Report

Preserved Acute Pain and Reduced Neuropathic Pain in Mice Lacking PKCγ

See allHide authors and affiliations

Science  10 Oct 1997:
Vol. 278, Issue 5336, pp. 279-283
DOI: 10.1126/science.278.5336.279

Abstract

In normal animals, peripheral nerve injury produces a persistent, neuropathic pain state in which pain is exaggerated and can be produced by nonpainful stimuli. Here, mice that lack protein kinase C gamma (PKCγ) displayed normal responses to acute pain stimuli, but they almost completely failed to develop a neuropathic pain syndrome after partial sciatic nerve section, and the neurochemical changes that occurred in the spinal cord after nerve injury were blunted. Also, PKCγ was shown to be restricted to a small subset of dorsal horn neurons, thus identifying a potential biochemical target for the prevention and therapy of persistent pain.

Neuropathic pain is a devastating consequence of nerve injury that is characterized by spontaneous, often burning, pain, an exaggerated response to painful stimuli (hyperalgesia), and pain in response to normally innocuous, for example touch, stimuli (allodynia). Neuropathic pain syndromes are among the most difficult to manage. Although the pain produced by tissue injury can usually be controlled by anti-inflammatory drugs and opioids, neuropathic pains such as postherpetic neuralgia, reflex sympathetic dystrophy, and phantom limb pain are often refractory to these treatments.

Some studies suggest that nerve injury leads to neuropathic pain because it triggers an N-methyl-d-aspartate (NMDA) receptor–mediated hyperexcitability of dorsal horn neurons in the spinal cord. Events downstream of the NMDA receptor, including activation of various protein kinases, have also been implicated; these are presumed to underlie the persistence of the pain (1). Conclusions from these studies, however, are very limited. For example, although there is evidence for a contribution of protein kinase C (PKC), those studies not only used inhibitors that are not specific for PKC but they also provided no information about the contribution of specific isoforms of PKC, at least 10 of which have been identified (2). In the present study we examined nerve injury–induced neuropathic pain in mice with a deletion of the gene that encodes for the neuronal-specific (gamma) isoform of PKC.

The deletion (knock-out) of PKCγ produces viable mice with normal appearance. The mice have a slight ataxia, modest impairments in tests of learning and memory (3), and some motor incoordination (4) that may be related to a defect in elimination of multiple climbing fiber innervation of Purkinje cells (5). Although synaptic transmission appears normal, long-term potentiation is impaired (3).

In the absence of nerve injury, we found no difference in paw withdrawal responses to thermal or mechanical stimulation in mutant and wild-type mice (Fig. 1). Thus, transmission of acute “pain” messages was intact in the mutant mice. To study pain behavior produced by nerve injury, we tightly ligated one-third to one-half of the diameter of the sciatic nerve; this partial nerve injury produces a neuropathic pain syndrome characterized by a marked and long-lasting reduction in the paw withdrawal threshold to both thermal and mechanical stimulation on the injured side (6). In the wild-type mice thermal response latencies were significantly decreased by the third day after nerve injury; this alteration persisted for the 14-day observation period (Fig. 1A). Compared with the wild-type mice, the mutant mice had a significantly decreased thermal allodynia at all time points after surgery. In the wild-type mice, the latency to withdraw from the heat stimulus decreased to 5 to 6 s from a baseline response of 10 to 11 s; the latency to withdraw in the mutant mice was 8 to 9 s. Nerve injury was not completely without effect in the mutant mice. Compared with values before nerve injury, we recorded a modest, albeit significant, decrease in thermal withdrawal latencies, but the magnitude of the change was much less than what we recorded in the wild-type mice.

Figure 1

Before surgery there were no differences in thermal or mechanical thresholds between wild-type and mutant mice. (A) In wild-type mice, nerve injury produced a significant reduction in paw withdrawal latencies to a heat stimulus on the injured side [P < 0.001, repeated measure analysis of variance (ANOVA)]. This decrease was significantly different from the paw withdrawal latencies in mutant mice. Asterisks indicate a significant difference on the injured side between wild-type and mutant mice (*P < 0.05, PLSD Fisher's test). The mutant mice also displayed a modest but significant thermal allodynia (P < 0.05, repeated measures ANOVA). No difference in paw withdrawal latencies were observed on the noninjured side in either group. (B) Injury to the sciatic nerve produced a significant decrease in paw withdrawal thresholds to von Frey hair stimulation on the injured side in wild-type (P < 0.05, Friedman test) but not in mutant mice (P > 0.05) compared with baseline thresholds. Asterisks indicate significantly lower thresholds on the injured side in wild-type mice compared with the injured side of mutant mice (*P < 0.05, Mann-Whitney test). No change in withdrawal latency was observed on the contralateral side.

Because mechanical hypersensitivity is a predominant symptom of neuropathic pain in patients (even contact of clothes is often intolerable), we also studied the response to mechanical stimulation. Partial sciatic nerve injury in the wild-type mice led to a profound decrease of the threshold for evoking hindpaw withdrawal to a mechanical stimulus (von Frey hair) (Fig. 1B). This mechanical allodynia appeared on the third day after surgery and persisted for the duration of the experiment. In contrast, we found no significant change of the mechanical threshold in the PKCγ mutant mice (Fig. 1B). Thus, PKCγ is essential for the production of mechanical allodynia after nerve injury.

Peripheral nerve injury not only produces a neuropathic pain syndrome but also significantly alters the neurochemistry of the ipsilateral dorsal root ganglion (DRG) and the spinal cord dorsal horn (7). Therefore, we also compared the neurochemical consequences of nerve injury in the wild-type and mutant mice. We focused on substance P (SP), a major neurotransmitter of small diameter nociceptive primary afferents, and on the neurokinin-1 (NK-1) receptor, which is targeted by SP. We also studied neuropeptide Y (NPY), which is found in the dorsal horn in normal animals but not in DRG cells; after nerve injury both NPY mRNA and peptide are expressed in DRG neurons (7). Consistent with many previous studies in the rat, in the wild-type mice we found that partial nerve injury produced a marked decrease in SP and an increase in NPY and NK-1 receptor immunoreactivity in laminae I and II of the dorsal horn ipsilateral to the sciatic nerve injury (Fig. 2). The alteration in the amounts of these neurotransmitter and receptor markers was greatest in the spinal segments that receive primary afferent input from the sciatic nerve (namely, lumbar segments L4 and L5), but the increase in the number of NK-1 receptors also extended several segments rostral and caudal to these sites. However, we found a significantly smaller nerve injury–induced alteration of SP, NK-1, and NPY immunoreactivity in the mutant compared with the wild-type mice (8) (Figs. 2 and 3). In other words, the almost complete failure of the mutant mice to develop the neuropathic pain syndrome after nerve injury was paralleled by a very limited neurochemical reorganization in the dorsal horn of the spinal cord.

Figure 2

SP (A andB), NPY (C and D), and NK-1 receptor (E and F) immunoreactivity at the L4 spinal segment of wild-type (A, C, E) and mutant mice (B, D, F). The nerve injury was on the right side. The largest change in SP, NPY, and NK-1 receptor immunoreactivity occurred in the wild-type mice. The increase in NK-1 receptor immunoreactivity was concentrated in the medial part of lamina I (arrowhead). Scale bar, 300 μm.

Figure 3

Nerve injury–evoked changes in SP, NPY, and NK-1 receptor immunoreactivity in the dorsal horn of the spinal cord in PKCγ mutant (white bars) and wild-type (black bars) mice. Data are presented as the mean ratio in percent of immunoreactivity ± SEM between the nerve-injured and the noninjured side at 14 days after the nerve injury (8). A value lower than 100% indicates that the injured side contained less immunoreactivity than the noninjured side; a value greater than 100% indicates that there is an increase of immunoreactivity on the injured side. Asterisks indicate significant differences between wild-type and PKCγ mutant mice with PLSD Fisher's test (*P < 0.05; **P < 0.01).

Because the NK-1 receptor is exclusively located in neurons that are postsynaptic to the nerve injury (9), it provides a marker of transneuronal changes produced by peripheral nerve injury. In contrast, changes in SP and NPY could occur both pre- and postsynaptically (7). It was thus important to specifically evaluate the DRG response to nerve injury. Furthermore, because partial nerve injury produces a variable effect in the DRG, it can be difficult to evaluate quantitatively. Thus, in a second series of experiments, we completely transected the sciatic nerve to produce a maximal response and then examined both the DRG and dorsal horn. As expected, this injury produced maximal neurochemical changes, including a significant decrease in SP and an increase in NPY immunoreactivity in the dorsal horn of the wild-type mice (10). In the mutant mice, however, despite there being a total nerve transection, we again found minimal change in the dorsal horn of the spinal cord (10). In contrast, the neurochemical response of the DRG to injury did not differ in the wild-type and mutant mice. Specifically, in the two groups of mice we recorded comparable decreases in the number of DRG neurons that expressed SP and a comparable up-regulation of NPY-immunoreactive neurons (10). The fact that we found differences in the dorsal horn but not in the DRG suggests that the response of the primary afferent to injury was not altered in the mice that lack PKCγ. Rather, the PKCγ deletion was manifest as a significant reduction of the neurochemical response of postsynaptic neurons to nerve injury.

Persistent pain states can arise from tissue as well as nerve injury (11). A selective deficit is found in the development of inflammation and tissue injury–induced “nociceptive” pain in mice that carry a null mutation in the gene that encodes the neuronal-specific isoform of the type I regulatory subunit (RIβ) of protein kinase A (PKA) (12). The latter animals, however, showed no change in the neuropathic pain behavior produced by partial nerve injury. To determine whether PKCγ also contributes to nociceptive pain, we studied the PKCγ mutant mice in an inflammation model produced by hindpaw injection of dilute formalin (13). Consistent with acute pain responses being unaffected by the PKCγ deletion, we found that pain behavior in the first phase of the formalin test, which is presumed to result from direct activation of small diameter primary afferent “pain” fibers, did not differ in wild-type and mutant mice (13). However, the second, prolonged phase of nociceptive pain behavior, which is driven largely by tissue inflammation, was attenuated (13). The PKCγ mutant mice also showed reduced swelling of the formalin-injected paw. In agreement with the reduced inflammation resulting in decreased nociceptive inputs to the central nervous system, in the dorsal horn of the spinal cord ipsilateral to the injury we recorded significantly less Fos immunoreactivity, a marker of neuronal activity, in mutant compared with wild-type mice (14). Finally, compared with wild-type mice, we found that the PKCγ mutant mice displayed a significant reduction (44%) of plasma extravasation induced by intradermal injection of capsaicin into the paw (15). These results underscore the difference in the pathophysiology of tissue injury– and nerve injury–evoked persistent pain states. Nociceptive pain that results from tissue injury involves both PKA and PKC (and possibly other second messenger systems). In contrast, a full-blown neuropathic pain state can be produced by nerve injury even when PKA RIβ is absent, but deletion of PKCγ prevents the development of this neuropathic pain condition.

Although PKCγ expression can only be detected 7 days after birth and reaches maximal levels by 28 days of age (16), the decreased neuropathic pain behavior and neurochemical reactivity in the mutant mice could have resulted from a developmental abnormality that reduced the number of small diameter primary afferent “pain” fibers. To address this possibility, we used electron microscopy to count the numbers of myelinated and unmyelinated axons in the L5 dorsal root of mutant and wild-type mice. We found that neither the morphology nor the numbers of myelinated or unmyelinated axons differed between mutant and wild-type mice (17). Thus, the behavioral and neurochemical phenotype is likely to be due to the absence of PKCγ rather than to a developmental change secondary to its deletion.

An important insight into the possible mechanism through which PKCγ influences spinal nociceptive processing was revealed in our subsequent immunocytochemical studies of the distribution of the classical set of PKC isozymes (α, βI, βII, and γ). Specifically, although the first three were distributed rather homogeneously in DRG, in sympathetic ganglia, and in all of the superficial layers of the dorsal horn of the spinal cord, the spinal cord distribution of PKCγ was highly restricted (18). It was only found in a subset of interneurons in the inner part of the substantia gelatinosa (lamina II) of the dorsal horn (Fig. 4). Because PKCγ was not detectable in DRG neurons, it is very unlikely that the phenotype observed resulted from long-term changes in the injured primary afferent. In fact, because the neurochemical response of the DRG to injury, namely, decreased amounts of SP and increased amounts of NPY, was not altered in the mutant mice, we conclude that nerve injury–evoked signals were transmitted faithfully to the DRG (presumably by means of retrograde axonal transport) in both wild-type and mutant mice. Furthermore, because primary afferents do not express the NK-1 receptor (9), it follows that nerve injury–induced changes in the neurochemistry of the dorsal horn (for example, NK-1 receptor up-regulation) involve PKCγ-containing interneurons that are downstream of the primary afferent. The functional target of these interneurons must include neighboring dorsal horn neurons that express the NK-1 receptor. Although we cannot rule out a contribution of other PKC isoforms, we suggest that PKCγ-mediated phosphorylation of substrate proteins in interneurons of the inner part of the substantia gelatinosa is critical and probably necessary for the full development of the neuropathic pain state produced by peripheral nerve injury.

Figure 4

Distribution of PKCα (A), PKCβI (B), PKCβII (C), and PKCγ (D) immunoreactivity in the spinal cord of wild-type mice (6). The arrowheads in (A), (B), and (C) point to axonal staining that probably originates in the DRG. Asterisks identify immunoreactivity of axons located in the corticospinal tract, which in rodents is found in the base of the posterior columns. Only the PKCγ staining is confined to interneurons of the inner part of lamina II. Scale bar, 200 μm.

The concomitant reduction of the behavioral and anatomical response to nerve injury points to the superficial dorsal horn as the critical locus of the PKCγ contribution to neuropathic pain, but it is not clear how a deletion of PKCγ in these interneurons resulted in inflammation deficits. Decreased central sensitization (19) could account for the reduction of tissue injury–induced pain, but it could not explain the reduction of neurogenic inflammation. It is conceivable that a deletion of PKCγ in the dorsal horn reduces interneuron-generated dorsal root reflexes (20); this would decrease the release of peptides from the peripheral terminals of primary afferents and thus reduce neurogenic inflammation. On the other hand, because inflammation can be influenced by sympathetic and hormonal factors (21, 22), including circulating corticotrophin releasing factor, it is possible that the reduction of tissue injury–evoked pain and neurogenic inflammation resulted from deletion of PKCγ at multiple sites in the central nervous system.

Interneurons of the inner part of lamina II differ considerably from those located dorsally, in lamina I and the outer part of lamina II. Those in the inner part of lamina II (where PKCγ is concentrated) receive a selective input from a neurochemically distinct population of unmyelinated primary afferents that express an adenosine triphosphate–sensitive P2X3 receptor, bind the lectin Bandeiraea simplicifolia, and uniquely contain a fluoride-resistant acid phosphatase in their central terminals (23). Most importantly, in contrast to the “pain”-responsive neurons of the overlying lamina I and outer lamina II, neurons of inner lamina II respond preferentially to non-noxious inputs (24). Thus, PKCγ-regulated changes in the processing of non-noxious inputs by dorsal horn neurons may be critical to the development of neuropathic pain after nerve injury.

From a clinical perspective, the very restricted spinal cord location of the PKCγ-containing interneurons is advantageous. If selective inhibitors of PKCγ can be developed, it may be possible to alleviate nerve injury–induced neuropathic pain states without the profound side effects that are inevitable with nonselective inhibitors of PKC. Moreover, because acute pain responses were not affected in the mutant mice, selective inhibitors of PKCγ would not interfere with the important, protective function that acute pain serves.

  • * To whom correspondence should be addressed. E-mail: annikam{at}phy.ucsf.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article