€ For information about subscribing or purchasing offprints of the published version, with commentaries and author's response, write to: journals_subscriptions@cup.org (North America) or journals_marketing@cup.cam.ac.uk (All other countries). €

CENTRAL INHIBITORY DYSFUNCTIONS: MECHANISMS AND CLINICAL IMPLICATIONS

Keywords

pain; neuropathy; axotomy; ischaemia; spinal cord; g-aminobutyric acid; GABA; cholecystokinin; morphine; opioids; dysinhibition.

Abstract

Injury to the central or peripheral nervous system is often associated with persistent pain. After ischemic injury to the spinal cord, rats develop severe mechanical allodynia-like symptoms, expressed as a pain-like response to innocuous stimuli. In its short-lasting phase the allodynia can be relieved by the g-aminobutyric acid (GABA)-B receptor agonist baclofen, which also reverses the hyperexcitability of dorsal horn interneurons to mechanical stimuli. Furthermore, there is a reduction in GABA immunoreactivity in the dorsal horn of allodynic rats.

Clinical neuropathic pain of peripheral and central origin is often not relieved by opiates at doses that do not cause side effects. The loss of sensitivity to opiates may be associated with the upregulation of endogenous anti-opioid substances, such as the neuropeptide cholecystokinin (CCK). CCK and its receptor (CCK-R) protein is normally not detectable in rat dorsal root ganglion cells. After peripheral nerve section both CCK and CCK-R are upregulated in the dorsal root ganglia. Furthermore, CI 988, an antagonist of the CCK-B receptor, chronically coadministered with morphine reduces autotomy, a behavior that may be a sign of neuropathic pain following peripheral nerve section. Thus, opiate insensitivity may be due to the release of CCK from injured primary afferents. Similarly, in the chronic phase of the spinal ischemic model of central pain, the allodynia-like symptom is not relieved by systemic morphine, but is significantly reversed by the CCK-B antagonist. Consequently, upregulation of CCK and CCK-R in the CNS may also underly opiate drug insensitivity following CNS injury. Thus, dysfunction of central inhibition involving GABA and endogenous opioids may be a factor underlying the development of sensory abnormalities and/or pain following injury to neural tissue.

Introduction

Injury of the peripheral (PNS) or central (CNS) nervous system is often associated with pain. Dysfunction of central inhibition involving the inhibitory neurotransmitters g-aminobutyric acid (GABA) and endogenous opioids, as well as the anti-opioid neuropeptide cholecystokinin (CCK), may be a factor underlying the development of sensory abnormalities and/or pain following injury to neural tissue. In this review the possible common dysfunction of these inhibitory systems in the generation of neuropathic pain as a consequence of PNS and CNS tissue injury will be discussed.

Models of neuropathic pain following PNS and CNS injury

Experimental neuropathic pain following peripheral nerve injury

After peripheral nerve section (axotomy), a neuroma develops when the proximal nerve stump has no possibility to regenerate and reinnervate the periphery. Wall and Gutnick (1974) were the first to demonstrate that an ongoing abnormal discharge originating in the neuroma develops rapidly after peripheral nerve section. The neuroma is also sensitive to mechanical pressure and stimulation by adrenergic agonists (Wall and Gutnick, 1974). Furthermore, the deafferented dorsal root ganglion (DRG) cells also become generators of ongoing discharges (Wall and Devor, 1983). Abnormal discharges can also be recorded from spinal cord interneurons following deafferentation by dorsal rhizotomy (Lombard and Larabi, 1983). Thus, there are functional changes at various levels of the nervous system following peripheral nerve injury that may underly the development of neuropathic pain.

A model of experimental neuropathic pain following injury to primary afferents is autotomy behaviour (self-mutilationon of the deafferented body region) after section of major peripheral nerves. This behaviour was proposed by Wall et al. (1979) to be a response to unpleasant and perhaps painful paresthesias referred to the denervated limb. The development of autotomy has also been observed after multiple dorsal rhizotomy (Albe-Fessard et al., 1979), after destruction of DRG (Wiesenfeld-Hallin et al., 1987) and after spinal nerve lesions ( Lombard et al., 1979; Wiesenfeld and Lindblom, 1980). An advantage of the axotomy model of neuropathic pain over other models of partial nerve injury that have more recently become available (e.g. Bennett and Xie, 1988; Seltzer et al., 1990) is its robustness and reproducibility in a large number of laboratories. However, autotomy does not represent all forms of neuropathic pain of peripheral origin, but only deafferentation pain corresponding to phantom limb pain and anesthesia dolorosa. Pain due to partial nerve injury has other characterstics, including hyperalgesia to thermal and mechanical stimuli (see Bennett, 1994 for review).

A model of central pain following spinal cord ischemia

Central pain is caused by a lesion or dysfunction of the CNS (Merskey, 1986). Application of excitatory agents on the dorsal horn or the trigeminal nucleus can induce hyperalgesic and allodynic responses in animals (Dyken, 1965; Black, 1974; Beyer et al., 1985; Yaksh, 1989). These models however do not involve a lesion to the CNS and because of the short duration of the pain-like symptoms they do not represent a clinical model of central pain. We have studied the effects of a photochemically-induced spinal ischemia which appears to be a useful model of central pain with a number of features that are similar to clinical pain. (Hao et al. 1991a,b; 1992a-e, Xu et al., 1992b, see Wiesenfeld-Hallin et al., 1994 for review). The major sensory symptom is the development of mechanical allodynia, a pain-like response to innocuous stimuli, which has a short-lasting (days) and chronic (permanent) phase. The response to thermal stimuli is unaffected, indicating that the allodynia-like symptom is mediated by low threshold mechanoreceptors, rather than heat nociceptors. The neuronal dysfunction related to the development of allodynia probably involves excitotoxicity through the activation of the N-methyl-D-aspartate (NMDA) receptor by glutamate since pretreatment with the NMDA-receptor antagonist MK-801 prevented the development of this sensory dysfunction (Hao et al., 1992a). There is also considerable evidence that the short-lasting allodynia is mediated by a loss of spinal GABA-ergic inhibition, whereas the chronic allodynia is due to a dysfunction of the endogenous opioid system (see below).

The role of GABA-ergic mechanisms in pain following injury to the PNS and CNS

Dysfunction of GABA-ergic inhibition after peripheral nerve injury

Immunohistochemical studies, using antisera to glutamate decarboxylase (GAD) and conjugates of GABA, have established that GABA-like immunoractive (GABA-LI) neurons are present in the mammalian dorsal horn of the spinal cord (see Todd and Spike, 1993 for review). The GABA-LI cells are particularly concentrated in Rexed's laminae I-III. It is well documented that GABAergic interneurons in dorsal spinal cord regulate the activity of many types of sensory afferents, predominantly through presynaptic inhibition on input from both A- and C-primary afferents, although postsynaptic actions of GABA agonists have also been reported (see Willis and Coggeshall, 1991 for review). Primary afferent depolarization (PAD) is an important mechanisms that reduces transmitter release from the terminals of primary afferents into the spinal cord through presynaptic inhibition (Eccles et al., 1963). PAD is believed to involve spinal GABA inhibitory interneurons since nonspecific GABA antagonists reduce PAD (Banna and Jabbur, 1969). Axotomy of peripheral nerve is associated with decreased capacity of the spinal cord to generate PAD (Wall and Devor, 1981) which may be associated with the possible dysfunction of GABA-ergic interneurons following peripheral nerve injury (Castro-Lopes et al., 1993). Thus, after nerve section loss of GABA-ergic inhibition may be one of the factors leading to the development of neuropathic pain. This was supported by the finding that the GABA-B receptor agonist baclofen reduced the persistent expression of the immediate early gene c-Fos in axotomized rats (Basbaum et al., 1991). Furthermore, degenerated ("dark") neurons have been found in spinal cord laminae I-II in rats with chronic constriction injury of the sciatic nerve. This effect was accentuated by the repeated administration of strychnine (Sugimoto et al., 1990). It is possible that these "dark neurons" are GABA-ergic and may underly loss of endogenous inhibition in this model of neuropathy following peripheral nerve injury since the GABA-B receptor agonist baclofen has been found to cause antinociception in this model (Smith et al., 1994).

Dysfunction of the GABA-ergic system after spinal cord ischemia

The pharmacological basis of short-lasting behavioral allodynia has been investigated (Hao et al., 1991b; 1992e; Xu et al., 1992b). A large number of analgesics applied systemically, including morphine at non-sedative doses, were ineffective, but the GABA-B receptor agonist baclofen dramatically reduced both tactile-evoked agitation and increased vocalization threshold to mechanical pressure in a dose-related manner. However, the GABA-A agonist muscimol was ineffective.

The effect of baclofen on the response of single dorsal horn wide dynamic range (WDR) neurons was tested in normal and allodynic rats (Hao et al., 1992c, d). In normal rats, subcutaneous electrical stimulation of the receptive field or the sciatic nerve at an intensity that activated both myelinated (A) and unmyelinated (C) fibers evoked a biphasic response in all WDR neurons with a short-latency A-fiber response and a long-latency C-fiber response. In the majority of WDR neurons recorded in allodynic rats, electrical stimulation evoked a single burst with no separation between responses to A- and C-fiber input. Detailed analysis of the poststimulus histogram showed that the number of discharges evoked in WDR neurons in response to electrical stimulation in allodynic rats was significantly higher than in normal rats at all post-stimulus intervals. The results also suggested that the myelinated A-fiber input was prolonged and enhanced during allodynia. The response of WDR neurons in normal rats to pressure by von Frey hairs increased linearly. Nearly all WDR neurons recorded in allodynic rats exhibited increased sensitivity to mechanical stimulation. The pressure-response curve was exponential and dramatically shifted to the left with a significant lowering of the threshold for evoking neuronal responses. In contrast, the response of WDR neurons to heat stimulation was similar in allodynic and normal rats, indicating a lack of involvement of C-afferent input to the spinal cord in the geneneration of abnormal responses and collaborating the results of behavioral tests.

Pretreatment with baclofen in normal rats did not alter the response pattern of WDR neurons to electrical stimulation. However, baclofen significantly depressed the responses of WDR neurons to intense, but not to innocuous, mechanical stimuli. The mechanical threshold of WDR neurons in normal animals pretreated with baclofen did not differ from that of rats without any drug treatment. In allodynic rats, pretreatment with baclofen normalized the response pattern of the neurons to electrical stimulation. The hypersensitivity of the WDR neurons to low-intensity mechanical stimulation in allodynic animals was totally reversed by baclofen and the depression of the response to intense stimuli was the same as in normal rats.

From the results of behavioral, physiological and pharmacological studies, it is apparent that allodynia following spinal cord ischemia is mediated predominantly by abnormal input from myelinated afferents, as both behavioral and elctrophysiological studies demonstrated that in the presence of mechanical hypersensitivity. In contrast, the behavioral response and the physiological response of dorsal horn WDR neurons to noxious heat stimulation was unchanged. Systemic low dose baclofen relieved both behavioral allodynia and neuronal hypersensitivity, suggesting that the neuronal hyperexcitability underlying short-term behavioral allodynia is induced by loss of GABA-ergic inhibitory control (Game and Lodge, 1975; Price et al., 1987), which may result from a high susceptibility of GABA-ergic neurons to EAA mediated neurotoxicity (Sloper et al., 1986). These results thus indicate that inihibition of myelinated, low threshold afferents is mediated through GABA-B receptors and operates tonically under normal conditions (figure 1).

[THESE ARE FIGURE CAPTIONS ONLY:
FIGURES THEMSELVES ARE ONLY AVAILABLE IN THE PAPER VERSION]

Figure 1: Illustration of the mechanisms underlying short-term allodynia. The normal organization of the inhibitory and excitatory mechanisms of the dorsal horn of the spinal cord that influence the response of wide dynamic range cells is shown. Fort the sake of simplicity, descending systems have been omitted. After reversible spinal cord ischemia the GABA-ergic interneurons become dysfunctional because of the excitotoxic effect of glutamate acting on NMDA receptors. The presynaptic inhibitory function of GABA-ergic interneurons on low threshold myelinated afferents is disrupted, leading to short-term mechanical allodynia. Hyperalgesia to thermal stimuli, mediated by unmyelinated afferents, does not occur because the endogenous opioid system is not disturbed by the ischemia that leads to short-term allodynia. Reproduced from Wiesenfeld-Hallin et al., 1994 with permission.

The role of the GABA-ergic system in short-lasting allodynia following spinal ischemia was recently examined with immunohistological methods (Zhang et al., 1994). The number of GABA-like immunoreactive cells in laminae I-III of the lumbar dorsal horn was significantly decreased bilaterally during the presence of allodynia compared to cervical levels and sham-operated controls. The number of GABA-immunoreactive cells was restored after recovery from allodynia, indicating a strong correlation between the spinal level of GABA and the presence of pain-like response in corresponding dermatomes. The recovery of GABA-immunoreactivity indicated that the dysfunction of GABA-ergic interneurons was temporary and recovery of function is possible. The reduction of GABA immunoractivity in the spinal cord following ischemia is paralleled by reduced spinal GABA levels following ischemia as measured with biochemical technique (Martiniak et al., 1991), indicating that the immunohistochemical results may actually reflect reduced amino acid levels.

The pharmacological basis of opiate insensitivity following injury to the PNS and CNS

Opiate insensitive forms of pain in the clinic

In the clinic it is desirable that a thorough examination and analysis of the patients' pain preceeds treatment. This may include quantitative sensory testing of peripheral nerve function (Lindblom, 1985; Linblom and Hansson, 1991), the morphine test (ArnŽr, 1991a), the phentolamine test (ArnŽr, 1991b) and a pain questionnaire (Carlsson, 1984), including pain drawings (Schwartz and DeGood, 1984). Consequently, nociceptive, neurogenic, idiopathic and psychogenic forms of pain can often be identified and treatment can be adjusted to suit the patient (ArnŽr, 1991a). For example, patients suffering from acute nociceptive and inflammatory pain after surgery are usually successfully treated with traditional drugs, including opiates. But pain originating from damage to peripheral nerve or the CNS is often long lasting or chronic and presents a severe problem in the pain clinic and has to be treated through alternative approaches. Thus, low efficacy of opiates in treating pain after infiltration of peripheral nerve or nerve plexus by cancerous tumors has been reported (ArnŽr and ArnŽr, 1985). Other examples of opiate insensitive pain are phantom limb pain following amputation (Sherman et al., 1980) and pain after spinal injury with deafferentation (Glynn et al., 1986). Although many different treatments have been suggested to reduce neurogenic pain, once established it is still a serious and difficult clinical problem (ArnŽr, 1991a).

The question of opiate insensitive forms of pain has been intensively debated in the clinical literature as some authors maintain that opiate treatment is useful if very high doses are used (Portenoy et al., 1990; McQuay, 1988). ArnŽr and Meyerson (1988, 1991) have proposed that opiates have less efficacy in neuropathic than nociceptive pain. Thus, the presence or absence of opiate insensitivity needs to be established for each patient. For patients unresponsive to analgesic doses of opiates that do not cause undesirable side effects, alternate methods of treatment must be considered. In milder forms of neurogenic pain transcutaneous nerve stimulation and antidepressant drugs are important therapeutic tools (Meyerson, 1990). In the pain clinic, intraspinal (epidural or intrathecal) administration of a2 adrenoceptor agonists like clonidine or local anaesthetics have been used as complements to morphine (Glynn et al., 1993).

Experimental evidence for opiate insensitivity

Although opiate insensitivity is an important clinical problem, there are relatively few experimental studies examining this issue. Furthermore, the results of these studies have been inconclusive. A high dose of morphine (240 µg/d) applied i.t. continuously, starting at the time of axotomy and the following 14 d was associated with decreased level of autotomy in rats (Wiesenfeld-Hallin, 1984). These results speak against the presence of opiate insensitivity. However, in this study the positive effect of morphine on autotomy may have been due to the preemptive analgesic effect of the drug (Puke and Wiesenfeld-Hallin, 1993) since morphine was injected in conjunction with the nerve injury. Indirect evidence for reduced sensitivity to morphine in neuropathic pain is the observation that after deafferentation a reduction of µ, d, and k opioid receptors have been reported in laminae I-III of the dorsal horn (Gouardres et al 1991). The reduction of opioid receptors seems to be correlated to the degree of deafferentation, rhizotomy > sciatic nerve section > constriction nerve injury (Besse at al., 1992). There are data from rats with nerve constriction injury indicating that pain-related behavior in this model can be effectively relieved with opiates (Jazat and Guilbaud, 1991; Attal et al., 1991). However, the downregulation of opioid receptors after peripheral nerve section seems to be temporary (Besse at al., 1992), suggesting that reduced opiate sensitivity after nerve section may only partly depend on downregulation of opioid receptors.

In support of the reduction of the analgesic effect of morphine following nerve section, the threshold dose of morphine required to depress the flexor reflex was 3-5 times higher in autotomizing rats following sciatic nerve section than in rats with intact nerves or axotomized rats that were not autotomizing (Xu and Wiesenfeld-Hallin, 1991). However, higher doses of morphine were equally effective in depressing the flexor reflex in all groups. These results indicate that the sensitivity to, rather than the effectivenes, of morphine-induced antinociception was reduced following axotomy. Evidence will be presented that this phenomenon may be a result of an intrinsic antiopioid mechanism involving the neuropeptide CCK.

The possible role of cholecystokinin in opiate insensitivity following PNS injury

CCK is normally present in the brain as the sulphated C-terminal fragment CCK-8 and fulfills many of the criteria for a neurotransmitter with multiple functions (Vanderhaegen and Crawley, 1985). CCK has been documented to have an opiate antagonistic property (Itoh et al., 1982.; Faris et al., 1983; Watkins et al., 1984; Wiesenfeld-Hallin and Duranti, 1987; Baber et al., 1989; Dourish et al., 1990; Wiesenfeld-Hallin et al., 1990a; Stanfa et al., 1994). CCK receptors have been found to be heterogeneous, in rodents the CCK-A receptor has been found primarily in peripheral tissue and the CCK-B receptor in the CNS (Moran et al., 1986). Potent and highly selective antagonists of the CCK-B receptor have made it possible to investigate the role of CCK in the CNS. Thus, CCK antagonists have been found to enhance the analgesic effect of morphine and to reduce the development of morphine tolerance (Baber et al., 1989; Dourish et al., 1990; Xu et al., 1992a). We have provided evidence that plasticity of the CCK system may be involved in the reduced effect of morphine in the autotomy model of experimental neuropathic pain.

We examined the effect of sciatic nerve section on the presence of CCK mRNA in rat DRG cells. Furthermore, the effect of CI 988, an antagonist of the CCK-B receptor (Hughes et al., 1990), which was previously shown to potentiate the analgesic effect of morphine (Wiesenfeld-Hallin et al., 1990a), on autotomy behaviour was also examined (Xu et al, 1993). After sciatic nerve section up-regulation of CCK mRNA in the ipsilateral L4 and L5 DRG was observed with in situ hybridisation technique. Only few CCK mRNA positive cells were seen in control ganglia and in ganglia contralateral to the nerve transection. Fourteen days after axotomy up to 30% of all DRG cells on the nerve sectioned side were CCK mRNA positive. Most of these cells were of the small type. CCK receptor protein also became expressed on DRG cells of all sizes (see Hškfelt et al., 1994 for review).

During an observation period of 15 d the occurrence, development and severity of autotomy were examined in five groups of rats injected with i.t. saline, s.c. saline, s.c CI 988, s.c. saline followed by i.t morphine or s.c. CI 988 followed by i.t. morphine. The drugs were administered twice daily, starting 24 h after axotomy. The rats injected with CI 988 and morphine autotomized significantly less than the other four groups during the 15 d observation period. The results of this behavioural study therefore suggested that coadministration of CI 988 and morphine produced significant suppression of autotomy behaviour in after peripheral nerve section.

As discussed above, the decreased sensitivity of the spinal cord to morphine is probably not just a result of a reduction of µ-receptors in the spinal cord. Therefore down-regulation of opioid receptors does not seem to be an important factor in decreased sensitivity to morphine after peripheral nerve section. Development of morphine tolerance could also decrease the effectiveness of morphine analgesia. Tolerance to morphine could be expected if an initially decreased autotomy was followed by increased autotomy. No such effect was observed. Thus, there may be other mechanisms involved in the lack of effect of morphine after nerve section.

CCK normally exists in spinal cord interneurones and descending pathways, but rarely in primary afferents, in rat (Hškfelt et al., 1988; Ju et al., 1986; Williams et al., 1987; Zhang et al., 1993). Verge et al. (1993) have recently observed that mRNA for CCK is upregulated in rat DRG after axotomy and even CCK-B receptor mRNA is increased in rat DRG after axotomy (Zhang et al., 1993). The results of a parallel behavioral study (Xu et al., 1993) indicated that the combination of morphine and CI 988 significantly suppressed autotomy behaviour after sciatic nerve section. The tendency to reduced rate of autotomy in the group that received only CI 988 may be the result of increased efficacy of the intrinsic opioid systems during treatment with the CCK-B antagonist, indicating that a tonic CCK-ergic control of the endogenous antinociceptive system may be present following peripheral nerve injury. This is supported by recent physiological data indicating that, like in normal animals, CI 988 potentates the antinociceptive effect of morphine in axotomized rats (Xu et al., 1994b).

These results suggest that after peripheral nerve section there is a markedly increased production of CCK and CCK-receptor protein in the DRG. The altered expression of CCK is in line with other reports showing complex changes in the levels of neuropeptides and their receptors in primary sensory neurons and spinal cord after peripheral nerve section (Hškfelt et al., 1994). The increased expression of CCK in DRG after nerve section and the observation that CI 988 in combination with morphine can reduce autotomy behaviour strongly suggest that the ineffectiveness of morphine in this experimental neuropathic pain model is related to the CCK system. Thus, after peripheral nerve injury it is possible that upregulation of CCK synthesis and an increased release of CCK from primary afferents in the spinal cord antagonizes the effect of exogenously administered or endogenously released opioids. CCK may be also involved in the development of tolerance to morphine because chronic co-administration of CI 988 with morphine prevented the development of tolerance to the analgesic effect of the opiate (Xu et al., 1992a) and CI 988 administered to tolerant rats reinstated the analgesic effect of morphine (Hoffmann and Wiesenfeld-Hallin, 1994).

The possible role of the endogenous opioid system and CCK in central pain

In addition to short-lasting allodynia as described above, we have observed a chronic pain-related syndrome in rats after spinal cord ischemia (Xu et al., 1992b). This syndrome only developed in a subpopulation of rats after severe ischemia and with extensive spinal cord lesion. The main symptom of this chronic pain-related syndrome is mechanical allodynia, which is more severe than during tonic allodynia. Similar symptoms have been described clinically in patients after spinal cord injury. We have also observed autotomy of the hindpaws in some allodynic animals, which may indicate the presence of phantom pain. In accordance with clinical experience, the chronic allodynia-like symptoms developed with a delay of several days to 1.5 months after the initial injury and persisted without signs of remission. Also in accordance with clinical experience, the chronic allodynia-like symptom was not responsive to most pharmacological treatments, including systemic morphine, clonidine, carbamazepine, pentobarbital, baclofen, muscimol and diazepam. Since baclofen effectively relieved short-lasting, but not chronic, allodynia the mechanisms underlying this pain-like state have presumably undergone a plastic change during transition from tonicity to chronicity.

Interestingly, in rats with spinal cord lesion, but without chronic allodynia, systemic naloxone consistently provoked allodynia with characteristics very similar to the symptoms observed in allodynic rats (Xu et al., 1994b). Thus, the endogenous opioid system may be tonically active in these animals and may suppress the expression of allodynia. However, naloxone did not influence the tail flick latency in these rats, suggesting that the action of the activated endogenous opioid system is selective, i.e. it did not raise the nociceptive threshold in general. Consequently, the endogenous opioid system may be involved in the suppression of abnormal pain-related sensations after spinal cord lesion and disruption of such control may lead to the emergence of chronic allodynia. The disruption of this control may involve the CCK system since the CCK-B antagonist CI 988, but not the CCK-A antagonist CAM 1481, increased the vocalization threshold in chronically allodynic rats (Xu et al., 1994b). This effect of CI 988 was not due to general antinociception, as it had no effect on vocalization threshold to mechanical pressure in normal rats and its effect on the tail flick latency was slight and only observed at a high dose. CI 988 did not produce sedation at the doses tested and since diazepam failed to relieve allodynia, the anxiolytic property of CI 988 (Hughes et al., 1990) cannot be responsible for the observed effects. Therefore, the effect of CI988 on the vocalization threshold of spinally injured rats reflects an analgesic effect against chronic allodynia. The effect of CI 988 was reversed by naloxone, suggesting that it may be mediated through an opioidergic link. Thus, the analgesic effect of CI 988 on chronic allodynia may reflect an ongoing antagonism by the CCK system on the endogenous opioid system, which could suppress the exhibition of allodynia, as in non-allodynic spinally injured rats (Xu et al., 1994b).

Based on these results, we can propose a mechanism for the emergence of chronic allodynia in rats after spinal cord injury. Injury to the spinal cord may interrupt normal transmission and integration of sensory information. These abnormal sensory inputs may activate the endogenous opioid system, resulting in enhanced inhibitory control and suppression of the development of allodynia-like symptoms. Such enhanced opioidergic control could, however, eventually be interrupted in some rats by an upregulated endogenous CCK system, leading to the development of chronic allodynia. Indeed, a recent study has demonstrated that when the activity of endogenous opioid system was enhanced by blocking the degradation of endogenous opioids, a subsequent increase in activation of CCK-B receptors was observed (Ruiz-Gayo et al. 1992). The involvement of the endogenous CCK system in the development of chronic allodynia may explain why morphine had little effect upon central pain as endogenous CCK may antagonize exogenously applied opiates as well. We are currently investigating where in the CNS changes in the endogenous opioid and CCK systems take place that may underly opioid insensitivity in chronic allodynia.

Conclusions

The expression of experimental neuropathic pain after peripheral nerve axotomy and spinal cord injury may involve dysfunction of the GABAergic system and an altered interaction between endogenous opioids and CCK. Increased understanding of the mechanisms underlying plasticity of neurotransmitter systems and their function following PNS and CNS injury may lead to improved treatment strategies for neuropathic pain.

Acknowledgements

The present work was supported by the Swedish Medical Research Council (grant no- 07913), Astra Pain Control AB, the Bank of Sweden Tercentenary Foundation and research funds of the Karolinska Institute.

References

Albe-Fessard, D., Nashold, B.S., Jr., Lombard, M.C., Yamaguchi, Y. & Boureau, F. (1979) Rat after dorsal rhizotomy, a possible animal model for chronic pain. In: J.J. Bonica, J.C. Liebeskind, D. Albe-Fessard (Eds.), Advances in Pain Research and Therapy Volume 3, Raven Press, New York. pp. 761-766.

ArnŽr, S. (1991a) Differentiation of Pain and Treatment Efficacy. Ph.D. Thesis, Karolinska Institute 1-57.

ArnŽr, S. (1991b) Intravenous phentolamine test: diagnostic and prognostic use in reflex sympathetic dystrophy Pain 46:17-22.

ArnŽr, S. & ArnŽr, B. (1985) Differential effects of epidural morphine in the treatment of cancer-related pain. Acta Anaesthesiologica Scandinaviska 29:332-336.

ArnŽr, S. & Meyerson, B.A. (1988) Lack of analgesic effect of opioids on neuropathic and idiopathic forms of pain. Pain 33:11-23.

ArnŽr, S. & Meyerson, B.A. (1991) Opioid sensitivity of neuropathic pain. A controversial issue. In: J.-M. Besson, G. Guilbaud (Eds.), Lesions of Primary Afferent Fibres as a Tool for the Study of Cinical Pain, Elsevier, Amsterdam pp. 259-275.

Attal, N., Chen, Y.L., Kayser, V. & Guilbaud, G. (1991) Behavioural evidence that systemic morphine may modulate a phasic pain-related behaviour in a rat model of peripheral mononeuropathy. Pain 47:65-70.

Baber, N.S., Dourish, C.T. & Hill, D.R. (1989) The role of CCK, caerulein, and CCK antagonists in nociception. Pain 39: 307-328.

Banna, N.R. & Jabbur, S.J. (1969) Pharmacological studies on inhibition in the cuneate nucleus of the cat. International Journal of Pharmacology 8:299-307.

Basbaum, A.I., Chi, S.-I. & Levine, J.D. (1991) Factors that contribute to peripheral nerve injury-evoked persistent expression of the C-fos proto- oncogene in the spinal cord of the rat. In: J.M. Besson and G. Guilbaud (Eds.), Lesions of Primary Afferent Fibers as a Tool for the Study of Clinical Pain, Elsevier, Amsterdam pp. 205-218.

Bennett, G.J. (1994) Animal models of neuropathic pain. In: G.F. Gebhart, D.L. Hammond and T.S. Jensen (Eds.) Porceedings of the 7th World Congress on Pain, Progress in Pain Research and Management, Vol 2 IASP Press, Seattle, 495-510.

Bennett, G.J. & Xie, Y.K. (1988) A peripheral mononeuropathy in rat produces disorders in pain sensation like those seen in man . Pain 33:87-107-

Besse, D., Lombard, M.C., Perrot, S. & Besson, J.M. (1992) Regulation of opioid binding sites in the superficial dorsal horn of the rat spinal cord following loose ligation of the sciatic nerve: comparison with sciatic nerve section and lumbar dorsal rhizotomy. Neuroscience 50:921-933.

Besse, D., Lombard, M.C., Zajac, J.M., Roques, B.P. & Besson, J.M. (1990) Pre- and postsynaptic distribution of µ, d and k opioid receptors in the superficial layers of the cervical dorsal horn of the rat spinal cord. Brain Research 521:15-22.

Beyer, C., Roberts, L.A. & Komisaruk, B.R. (1985) Hyperalgesia induced by altered glycinergic activity at the spinal cord. Life Science 37:875-882.

Black, R.G. (1974) A laboratory model for trigeminal neuralgia, in: J.J. Bonica (Ed) Advances in Neurology, Vol. 4 Raven Press, New York pp. 651-658.

Castro-Lopes, J.M., Tavares, I. & Coimbra, A. (1993) GABA decreases in the spinal cord dorsal horn after peripheral neurectomy. Brain Research 620:287-291.

Dourish, C.T., O'Neill, M.F., Coughlan, J., Kitchener, S.J., Hawley, D. & Iversen, S.D. (1990) The selective CCK-B receptor antagonist L-365,260 enhances morphine analgesia and prevents morphine tolerance in the rat. European Journal of Pharmacology 176:35-44.

Dyken, P.R. (1965) Hyperpathic disorder from intrathecal alumina gel injection, Archives of Neurology 11: 521-528.

Eccles, J.C., Scmidt, R.F. & Willis, W.D. (1963) Pharmacological studies on presynaptic inhibition. Journal of Physiology 168:500-530.

Faris, P.L., Komisaruk, B.R., Watkins, L.R. & Mayer, D.J. (1983) Evidence for the neuropeptide cholecystokinin as an antagonist of opiate analgesia. Science 219:310- 312.

Game, C.J.A. & Lodge, D. (1975) The pharmacology of the inhibition of dorsal horn neurones by impulses in myelinated cutaneous afferens in the cat. Experimental Brain Research 23:75-84.

Glynn, C.J., Jamous, M.A., Teddy, P.J., Moore, R.A. & Lloyd, J.W. (1986) Role of spinal noradrenergic system in transmission of pain in patients with spinal cord injury, Lancet 29:1249-1250.

Glynn, C., Smith, S. &Teddy, P. (1993) An audit of the effect of epidural clonidine and lidocaine on pain and the quality of life of patients with neuropathic pain. Regional Anesthesiology, Suppl. 18:21.

GouardŽres, C., Beaudet, A., Zajac, J.-M., Crost, J. & Quirion, R. (1991) High resolution radioautographic localization of [125I]FK-33-824-labelled µ opioid r eceptors in the spinal cord of normal and deafferented rats. Neuroscience 43:197- 209.

Hao, J.-X., Watson, B.D., Xu, X.-J., Wiesenfeld-Hallin, Z., Seiger, . Sundstršm, E. (1992a) Protective effect of the NMDA antagonist MK-801 on

photochemically induced spinal lesions in the rat. Experimental Neurology 118: 143-152.

Hao, J.-X., Xu, X.-J., Aldskogius, H., Seiger,  & Wiesenfeld-Hallin, Z. (1991a) The excitatory amino acid receptor antagonist MK-801 prevents the hypersensitivity induced by spinal cord ischemia in rats. Experimental Neurology 113:182-191.

Hao, J.-X., Xu, X.-J., Aldskogius, H., Seiger, . & Wiesenfeld-Hallin, Z. (1991b) Allodynia-like effect in rat after ischemic spinal cord injury photochemically induced by laser irradiation. Pain 45:175-185.

Hao, J.-X., Xu, X.-J., Aldskogius, H., Seiger, . & Wiesenfeld-Hallin, Z.(1992b) Photochemically induced transient spinal ischemia induces behavioral

hypersensitivity to mechanical and cold, but not to noxious heat, stimuli in the rat. Experimental Neurology 118:187-194. Experimental Neurology 118:187-194.

Hao, J.-X., Xu, X.-J., Yu, Y.-X., Seiger, . & Wiesenfeld-Hallin, Z. (1992c) Transient spinal cord ischemia induces temporary hypersensitivity of dorsal horn wide dynamic range neurons to myelinated, but not unmyelinated, fiber input. Journal of Neurophysiology 68: 384-391.

Hao, J.-X., Xu, X.-J., Yu, Y.-X., Seiger, . & Wiesenfeld-Hallin, Z. (1992d) Baclofen reverses the hypersensitivity of dorsal horn wide dynamic range neurons to mechanical stimulation after transient spinal cord ischemia; implications for a tonic GABAergic inhibitory control of myelinated fiber input. Journal of Neurophysiology 68:392-396.

Hao, J.-X., Yu, Y.-X., Seiger, . & Wiesenfeld-Hallin, Z. (1992e) Systemic tocainide relieves mechanical hypersensitivity and normalizes the responses of hypersensitive dorsal horn wide dynamic range neurons after transient spinal cord ischemia in rats. Experimental Brain Research 91:229-235.

Hoffmann, O. & Wiesenfeld-Hallin, Z. (1994) The CCK-B receptor antagonist CI 988 reverses tolerance to morphine in rats. NeuroReport 5:2565-2568.

Hškfelt, T., Herrera-Marschitz, M., Seroogy, K., Ju, G., Staines, W.A., Holets, V., Schalling, M., Ungerstedt, U., Post, C., Rehfeld, J.F., Frey, P., Fischer, J., Dockray, G., Hamaoka, T., Walsh, J.H. & Goldstein, M.(1988) Immunohistochemical studies on cholecystokinin (CCK)-immunoreactive neurons in the rat using sequence specific antisera and with special reference to the caudate nucleus and primary sensory neurons. Journal of Chemical Neuroanatomy 1:11-52.

Hškfelt, T., Zhang, X. & Wiesefeld-Hallin, Z. (1994) Messenger plasticity in primary sensory neurons following axotomy and its functional implications. Trends in Neuroscience 17:22-30.

Hughes, J., Boden, P., Costall, B., Domeney, A., Kelly, E., Horwell, D.C., Hunter, J.C., Pinnock, R.D. & Woodruff, G.N. (1990) Development of a class of selective cholecystokinin type B receptor antagonists having potent anxiolytic activity. Proceedings of the National Academy of Sciences of the U.S.A. 87:6728-6732.

Itoh, S., Katsuura, G. & Maeda, Y. (1982) Caerulein and cholecystokinin suppress beta-endorphin-induced analgesia in the rat. European Journal of Pharmacology 80:421-425.

Jazat, F. & Guilbaud, G. (1991) The 'tonic' pain-related behaviour seen in mononeuropathic rats is modulated by morphine and naloxone. Pain 44:97-102.

Ju, G., Hškfelt, T., Fischer, J.A., Frey, P., Rehfeld, J.F. & Dockray, G.J. (1986) Does cholecystokinin-like immunoreactivity in rat primary sensory neurons represent calcitonin gene-related peptide? Neuroscience Letters 68:305-310.

Lindblom, U. (1985) Assessment of abnormal evoked pain in neurological pain patients and its relation to spontaneous pain: a descriptive and conceptual model with some analytical results. In: H.L. Fields, R. Dubner, F. Cervero. (Eds.), Advances in Pain Research and Therapy, Vol. 9, Raven Press, New York pp. 409- 423.

Lindblom, U. & Hansson, P. (1991) Sensory dysfunction and pain after clinical nerve injury studied by means of graded mechanical and thermal stimulation. In: J.-M. Besson, G. Guilbaud, (Eds.), Lesions of Primary Afferent Fibres as a Tool for the Study of Clinical Pain, Elsevier, Amsterdam pp.1-18.

Lombard, M.C. & Larabi, Y. (1983) Electrophysiological study of cervical dorsal horn cells in partially deafferented rats. In: J.J. Bonica, U. Lindblom, A. Iggo (Eds.), Advances in Pain Research and Therapy, Vol. 5, Raven Press, New York pp. 147-154.

Lombard, M., Nashold, B.S. & Albe-Fessard, D. (1979) Deafferentation hypersensitivity in the rat after dorsal rhizotomy: a possible animal model of chronic pain. Pain 6:163-174.

Martiniak, J., Saganov‡, K. & Chavko, M. (1991). Free and peptide-bound amino acids as indicators of ischemic damage of the rabbit spinal cord. Journal of Neuropathology and Experimental Neurology 50:73-81.

McQuay, H.J. (1988) Pharmacological treatment of neuralgic and neuropathic pain, Cancer Surveys, 7:141-159.

Merskey, H. (Ed) (1986) Classification of chronic pain: description of chronic pain syndromes and definitions of pain terms. Pain (Suppl.) 3: S1-S220.

Meyerson, B.A. (1990) Neuropathic pain - an overview. In: S. Lipton, E. Tunks, M. Zoppi (Eds.), Advances in Pain Research and Therapy, Vol. 13, Raven Press, New York, pp. 193-199.

Moran, T.H., Robinsson, P., Goldrich, M.S. & McHugh, P.R. (1986) Two brain cholecystokinin receptors: implications for behaviour actions. Brain Research 362:175-179.

Portenoy, R.K., Foley, K.M. & Inturrisi, C. (1990) The nature of opioid responsiveness and its implications for neuropathic pain. New hypotheses derived from studies of opioid infusions.Pain 43:273-86.

Price, G.W., Kelly, J.S. & Bowery, N.G. (1987) The location of GABAB receptor binding sites in mammalian spinal cord. Synapse 1:530-538.

Puke, M.J.C. & Wiesenfeld-Hallin, Z. (1993) The differential effects of morphine and the a2-adrenoceptor agonists clonidine and dexmedetomidine on the prevention and treatment of exprimental neuropathic pain. Anesthesia and Analgesia 77:104-109

Ruiz-Gayo, M., Durieux, C., FourniŽ-Zaluski, M.C. & Roques, B.P. (1992) Stimulation of d-opioid receptors reduces the in vivo binding of the cholecystokinin (CCK)-B- selective agonist (3H)pBC 264: evidence for a physiological regulation of CCKergic systems by endogenous enkephalins. Journal of Neurochemistry 59:1805-1811.

Schwartz, D.P. & De Good, D.E. (1984) Global appropriateness of pain drawings. Blind ratings predict patterns of psychological distress and litigation status. Pain 19:383-388.

Seltzer, Z., Dubner, R. & Shir, Y. (1990) A novel behavioral model of neuropathic pain disorders produced in rats by partial sciatic nerve injury. Pain 43:205-218.

Sherman, R.A., Sherman, C.J. & Gall, N.G. (1980) A survey of current phantom limb pain treatment in the United States. Pain 8: 85-99.

Sloper, J.J., Johnson, P. & Powell, T.P.S. (1986) Selective degeneration of interneurons in the motor cortex of infant monkeys following controlled hypoxia: a possible cause of epilepsy. Brain Research 198: 204-209.

Smith, G.D., Harrison, S.M., Birch, P.J., Elliott, P.J., Malcangio, M. & Bowery, N.G. (1994) Increased sensitivity to the antinociceptive activity of (+/-)-baclofen in an animal model of chronic neuropathic, but not chronic inflammatory hyperalgesia. Neuropharmacology 33:1103-1108.

Stanfa, L., Dickenson, A., Xu, X.-J. & Wiesenfeld-Hallin, Z. (1994) Cholesytokinin and morphine analgesia: variation on a theme. Trends in Pharmacological Sciences 15:65-67.

Sugimoto, T., Bennett, G.J. & Kajander, K.C. (1990) Transsynaptic degeneration in the superficial dorsal horn after scietic nerve injury: effects of a chronic constriction injury, transection, and strychnine. Pain 42:205-213.

Todd, A.J. & Spike, R.C. (1993) The localization of classical transmitters and neuropeptides within neurons in laminae I-III of the mammalian spinal dorsal horn. Proceedings in Neurobiology 41:609-645.

Vanderhaeghen, J-J. & Crawley, J.C. (Eds) (1985) Neuronal Cholecystokinin. Annals of the New York Academy of Sciences 448.

Verge, V.M.K., Wiesenfeld-Hallin, Z. & Hškfelt, T. (1993) Cholecystokinin in mammalian primary sensory neurons and spinal cord: in situ hybridization studies in rat and monkey. European Journal of Neuroscience 5:240-250.

Wall, P.D. & Devor, M. (1981) The effect of peripheral nerve injury on dorsal root potentials. Brain Research 209:95-111.

Wall, P.D. & Devor, M. (1983) Sensory afferent impulses originate from dorsal root ganglia as well as from the periphery in normal and nerve injured rats. Pain 17:321-339.

Wall, P.D., Devor, M., Inbal, R., Scadding J.W., Schonfield, D., Seltzer, Z. & Tomkiewicz, M.M. (1979) Autotomy following peripheral nerve lesions: experimental anaesthesia dolorosa. Pain 7:103-111.

Wall, P.D & Gutnick, M. (1974) Ongoing activity in peripheral nerves: the physiology and pharmacology of impulses originating from a neuroma. Experimental Neurology 43:580-593.

Watkins, L.R., Kinscheck, I.B. & Mayer, D.J. (1984) Potentiation of opiate analgesia and apparent reversal of morphine tolerance by proglumide. Science 224:395-396.

Wiesenfeld, Z. & Lindblom, U. (1980) Behavioural and electrophysiological effects of various types of peripheral nerve lesions in the rat: a comparison of possible models for chronic pain. Pain 8:285-298.

Wiesenfeld-Hallin Z. (1984) The effect of intrathecal morphine and naltrexone on autotomy in sciatic nerve sectioned rats. Pain 18:267-278.

Wiesenfeld-Hallin, Z..& Duranti, R. (1987) Intrathecal cholecystokinin interacts with morphine but not substance P in modulating the nociceptive flexion reflex in the rat. Peptides 8:153-158.

Wiesenfeld-Hallin, Z., Hao, J.-X., Aldskogius, H., Seiger, . & Xu, X.-J. (1994) Allodynia-like symptoms in rats after spinal cord ischemia: An animal model of central pain. In: J. Boivie, P. Hansson and U. Lindblom (Eds.) Touch, Temperature, and Pain in Health and Disease: Mechanisms and Assessments, Progress in Pain Researcha dn Management, Vol. 3, IASP Press, Seattle, 355-372.

Wiesenfeld-Hallin, Z., Nennesmo, I. & Kristensson, K. (1987) Autotomy in rats after nerve section compared with nerve degeneration following intraneural injection of Ricinus communis agglutinin I, Pain, 30:93-102.

Wiesenfeld-Hallin, Z., Xu, X.-J., Hughes, J., Horwell, D.C., Hškfelt, T. (1990) PD134308, a selective antagonist of cholecystokinin type-B receptor, enhances the analgesic effect of morphine and synergistically interacts with intrathecal galanin to depress spinal nociceptive reflexes. Proceedings of the National Academy of Sciences of the U.S.A.. 87:7105-7109.

Williams, R.G., Dimaline, R., Varro, A., Isetta, A.N., Trizio, D. & Dockray, G.J. (1987) Cholecystokinin octapeptide in rat central nervous system: immunocytochemical studies using a monoclonal antibody that does not react with CGRP. Neurochemistry Intternational 11: 433-442.

Willis, W.D. & Coggeshall, R.E. (1991) Sensory Mechanisms of the Spinal Cord. Plenum: New York,

Xu, X.-J., Hao, J.-X., Aldskogius, H., Seiger, . & Wiesenfeld-Hallin, Z. (1992b) Chronic pain-related syndrome in rats after ischemic spinal cord lesion: a possible model for pain in patients with spinal cord injury. Pain 48:279-290.

Xu, X.-J., Hao, J.X., Seiger, ., Hughes, J., Hškfelt, T. & Wiesenfeld-Hallin, Z. (1994a) Chronic pain-related behaviour in spinally injured rats: evidence for functional alterations of the endogenous cholecystokinin and opioid systems. Pain 56:271-298.

Xu, X.-J., Hškfelt, T., Hughes, J. & Wiesenfeld-Hallin, Z. (1994b) The CCK-B antagonist CI 988 enhances the reflex depressive effect of morphine in axotomized rat, NeuroReport 5:718-720.

Xu, X.-J., Puke, M.J.C., Verge, V.M.K., Wiesenfeld-Hallin, Z., Hughes, J. & Hškfelt, T. (1993) Up-regulation of cholecystokinin in primary sensory neurons is associated with morphine insensitivity in experimental neuropathic pain in the rat. Neuroscience Letters 59:129-132.

Xu, X.-J. & Wiesenfeld-Hallin, Z. (1991) The threshold for the depressive effect of intrathecal morphine on the spinal nociceptive flexor reflex is increased during autotomy after sciatic nerve section in rats. Pain 46:223-229.

Xu, X.-J., Wiesenfeld-Hallin, Z., Hughes, J., Horwell, D.C. & Hškfelt, T. (1992c) PD134308, a selective antagonist of cholecystokinin type-B receptor, prevents morphine tolerance in the rat. British Journal of Pharmacology 105:591-596.

Yaksh, T.L. (1989) Behavioral and autonomic correlates of the tactile evoked allodynia produced by spinal glycine inhibition: effect of modulatory receptor systems and excitatory amino acid antagonists. Pain 37:111-123.

Zhang, A.-L., Hao, J.-X., Seiger, ., Xu, X.-J., Wiesenfeld-Hallin, Z., Grant, G. and Aldskogius, H. (1994) Decreased GABA immunoreactivity in spinal cord dorsal horn neurons after transient spinal cord ischemia in the rat. Brain Research 656:187-190..

Zhang, X., Dagerlind, ., Elde, R.P., Castel, M.N., Broberger, C., Wiesenfeld-Hallin, Z. & Hškfelt, T. (1993) Marked increase in cholecystokinin B receptor messenger RNA levels in rat dorsal root ganglia after peripheral axotomy Neuroscience 57:227-233.LEGEND