reviews
Nervegrowthfactorandnociception
G a r y R. L e w i n a n d Lorne M . M e n d e l l
Nerve growth factor (NGF) is thought of as a targetderived factor responsible for the survival and maintaining the phenoOrpe of specific sets of peripheral and
central neurons during development and maturation.
Recently, using physiological techniques, we have
shown that specific functional b'pes of nociceptive
sensory neurons require NGF, first for survival during
development in utero and then for their normal
phenob,pic development (but not survival) in the early
postnatal period. In adulthood, the physiological role of
NGF changes dramatically and here it may serve as a
link between inflammation and hyperalgesia. Despite
apparent changes in NGF's mode of action as the
animal matures, it always interacts specifically with
nociceptive sensory neurons.
The neurotrophic hypothesis states that, during
development, neurons are critically dependent for
survival on target-derived factors. The presence of
limiting amounts of such factors ensures that only a
proportion of neurons survive naturally occurring cell
death and that the appropriate innervation density of
the target is attained. Nerve growth factor (NGF) is
the prototypical neurotrophic factor and is a survival
factor for both sympathetic and sensory neurons
during development1. Many sympathetic neurons
continue to depend on NGF for survival into adulthood, whereas sensory neurons cease to rely on this
molecule in the postnatal period2~. Furthermore, it
appears that at no point during development are all
sensory neurons dependent on NGF for survival.
Indeed, the number of neurons that need NGF for
survival gets smaller as development proceeds (see
Box 1). This change in dependence on NGF for
survival occurs despite the fact that many smalldiameter peptide-containing sensory neurons in the
adult have high-affinity NGF receptors and retrogradely transport NGF from their target tissues 5'6.
This suggests that NGF has a physiological role in the
adult that is distinct from that in development.
Several lines of evidence have indicated that the
action of NGF on sensory neurons might be directed
to specific functional groups of nociceptive neurons.
For example, small sensory neurons containing
neuropeptides, such as substance P, do not survive
NGF deprivation during development (see Box 1);
and in adults, the levels of these peptides are
downregulated after NGF deprivation 7'8. High-affinity
NGF receptors are present primarily on this same
population of neuropeptide-containing neurons6. Conversely, large cells devoid of peptides do not require
NGF for survival and do not have high-affinity NGF
receptors 6'9'1°. It is widely assumed that peptidecontaining neurons are nociceptive sensory neurons
with unmyelinated or thinly myelinated axons 13 (see
Box 2). The fact that they are susceptible to NGF
deprivation suggests that NGF regulates a functional
group of sensory neurons. In addition, the sensitivity
of adult sensory neurons to capsaicin, a compound
TINS, Vol. 16, No. 9, 1993
that elicits pain in humans and excites C-mechanoheat
fibers (see Box 2) in vivo n, can be regulated in
culture by the supply of NGF (Ref. 12). Animals
subjected to pre- or postnatal anti-NGF treatments
appear to suffer from a decreased sensitivity to
noxious mechanical and heat stimuli14'~5. In these
cases, the death of sensory neurons (perhaps specifically nociceptive neurons) has been proposed to be
responsible for the observed hypoalgesia 14'15. However, nociceptive neurons are very heterogeneous in
their functional properties (see Box 2), and not all
small peptidergic neurons can be associated with
physiologically identified nociceptors 16. Thus the
relationship between NGF and nociceptors can be
determined in a definitive fashion only by using
physiological analysis rather than by inference from
anatomical, behavioral and immunocytochemical
studies.
GaryR. Lewinand
LomeM. Mendel/are
at the Deptof
NeurobioloD, and
Behavior,SUNYat
StonyBrook,
StonyBrook,
New York,
NY 11794, USA.
Role of NGF in the postnatal development of
nociceptors
Anti-NGF treatment. The innervation of hindlimb
skin by sensory neurons in adult rats subjected to
anti-NGF treatments has been analysed by recording
from the axons of single afferents in dorsal root
filaments and characterizing their receptive fields4'17.
By using an electrical search stimulus on the sural
nerve it is possible to obtain an unbiased estimate of
the types of primary afferents innervating skin. After
treatment with anti-NGF in the first two weeks after
birth we observed a loss of just one myelinated
afferent type, the thinly myelinated high-threshold
mechanoreceptor (A6-HTMR) 4' 17. This selective loss
might have been due to the anti-NGF-induced death of
these fibers. However, this is unlikely for two
reasons. First, the selective depletion of HTMRs can
be obtained even when anti-NGF is administered too
late to cause cell death (after postnatal day 2; see
Box 1). Second, one would have expected the
proportions of the two other afferent types present
within the A6 range (low-threshold D-hair afferents,
and afferents innervating subcutaneous structures,
deep afferents) to increase as HTMRs are lost.
However, only D-hairs exhibited this proportional
increase (see Fig. 1). By using more circumscribed
anti-NGF treatments in the postnatal period, we
determined a critical period for the effects of antiNGF on A6 nociceptors. This period, from postnatal
days 4-11, coincides with a period of a normally
occurring rearrangement of afferent terminals in skin
(see below).
We have proposed that after neonatal anti-NGF
treatment AS-HTMRs are forced to assume the
phenotype of D-hair afferents4'17'18. This would
account for the reciprocal changes in the proportions
of D-hairs and HTMRs described above. It also
implies that sensory neurons may be subject to respecification of their phenotype by environmental
cues well into the postnatal period.
© 1993,ElsevierSciencePublishersLtd,(UK)
353
Box 1. How and when can you kill sensory neurons?
One can kill sensory neurons in vivo by depriving them
of neurotrophic support. This can be achieved in two
ways: first, by systemically treating animals with high
titre antibodies raised against NGF; second, by making
auto-immunized animals that will produce antibodies
to their own NGF. Auto-immunized females can
produce litters of rat pups that have been deprived of
NGF in utero by the transfer of maternal anti-NGF to
the foetus a. The effect of neurotrophin deprivation on
the survival of sensory neurons varies according to the
time at which it is implemented.
Deprivation of NGF in utero
During development up to 80% of sensory neurons
can be killed by depriving them of NGF in utero a'b. The
cells that are lost after these treatments all appear to be
small cells that probably expressed the trk A receptorc
(high-affinity NGF receptor). Many contain neuropeptides and most project to the superficial dorsal horn c-e.
The large cells that remain are probably the precursors
of mature low-threshold mechanoreceptors innervating
skin and muscle targets f. The proportion of neurons in
the mature dorsal root ganglion with large cell bodies
and thickly myelinated axons is around 20% of the
total in the lumbar ganglia g.
Interestingly, administration of anti-NGF after postnatal day 2 fails to produce any cell death g.
Post-natal
Control
Anti-NGF
Adult NGF deprivation
Both methods of NGF deprivation (autoimmune and
systemic antibodies) fail to result in the death of any
sensory neurons in the adult or after postnatal day 2
(Refs g, j).
Adult
Control
Anti-NGF
In utero
Control
Anti-NGF
Perinatal deprivation of NGF
In the early postnatal period, studies have shown
that 20-35% of sensory neurons could be killed after
administration of anti-NGF g-~. The higher percentage
was obtained in thoracic ganglia where the proportion
of afferents innervating cutaneous targets is greater.
The cells that are killed are all small; however,
appreciable numbers of these small neurons still
remain. This more resilient population might represent
a functional group or neurons that are more mature.
The development of sensory innervation provides
certain important clues as to how this might occur.
Although peripheral afferent fibers have innervated
the skin by the time that the rat is born, this
innervafion undergoes fundamental changes during
the immediate postnatal period 19'2°. From around
embryonic day 17 (El7) to around postnatal day 5, the
sensory innervation of the skin appears to be largely
epidermal with virtually no fibers terminating in the
dermis. The innervafion takes on the adult pattern in
the hindlimb with the appearance of hair follicles in the
dermis from around postnatal days 5-10, and so some
fibers must move fl'om epidermis to dermis. In the
adult animal, the terminations of A6--HTMR afferents
appear to be in the epidermis 21. This suggests that
when NGF is eliminated by anti-NGF, HTMRs are
unable to sustain their normal epidermal terminals and
'drop down' into the dermis to innervate receptors
characteristic of D-hairs.
354
References
a Johnson, E. M., Jr, Rich, K. M. and Yip, H. K. (1986)
Trends NeuroscL 9, 33-37
b Johnson, E. M., Jr, Gorin, P. D., Brandeis, L. D. and
Pearson, J. (1980) Science 210, 916-918
c Carroll, S. L. etal. (1992) Neuron 9, 779-788
d Otten, U., GoederL M., Mayer, N. and Lernbeck, F.
(1980) Nature 287, 158--159
e GoederL M. et aL (1984) Proc. Natl Acad. Sci. USA 81,
1580-1584
f Ruit, K. G., Elliot-[, J. L., Osborne, P. A., Yah, Q. and
Snider, W. D. (1992) Neuron 8, 573-587
g Lewin, G. R., Ritter, A. M. and Mendell, L. M. (1992)
J. Neurosci. 12, 1896-1905
h Yip, H. K., Rich, K. M., Larnpe, P. A. and Johnson,
E. M., Jr (1984) J. Neurosci. 4, 2986-2992
i Hulsebosch, C. E., Perez-Polo,J. R. and Coggeshall, R. E.
(1987) J. Comp. Neurol. 259, 445-451
j Gorin, P. D. and Johnson, E. M., Jr (1980) Brain Res. 198,
27-42
The phenotype of A6 afferents after anti-NGF
treatment was determined by characterizing their
adequate stimulus. It is possible that the D-hair
afferents recorded had only partially changed so that
their cell bodies still expressed phenotypes more
characteristic of HTMRs. However, an important
characteristic difference between D-hair and HTMR
afferents is the configuration of their somatic spikes
recorded in the dorsal root ganglion. D-Hairs have
a very brief spike with a smooth falling phase,
whereas HTMRs have a much broader spike with a
characteristic inflection on the falling phase 22. In antiNGF-treated animals no D-hairs are observed with
broad HTMR-like spikes a3, suggesting that the conversion occurs not only in the skin terminals, but also
in the cell body. This finding argues more convincingly
for a true phenotypic conversion as a result of the
withdrawal of NGF during the postnatal critical
period.
TINS, Vol. 16, No. 9, 1993
Among the population of unmyelinated C-fibers, at
least five different types of nociceptor have been
described (see Box 2) z4'25. After neonatal anti-NGF
treatment (postnatal days 2-14) it was found that the
numbers of C-mechanoheat fibers (see Box 2) in adult
animals had been severely reduced (by about 60%).
These lost fibers appeared to be replaced by a new
population of C-fibers that responded exclusively to
mechanical stimuli. The new population of mechanoreceptors was unusual in that their mechanical
thresholds were intermediate between those of nociceptors and low-threshold mechanoreceptors. This
change in threshold is unlikely to be due to physical
changes in the skin, since the mechanical sensitivity of
the few remaining C-mechanoheat fibers was normal.
That the increase in the fraction of C-mechanoheat
afferents seen after excess NGF is given neonatally2s
is additional evidence that the fibers conferring
noxious heat sensitivity are selectively susceptible to
the supply of NGF.
The ability of A6 and C-fiber afferents to elicit
antidromic vasodilation was found to be impaired in
neonatal anti-NGF-treated animals26. This result
suggests that the sensory neurons lost (i.e. converted) after anti-NGF treatment (A6--HTMRs and
C-mechanoheat fibers) probably mediated the antidromic vasodilation, and this is in agreement with data
from other studies 27'2s.
NGF treatment. We have administered excess
amounts of NGF to neonatal animals in the first two
weeks after birth. It is possible that large amounts of
systemic NGF might lead to changes that do not
reflect this factor's normal function. However, for the
most part, the physiological consequences were
specific for the same types of fibers affected by lack of
NGF (Refs 18, 29). In some ways the results of NGF
administration were less dramatic than anti-NGF
treatments. For instance, the relative proportions of
Ab-HTMRs, D-hair and deep afferents in the sural
nerve were unchanged. If our interpretation of the
phenotypic switch is correct, this implies that the
naturally occurring amount of NGF present in skin is
sufficient to ensure that all the A6 nociceptors
develop normally. Excess NGF does not apzpear to
force more sensorv neurons to become ASs~L
The physiological properties of the AS-HTMRs in
neonatally NGF-treated animals were different from
those of controls. A6-HTMRs in NGF-treated fibers
were about twice as sensitive to mechanical stimulation as normal control fibers. This increased sensitivity after neonatal treatment with NGF is not
permanent, and wears off when the animal is about
nine weeks of age 29. However, if animals were
treated with NGF after the second postnatal week,
this change in AS-HTMR receptor sensitivity was
never observed 29. This suggests that the mechanical
thresholds of the A6 fibers are in part regulated by
NGF during the postnatal critical period. Further
evidence for this suggestion comes from the anti-NGF
experiments in which HTMRs not converted by
treatment with anti-NGF during the critical period
have a decreased mechanical sensitivity (thresholds
are about double those of control units) 4.
Box 2. Physiological heterogeneity of sensory
neurons
Sensory neurons are normally classified physiologically
using two main criteria, their axonal conduction vel-
ocities and their adequate stimuli. Sensory neurons
with thickly myelinated axons (A~-fibers) almost all
respond to very low-threshold stimuli, such as hair
movement in skin or length changes in skeletal muscle.
Most of the slower conducting thinly myelinated
sensory neurons (A6-fibers) innervating skin are highthreshold mechanoreceptors (HTMRs) as they respond
exclusively to noxious stimuli such as pinching the skin.
However, a significant proportion of Ai5 sensory
neurons respond to very-low-threshold stimuli such as
hair movement; these are often called D-hairs1'2.
The majority of sensory neurons have unmyelinated
axons (C-fibers) and the majority of these neurons in
rodents, monkeys and man respond to noxious
stimulia-c. However, this does not mean that C-fibers
are all functionally equivalent. These neurons can
respond to noxious mechanical, heat, cold and
chemical stimuli, either exclusively or in combinationd.
Thus the most common type has been termed
C-mechanoheat (C-MH) as it responds to both these
modalities. Other types include C-mechanoreceptors
(C-M), C-mechanocold (C-MC) and C-Cold (C-C) d.
Finally, some C-fibers innervating skin, viscera and
joints appear to be unresponsive to any of these stimuli
and these have been termed silent or mechanically
insensitive fiberse.
For both A6 and C-fibers, no clear correspondence
between cytochemical markers and the physiological
phenotype of the sensory neurons has been foundf.
References
a Burgess, P. R. and Perl, E. R. (1973) in Handbook of
Sensory Physiology, Vol. 2 (Iggo, A., ed.), pp. 29-78,
Springer
b Lynn, B. and Carpenter, S. E. (1982) Brain Res. 238,
29-43
c Treede, R. D., Meyer, R. A., Raja, S. N. and Campbell,
J. N. (1992) Prog. Neurobiol. 38, 397-421
d Kress,M., Koltzenburg, M., Reeh, P, W. and Handwerker,
H. O. (1992) J. Neurophysiol. 68, 581-595
e McMahon, S. B. and Kolt.zenburg, M. (1990) Trends
Neurosci. 13, 199-201
f Leah, J. D., Cameron, A. A. and Snow, P. J, (1987)
Neurosci. Lett. 56, 257-263
the physiology of A6 fibers. However, since adult
sensory neurons possess functional high-affinity NGF
receptors s'6, NGF must have some functional role.
Behavioral experiments where the effects of excess
NGF were tested on adult animals may provide some
insights. After just one systemic injection of NGF
(1 ~tg/g, intraperitoneal), adult rats develop a profound
hypersensitivity to noxious heat and mechanical
stimuli29, a so-called hyperalgesia. The mechanical
hyperalgesia that develops is particularly profound as
the mechanical threshold needed to elicit a flexion
reflex drops from around 150 g in controls to as low as
20 g in NGF-treated animals29. This hyperalgesia does
not appear to be accompanied by any ongoing pain, or
pain elicited by stimuli sufficient to excite lowthreshold AI5 afferents (the latter is often called
allodynia). The mechanical hyperalgesia takes around
six hours to develop and there are no visible signs of
skin inflammation. We have recorded from A6R o l e of N G F in a d u l t a n i m a l s
Under normal circumstances, treatment of adult HTMRs in animals displaying this NGF-induced
animals with anti-NGF does not lead to any change in hyperalgesia and they are not sensitized to mechanical
TINS, Vol. 16, No. 9, 1993
355
A
Control
Anti-NGF
Cell death?
or
Phenotypic
conversion?
OI
mice displaying behavioral changes (Johnson, E. M.,
Davis, B. M. and Albers, K., unpublished observations).
In adult rats, the same injection of NGF gave rise to
a heat hyperalgesia in addition to the mechanical
hyperalgesia. The heat hyperalgesia develops within
minutes of the injection and thus may be due to the
sensitization of C-mechanoheat fibers to noxious
heat. However, as yet we have no direct evidence for
this.
Is NGF t h e h y p e r a l g e s i a - i n f l a m m a t i o n
linchpin?
The hyperalgesia that follows an NGF injection is
qualitatively very similar to that which often accompanies inflammatory tissue injury 31. Could over• HTMR • D-Hair (~ Deep
expression of NGF in inflamed tissue be a critical link
between tissue injury and the hyperalgesia that
B
ensues? There is, in fact, abundant evidence supporting a critical role for NGF in this process. The
O
HTM
Rs
8o
levels of NGF in damaged or inflamed tissue have
(t)
[] D-Hairs
p,,,,
been shown to be increased many fold above normal
Q.
~
60
levels32-34, and this effect can be seen within hours
E
of the initiation of inflammation 3z. An enhanced retrograde transport of NGF by the sciatic nerve occurs in
~ 40
conjunction with this peripheral increase 35 (Fig. 2).
Many cytokines and growth factors, levels of which
•
20
are probably elevated at the site of inflammation, can
cpotently upregulate the production of NGF by skinI
I
i
I
|
I
I
|
i
and nerve-derived fibroblasts36'37. These mediators
£
0
Cont. 2 - 5 7 - 1 4 2 - 9
4-11 2 - 1 4 0 - 1 4 0 - 5
0-5*
include acidic and basic fibroblast growth factor
wks days days days days days wks wait
(FGF), interleukin 113 (IL-I[3), tumor necrosis factor
Duration of treatment
(TNF-00, platelet-derived growth factor (PDGF),
transforming growth factor (TGF-~t and TGF-[31), and
Fig. 1. (A) AAodel of two possible scenarios to account for epidermal growth factor (EGF) 36'37. Skin keratinodepletion of high-threshold mechanoreceptors (HTAARs) cytes appear to be major sources of NGF in viv03°'38;
after administration of anti-NGF antibodies over the
however, it is not known if the above cytokines can
neonatal critical period (postnatal days 4-11). Each of
regulate
NGF expression in these cells. In recent
the four dorsal root ganglia shown contains three types of
A6 sensory neurons with axons in the sural nerve: years it has become increasingly clear that NGF may
HTAARs, D-hairs and deep (subcutaneous) receptors4,17,18. have an important immunomodulatory role and it
Upper panel shows selective loss of HTAARs; lower panel might potently affect the responsiveness of immune
shows conversion of HTAARs to D-hairs. (B) Relative competent cells during inflammation39'4°; in short,
proportions of the three types of A6 neurons after NGF may be involved in the process of inflammation.
different times and durations of postnatal anti-NGF We will not review this large literature here, but it is
treatment. Critical period is from postnatal days 4-11,
shown by the fact that HTAARs are converted to D-hairs
only when anti-NGF is administered over this period.
(Reproduced, with permission, from Ref. 4.) Note that the
actual changes in the proportion of each type observed
conforms to the predictions of the conversion model
shown schematically above in (A).
5
4
E
stimuli. Stimuli sufficient to excite low-threshold A[3
afferents did not lead to any behaviors indicative of
pain, i.e. there was no allodynia present.
Recently, Albers et al. 3° described a transgenic
mouse where an overproduction of NGF in skin was
achieved by attaching the K14 keratin promoter to the
NGF gene. These mice exhibited a mechanical
hyperalgesia very similar to that found in rats after a
systemic NGF injection (Davis, B. M., Lewin, G. R.,
Mendell, L. M., Johnson, M. and Albers, K., unpubfished observations); this result suggests that an
increased level of NGF restricted to the skin is
sufficient to produce mechanical hyperalgesia. Preliminary data indicate that biologically active NGF is
not present in the systemic circulation of many of the
356
Q.
LL
3
2
1
I
I
I
I
1
2
3
4
Time (days)
Fig. 2. Levels of NGF in sciatic nerve after inflammation
induced by injection of 0.15ml of complete Freunds
adjuvant. Note the increase in the NGF content in the
ipsilateral sciatic nerve (filled squares) compared to contralateral sciatic nerve (open squares). Statistical significance is indicated by * (p <0.05). (Reproduced, with
permission, from Ref. 35.)
~NS, Vo1.16, No. 9,1993
Periphery
Cell body
Spinal cord
Sensory consequence
clear that NGF seems to be particularly important for the action
of histamine-secreting mast cells
and circulating basophils 4v-44.
An increase in NGF might have
I
Injury
t Neuropeptides
two major effects on the sensory
e.g. SP + CGRP
neurons
innervating
inflamed
1
tissue. The first effect may be
¢
indirect as NGF is a potent stimuCentral ~
Mechanical
Mechanical
lator of mast cells and can cause
sensitization [ I
hyperatgesi~
hyperatgesia
their degranulation 42'45. Rat peritoneal mast cells have recently
NGF
been found to possess the putadependent
tive high-affinity NGF receptor trk
\
A (Ref. 46), and they require NGF
for survival in vitro 47. Substances
Heat
L hyperalgesia
released by stimulated mast cells
could directly sensitize the peripheral endings of C-mechanoheat
fibers, which might explain the
~
] [ LTMR-mediated
heat hyperalgesia that immediately
' [ sensptiz~ion I'--[-~"1
pain
I
] H l I=io~y~l I NGF
follows treatment with exogenous
independent
NGF. In the rat, mast cells release
5-hydroxytryptamine (5-HT) as
well as histamine upon stimulation48.
However, neither of these substances can directly sensitize C- Fig. 3. Schematic diagram of possible mechanisms for NGF-dependent (above) and NGFmechanoheat fibers to heat 49. In independent (below) sensory consequences of tissue damage and inflammation. Increased levels of
preliminary studies we have found NGF can lead to both mechanical and heat hyperalgesia29. Two major pathways are proposed to be
that animals with degranulated involved in these two types of hyperalgesia. The long-lasting heat and mechanical hyperalgesia
may come about via the upregulation of neuropeptide levels in the dorsal root ganglion; the central
mast cells do not display heat release of these neuropeptides might lead to central sensitization 51. We have obtained evidence
hyperalgesia in the first three that the long-lasting heat hyperalgesia may be dependent on an increase in NMDA-mediated
hours following treatment with neurotransmission. The rapid heat hyperalgesia after treatment with exogenous NGF might result
exogenous NGF (Ref. 50); thus from the sensitization of C-fibers to heat in the periphery. This may be initiated by the NGF-induced
the nature of the peripheral mech- release of agents from mast cells5~. Low-threshold mechanoreceptor (LTMR)-mediated pain
(allodynia) and spontaneous pain do not appear to be produced by the administration of
anism requires further study.
The second effect of increased exogenous NGF. These sensory consequences may be produced via other mechanisms of central
NGF on sensory neurons in in- sensitization or by the continuous activation of nociceptors or the recruitment of silent fibers (see
flamed tissue is more direct. NGF Box 2). Note that where the boxes are tinted, the associated processes have not been definitively
might bind to its receptor on sen- proven to be involved in the mechanism indicated. Abbreviations: SP, substance P; CGRP,
calcitonin gene-related peptide.
sory neurons and be retrogradely
transported to the cell body where
it can lead to a rapid and large increase in the the inflammation, can be abolished by concurrent
production of neuropeptides 5 such as substance P and administration of anti-NGF (Ref. 57). Unfortunately,
calcitonin gene-related peptide (CGRP) at the level of the use of antibodies presents problems as very large
gene expression 5L52. Increases in the levels of these amounts of these molecules are required to have any
two peptides in sensory neurons and their peripheral effect. In future, the development of non-pepride
axons innervating inflamed tissue have been reported antagonists to the high-affinity NGF receptor may
In
• many models of inflammatory pain 3 5 ' 5 3 - 5 6 . For provide a better tool to address this question 58.
example, increases in these peptides accompany skin
Hyperalgesia following tissue injury has in recent
inflammation induced by complete Freunds adjuvant years been increasingly attributed to the sensitization
(CFA) or carrageenin 35'54, as well as damage to the of spinal cord circuits in response to an afferent
skin by ultraviolet light (sunburn) 56 and joint inflam- barrage from the injured tissue al. The increased
mation or experimental arthritis 5a'55. In one of these synthesis of neuropeptides that follows tissue inflammodels, where CFA is used to inflame the foot pad of marion may well lead to an increased release of these
a rat, there is now direct evidence that NGF is the substances from the central terminals of primary
cause of these changes in peptide levels. Donnerer et afferents, and indeed there is evidence to support this
aL 35 demonstrated that the increases in the levels of contention 54'55'59. Neuropeptides released in the
substance P and CGRP in the skin and dorsal root dorsal horn may sensitize central neurons, perhaps by
ganglion after CFA-induced inflammation of the skin enhancing neurotransmission via the N-methyl-Dcould be blocked with the exogenous administration of aspartate (NMDA) receptor by producing long-lasting
anti-NGF. Unfortunately, these workers did not test depolarizations 6° or by directly modulating the recepto see if this blockade of increases in the peptides also tor complex61'62. The NMDA receptor has been
blocked hyperalgesia associated with the inflam- shown to be necessary for the generation and
marion. Preliminary experiments carded out in this maintenance of many central sensitization events 63. In
laboratory using the same CFA model have shown recent experiments we have obtained direct evidence
that heat hyperalgesia, which normally accompanies that the long-lasting heat hyperalgesia that follows
mS
m
TINS, VoL 16, No. 9, 1993
/
357
Hyperalgesia
Mechanical (central?)
Heat (peripheral?)
Maintenance of
HTMR phenotype
I
I
Survival of small
peptidergic cells
'"
I
i
I
i
i
}
}'
I
Adult
Neonate
Fetus
Fig. 4. Actions of NGF depend on the developmental stage. Up to postnatal
day 2, NGF acts as a survival factor. Between postnatal days 2-14, NGF
influences the phenotype of the sensory neuron. In the adult, NGF can lead to
hyperalgesia.
NGF with a delay of around seven hours is dependent
on activation of the NMDA receptor, since administration of a noncompetitive NMDA antagonist blocks
the hyperalgesias°. However, the mechanical hyperalgesia that is induced by NGF appears to be
independent of NMDA receptor activation~°. Whatever the central mechanism of sensitization, there is
direct evidence that artificially increased levels of
NGF in a peripheral target can sensitize central
neurons to afferent barrages from afferents innervating that target c~. The long latency of NGF-induced
mechanical hyperalgesia is consistent with the initiation of increased synthesis of peptides in sensory
neurons and their transport to the central terminals
in the spinal cord (Fig. 3). Changes in the functional
connectivity of these afferents in the spinal cord may
then underlie the mechanical hyperalgesia.
Neuropepfides that are upregulated by NGF are
also transported in large amounts to the periphery 11'3S. Their release in skin may contribute to the
inflammationl~; however, it is unlikely that sensory
neurons would be further sensitized since neurogenic
extravasation does not sensitize peripheral fibers to
heat 65.
This model suggests that inflammation-induced
increases in NGF in the periphery may be the linchpin
that connects tissue damage to the accompanying
hyperalgesia. The hyperalgesia that follows NGF
injections in normal adult rats is not accompanied by
detectable inflammation29,34, and this suggests that
NGF has in this case short-circuited the normal
physiological process. The fact that abnormal sensations such as spontaneous pain and allodynia do not
result from exogenous NGF suggests that these
sensations arise via a pathway that is independent
of NGF (see Fig. 3). More experiments designed to
test this model specifically will be needed before
NGF's role in generating the hyperalgesia that follows
tissue injury is more clearly defined. Nevertheless,
the existing evidence suggests that NGF may be an
important target of drug therapies aimed at alleviating
hyperalgesia following tissue injury.
358
In summary, it appears that the physiological role of
NGF in sensory neurons is highly dependent on their
developmental context (see Fig. 4). The presence of
NGF is critically important for the normal development of important subsets of nociceptive primary
afferent neurons. The developmental mechanisms by
which NGF regulates nociceptor development in
some cases do not include a role as a survival factor,
but rather as a molecule that maintains certain
neuronal phenotypes. In the adult, NGF may act
through entirely different mechanisms to serve as a
link between the physiological state of nociceptors and
the health of the tissue they innervate. The growing
complexity of NGF's physiological role may serve as a
model by which the other largely uncharacterized
neurotrophins comprising the NGF family can be
studied.
Selected references
1 Barde, Y-A. (1989) Neuron 2, 1525-1534
2 Gorin, P. D. and Johnson, E. M., Jr (1980) Brain Res. 198,
27-42
3 Ruit, K. G., Osborne, P. A., Schmidt, R. E., Johnson, E. M., Jr
and Snider, W. D. (1990) ]. NeuroscL I0, 2412-2419
4 Lewin, G. R., Ritter, A. M. and Mendell, L. M. (1992)
J. Neurosci. 12, 1896-1905
5 Goedert, M., Stoeckel, K. and Otten, U. (1981) Proc. Natl
Acad. 5ci. USA 78, 5895-5898
6 Verge, V. M., Richardson, P. M., Benoit, R. and Riopelle, R. J.
(1989) J. Neurocytol. 18, 583-591
7 Goedert, M. eta/. (1984) Proc. Natl Acad. ScL USA 81,
1580-1584
8 0 t t e n , U. (1984) Trends Pharmacol. Sci. 5, 307-310
9 Johnson, E. M., Jr, Rich, K. M. and Yip, H. K. (1986) Trends
Neurosci. 9, 33-37
10 Carroll, S. L. eta/. (1992) Neuron 9, 779-788
11 Holzer, P. (1991) Pharmacol. Rev. 43, 143-201
12 Winter, J., Forbes, C. A., Sternberg, J. and Lindsay, R. M.
(1988) Neuron 1,973-981
13 Hunt, S. P., Mantyh, P. W. and Priestley, J. V. (1992) in
Sensory Neurons: Diversity, Development and Plasticity
(Scott, S. A., ed.), pp. 60-76, Oxford University Press
14 Aloe, L., Cozzari, P., Calissano, P. and Levi-Montalcini, R.
(1981) Nature 291,413-415
15 Urschel, B. A., Brown, P. N. and Hulsebosch, C. E. (1991) Exp.
Neurol. 114, 44-52
16 Leah, J. D., Cameron, A. A. and Snow, P. J. (1987) Neurosci.
Lett. 56, 257-263
17 Ritter, A. M., Lewin, G. R., Kremer, N. E. and Mendell, L. M.
(1991) Nature 350, 500-502
18 Ritter, A. M., Lewin, G. R. and Mendell, L. M. (1993) Brain
Res. Bull. 30, 245-249
19 Fitzgerald, M. J. T. (1965) J. Comp. Neurol. 126, 37-42
20 Reynolds, M. L., Fitzgerald, M. and Benowitz, L I. (1991)
Neuroscience 41,200-211
21 Kruger, L., Perl, E. R. and Sedivec, M. J. (1981) J. Comp.
Neurol. 198, 137-154
22 Koerber, H. R. and Mendell, L. M. (1992) in Sensory
Neurons: Diversity, Development and Plasticity (Scott, S. A.,
ed.), pp. 77-97, Oxford University Press
23 Ritter, A. M. and Mendell, L. M. (1992) J. Neurophysiol. 68,
2033-2041
24 Kress, M., Koltzenburg, M., Reeh, P. W. and Handwerker,
H. O. (1992) J. NeurophysioL 68, 581-595
25 Lewin, G. R. and Mendell, L. M. (1992) J. Physiol. 452, 43P
26 Lewin, G. R., Lisney, S. J. W. and Mendell, L. M. (1992) Eur. J.
Neurosci. 4, 1213-1218
27 Kenins, P, (1981) Neurosci. Lett. 25, 137-141
28 J~nig, W. and Lisney, S. J. W. (1989) J. Physiol. 415,477-486
29 Lewin, G. R., Ritter, A. M. and Mendell, L. M. (1993)
J. Neurosci. 13, 2136-2148
30 Albers, K., Wright, B. M. and Davis, B. M. (1992) 5oc.
Neurosci. Abstr. 18, 1288
31 Treede, R. D., Meyer, R. A., Raja, S. N. and Campbell, J. N.
(1992) Prog. Neurobiol. 38, 397-421
32 Weskamp, G. and Otten, U. (1987) J. Neurochem. 48,
1779-1786
33 Varilek, G. W., Weinstock, J. V. and Pantazis, N. J. (1991)
TINS, Vol. 16, No. 9, 1993
Infect. Immun. 59, 4443-4449
34 Aloe, L., Tuveri, M. A. and Levi-Montalcini, R. (1992)
RheumatoL Int. 12, 213-216
35 Donnerer, J., Schuligoi, R. and Stein, C. (1992) Neuroscience
49, 693-698
36 Lindholm, D., Heumann, R., Meyer, M. and Thoenen, H.
(1987) Nature 330, 658-659
37 Yoshida, K. and Gage, F. H. (1992) Brain Res. 569, 14-25
38 DiMarco, E. etaL (1991)J. BioL Chem. 266, 21718-21722
39 Gee, A. P., Boyle, M. D. P., Munger, K. L., Lawman, J. P. and
Young, M. (1983) Proc. NatlAcad. 5ci. USA 80, 7215-7218
40 Often, U., Ehrhard, P. and Peck, R. (1989) Proc. NatlAcad.
Sci. USA 86, 10059-10063
41 Aloe, L. and Levi-Montalcini, R. (1977) Brain Res. 133,
358-366
42 Bienenstock, J. et al. (1987) Int. Arch. Allergy Appl. Immun.
82,238-243
43 Takahashi, T. et al. (1991) Blood 77, 971-979
44 Bischoff, S. C. and Dahinden, C. A. (1992) Blood 79,
2662-2669
45 Mazurek, N., Weskamp, G., Erne, P. and Otten, U. (1986)
FEBS Lett. 198, 315-320
46 Pryor, J. C., Horigome, K. and Johnson, E. M., Jr (1992) Soc.
Neurosci. Abstr. 18, 950
47 Horigome, K., Lampe, P. A. and Johnson, E. M., Jr (1992)
5oc. Neurosci. Abstr. 18, 950
48 Bergendorff, A. and Uvnas, B. (1972) Acta Physiol. Scand. 84,
320-331
49 Raja, S. N., Meyer, R. A. and Campbell, J. N. (1988)
Anesthesiology 68, 571-590
50 Rueff, A., Lewin, G. R. and Mendell, L. M. Soc. Neurosci.
Abstr. (in press)
51 Lindsay, R. M. and Harmer, A. J. (1989) Nature 337,
362-364
52 Lindsay, R. M., Lockett, C., Sternberg, J. and Winter, J.
(1989) Neuroscience 33, 53--65
53 Kuraishi, Y. et al. (1989) Peptides 10, 447-452
54 Garry, M. G. and Hargreaves, K. M. (1992) Brain Res. 582,
139-142
55 Hope, P. J., Jarrott, B., Schaible, H-G., Clarke, R. W. and
Duggan, A. W. (1990) Brain Res. 533, 292-299
56 Gillardon, F., Schr6ck, H., Morano, I. and Zimmermann, M.
(1992) Ann. NY Acad. Sci. 657, 493-496
57 Lewin, G. R. and Mendell, L. M. (1992) Soc. Neurosci. Abstr.
18, 130
58 Davis, R. E. etal. (1992) 5oc. Neurosci. Abstr. 18, 1288
59 Donnerer, J. and Stein, C. (1992) Ann. NYAcad. Sci. 657,
Acknowledgements
Thisresearch was
supported by a
subproject of NIH
Program Project
Grant PC)1 N514899.
Additional support
was derived from NIH
R01 N516966 to
L. M. M. (Javits
NeuroscienceA ward).
505-506
60 Thompson, S. W. N., King, A. E. and Woolf, C. J. (1990) Eur.
J. Neurosci. 2, 638-649
61 Randic, M., Hecimovic, H. and Ryu, P. D. (1990) Neurosci.
Lett 117, 74-80
62 Rusin, K. I., Ryu, P. D. and Randic, M. (1992) J. Neurophysiol.
68, 265-286
63 Dubner, R. and Ruda, M. A. (1992) Trends Neurosci. 15,
96-103
64 Lewin, G. R., Winter, J. and McMahon, S. B. (1992) Eur. J.
Neurosci. 4, 700-707
65 Reeh, P., Kocher, L. and Jung, S. (1986) Brain Res. 384,
42-50
TheTINS/TiPSLecture
Themolecularbiologyof mammalianglutamatereceptor
channels
Peter H. Seeburg
At the 1992 ENA meeting, held in Munich, the plenary
lecture given by Peter Seeburg was sponsored jointly by
TINS and TIPS. The following article is adapted from
this lecture.
In native brain membranes the principal excitatory
neurotransmitter L-glutamate activates cation-conducting channels with distinct biophysical and pharmacological properties. Molecular cloning has revealed the
existence of 16 channel subunits that can assemble in
homomeric or heteromeric configurations in vitro to
form receptor channels with disparate functional properties. This review describes the different channel types
obtained by recombinant means and the genetic mechanisms controlling the expression of functionally important channel structures.
Many actions of the neurotransmitter L-glutamate,
ranging from fast excitatory synaptic transmission to
the regulation of developmental processes, are mediated by ionotropic receptors 1-3. Biophysical analysis
using membrane patches taken from the soma of
neurones indicates that different glutamate-sensitive
channel types coexist in many cells4. These channels
differ with respect to activation and deactivation timecourses, display distinct desensitization kinetics, and
have differing ion permeabilities and conductance
properties. Selective agonists and antagonists permit
a distinction between ionotropic glutamate receptor
(GluR) classes 5'6. One selective agonist is c~-amino-3TINS, Vol. 16, No. 9, 1993
hydroxy-5-methyl-4-isoxazole propionic acid (AMPA),
which activates channels with fast kinetics 7-9 (onset,
offset and desensitization time-courses are in the
order of ms); in most neurones these channels are
characterized by very low Ca2+ permeability 1°. NMethyl-i)-aspartate (NMDA) selectively gates a
channel with comparatively much slower kinetics 4'11
and high Ca2+ permeability 12. These unique properties, together with a voltage-dependent block by
Mg2+ (Refs 13, 14), a requirement for glycine as coagonist 15, and a large single channel conductance 14,
collectively determine the functional aspects of
NMDA receptors. The potent neurotoxin kainate
generates currents with different properties in different receptor channels: through AMPA receptors
kainate activates a current that persists in the
continued presence of this agonist16'17; in another
receptor type, called high-affinity kainate receptors
(referred to here as kainate receptors) kainate
activates fast desensitizing currents 18.
The neurophysiological functions of these three
channel classes (AMPA, NMDA, kainate) can be
summarized as follows. AMPA receptors, found in the
majority of excitatory synapses, mediate the majority
of all fast excitatory neurotransmission. Their rapid
kinetics render them ideally suited for this purpose.
The generally very low Ca 2+ permeability of AMPA
receptor channels means that glutamate-activated
ionic currents mediated by these channels do
not carry sufficient Ca 2÷ ions into cells to initiate
© 1993,ElsevierScience Publishers Ltd,(UK)
Peter H. Seeburgis at
the Universityof
Heidelberg, Centre of
Molecular Biology
(ZMBH), Laboratory
of Molecular
Neuroendocrinology,
POBox 106249, Im
NeuenheimerFeld
282, 69052
Heidelberg, Germany.
359