletal muscle. J. Physiol. 282, 571-582. O'BRIEN, R. A., t~STBERG, A. J., VRBOV~,G. (1980). The effect of acetylcholine on the function and structure of the developing mammalian neuromuscular junction. Neuroscience, 5, 1367-1379. O'BRIEN, R. A., PURVES, R. D. and VRBOV/,, G. (1977). Effect of activity on the elimination of multiple inervation in soleus muscles of rats. J. Physiol. 271, 54P-55P. OLEK, A. J. (1980). Effects of alpha and beta bungarotoxin on motor neuron loss in Xenopus larvae. Neuroscience, :5, 1557-1563. OLIVEIRA, A. C., DESHPANDE,S. S. and ALBUQUERQUE.E. X. (1978). Effects of temperature and denervation on endplate currents of awake and hibernating ground squirrels. Fed. Proc. 37, 525. ONTELL, M. and DUNN, R. F. (1978). Neonatal muscle growth: a quantitative study. Am. J. Anat. 152, 539 556. OPPENHEIM, R. W. (1981). Cell death of motoneurons in the chick embryo spinal cord. V. Evidence on the role of cell death and neuromuscular function in the formation of specific peripheral connections. J. Neurosei. I, 141 151. OPPENHEIM, R. W. and MAJORS-WILLARD,C. (1978). Neuronal cell death in the brachial spinal cord of the chick is unrelated to the loss of polyneuronal innervation in wing muscle. Brain Res. 154, 148 152. ORBELLI, L. A. (1923). Die sympathische innervation der Skelettmuskeln. Bull. Inst. sci. Leshaft. 6, 194-197. ORKAND, R. K. (1963). A further study of electrical responses in slow and twitch fibres of the frog. J. Physiol. 167, 181-191. ORKAND, R. K. (1980). Extracellular potassium accumulation in the nervous system. Fed. Proc. 39, 1515 1518. ORKAND, R. K., NICHOLLS,J. G. and KUFFLER,S. W. (1966). Effect of nerve impulses on the membrane potential of glial cells in the central nervous system of amphibia. J. Neurophys. 29, 788-806. OTSUKA, M., ENDO, M. and NONOMURA,Y. (1962). Presynaptic nature of neuro-muscular depression. Jap. J. Physiol. 12, 573-584. PARRY, D. J. and MELENCHUK, S. (1981). Rate and extent of functional reinnervation in fast-twitch and slow-twitch muscles of the dystrophic mouse. (C57Bl/6Jdy2j/dy2j). Exp. Neurol. 72, 446-461. PATON, W. D. M. and WAUD, D. R. (1967). The margin of safety of neuromuscular transmission. J. Physiol. 191, 59 90. PATRICK, J., HEINEMANN,S. and SCHUBERT,D. (1978). Biology of cultured nerve and muscle. Ann. Rer. Neurosci. I, 417~143. PESTRONK, A. and DRACHMAN,D. B. (1978). Motor nerve sprouting and acetylcholine receptors. Science, 199, 1223 1225. PESTRONK, A., DRACHMAN,D. B. and GRIFFIN, J. W. (1976). Effect of muscle disuse on acetylcholine receptors• Nature, 260, 352-353. PESTRONK, A., DRACHMAN, D. B. and GRWFIN, J. W. (1980). Effects of aging on nerve sprouting and regeneration. Exp. Neurol. 70, 65-82. PETTIGREW, A. G., LINDEMAN,R. and BENNETT, M. R. (1979). Development of the segmental innervation of the chick forelimb. J. Embryol. Exp. Morph. 49, 115 137. PFENNINGER, K. H. and MAVLI~:-PFENN1NGER.M.-F. (1978). Characterization, distribution, and appearance of surface carbohydrates on growing neurites. In: Neuronal Information Transfer, pp. 373-386. Eds. A. KARLIN, V. M. TENNYSON and H. J. VOGELAcademic Press, New York. PFENN1NGER, K. H. and MAYLII~-PFENNINGER,M.-F. (1981). Lectin labeling of sprouting neurons. I. Regional distributions of surface glycoconjugates. J. Cell. Biol. 89, 536 546. PILAR, G. and LANDMESSER,L. (1972). Axotomy mimicked by localized colchicine application. Sciem'e, 177, Ill6-1118. PILAR, G., LANDMESSER,L. T. and BURSTEIN, L. (1980). Competition for survival among developing ciliary ganglion cells. J. Neurophys. 43, 233-254. PITTMAN, R. H. and OPPENHE1M, R. W. (1978). Neuromuscular blockade increases motoneurone survival during normal cell death in the chick embryo. Nature, 271,364-366. POBERAI, M., SJtVAY,G. and CSILLIK, B. (1972). Function-dependent proteinase activity in the neuromuscular synapse. Neurobiolo#y, 2, 1-7. POCKETT. S., SANDSET. P. M. and JANSEN,J. K. S. (1979). Developmental changes in the pattern of innervation of focally innervated chick skeletal muscle. Acta Physiol. Seand. 105, 19-20A. PRESTIGE, M. C. (1967). The control of cell number in the lumbar ventral horn during the development of Xenopus laeuis tadpoles. J. Embryol. Exp. Morph. 18, 359-387. PRESTIGE, M. C. (1970). Differentiation, degeneration, and the role of the periphery: quantitative considerations. In: The Neurosciences: Second Study Pro#ram, pp. 73-82. Ed. F. O. SCHMITT. Rockefeller University Press, New York. PRESTIGE, M. C. (1973). Gradients in the time of origin of tadpole motoneurones. Brain Res. 59, 400-404. PURVES, O. (1975). Functional and structural changes in mammalian sympathetic neurones following interruptions of their axons. J. Physiol. 252, 429-463. PURVES, D. (1976a). Functional and structural changes in mammalian sympathetic neurones following colchicine application to postganglionic nerves. J. Physiol. 259, 159-175. PURVES, D. (1976b). Long-term regulation in the vertebrate peripheral nervous system. Int. Rev. of Physiol. 10, 125-177. PURVES, D. (1976c). Competitive and non-competitive re-innervation of mammalian sympathetic neurones by native and foreign fibres. J. Physiol. 261, 453-475. PURVES, D. and HUME, R. I. (1981). The relation of postsynaptic geometry to the number of presynaptic axons that innervate autonomic ganglion cells. J. Neurosci. 1, 441-452. PURVES, D. and LICrtTMAN, J. W. (1978). Formation and maintenance of connections in autonomic ganglia. Physiol. Ret,. 58, 821-862. PURVES, D. and LICHTMAr4, J. W. (1980a). Elimination of synapses in the developing nervous system. Science, 210, 153-157. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 279 PURVE& D. and LICHTMAN, J. W. (1980b). The elimination of some synaptic connections in autonomic ganglia during early postnatal life. In: Ontogenesis: Funetional Mechanisms of Peripheral Synapses, pp. 15-26. Ed. J. TAXl. Elsevier, Amsterdam, North Holland. PURVES, D. and NJL A. (1976). Effect of nerve growth factor on synaptic depression after axotomy. Nature, 260, 533-536. PURVES, D. and THOMPSON, W. (1979). The effects of postganglionic axotomy on selective synaptic connexions in the superior cervical ganglion of the guinea pig. J. Physiol. 297, 95-110. RACE, J., JR. (1961). Thyroid hormone control of development of lateral motor columnar cells in the lumbosacral cord in hypophysectomized Rana pipiens. Gen. Comp. Endocrinol. 1, 322-331. RAKIC, P. (1977). Prenatal development of the visual system in rhesus monkey. Phil. Trans. R. Soc. B 278, 245-260. RALSTON, H. J. and LIBET, B. (1953). Effect of stretch on action potential of voluntary muscle. Amer. J. Physiol. 173, 449-455. REDFERN, P. A. (1970). Neuromuscular transmission in newborn rats. J. Physiol. 209, 701-709. REICHERT, R. and ROTSHENgER, S. (1979). Motor axon terminal sprouting in intact muscles. Brain Res. 170, 187-189. RENAUD, D., LEDOUARIN, G. H. and KHAS~IYE A. (1978). Spinal cord stimulation in chick embryo: effects on development of the posterior latissimus dorsi muscle and neuromuscular junctions. Exp. Neurol. 60, 189-200. RICHARDSON, R. L., GRECO, F. A., OLDHAM,R. K., LIDDLE,G. W. (1978). Tumor products and potential markers in small cell lung carcinoma. Semin. Oncol. 5, 253-262. RIDGE, R. M. A. P. and THOMSON, A. M. (1980). Electrical responses of muscle fibres in a small foot muscle of Xenopus laevis. J. Physiol. 306, 41-49. RmER, W. F. and OKAMOTO, M. (1969). Pharmacology of motor nerve terminals. A. Rev. Pharmacol. 9, 173-208. RILEY, D. A. (1977). Spontaneous elimination of nerve terminals from the endplates of developing skeletal myofibers. Brain Res. 134, 279-285. ROBBINS, N. (1974). Long term maintenance and plasticity of the neuromuscular junction. In: The Neurosciences: Third Stud)" Pro#ram, pp. 953-960. Eds. F. O. SCHMITT and F. G. WORDEN. M.I.T. Press, Cambridge. ROBBINS. N. (19801. Plasticity at the mature neuromuscular junction. Trends in Neurosci. May 1980: 120-122. ROBBINS, N., ANTOSIAK, J., GERDING, R. and UCHITEL, O. D. 11977). Nonacceptance of innervation by innervated neonatal rat muscle. Def. Biol. 61, 166-176. ROBBINS, N. and FISCHBACH, G. D. (1971). Effect of chronic disuse of rat soleus neuromuscular junctions on presynaptic function. J. Neurophysiol. 34, 570--578. ROBERT, E. D. and OESTER, Y. T. (1970). Absence of supersensitivity to acetylcholine in innervated muscle subjected to a prolonged pharmacologic nerve block. J. Pharmacol. Exp. Ther. 174, 133-140. ROMANES,G. J. (1946). Motor localization and the effects of nerve injury on the ventral horn cells of the spinal cord. J. Anat. 80, 117-131. ROMANES,G. J. (1964). The motor pools of the spinal cord. In: Pro#ress in Brain Research II, pp. 93-119. Eds. J. C. ECCLES and J. P. SHADE. Elsevier, New York. RONG, X. and XING, W. (1981). Changes in the electrophysiological properties of the spinal motoneurones after intramuscular injection of botulinum toxin. Scientia Sinica (in press). RONNEVl, L.-O. (1977). Spontaneous phagocytosis of boutons on spinal motoneurons during early postnatal development. An electron microscopical study in the cat. J. Neurocytol. 6, 487-504. RONNEVI, L.-O. and CONRADI, S. (1974). Ultrastructural evidence for spontaneous elimination of synaptic terminals on spinal motoneurons in the kitten. Brain Res. 80, 335-359. ROOKE, E. D., EATON, L. M., LAMBERT,E. H. and HODGSON, C. H. (1960). Myasthenia and malignant intrathoracic tumor. Med. Clin. N. Amer. 44, 977-988. ROSENTrlAL, J. L. (1977). Trophic interactions of neurons. In: Handbook of Physiolo#y. The Nervous System, Sect. 1, Vol. 1, pp. 775-801. Am. Physiol. Soc., Bethesda, MD. ROSENTHAL, J. L. and TARASKEVICH,P. S. (1977). Reduction of multiaxonal innervation of the neuromuscular junction of the rat during development, d. Physiol. 270, 299-310. ROTSHENKER, S. (1978a). De novo formation of synapses on innervated muscle fibers. Israel J. Med. Sci. 14, 1002. ROTSHENKER, S. (1978b). Sprouting of intact motor neurons induced by neuronal lesion in the absence of denervated muscle fibers and degenerating axons. Brain Res. 155, 354-356. ROTSHENKER. S. (1979). Synapse formation in intact innervated cutaneous-pectoris muscles of the frog following denervation of the opposite muscle. J. Physiol. 292, 535-547. ROTSHENKER, S. (1980). Sprouting and synapse formation produced by colchicine. Soc. for Neurosei. Abstr. 6, no. 59.1. ROTSHENKER, S. and MCMAHAN, U. J. (1976). Altered patterns of innervation in frog muscle after denervation. J. Neurocytol. 5, 719-730. ROTSHENKER, S. and REICHERT, F. (1980). Motor axon sprouting and site of synapse formation in intact innervated skeletal muscle of the frog. J. Comp. Neurol. 193, 413-422. SAITO, A. and ZACKS, S. I. (1969). Fine structure of neuromuscular junctions after nerve section and implantation of nerve in denervated muscle. Exp. Mol. Pathol. 10, 256-273. SANES, J. R. and HALL, Z. W. (1979). Antibodies that bind specifically to synaptic sites on muscle fiber basal lamina. J. Cell Biol. 83, 357-370. SANES, J. R., MARSHALL, L. M. and MCMAHAN, U. J. (1978). Reinnervation of muscle fiber basal lamina after removal of myofibers. Differentiation of regenerating axons at original synaptic sites. J. Cell Biol. 78, 176-198. SANTA, T., ENGEL, A. G. and LAMBERT, E. H. (1972a). Histometric study of neuromuscular junction ultrastructure. I. Myasthenia gravis. Neurolo#y (Minneap.) 22, 71-82. 280 ALAN D. GRINNELL AND ALBERT A. HERRERA SANTA, T., ENGEL, A. G. and LAMBERT, E. H. (1972b). Histometric study of neuromuscular junction ultrastructure. II. Myasthenic syndrome. Neurology (Minneap.) 22, 370-376. SAR, M. and STUMPF,W. E. (1977). Androgen concentration in motor neurons of cranial nerves and spinal cord. Science, 197, 77-79. SARGENT. P. B. and DENNIS, M. J. (1981). The influence of normal innervation upon abnormal synaptic connections between frog parasympathetic neurons. Dev. Biol. 81, 65-73. SCHMIDT, H. and STEFANI, E. (1976). Reinnervation of twitch and slow muscle fibres of the frog after crushing the motor nerves. J. Physiol. 2f~, 99-123. SCHMtDT, H. and STEFANI, E. (1977). Action potentials in slow muscle fibres of the frog during regeneration of motor nerves. J. Physiol. 270, 507-517. SCHNEIDER, G., RUBINSTEIN.N., ZEMLAN,F. P. and ERULKAR,S. D. (1980). Hormonal modulation of central and peripheral components of "'fast" and "'slow" skeletal nerve-muscle systems in Xenopus laet'is. Soc. for Neurosci. Ahstr. 6, no. 189.3. SCOTT, S. A. (1975). Persistence of foreign innervation of reinnervated goldfish extraocular muscles. Science, 189, 644-646. SCOTT, S. A. (1977). Maintained function of foreign and appropriate junctions on reinnervated goldfish extraocular muscles. J. Physiol. 268, 87-109. SHALTON. P. M. and WAREHAM,A. C. (1980). Some factors affecting spontaneous transmitter release in dystrophic mice. M,scle aml Ner~'e 3, 120-127. SHARPLESS,S. K. (1964). Reorganization of function in the nervous system--use and disuse. Ann. Rev. Physiol. 26, 357-388. SHATZ, C. J. and STRYKER,M. P. (1978). Ocular dominance in layer IV of the cat's visual cortex and the effects of monocular deprivation. J. Physiol. 281,267-283. SHIMADA, Y. and FISCHMAN, D. A. (1973). Morphological and physiological evidence for the development of functional neuromuscular junctions in vitro. Dev. BioL 31,200-225. SHOREY, M. L. (1909). The effects of the destruction of the peripheral areas on the differentiation of the neuroblasts. J. Exp. Zool. 7, 25-63. SHUKLA, P. L. and AITREN, J. T. (1963). Formation of motor endplates in denervated voluntary muscles of the rat. J. Anat. 97, 152P. SLACK, J. R. {1978). Interaction between foreign and regenerating axons in axolotl muscle. Brain Res. 146, 172-176. SLACK, J. R.. HOPKINS, W. G. and WILLIAMS,M. N. (1979). Nerve sheaths and motoneurone collateral sprouting. Nature, 2112, 506-507. SMITH, D. O. (1979). Reduced capabilities of synaptic transmission in aged rats. Exp. Neurol. 66, 650-666. SMOLEN, A. and RAISMAN,G. (1980). Synapse formation in the rat superior cervical ganglion during normal development and after neonatal deafferentation. Brain Res. 181,315-323. SNtDER, W. D. and HARRIS,G. L. (1979). A physiological correlate of disused-induced sprouting at the neuromuscular junction. Nature, 281, 69-71. SNVDER, D. H., RIFENBERICg, D. H. and MAX, S. T. (1973). Effects of neuromuscular activity on choline acetyltransferase and acetylcholinesterase. Exp. Neurol. 40° 36-42. SOLLER, R. W. and ERULgAR, S. D. (1978). Morphological basis for crossed motoneuronal interactions in frog spinal cord. Soc. for Neurosci. Abstr. 4, no. 1069. SORENSON, G. D., PETTENGILL, O. S., BRINCK-JOHNSEN,T., CARE, C. C. and MAURER, L. H. [1981). Hormone production by cultures of small-cell carcinoma of the lung. Cancer 47, 1289-1296. SOUTH, F. E. (1961). Phrenic nerve-diaphragm preparations in relation to temperature and hibernation. Amer. J. Physiol. 200, 565-571. SPERRY, R. W. (1954). The problem of central nervous reorganization after nerve regeneration and muscle transposition. Quart. Rec. Biol. 20, 311-369. SPERRY, R. W. 11963). Chemoaffinity in orderly growth of nerve fiber patterns and connections. Proc. Nat. Acad. Sci. 50, 703-710. SPERRY. R. W. and ARORA, H. L. (1965). Selectivity in regeneration of the oculomotor nerve in the cichlid fish, Astronotus ocellatus. J. Erabryol. Exp. Morphol. 14, 307-317. SPITZER, N. [1972). Miniature end-plate potentials at mammalian neuromuscular junctions poisoned by botulinum toxin. Nature, 237, 26-27. SRIHARI. T. and VRBOVX,G. (1978). The role of muscle activity in the differentiation of neuromuscular junctions in slow and fast chick muscles. J. Ne,rocytol. 7, 529 540. STANDAERT, F. G. (1964). The action of d-tubocurarine on the motor nerve terminal. J. Pharmacol. Exp. Ther. 143, 181-186. STEtNBACH, J. H. (1981). Neuromuscular junctions and ,y-bungarotoxin-binding sites in denervated and contralateral cat skeletal muscles. J. Physiol. 313, 513-528. STEPHENSON, R, S. (1979). Axon reflexes in axolotl limbs: evidence that branched motor axons reinnervate muscle selectively. Exp. Neurol. 64, 174-189. STmLING, R. V. and SUMMERBELL, D. (1977). The development of functional innervation in the chick wing-bud following truncations and deletions of the proximo-distal axis. J. Embryol. Exp. Morph. 41, 189-207. STIRL1NG.R. W. aand SUMMERBELL,D. 11979]. The segmentation of axons from the segmental nerve roots to the chick wing. Nat,re. 278, 640-642. STgAZNICKY, K. {1963). Function of heterotopic spinal cord segments investigated in the chick. Acta. Biol. Acad. Sci. Hung. 14, 145-155. SULLIVAN,G. E. (1962). Anatomy and embryology of the wing musculature of the domestic fowl (Gallus). Aust. J. Zool. 10, 458-518. SUMNER, B. E. H. (1975). A quantitative analysis of the response of presynaptic boutons to postsynaptic motor neuron axotomy. Exp. Neurol. 46, 605-615. SPECIFICITY AND PLASTICITYOF NEUROMUSCULARCONNECTIONS 281 SUMNEg, B. E. H. and SUTHERLAND,F. I. (1973). Quantitative electron microscopy on the injured hypogiossal nucleus in the rat. J. Neurocytol. 2, 315-328. SZI~KELY,G. (1976). The morphology of motoneurons and dorsal root fibers in the frog's spinal cord. Brain Res. 103, 275-290. SZ~KELY, G. and Czr~n, G. 11967). Localization of motoneurons in the limb moving spinal segments of Amblystoma. Acta Physiol. Acad. Sci. Hung. 32, 3-18. TABARY, J. C., TABARY, C., TARDIEU, C., TARDIEU, G. and GOLDSPINK, G. 11972). Physiological and structural changes in the cat's soleus muscle due to immobilization at different lengths by plaster casts. 2. Physiol. 224; 231-244. TADA, K., OHSHITA, S., YONENOBU, K., SATOH, K. and SHIMIZU, N. (1979). Experimental study of spinal nerve repair after plexus brachialis injury in newborn rats: a horseradish peroxidase study. Exp. Neurol. 65, 301-314. TAKANO, KI 11976). Absence of the gamma-spindle loop in the reinnervated hind leg muscles of the cat: "alpha-muscle". Exp. Brain Res. 26, 343-354. TAMAKI, K. (1933). The effect of unilateral section of the peroneal nerve of the albino rat on the number of myelinated fibres in the intact nerve of the opposite side. Anat. Rec. 56, 219-228. TAMAKI, K. (1936). Further studies on the effect of section of One peroneal nerve of the albino rat on the intact nerve of the opposite side. J. Comp. Neurol. 64, 437--448. TANG, C.-MI, COHEN, M. W. and ORKANO, R. K. 11980). Electrogenic pumps in axons and neuroglia and extraceUular potassium homeostasis. Brain Res. 194, 283-286. THOMPSON, W. 11978). Reinnervatidn of partially denervated rat soleus muscle. Acta Physiol. Scand. 103, 81-91. THOMPSON, W. and JANSEN, J. K. S. 11977). The extent o f sprouting of remaining motor units in partly denervated immature and adult rat soleus muscle. Neuroscience, 2, 523-535. THOMPSON, W., KUFFLER, D. P. and JANSEN, J. K. S. 11979). The effect of prolonged reversible block of nerve impulses on the elimination of polyneuronal innervation of new-born rat skeletal muscle fibres. Neuroscience, 4, 271-281. THOMSEN, P. and Luco, J. V. (1944). Changes of weight and neuromuscular transmission in muscles of immobilized joints. 3, Neurophysiol. 7, 245-251. TONGL D. A. 11974). Physiological characteristics of re-innervation Of skeletal muscle in the mouse. J. Physiol. 241, 141-153. TORVIK, A. E. and Srd6RTEN, F. 11971). Electron microscopic observations on nerve cell regeneration and degeneration after axon lesions. I. Changes in the nerve cell cytoplasm. Acta Neuropathol. 17, 248-264. TOUTANT, M., TOUTANT, J. P., RENAUD, D. and LE DOUARIN, G. (1981). Effects of spinal cord stimulation on the differentiation of posterior latissimus dorsi nerve in the chicken embryo. Exp. Neurol. 72, 267280. TOYKA, K. V., DRACfiMAN,D. B., GRIFFIN, D. E., PESTRONK, A., WINKELSTEIN,J. A., FISCHBECK, K. H. and KAO, I. (1977). Myasythenia gravis: study of hormonal immune mechanisms by passive transfer in mice. N. Eng. d. Med. 296, 125-131. TSUKAHARA, N. (1981). Synaptic plasticity in the mammalian central nervous system. Ann. Rev. Neurosci. 4, 351-379. TU~EK, S., KOSTIOVA, D. and GUTM^NN, E. (1976). Testosterone-induced changes of choline acetyltransferase and cholinesterase activities in rat levator ani muscle. J. Neurol. Sci. 27, 353-362. TUFFERY, A. R. (1971). Growth and degeneration of motor end-plates in normal cat hind limb muscles. J. Anat. 110, 221-247. TURKANIS, S. A. (1973). Effects of muscle stretch on transmitter release at endplates of rat diaphragm and frog sartorius muscle• J. Physiol. 230, 391-403. TWEEDLE, C. D'. and STEPrlENS K. S. (1980). Development of complexity in rat neur0muscu-iarjunctions. Soc.for Neurosci. Abstr. 6, no. 277.1. TWEEDLE, C. n. and STEPHENS, K. E. (1981). Development of complexity in motor nerve endings at the rat neuromuscular junction. Neuroscience (in press). VENABLE, J. H. (1966). Morphology of the cells of normal, testosterone-deprived and testosterone-stimulated levator ani muscles. Am. J. Anat. 119, 271-302. VIZl, E. S. (1979). Presynaptic modulation of neurochemical transmission. Prog. Neurobiol. 1~, 181-290. VRBOVA, G. (1963). Changes in motor reflexes produced by tenotomy. J. Physiol. 166, 241-267. VRBOVA, G., GORDON, T. and JONES, R. (1978). Nerve-Muscle Interaction. Chapman and Hall, London• VRBOVA, G. and WAREHAM, A. C. (1976). Effects of nerve activity on the postsynaptic membrane of skeletal muscle• Brain Res. 118, 371-382. VYSKOgTIL, F. (1976). Miniature end-plate potentials and sensitivity to acetylcholine in the fast and slow limb muscles of hibernating golden hamsters• Pfliigers Arch. 361, 165-167. VYSKOt~IL, F. and GUTMANN, E. (1969). Spontaneous transmitter release from motor nerve endings in muscle fibres of castrated and old animals. Experientia, 25, 945-946. VYSKO~IL, F. and GUTMANN, E. (1972). Spontaneous transmitter release from nerve endings and contractile properties in the soleus and diaphragm muscles of senile rats. Experientia, 28, 280-281. VYSKO(?IL, F. and GUTMANN, E. (1977). Electrophysiological and contractile properties of the levator ani muscle after castration and testosterone administration. Pfliigers Arch. 368, 105-109. VYSKO~'IL, F. and MORAVEC, J. (1978). Effect of 2,4-dinitrophenol and rotenone on frequency of miniature end-plate potentials of hibernating golden hamsters. Physiol. Bohemoslav. 27, 94-96. VYSKO~IL, F., MORAVEC,J. and JANSK'~',L. (1971). Resting state of the myoneural junction in a hibernator. Brain Res. 34, 381-384. VYSKOt~tL, F., VYKLICK~', L. and'HuSTON, R. (1971). Quantum content at the neuromuscular junction of fast muscle after cross-union with the nerve of slow muscle in the chick. Brain Res. 26, 443-445. WAKABAYASHI,T. and IWASAKI,S. (1962). Neuromuscular facilitation and asthenic synapses. Jap. J. Physiol. 12, t-13. 282 ALAN D. GRINNELL AND ALBERT A. HERRERA WARNtCK, J. E. and ALBUQUERQUE,E. X. (1979). Changes in genotypic expression, development and the effects of chronic penicillamine treatment on the electrical properties of the posterior latissimus dorsi muscle in two lines of normal and dystrophic chickens. Exp. Neurol. 63, 135-162. WARNICK, J. E., LEBEDA, F. J. and ALBUQUERQUE,E. X. (1979). Junctional and extrajunctional aspects of inherited muscular dystrophy in chickens: development and pharmacology. Ann. N . Y Acad. Sci. 317, 263-284. WATSON, W. E. (1965). An autoradiographic study of the incorporation of nucleic acid precursors by neurones and glia during nerve regeneration. J. Physiol. 180, 741-753. WATSON, W. E. (1968). Observations on the nucleolar and total cell body nucleic acid of injured nerve cells. J. Physiol. 196, 655-676. WATSON, W. E. (1969). The response of motor neurones to intramuscular injecton of botulinum toxin. J. Physiol. 202, 611-630. WATSON, W. E. (1970). Some metabolic responses of axotomized neurones to contact between their axons and denervated muscle. J. Physiol. 210, 321 343. WATSON, W. E. (1972). Some quantitative observations upon the response of neuroglial cells which follow axotomy of adjacent neurones. J. Physiol. 225, 415-435. WATSON, W. E. (1974). Cellular responses to axotomy and to related procedures. Brit. Med. Bull. 30, 112-115. WAUD, D. R. (1971). A review of pharmacological approaches to the acetylcholine receptors at the neuromuscular junction. Ann. N.Y. Acad. Sci. 183, 147-157. WEAKLY, J. N. (1978). Similarities in synaptic efficacy along multiply innervated twitch muscle fibers of the frog: a possible muscle-to-motoneuron interaction. Brain Res. 158, 235-239. WEAKLY, J. N. and YAO, Y. M. (1981). Section of lumbar spinal roots fails to induce synapse formation in contralateral, innervated sartorius muscles of the frog. Brain Res. 2114, 421-423. WEINBERG,C. B., REINESS, C. G. and HALL, Z. W. (1981). Topographical segregation of old and new acetylcholine receptors at developing ectopic endplates in adult rat muscle. J. Cell Biol. 88, 215-218. WEINSTEIN,S. P. (1980). A comparative electrophysiological study of motor endplate diseased skeletal muscle in the mouse. J. Physiol. 307, 453-464. WEISS, P. (1936). Selectivity controlling the central-peripheral relations in the nervous system. Biol. Rer. II, 494-531. WEISS, P. (1941). Self-differentiation of the basic patterns of coordination. Comp. Psychol. Monogr. 17, 1 96. WEISS, P. (1950). Experimental analysis of coordination by the disarrangement of central-peripheral relations. Syrup. Soc. Exp. Biol. 4, 92 I I I. WEISS, P. and EDDS, M. V. (1946). Spontaneous recovery of muscle following partial denervation. Am. J. Physiol. 145, 587-607. WEISS, P. and HOAG, A. (1946). Competitive re-innervation of rat muscles by their own and foreign nerves. J. Neurophys. 9, 413-418. WENGER, E. L. 0951). Determination of structural patterns in the spinal cord of the chick embryo studied by transplantations between brachial and adjacent levels. J. Exp. Zool. 116, 123 163. WERNIG. A.. PI~COT DECHAVASSINE.M. and ST6VER. H. (1980a). Signs of nerve regression and sprouting in the frog neuromuscular synapse. In: Ontogene~i.~ and F,netional Mechanisms r~f Peripheral Synapses. pp. 225 238. Ed. J. TAXL I.N.S.E.R.M., Symp. 13. Elsevier Press. Amsterdam, North Holland. WERNtG. A.. Pt~COT-DECHAVASStNE,M. and ST6VER. H. (1980b). Sprouting and regression of the nerve at the frog neuromuscular junction in normal conditions and after prolonged paralysis with curare, d. Ne,rocytol. 9, 277 303. WtGSTON, D. J. (1979a). Suppression of foreign innervation in axolotl muscle may not be dependent on juxtaposition of native and foreign nerve terminals. Neurosci. Lett. I1, 165-170. WtGSTON, D. J. (1979b). Suppression of foreign nerve transmission in axolotl muscle. Ph.D. thesis. WIGSTON, D. J. (1980). Suppression of sprouted synapses in axolotl muscle by transplanted foreign nerves. J. Physiol. 307, 355 366. WtLSON, D. F. (1980). Presynaptic effects of anticholinesterases. Soc. for Neurosci. Abstr. 6, no. 206.9. WILSON, B. W., RANDALL, W. R., PATTERSON, G. T. "and EN'n~t~:IN, R. K. (1979). Major physiologic and histochemical characteristics of inherited dystrophy of the chicken. Ann. N.Y. Acad. Sci. 317, 224-246. YEAGLE. S.~ WARNlC~:. J_ E. and ALBUQUERQUE,E. X. (1979). Characterization of spontaneous transmitter release and the effects of temperature in normal and dystrophic chickens. Soc. Neurosci. Ahstr. 5, no. 2595. Yw, J. W. and DENNtS, M. J. (1976). Suppression of transmission at foreign synapses in adult newt muscle involves reduction in quantal content. Nature, 260, 350-352. ZACrZS, S. 1.. BAUER. W. C. and BLt:MBERG.J. J. (1961). Abnormalities in the fine structure of the neuromuscular junction in patients with myasthenia gravis. Nature. 190, 280-281. ZAC~S, S. I.. BAUER. W. C. and BLUMBERG,J. M. (1962). The fine structure of the myasthenic neuromuscular junction. J. Neuropath. Exp. Neurol. 21,335 347. ZELENA. J.. JtRMANOVi,, I. and VV~:LICKq. L. (1967). Motor endplates in fast and slow muscles after cross union of their nerves. Nat,re. 214, 1010-1011. ZELEN/~. J.. VVSKO(tL. F. and JXRMANOV/,. I. (1979). The elimination of polyneuronal innervation of endplate in developing rat muscles with altered function. In: The Choliner.qic Synapses, pp. 365 372. Ed. S. TUg'Eg.. Elsevier. Amsterdam. ZIPSER, B. and MCKAV. R. (1981). Monoclonal antibodies distinguish identifiable neurones in the leech. Nat,re. 289, 549- 554. ZUCKER, R. S. (1977). Synaptic plasticity at crayfish neuromuscular junctions. In: hlent!lied neurons anti heha~'ior ~farthropods. p. 49. Ed. G. HOYLE. Plenum Press. New York.