Loss of GABA-immunoreactivity in the spinal dorsal horn of rats with peripheral nerve injury and promotion of recovery by adrenal medullary grafts

Pergamon

PII:

Neuroscience Vol. 76, No. 3, pp. 845–858, 1997 Copyright ? 1996 IBRO. Published by Elsevier Science Ltd Printed in Great Britain 0306–4522/97 $17.00+0.00 S0306-4522(96)00341-7

LOSS OF GABA-IMMUNOREACTIVITY IN THE SPINAL DORSAL HORN OF RATS WITH PERIPHERAL NERVE INJURY AND PROMOTION OF RECOVERY BY ADRENAL MEDULLARY GRAFTS T. IBUKI,*‡ A. T. HAMA,†‡ X.-T. WANG,‡ G. D. PAPPAS‡ and J. SAGEN‡§ ‡Department of Anatomy and Cell Biology, University of Illinois at Chicago, Chicago, IL 60612, U.S.A. Abstract––Abnormal pain-related behaviour that accompanies peripheral nerve injury may be the result of altered spinal neuronal function. The long-term loss of inhibitory function by GABA neurons in particular may be a mechanism by which abnormal neural hyperactivity occurs, leading to exaggerated sensory processing following nerve injury. In order to assess this, changes in spinal GABA immunoreactivity at several time points following constriction nerve injury were quantified in parallel with behavioural assessments of abnormal sensory responses to noxious and innocuous stimuli. In addition, the effects of spinal adrenal medullary transplants were determined since previous findings have demonstrated alleviation of behavioural pain symptoms by such transplants. In response to unilateral sciatic nerve injury, GABAergic profiles normally found in lumbar dorsal horn laminae I–III significantly decreased. The decrease was apparent three days following ligation, particularly on the side ipsilateral to the nerve injury. By two weeks, no GABAergic profiles could be seen, with the deficit appearing in the spinal dorsal horn both ipsilateral and contralateral to the unilateral peripheral nerve injury. Marked decreases in GABA-immunoreactive profiles persisted for at least up to five weeks post-injury, with partial restoration occurring by seven weeks. However, even at seven weeks, losses in GABA-immunoreactive profiles persisted in the dorsal horn ipsilateral to peripheral nerve injury. These findings were comparable in animals receiving control striated muscle transplants. In contrast, adrenal medullary transplants markedly reduced the loss in GABA-immunoreactive profiles at all time-points examined. In addition, GABAimmunoreactive profile levels were normalized near that of intact animals by five to seven weeks following nerve injury in animals with adrenal medullary transplants. Parallel improvements in sensory responses to innocuous and noxious stimuli were also observed in these animals. The results of this study indicate that peripheral nerve injury can result in severe losses in spinal inhibitory mechanisms, possibly leading to exaggerated sensory processes in persistent pain states. In addition, adrenal medullary transplants may provide a neuroprotective function in promoting recovery and improving long-term survival of GABAergic neurons in the spinal dorsal horn which have been damaged by excitotoxic injury. Copyright ? 1996 IBRO. Published by Elsevier Science Ltd. Key words: inhibitory neurons, nociception, spinal cord, chronic pain, neural transplants, chromaffin cells.

Injury to peripheral nerves often leads to a persistent state of aberrant sensory processing which can be characterized by exaggerated responses to both innocuous and noxious stimuli, and abnormal pain sensations. While these abnormal sensations may be mediated in part by ectopic discharges from injured peripheral nerves, recent evidence suggests that persistent pathological pain involves dysfunctional *Present address: Department of Anesthesiology, Kyoto Prefectural University School of Medicine, Kyoto, 602, Japan. †Present address: Anesthesia Research, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115, U.S.A. §To whom correspondence should be addressed at: Cytotherapeutics Inc., 2 Richmond Square, Providence, RI 02906, U.S.A. (present address). Abbreviations: CCI, chronic constriction injury; GABA-IR GABA-like immunoreactivity; PBS, phosphate-buffered saline.

sensory processing within the CNS.9,26,34 The chronic state of central hyperexcitability following peripheral tissue damage or nerve injury appears to involve neuroplastic remodelling in the spinal cord, and can result in long-term spinal pathology and altered sensory transmission, including persistent increases in spontaneous neuronal activity, expansion of dorsal horn receptive fields, and reduced thresholds to afferent input.9,11,26,41,44,47 A possible mechanism for the maintenance of long-term hyperexcitability and exaggerated sensory processing is an impairment in spinal inhibitory function. It is likely that the barrage of activity in damaged primary afferents and consequent excessive excitatory amino acid release results in excitotoxic insult to particularly vulnerable small inhibitory interneurons.10 In support of this, an increased incidence of hyperchromatic ‘‘dark neurons’’ in the spinal or medullary dorsal horn is consistently found following peripheral nerve injury,

845

846

T. Ibuki et al.

and this can be further exacerbated by pharmacologic blockade of inhibitory neurotransmission.34,35 These dark neurons may be indicative of trans-synaptic degeneration or atrophy and are likely to include functionally impaired inhibitory interneurons.27,34,35 GABA is a major inhibitory neurotransmitter in the CNS. In the spinal cord, both the synthetic enzyme glutamic acid decarboxylase and GABAimmunoreactivity (GABA-IR) are concentrated in the superficial laminae of the dorsal horn (laminae I–III), where sensory, particularly nociceptive, processing predominates.5,9,19,24,25,37 An important role for GABA in sensory processing is suggested by both physiological and behavioural studies, which indicate primarily an inhibitory function in the transmission of noxious stimulation.1,3,18,20,21,23,42,43 The blockade of spinal GABAergic neurotransmission by intrathecal antagonists produces hypersensitivity to innocuous tactile stimuli as if perceived as painful (tactile allodynia), a common symptom following peripheral nerve injury.16,45 Further, tactile allodynia is exacerbated in peripheral nerve-injured animals by GABA antagonists.46 Thus, it is conceivable that a loss of GABAergic inhibitory mechanisms in the spinal dorsal horn leads to sustained hyperexcitability in persistent pain states. In support for this, recent findings indicate a significant reduction in laminae I–III GABA-IR neurons after sciatic nerve transection. Interestingly, an increase in GABA-IR cells in the dorsal horn is induced by peripheral inflammation.7,8 A fairly well characterized model of peripheral nerve injury, the chronic constriction injury model2 (CCI), has been shown to produce many parallels with human peripheral neuropathy syndromes in abnormal sensory processing, including spontaneously generated pain, allodynia, and hyperalgesia. Thus, one goal of the present study was to determine whether parallel changes in GABA-IR in the spinal dorsal horn could be detected during the phases following constriction nerve injury. In addition, work in our laboratory has demonstrated that implantation of adrenal medullary cells into the spinal subarachnoid space can reduce sensory abnormalities following peripheral nerve injury.12,14 Recent findings have suggested that a mechanism of these grafts is reduction of spinal hyperexcitability13,32 and consequent neuropathological processes including the incidence of dark neurons.28 Thus, an additional goal of these studies was to determine whether adrenal medullary grafts can reduce the loss in GABA-IR and promote recovery of spinal inhibitory processes following peripheral nerve injury. EXPERIMENTAL PROCEDURES

proved by the Animal Care Committee, University of Illinois at Chicago. Unilateral chronic constriction nerve injury was induced according to methods originally described by Bennett and Xie.2 Animals were anaesthetized with sodium pentobarbital (40 mg/kg, i.p., supplemented as necessary), and the right common sciatic nerve was exposed on one side at the mid-thigh level using aseptic surgical techniques. Four 4-0 chromic gut ligatures spaced about 1 mm apart were loosely tied around the sciatic nerve proximal to the trifurcation. No surgery was performed on the left side. Following ligation, the musculature was sutured in layers, and the skin was closed with wound clips. Animals were returned to their cages and food and water were available ad libitum. Immunocytochemical procedures In order to visualize GABA-LI, animals were deeply anaesthetized (pentobarbital, 60 mg/kg, i.p.) and immediately perfused intracardially with saline followed by 300 ml 4.0% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). After perfusion, spinal cords were removed, segments at L4–L5 were blocked and postfixed overnight in 4% paraformaldehyde, and immersed in 20% sucrose (phosphate buffered) for cryoprotection. The L4–L5 segments were chosen as these segments receive the majority of sciatic nerve afferents. 10 µm frozen sections were serially cut on a cryostat (Hacker–Bright) and mounted on gelatin-coated glass slides. Sections were incubated in phosphate-buffered saline (PBS; pH 7.4) containing 2.0% normal goat serum for 1 h at room temperature to reduce background staining. Sections were then incubated in a rabbit polyclonal GABA antibody (diluted 1:2000; Sigma, St.Louis, MO. USA) in PBS for 5 h at room temperature and overnight at 4)C. Following incubation in primary antisera, sections were washed three times with PBS, and incubated with fluorescein-conjugated goat anti-rabbit IgG (1:100; Cappel, Durham, NC) for 1 h at room temperature. Primary antisera was eliminated in random sections (approximately every 200 µm) to estimate non-specific staining, which was barely discernible at these concentrations. Following additional washes with PBS, slides were coverslipped with Fluoromount and viewed with an epifluorescence microscope (Zeiss Axiophot). In order to quantitate GABA-IR-positive neuronal profiles, ten sections from each animal were examined. The sections were chosen randomly with a minimum of 100 µm between selected samples, in order to avoid double counting. Cell profiles were counted by a blind observer who had no knowledge of treatment group. The number of GABAIR-positive profiles were counted separately on the sides ipsilateral and contralateral to nerve injury in order to detect any bilateral effects of unilateral peripheral nerve injury. Only cell profiles in Lamina I–III were included, because most positive GABA-IR was concentrated in this area. It should be noted that, using this technique, no attempt was made to analyse changes in cell sizes and shapes. Since portions of cells or cells that may be shrunken (e.g., due to nerve lesion) were not distinguished and would not have been counted, it is possible that reduced counts of GABA-IR cell profiles may reflect cell shrinkage rather than cell loss per se. For the first part of the study, in order to determine the time-course and dynamic nature of alterations in GABA-IR following peripheral nerve injury, animals were perfused at three days, or one, two, three, four, five and seven weeks post-injury (n=2–4 animals/time-point). Adrenal medullary transplants

Peripheral nerve injury Adult male Sprague–Dawley rats (Sasco, Inc., WI) weighing 200–250 g at the beginning of the study were used as both hosts and donors. All animal procedures were ap-

A second group of animals were used to assess the effect of adrenal medullary grafting on GABA-IR losses in animals with chronic peripheral nerve injury. One week following nerve ligation, animals were transplanted with

GABA-IR in the spinal dorsal horn of rats

847

Fig. 1. Low magnification fluorescence photomicrograph of GABA-IR profiles in the dorsal horn of spinal cord. GABA-IR profiles (arrowheads) are oval and are numerous in laninae I–III. A dense network of immunoreactive fibres is apparent in these regions as well. Hatched line indicates medial border of dorsal horn gray matter, dotted line indicates lateral border.

either adrenal medullary or striated muscle (control) tissue in the subarachnoid space of the spinal cord at the level of the lumbar enlargement as described in detail previously.12,29,30 Adrenal medullary tissue for transplantation was obtained from the adrenal glands of adult rats. To prepare adrenal medullary tissue for transplantation, adrenal glands were rapidly removed from donor animals using aseptic techniques. Adrenal medullae were carefully dissected from cortical tissue in ice-cold Hank’s buffer under a dissecting microscope, and cut into small pieces (about 0.5 mm3). Medullary tissue from two adrenal glands were used, as this amount has been shown to consistently reduce neuropathic pain symptoms.12,13,40 The dorsal surface of the spinal cord was exposed via laminectomy at L1–L3. Tissue pieces were transplanted into the spinal cord subarachnoid space of host animals via a slit in the dura and arachnoid membranes. Control animals received equal volumes of striated muscle tissue. Following transplantation, the musculature was closed in layers and the skin was closed with wound clips. Behavioural assessment To confirm reduction in sensory abnormalities in animals with adrenal medullary transplants, a series of tests for allodynia and hyperalgesia was conducted, before nerve injury and one week following nerve injury, and repeated at two, three, five and seven weeks after CCI (equivalent to one, two, four and six weeks post-transplant, respectively). For assessment of cold allodynia, rats were placed on a cold copper plate (5.0&1.0)C) enclosed in a plexiglass cylinder for a 20 min observation period. The cold surface did not evoke hind paw lifting in unoperated rats. However, marked lifting and guarding of the hind paw on the ligated side is characteristic of nerve-injured rats in this model.2,12 The number of hind paw lifts and total duration of lifting

(in seconds) were recorded for both sides. Difference scores were calculated by subtracting the left control hind paw scores from the right nerve ligated hind paw scores. Since there is essentially no lifting on the intact side, positive difference scores are indicative of cold allodynia on the right nerve-injured side.2 The animal’s threshold level to an innocuous tactile stimulus was measured with calibrated von Frey hairs ranging from 0.69 to 75.86 g of force using procedures similar to that described by Tal and Bennett.36 Animals were placed beneath an inverted clear plastic cage (17.5#28.0#12.5 cm) on an elevated wire mesh floor. Von Frey hairs were indented on the hind paw mid-planter skin until they just bent three times at a frequency of about 2/s. At threshold, rats responded with a paw withdrawal. The lowest hair in the series that evoked at least one withdrawal response was considered to be threshold, measured in g. Animals were tested sequentially on the ligated and control sides through the range of von Frey hairs until threshold was reached. To assess the animal’s response to a noxious thermal stimulus, rats were placed beneath an inverted clear plastic cage on an elevated glass floor and allowed to acclimatize for 5 min. A radiant heat source beneath the glass was aimed at the plantar hind paw.15 Withdrawal latencies were determined electronically as the time between onset of the heat source and the hind paw withdrawal response. To avoid tissue damage in the absence of a withdrawal, the apparatus automatically shut off at 15 s. Testing was alternated between hind paws for three trials each with at least 30 s intervals, the average values used for statistical analysis. In addition to raw withdrawal scores, difference scores were calculated by subtracting the withdrawal latencies of the control side from withdrawal latencies of the ligated side, so that negative difference scores indicate thermal hyperalgesia on the nerve-injured side.

848

T. Ibuki et al.

Fig. 2. Fluorescence photomicrographs of GABA-IR in the dorsal horn of lumbar spinal cord. A and B are taken from an intact, unligated animal. GABA-IR neuronal profiles (arrowheads) were found primarily in superficial laminae of the dorsal horn, both in small neuronal cell bodies and within dense fibre plexi. C and D shows reduced GABA-IR in the dorsal horn of an animal three days following constriction injury of the sciatic nerve. C is ipsilateral and D is contralateral to the nerve injury. To assess the effects of grafting on GABA-IR changes in the dorsal horn of nerve-injured animals, rats with transplants were perfused at several time points after CCI: three weeks (n=8, four animals with adrenal and four with muscle transplants), five weeks (n=8, four animals with adrenal and four with muscle transplants) and seven weeks (n=8, four animals with adrenal and four with muscle transplants) after nerve ligation. Tissue was processed for GABA-IR and quantitated as described above. Statistical analysis For each animal, ipsilateral or contralateral GABA-IR profile counts were pooled, and mean&S.E.M. GABA-IR profiles/section were determined. The means were expressed as percentages compared to the respective sides of normal intact animals. GABA-IR profile counts between treatment groups and at various times following nerve injury were compared using ANOVA and the Newman–Keuls test for multiple post hoc comparisons. Behavioural data was analysed using ANOVA for repeated measures and the Newman–Keuls test. In addition, differences in the distribution of withdrawal thresholds to von Frey hairs were compared using the chi-square test.

RESULTS

Time-course of changes in dorsal horn GABA-like immunoreactive neuronal profiles after nerve ligation The appearance and distribution of GABA-IR in control unligated animals is shown in Fig. 1 and Fig. 2A,B. Fig. 1 shows the distribution of GABA-IR in the dorsal horn region. Note that GABA-positive

neuronal profiles were primarily observed in the superficial laminae (laminae I–II), with few profiles observed in deeper laminae (Fig. 1). Thus, follow-up studies were performed using only the superficial laminae for quantification. Most of the GABA-IR cells in this region were small and roundish-shaped, with one or a few apparent processes. In the neuropil surrounding the GABA-positive cell bodies, dense staining of fibres and varicosities was notable (Fig. 2A,B). These findings in control animals were similar to those described by others.7,37 Neuropil staining in the deeper lamina and in the ventral horn were moderate. Motoneurons showed no GABA-IR. In contrast to intact animals, a substantial loss in GABA-IR was already apparent by three days postnerve injury. The loss in GABA-IR neuronal profiles in the superficial laminae was particularly severe on the side ipsilateral to nerve ligation (Fig. 2C). On the contralateral side, GABA-IR cell profiles were still present (Fig. 2D, although the dense plexus of GABA-IR fibres and varicosities was markedly reduced compared to control animals with no peripheral nerve injury. At three weeks following peripheral nerve injury, GABA-IR neuronal profiles were nearly completely absent in the superficial dorsal horn, both ipsilateral and contralateral to the nerve ligation (Fig. 3A,B). In spite of the absence of neuronal profiles, there appear to be some stained elements in the dorsal horn three

GABA-IR in the spinal dorsal horn of rats

849

Fig. 3. Dynamic changes in GABA-IR in the spinal dorsal horn following unilateral peripheral nerve injury. The left side panels (A,C,E) are ipsilateral to the constriction nerve injury. The right side panels (B,D,F) are contralateral to the injury. A and B are taken from an animal at three weeks following injury, when virtually no GABAergic neurons could be identified. C and D are taken from an animal at five weeks post-injury when slight recovery is apparent, although few GABAergic neurons (arrowheads) are identifiable. E and F are taken from an animal at seven weeks following injury, when there is substantial recovery in GABA-IR, particularly on the contralateral side (F), although the recovery is only partial on the side ipsilateral to the nerve injury (E).

weeks following nerve injury. It is possible that these are glial cell profiles, processes, or staining artifact due to cellular damage. The striking loss in GABA-IR neurons was maintained for at least four to five weeks post-injury, when some recovery began to appear (Fig. 3C,D). At five weeks following CCI, a few small GABA-positive cell profiles were identified, primarily on the side contralateral to nerve injury. At seven weeks post-injury, GABA-IR was nearly recov-

ered on the unligated side (Fig. 3F), although there still seemed to be some paucity in GABA-IR fibres and processes. In addition, on the ipsilateral side, an increase in GABA-IR cell profiles was found, but the recovery was not as extensive as on the side contralateral to injury, and did not appear to be completely reversed to normal levels in intact animals. When GABA-IR cell profiles were quantified, significant differences were found both between

850

T. Ibuki et al.

Fig. 4. Changes in GABAergic neuronal profiles in the superficial dorsal horn over time following nerve ligation. Counts prior to nerve injury in normal intact animals are indicated as ‘‘unoperated’’ Times following peripheral nerve ligation are indicated on the horizontal axis. Data are presented as the mean percentages (&S.E.M.) of GABA-IR neuronal profiles compared with unoperated controls. Counts were taken from 10 sections per animal (n=2–4 animals/time point). Changes in GABA-IR neuronal profiles are shown both contralateral (Contra., cross-hatched bars) and ipsilateral (Ipsil., filled bars) to nerve ligation.

contralateral and ipsilateral sides, and throughout the time-course of the study (Fig. 4; overall F1,41 = 125.4; P<0.01). In intact controls, mean numbers of GABA-IR neuronal profiles in Lamina I–III were 18.4&0.3 on the left side and 17.3&0.3 on the right side, with no significant difference between the sides (P>0.05). By three days post-ligation, there was a significant decrease in GABA-positive neuronal profiles on the side ipsilateral to the lesion (P<0.05). GABA-IR cell profiles were reduced to 16.8&5.6% of the numbers found in intact non-lesioned animals. In contrast, on the contralateral side, the decrease in GABA-IR cell profiles (to 74.6&6.8% intact animals) was not significant at three days after nerve injury (P0.05). However, by one week following nerve injury, there were significant decreases (P<0.05) in numbers of GABA-positive neuronal profiles in both ipsilateral and contralateral dorsal horn. The loss in GABA-IR was so severe during the one to three week post-injury period that GABA-positive neuronal profiles were rarely observed. There was some recovery observed on the side contralateral to nerve injury by four to five weeks post-injury (P<0.05 compared to two to three weeks), however, GABA-positive neuronal profiles were still reduced compared to intact controls (P<0.05). At four to five weeks postinjury, the number of GABA-IR neuronal profiles on the ipsilateral side did not show significant recovery compared to two to three weeks (P>0.05). Finally, at seven weeks following CCI, there was a complete recovery of GABA-IR neuronal profiles contralateral to nerve injury (92.3&3.1%; P>0.05 compared to intact controls). There was significant recovery on the ipsilateral side also (P<0.05 compared to one to five week post-injury time-points); however the recovery on the ipsilateral side was still incomplete at this

Fig. 5. Degree of cold allodynia over time following nerve ligation. The number of hind paw withdrawals (A) and cumulative duration (B) of lifting (in s) from a cold surface were measured. Difference scores were calculated by subtracting the scores of the left unoperated hind paw from the right nerve-ligated hind paw. Thus, positive difference scores indicate cold allodynia. Adrenal medullary or control tissue was transplanted in animals one week following nerve ligation. Before nerve ligation (‘B’), animals did not respond to the cold surface, hence difference scores are 0. Times after unilateral nerve ligation are indicated on the horizontal axis. Data are presented as mean&S.E.M. for each transplant group. PRE-ADR TP, cold allodynia scores one week following nerve ligation prior to adrenal transplantation; PRE-CON TP, cold allodynia scores one week following nerve ligation prior to control muscle transplantation; POST-ADR TP, cold allodynia scores in nerve ligated animals at various times following adrenal medullary transplantation; POST-CON TP, cold allodynia scores in nerve ligated animals at various times following control striated muscle transplantation. n=12 animals/transplant group at the beginning of the study (from baseline up to three weeks), eight animals/transplant group at five weeks, and four animals/transplant group at seven weeks.

point (44.9&3.3%; P<0.05 compared to intact controls). Furthermore, a significant difference in the numbers of GABA-positive neuronal profiles was found between the two sides (P<0.05).

Effects of adrenal grafting on sensory behaviours and GABA-like immunoreactivity Prior to peripheral nerve injury, animals did not exhibit hind paw lifting in response to the cold surface (Fig. 5, at baseline B). However, at one week after CCI prior to transplantation, hind paw withdrawals and prolonged lifting was apparent on the

GABA-IR in the spinal dorsal horn of rats

side ipsilateral to ligation in both groups. Animals showed no signs of cold allodynia in the contralateral hind paw. This was reflected in positive difference scores, both in the number of hind paw withdrawals (Fig. 5A; overall F1,5=10.82; P<0.05)) and the cumulative withdrawal duration (Fig. 5B; overall F1,5=7.39; P<0.05)). This cold allodynia continued in the muscle transplant group through five weeks following CCI (P<0.05 compared to pre-injury baselines), with no significant reduction in severity following transplantation (P>0.05 compared to one week after CCI, before transplantation). There appeared to be some recovery in these animals by seven weeks post-injury, particularly in terms of cumulative withdrawal duration (P>0.05 compared to pre-injury

851

baselines), although cold allodynia still persisted when assessed in terms of withdrawal numbers (P>0.05 compared to one week post-CCI). In contrast, in animals with adrenal medullary transplants, cold allodynia was markedly reduced by one week following transplantation (P<0.05 compared to one week post-CCI, before transplantation and compared to muscle implanted animals). Cold allodynia remained reduced in these adrenal medullary grafted animals throughout the remainder of the study. Responses to a noxious thermal stimulus is shown in Fig. 6. In both transplant groups, significant thermal hyperalgesia was induced ipsilateral to the peripheral nerve injury by one week post-ligation, prior to transplantation (Fig. 6A,B, P<0.05 compared to both pre-injury baseline and left unligated hind paw responses). In adrenal medullary transplanted animals, hyperalgesia was reversed by one week following grafting, and no significant differences between responses on the injured compared to non-injured sides were noted throughout the remainder of the study (Fig. 6A; P>0.05). Although some hypoalgesia may have occurred following adrenal medullary transplantation on the side ipsilateral to nerve injury, this was not statistically significant (P>0.05 compared to baseline). In contrast, in control transplanted animals, thermal hyperalgesia on the ligated side was apparent throughout the study (Fig. 6B; P<0.05 compared to baseline and noninjured side), and did not recover even at seven weeks post-CCI (P>0.05 compared to one week following

Fig. 6. Degree of thermal hyperalgesia over time following nerve ligation. 0 indicates responses to noxious thermal stimuli prior to nerve injury in intact animals. Scores were obtained at one week following unilateral nerve ligation, followed by transplantation (at arrows) of either adrenal medullary (A) or control striated muscle (B) tissue in the spinal subarachnoid space. The length of time between presentation of a noxious thermal stimulus and the reflex withdrawal response was recorded as the withdrawal latency (s), at several time-points post-nerve injury as indicated on the horizontal axis. Responses were measured both ipsilateral (ipsil.) and contralateral (contra.) to nerve injury. C shows the withdrawal latencies as difference scores, calculated by subtracting the left unoperated hind paw latency from the right nerve-ligated hind paw latency. Thus, negative difference scores suggest thermal hyperalgesia. before (pre-) and following (post-) adrenal (adr) or control (con) tissue transplantation, presented as difference scores. PREADR TP, thermal hyperalgesia scores one week following nerve ligation prior to adrenal transplantation (open bars); PRE-CON TP, thermal hyperalgesia scores one week following nerve ligation prior to control muscle transplantation (diagonal bars); POST-ADR TP, thermal hyperalgesia scores in nerve ligated animals at various times following adrenal medullary transplantation (solid bars); POST-CON TP, thermal hyperalgesia scores in nerve ligated animals at various times following control striated muscle transplantation (cross-hatched bars). All data are presented as mean&S.E.M. n=12 animals/transplant group at the beginning of the study (from baseline up to three weeks), eight animals/transplant group at five weeks, and four animals/ transplant group at seven weeks.

852

T. Ibuki et al.

Fig. 7. Degree of tactile allodynia over time following nerve ligation. The animal’s threshold response to an innocuous mechanical stimulus, von Frey hairs, was measured in g. Intact unoperated hindpaws (left panels) and nerve-ligated hind paws (right panels) were evaluated prior to nerve injury (BASELINE), and at several weeks following CCI. Following testing at one week post-CCI, animals received either adrenal medullary transplants (ADRENAL TP GROUP, solid bars) or control striated muscle transplants (CONTROL TP GROUP, cross-hatched bars) and were tested again after transplantation at three and five weeks following CCI. The data presented are the frequency (%) of animals that responded to a range of stimuli (vertical axis). n=12 animals/transplant group at the beginning of the study (from baseline up to three weeks) and eight animals/transplant group at five weeks.

GABA-IR in the spinal dorsal horn of rats

nerve injury, prior to transplantation). No significant changes in hind paw withdrawal latencies were noted on the side contralateral to peripheral nerve injury (P>0.05). These findings are shown as difference scores in Fig. 6C. Negative difference scores are indicative of thermal hyperalgesia on the nerveinjured side. After transplantation, the thermal hyperalgesia induced by peripheral nerve injury remained severe in animals with control transplants compared to those with adrenal medullary transplants (overall F1,5=24.18; P<0.01). The distribution of responses to innocuous tactile stimuli before (baseline), one week following CCI (before transplantation), and three and five weeks following CCI (after transplantation) is shown in Fig. 7. Data is only shown up to five weeks following CCI, since the number of animals after this point has been reduced due to perfusion of four animals/group for immunological analyses at three and five weeks postCCI. As a result, it is difficult to present a meaningful frequency distribution at seven weeks following CCI (with only four animals/group still remaining in the study at that time). No significant differences were observed in the response frequency distributions to von Frey hairs prior to peripheral nerve injury (P>0.05). At one week following peripheral nerve injury, there was a significant shift in the response frequency distributions, with more responding to lower range von Frey hairs, on the nerve-injured side compared to both baseline and the intact side in both groups of animals (P<0.01), indicating tactile allodynia. After control muscle transplantation, this shift remained significant at three weeks following peripheral nerve injury (P<0.01), but returned toward baseline levels by five weeks post-injury (P>0.05). In contrast, in adrenal medullary transplanted animals, the tactile allodynia produced by peripheral nerve injury showed partial recovery toward baseline by three weeks post-CCI (P>0.05). The response frequency distribution in these animals was significantly improved at three weeks following peripheral nerve injury compared to control transplanted animals (P<0.01). These findings are shown as cumulative response curves in Fig. 8. In control transplanted animals with peripheral nerve injury, the cumulative response curve is shifted leftward compared to both intact and adrenal medullary transplanted animals (P<0.01). While the response thresholds of less than 30% of the animals with adrenal transplants was below 28.80–75.86 g, over 60% in the control transplant group had thresholds below this range, with approximately 1/3 responding to very light von Frey filaments. The appearance of GABA-IR in the dorsal horns of nerve-injured animals with control transplants or adrenal transplants is shown in Figs 9, 10. At three weeks following unilateral CCI (one week posttransplantation), there is a marked bilateral reduction in GABA-IR, particularly in animals with control transplants (Fig. 9A,B). Both the numbers of

853

Fig. 8. Comparisons between Von Frey responses in normal intact animals (INTACT) and in nerve-injured animals with either adrenal medullary transplants (CCI-ADR TP) or control striated muscle transplants (CCI-CON TP). The graph shows the cumulative percentage of animals (%) with threshold responses to a particular von Frey hair. Data shown is from determinations taken at three weeks following nerve ligation, two weeks following transplantation. n=8 animals/transplant group.

GABA-positive neuronal profiles as well as fibres appear reduced in these animals, similar to that found above in nerve-injured animals without transplants. While GABA-IR was also reduced in nerveinjured animals with adrenal medullary transplants (Fig. 9C,D) compared to intact animals, the reduction was less severe, suggesting some promotion of recovery of GABA-IR in these animals. The enhanced recovery of GABA-IR in animals with adrenal medullary transplants was even more robust at five weeks following peripheral nerve injury, when GABA-IR in these animals was similar to noninjured controls (Fig. 10C,D). In contrast, GABA-IR was still severely reduced bilaterally in nerve-injured animals with control transplants (Fig. 10A,B). When quantified, significant differences were found between the adrenal medullary and muscle transplanted groups (Fig. 11; overall F3,47=21.86; P<0.01). By three weeks following CCI (after transplantation), increases of GABA-IR neuronal profiles were already apparent bilaterally in adrenal medullary transplanted animals (P<0.05 compared to control transplanted animals). Mean numbers of GABApositive cell profiles were further increased in these animals at five weeks following nerve injury, 70.7&2.8% (contralateral) and 76.3&2.6% (ipsilateral) of the levels found in intact animals. In contrast, there was little recovery of GABA-IR profiles in control transplanted animals, which were 21.7+&1.0% (contralateral) and 7.5&7.5% (ipsilateral) of unoperated controls (P<0.05 compared to unoperated controls and adrenal transplanted animals). By seven weeks the number of GABA-positive neurons was similar to levels found in normal noninjured animals (97.6&19.4% and 84.4&10.4% on the contralateral and ipsilateral side, respectively). Some increase in GABA-IR neuronal profiles was also found in control transplanted animals five weeks

854

T. Ibuki et al.

Fig. 9. Photomicrographs of GABA-IR in the spinal dorsal horn of animals with unilateral peripheral nerve injury following control striated muscle transplantation (A,B) or adrenal medullary transplantation (C,D). A and C are ipsilateral to the nerve injury and B and D are contralateral. These are taken from animals three weeks following nerve ligation (two weeks post-transplantation).

following CCI, but this was not statistically significant (P>0.05 compared to three weeks post-injury). By seven weeks, there was full recovery of GABApositive neurons on the side contralateral to nerve injury (91.7&1.8%). However, on the ipsilateral side in control transplanted animals, while there was some recovery (P<0.05 compared to three weeks), it was incomplete even at seven weeks post-injury, being 38.1&12.4% of levels found in intact non-operated animals (P<0.05 compared to both contralateral side and bilaterally in adrenal transplanted animals). These findings of only partial recovery in control transplanted animals was similar to non-transplanted animals described above. DISCUSSION

Findings from this study indicate that peripheral nerve injury can produce severe losses in GABA-IR in the spinal cord of rats, and that recovery can be promoted by adrenal medullary transplants in the spinal subarachnoid space. The dynamic changes in spinal GABA-IR parallel to some degree alterations in behavioural responses to sensory stimuli following peripheral nerve injury. In particular, heightened sensitivity to innocuous stimuli, especially tactile allodynia; peaks in severity during post-injury phases when loss in GABA-IR is maximal, and recovers at

periods when spinal GABA-IR increases. In contrast, there appears to be less association with alterations in sensitivity to noxious stimuli, as heat hyperalgesia is maintained for longer post-injury periods, even when there is some recovery in spinal GABAergic neurons. Thus, a loss in central GABA-IR following peripheral nerve injury may underlie exaggerated sensory processing, particularly of tactile allodynia, a common clinical symptom of peripheral nerve pathologies. In support for this, intrathecal administration of GABA antagonist bicuculline results in prominent tactile evoked allodynia45 and damage to dorsal horn GABAergic neurons by transient ischemia produces hypersensitivity to innocuous stimuli, reversible by administration of GABAB agonist baclofen.16–18 It is interesting that, while bilateral reductions in spinal GABA-IR were found, especially during the one to three week post-injury phases, changes in behavioural responses to both noxious and innocuous stimuli appear to be limited primarily to the side ipsilateral to peripheral nerve injury. In contrast, near normal responses were obtained from the intact hind paw in spite of marked losses in spinal GABA-IR. Similarly, bilateral changes in the spinal cord, such as the appearance of dark neurons or induction of nitric oxide synthase, have been reported following unilateral injury.13,21,33,38 The present findings may indicate that the loss in GABA-IR is merely

GABA-IR in the spinal dorsal horn of rats

855

Fig. 10. Photomicrographs of GABA-IR in the spinal dorsal horn of animals with unilateral peripheral nerve injury following control striated muscle transplantation (A,B) or adrenal medullary transplantation (C,D). A and C are ipsilateral to the nerve injury and B and D are contralateral. These are taken from animals five weeks following nerve ligation (four weeks post-transplantation).

an epiphenomenon consequent to peripheral nerve damage or a more generalized excitability state in the spinal cord rather than an underlying mechanism for altered sensory processing. Alternatively, the production of exaggerated sensory responses may require simultaneous alterations in both the damaged peripheral nerve and the CNS; i.e. a loss in spinal inhibitory processes such as GABAergic neurons could ‘‘set the tone’’ for hyperexcitability to sensory stimuli, which becomes apparent in the presence of abnormal input such as that from injured peripheral nerves.27 In addition, the role of glycinergic neurons, another source of potent inhibitory effects on dorsal horn function, has not been addressed by the present studies, and the presence (or absence) of these neurons may explain laterally distinct pain-related behaviours. In contrast to the findings of the present study using chronic constriction nerve injury, peripheral neurectomy has been reported to produce decreases in spinal GABA-IR restricted to the side ipsilateral to the lesion.6 Further, the severity in loss of GABA-IR at the peak post-nerve injury phases in the present study was quite marked following constriction injury (nearly 100%) compared with that following peripheral neurectomy (approximately 30%6). A possible explanation for these differences is that constriction injury produces a more prolonged barrage of abnormal primary afferent activity than complete neurec-

tomy, resulting in persistent excitotoxic insult to spinal cord neurons. Thus, the loss in particularly vulnerable neurons, such as small inhibitory interneurons may be greater, both in terms of magnitude and extent, with a constriction injury vs a complete neurectomy. In contrast to peripheral nerve injury, peripheral inflammation apparently results in increased GABA-IR in the spinal cord.7 The authors suggest that this may be indicative of the ability to up- or down-regulate GABA content of dorsal horn cells depending on variations in afferent input. Thus, the reduction in GABA-IR following peripheral nerve injury may reflect a down-regulation in GABA synthesis rather than a loss in GABAergic neurons per se. Results of the present study indicate that the decrease in spinal GABA-IR is transient, and recovers to a large extent, at least on the contralateral side, by seven weeks post-nerve injury. This may reflect a temporary down-regulation in GABA synthesis, and argues against cell loss. Similarly, other studies in our laboratory have demonstrated that the appearance of abnormal dark neurons following peripheral nerve injury is transient, as these neurons are no longer apparent at later phases.28 However, in the latter case, it is not possible to distinguish dark neuron recovery from cellular death and removal in the late post-injury phases. In the present studies, there appears to be a general restoration in spinal inhibitory

856

T. Ibuki et al.

Fig. 11. Changes in GABA-IR neuronal profiles at various times following peripheral nerve injury (horizontal axis) and transplantation of either adrenal medullary or control striated muscle tissue. Contra. CON TP, GABA-IR neuronal profiles in dorsal horn of animals with control transplants contralateral to nerve injury (cross-hatched bars); Ipsil. CON TP, GABA-IR neuronal profiles in dorsal horn of animals with control transplants ipsilateral to nerve injury (solid bars); Contra. ADR TP, GABA-IR neuronal profiles in dorsal horn of animals with adrenal medullary contralateral to nerve injury (diagonal bars); Ipsil. ADR TP, GABA-IR neuronal profiles in dorsal horn of animals with adrenal medullary transplants ipsilateral to nerve injury (stippled bars). Data represent percentage (mean&S.E.M.) compared to intact unoperated controls; 10 sections/animal, n=4 animals/group.

processes after a period of time. This may not be complete, since GABA-IR only partially recovers on the ipsilateral side, even seven weeks following injury. Thus, some loss in GABA-IR due to spinal neuronal cell death still remains a possibility. Regardless of whether the loss in GABA-IR following peripheral nerve injury is due to transient down-regulation in GABA synthesis, permanent loss of a subgroup of GABAergic neurons over time, or a combination of both, the transplantation of adrenal medullary tissue in the spinal subarachnoid space results in improved outcome, both in the short run by reducing the recovery period, and in the long run by decreasing the apparent long-term loss in GABAergic neurons. Thus, during intermediate phases following peripheral nerve injury (three to five weeks), adrenal medullary transplants appear to promote recovery in GABA-IR neurons, compared to control transplants. This recovery parallels improvements in allodynia and hyperalgesia by adrenal medullary during periods of peak severity in nonimplanted nerve injured animals, similar to previous findings in our laboratory.12,13 In addition, at later post-injury phases

(seven weeks) in adrenal transplanted animals, the levels of GABAergic neuronal profiles appeared to be fully restored to levels found in non-injured animals, compared to only a partial restoration in control transplanted animals. Similarly previous findings in our laboratory have shown that adrenal medullary implants can reverse the appearance of abnormal dark neurons which may be in the process of degeneration following peripheral nerve injury.28 The precise mechanism for these apparently beneficial effects is unclear, although it most likely involves the release of neuroprotective agents from the transplanted tissue. Adrenal medullary chromaffin cells produce a variety of neurotrophic factors, neuropeptides, and cytokines with neurotrophic activity, a ‘‘trophic cocktail’’.39 While the agent responsible for neuroprotective effects has not been identified, chromaffin cells can apparently constitutively release neurotrophic factors, since media conditioned by purified chromaffin cells can promote the survival of several neuronal cultures of CNS and PNS origin.22 The provision of neurotrophic factors by adrenal medullary or chromaffin cell grafts has also been suggested as a possible mechanism of some of the functional improvements in Parkinson’s and Huntington’s disease models.4,31 Thus, it is conceivable that transplanted chromatin cells can exert a neuroprotective effect on vulnerable inhibitory GABAergic neurons which have been damaged by excitotoxic injury.

CONCLUSIONS

In summary, the results of this study indicate that peripheral nerve injury can result in central changes in spinal neurocircuitry, notably a loss in dorsal horn inhibitory mechanisms which may account for exaggerated sensory processes in persistent pain states. To some extent, the reduction in GABA-IR appears to be transient, indicative of a down-regulation in GABA synthesis, although some GABAergic neurons may be permanently destroyed since long-term recovery is only partial. Transplants of adrenal medullary tissue can promote recovery and improve long-term survival of GABAergic neurons in the spinal dorsal horn, possibly by providing a source of neurotrophic factors during critical post-injury phases. Acknowledgements—This work was supported in part by NIH grant NS25054.

REFERENCES

1.

Aanonsen L. M. and Wilcox G. L. (1989) Muscimol, GABAA receptors and excitatory amino acids in the mouse spinal cord. J. Pharmac. exp. Ther. 248, 1026–1033. 2. Bennett G. J. and Xie Y.-K. (1988) A peripheral mononeuropathy in rat that produces disorders of pain sensation like those seen in man. Pain 33, 87–107. 3. Beyer C., Banas C., Gonzalea-Florez O. and Komisaruk B. R. (1989) Blockage of substance P-induced scratching behavior in rats by the intrathecal administration of inhibitory amino acid antagonists. Pharmac. Biochem. Behav. 34, 491–495.

GABA-IR in the spinal dorsal horn of rats 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

18.

19.

20. 21. 22. 23. 24.

25. 26. 27. 28. 29. 30. 31.

32. 33. 34. 35.

857

Bohn M. C., Cupit C., Marciano F. and Gash D. M. (1987) Adrenal medullary grafts enhance recovery of striatal dopaminergic fibers. Science 237, 913–916. Carlton S. M. and Hayes E. S. (1991) Light microscopic and ultrastructural analysis of GABA-immunoreactive profiles in the monkey spinal cord. J. comp. Neurol. 300, 162–182. Castro-Lopes J. M., Tavares I. and Coimbra A. (1993) GABA decreases in the spinal cord dorsal horn after peripheral neurectomy. Brain Res. 620, 287–291. Castro-Lopes J. M., Tavares I., Tolle T. R. and Coimbra A. (1994) Carageenan-induced inflammation of the hind foot provokes a rise of GABA-immunoreactive cells in the rat spinal cord that is prevented by peripheral neurectomy or neonatal capsaicin treatment. Pain 56, 193–201. Castro-Lopes J. M., Tavares I., Tolle T. R., Coito A. and Coimbra A. (1992) Increase in GABAergic cells and GABA levels in the rat spinal cord in unilateral inflammation of the hindlimb in the rat. Eur. J. Neurosci. 4, 296–301. Coderre T. J., Katz J., Vaccarino A. L. and Melzack R. (1993) Contribution of central neuroplasticity to pathological pain: review of clinical and experimental evidence. Pain 52, 259–285. Dubner R. (1991) Neuronal plasticity in the spinal arid medullary dorsal horns: a possible role in central pain mechanisms. In Pain and Central Nervous System Disease: The Central Pain Syndromes (ed. Casey K. L.), pp. 143–155. Raven Press, New York. Dubner R. and Ruda M. A. (1992) Activity-dependent neuronal plasticity following tissue injury and inflammation. Trends Neurosci. 15, 96–103. Hama A. T. and Sagen J. (1993) Reduced pain-related behavior by adrenal medullary transplants in rats with experimental painful peripheral neuropathy. Pain 52, 223–231. Hama A. T. and Sagen J. (1994) Induction of spinal NADPH-diaphorase by nerve injury is attenuated by adrenal medullary transplants. Brain Res. 640, 345–351. Hama A. and Sagen J. (1994) Alleviation of neuropathic pain symptoms by xenogeneic chromaffin cell grafts in the spinal subarachnoid space. Brain Res. 651, 183–193. Hargreaves K. M., Dubner R., Brown F., Flores C. and Joris J. (1988) A new and sensitive method for measuring thermal nociception in cutaneous hyperalgesia. Pain 32, 77–88. Hao J.-X, Xu X.-J. and Wiesenfeld-Hallin Z. (1994) Intrathecal ô-aminobutyric acidB (GABAB) receptor antagonist CGP 35348 induces hypersensitivity to mechanical stimuli in the rat. Neurosci. Lett. 182, 299–302. Hao J.-X., Xu X.-J., Yu Y.-X., Seiger A. and Wiesenfeld-Hallin Z. (1991) Hypersensitivity of dorsal horn wide dynamic range neurons to cutaneous mechanical stimuli after transient spinal cord ischemia in the rat. Neurosci. Lett. 128, 105–108. Hao J.-X, Xu X.-J., Yu Y.-X., Seiger A } . and Wiesenfeld-Hallin Z. (1992) Baclofen reverses the hypersensitivity of dorsal horn wide dynamic range neurons to mechanical stimulation after transient spinal cord ischemia; implications for a tonic GABA-ergic inhibitory control of myelinated fiber input. J. Neurophysiol. 68, 392–396. Hunt S. P., Kelly J. S., Emson P. C., Kimmel J. R., Miller R. J. and Wu J.-Y. (1981) An immunohistochemical study of neuronal populations containing neuropeptides or ô-aminobutyrate within the superficial layers of the rat dorsal horn. Neuroscience 6, 1883–1898. Hwang A. S. and Wilcox G. L. (1989) Baclofen, ô-aminobutyric acidB and substance P in the mouse spinal cord. J. Pharmac. exp. Ther. 248, 1026–1033. Kangraga I., Jiang M. and Randic M. (1991) Actions of (")-baclofen on rat dorsal horn neurons. Brain Res. 562, 265–275. Lachmund A., Gehrke D., Krieglstein K. and Unsicker K. (1994) Trophic factors from chromaffin granules promote survival of peripheral and central nervous system neurons. Neuroscience 62, 361–370. Levy R. A. and Proudfit H. K. (1977) The analgesic action of baclofen (â-(4-chlorophenyl)-ô-aminobutyric acid). J. Pharmac. exp. Ther. 202, 437–445. Magoul R., Onteniente B., Geffard M. and Calas A. (1987) Anatomical distribution and ultrastructural organization of the GABAergic system in the rat spinal cord. An immunocytochemical study using anti-GABA antibodies. Neuroscience 20, 1001–1009. McLaughlin B. J., Barber R., Saito K., Roberts E. and Wu J.-Y. (1975) Immunocytochemical localization of glutamate decarboxylase in rat spinal cord. J. comp. Neurol. 164, 305–322. McMahon S. B., Lewin G. R. and Wall P. D. (1993) Central hyperexcitability triggered by noxious inputs. Curr. Opin. Neurobiol. 3, 602–610. Nachemson A. K. and Bennett G. J. (1993) Does pain damage spinal cord neurons? Transsynaptic degeneration in rat following a surgical incision. Neurosci. Lett. 162, 78–80. Sagen J., Pappas G. D. and Hama A. T. (1994) Incidence of dark neurons in spinal cord of animals with neuropathic pain is reduced following adrenal medullary tissue transplantation. Soc. Neurosci. Abstr. 20, 552. Sagen J., Pappas G. D. and Perlow M. J. (1986) Adrenal medullary tissue transplants in the rat spinal cord reduce pain sensitivity. Brain Res. 384, 189–194. Sagen J., Wang H. and Pappas G. D. (1990) Adrenal medullary implants in the rat spinal cord reduce nociception in a chronic pain model. Pain 42, 69–79. Sanberg P. R., Emerich D. F., McDermott P. E., Frydel E. R., Bertine A. M., Kaplan F. A., Palmatier M. S. and Christenson L. (1991) Transplantation of polymer encapsulated bovine adrenal cells prevent quinolinic acid induced lesions of the striatum. Soc. Neurosci. Abstr. 17, 903. Siegan J., Hama A. T. and Sagen J. (1995) Attenuation of NMDA-induced spinal hypersensitivity by adrenal medullary transplants. Brain Res. 680, 88–98. Solodkin A., Traub R. J. and Gebhart G. F. (1992) Unilateral hindpaw inflammation produces a bilateral increase in NADPH-diaphorase histochemical staining in the rat lumbar spinal cord. Neuroscience 51, 495–499. Sugimoto T., Bennett G. J. and Kajander K. C. (1990) Transsynaptic degeneration in the superficial dorsal horn after sciatic nerve injury: effects of a chronic constriction injury, transection, and strychnine. Pain 42, 205–213. Sugimoto T., Takemura M., Sakai A. and Ishimaru M. (1987) Rapid transneuronal destruction following peripheral nerve transaction in the medullary dorsal horn is enhanced by strychnine, picrotoxin and bicuculline. Pain 30, 385–393.

858 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47.

T. Ibuki et al. Tal M. and Bennett G. J. (1994) Extra-territorial pain in rats with a peripheral mononeuropathy:mechanohyperalgesia and mechano-allodynia in the territory of an uninjured nerve. Pain 57, 375–382. Todd A. J. and McKenzie J. (1989) GABA-immunoreactive neurones in the dorsal horn of the rat spinal cord. Neuroscience 31, 799–806. Traub R. T., Solodkin A. and Gebhart G. F. (1994) NADPH-diaphorase: histochemistry provides evidence for a bilateral, somatotopically inappropriate response to unilateral hindpaw inflammation in the rat. Brain Res. 647, 113–123. Unsicker K. (1993) The trophic cocktail made by adrenal chromatin cells. Expl Neurol. 123, 167–173. Wang H. and Sagen J. (1994) Optimization of adrenal medullary allograft conditions for pain alleviation. J. Neural Transpl. Plastic. 5, 49–64. Wilcox G. L. (1991) Excitatory neurotransmitters and pain. In Proceedings of The VIth World Congress on Pain (eds Bond M. R., Carlton J. E. and Woolf C. J.), pp. 97–117. Elsevier, Amsterdam. Willcockson W. S., Chung J. M., Hori Y., Lee K. H. and Willis W. D. (1984) Effects of iontophoretically released amino acids and amines on primate spinothalamic tract neurons. J. Neurosci. 4, 732–740. Wilson P. R. and Yaksh T. L. (1978) Baclofen is antinociceptive in the spinal intrathecal space of animals. Eur. J. Pharmac. 51, 323–330. Woolf C. J. and Thompson S. W. N. (1991) The induction and maintenance of central sensitization is dependent on N-methyl--aspartic acid receptor activation;implications for the treatment of post-injury pain hypersensitivity states. Pain 44, 293–299. Yaksh T. L. (1989) Behavioral and anatomic correlates of the tactile evoked allodynia produced by spinal glycine inhibition: effects of modulatory receptor systems and excitatory amino acid antagonists. Pain 37, 111–123. Yamamoto T. and Yaksh T. L. (1993) Effects of intrathecal strychnine and bicuculline on nerve compression-induced thermal hyperalgesia and selective antagonism by MK-801. Pain 54, 79–84. Zieglga¨nsberger W. and To¨lle T. (1993) The pharmacology of pain signalling. Curr. Opin. Neurobiol. 3, 611–618. (Accepted 24 July 1996)