Previous work carried out in our own and other laboratories has indicated that chemokine signaling may contribute to the genesis and maintenance of neuropathic pain [11, 17, 18]. Thus, the present studies were designed to investigate a potential association between focal nerve demyelination, neuropathic pain behavior and chemokine signaling in DRG neurons. Based on our previous studies, we hypothesized that the focal demyelination of the sciatic nerve, a known rodent model of neuropathic pain , would result in upregulated chemokine expression and chemokine receptor signaling in DRG neurons. Indeed, we observed the predicted increases for several chemokines and their receptors. Importantly, as the PWT decreased over 14 days post-injury, chemokines, chemokine receptors and chemokine/receptor signaling increased significantly. This occurred not only in the DRG directly associated with the injured nerve, but also to a lesser degree, in the DRG directly contralateral to the nerve injury. Bilateral nociceptive behavior was apparent between days 3–28 and could be stereospecifically attenuated with a CCR2 receptor antagonist on days 14 and 28. Moreover, five weeks following injury, both the neuropathic pain behavior and the incidence of chemokine signaling were greatly diminished. Together with the known cellular effects produced by chemokines on sensory neurons [3, 7], these results suggest that the changes in sensory neuron chemokine/receptor signaling may be central to the maintenance phase of neuropathic pain behavior in particular.
The results of our previous studies, together with the present results on LPC-associated neuropathy, point to a significant role for chemokine signaling expressed directly by peripheral nerves. For example, we have previously demonstrated that activation of chemokine receptors expressed by cultured or acutely isolated DRG neurons produces increased (Ca2+)i, or neural excitation [4, 38]. Indeed, following upregulation of CCR2 by sensory neurons in whole DRG derived from animals exhibiting neuropathic pain, application of MCP-1 produces powerful excitation . The mechanism underlying this response probably involves activation of phospholipase C-induced degradation of PIP2, production and concomitant transactivation of TRPV1 and/or TRPA1 together with inhibition of K+ conductance [9, 10].
The experiments reported here suggest a model in which focal nerve demyelination produces a concomitant upregulation of several chemokines and their receptors in the cell bodies of sensory neurons in the DRG. We have observed that chemokines expressed by DRG neurons, including MCP1, IP10 and SDF1, can be packaged into secretory vesicles and released upon depolarization . Presumably, chemokines released in this fashion may influence neural cells in the local vicinity eliciting excitation as described above. Such activation would produce further chemokine release and excitation driving the overall excitability of the chemokine sensitive neurons to new levels. The resulting neuronal behavior may explain certain aspects of pathologically maintained neuronal states of depolarization or electrical hyperexcitability of peripheral sensory neurons [39–41]. In addition to the chronic maintenance of sensory neuron hyperexcitability, release of chemokines such as MCP-1 and fracktalkine from central axon terminals in the spinal cord may initiate microglial-mediated neuropathic pain states [7, 11, 42–44]. However, it is important to note that pharmacological therapies which inhibit microglial activation and effectively attenuate the development of hyperalgesia and allodynia have no effects on preexisting nociceptive pain behavior .
As we have demonstrated, the exact pattern of changes in chemokine signaling observed following focal nerve demyelination depends on the particular chemokine receptor and ligand examined. There are over 50 known chemokines and 20 chemokine receptors , and it is obviously not feasible to study all of these simultaneously. However, the receptors studied in the present experiments represent obvious candidates for a role in peripheral neuropathy. Chemokines that signal via the CCR2, CCR5, CXCR3 and CXCR4 receptors have previously been shown to influence the behavior of sensory neurons [3, 4, 6, 8, 11, 17]. Furthermore, many of these receptors can be upregulated in leukocytes by mechanisms suggesting that regulation of their expression may often be coordinated through the same transcriptional control mechanisms .
The four chemokines/chemokine receptors that we studied all displayed different patterns of expression in response to focal nerve demyelination, suggesting different roles in the genesis of pain or other functions in the DRG. The upregulation of MCP-1 and the CCR2 chemokine receptor observed in association with focal nerve demyelination is similar to the pattern we previously observed using a spinal stenosis model of neuropathic pain . Indeed, CCR2 receptor deficient mice are resistant to the induction of some sensory neuropathies, highlighting the potential importance of this chemokine signaling system . In the current experiments, we utilized a Ca2+ imaging paradigm in lieu of electrophysiological recording, as chemokine-induced increased neuronal excitability would be expected to be correlated with a chemokine induced increase in (Ca2+)I. The observed increase and subsequent decrease in MCP-1-induced Ca2+responsiveness in acutely isolated DRG neurons over time generally correlated with the anatomical observations of receptor expression, and both effects returned to baseline by POD 35. Importantly, the ability of the CCR2 receptor antagonist to attenuate bilateral nociceptive behavior at both 14 and 28 days after nerve injury strongly suggests an integral role for MCP-1/CCR2 signaling in maintaining this phase of pain hypersensitivity.
Although the CCR2 antagonist was effective in blocking pain hypersensitivity, more than one chemokine or chemokine receptor was upregulated in this neuropathic pain model. The particular effectiveness of CCR2 block could be due to the fact that upregulation of chemokines can be bilaterally expressed in different populations of sensory neurons following nerve injury, as is this case of cholecystokinin vasoactive inhibitory peptide and neuropeptide Y [43, 47]. In our experiments over 50% of the cells that upregulated MCP-1 also expressed IB-4, which is a marker for C-fiber nociceptors that are responsible for transmitting pain information. This differs from the case of IP-10, where a majority of the neurons upregulating this chemokine did not co-localize with IB-4. As such, it is possible that the population of neurons that upregulates CCR2 signaling is particularly linked to the production of neuronal hyperexcitability. It is also likely that the CCR2R antagonist may impact non-neuronal cells within the CNS, such as microglial cells, which are known to express CCR2 in the spinal cord and contribute to the development of chronic pain states [11, 48, 49].
The precise location of action of the CCR2 antagonist is not known. However, it has been shown that the blood nerve barrier is less restrictive than the blood brain barrier , with the cell body rich area of the DRG being vulnerable to extravascular leakage. Given these studies, it is likely that the CCR2 R antagonist reached the cell bodies of the DRG. As activation of CCR2 receptors in the DRG is probably of considerable importance in the production of pain behavior it is likely that block of these receptors contributes to the antinociceptive effects of the CCR2 antagonist.
In the face of the effectiveness of CCR2 receptor block either pharmacologically (Fig 9) or genetically , the function of other types of upregulated chemokine signaling to chronic pain behavior is not immediately obvious. Like CCR2, the CCR5 receptor function and its ligand, RANTES, were also strongly upregulated in DRG neurons in response to focal demyelination. It has previously been shown that RANTES may be important in other chronic pain situations . Our findings in this sciatic nerve injury model differ from a report by Taskinen and Royotta  which demonstrated bilateral upregulation of non-neuronal RANTES for up to four weeks following sciatic nerve transection in the rat. The apparent differences may be due to the nature of the nerve injuries. Alternatively, CCR5 may also be activated by a number of other chemokine ligands which we did not measure . Chemokine interactions with CCR5 may also depress the analgesic action of endogenous opioids and/or sensitize TRPV1 [10, 53, 54] thereby generally promoting hyperalgesia.
The signaling pattern of IP-10 and CXCR3 receptors in the DRG differs in some respects as there is appreciable basal neuronal expression of both CXCR3 receptors and IP-10. In spite of this, few Ca2+ neuronal responses were observed in the naïve or sham animals, perhaps because of desensitization resulting from ongoing receptor activation induced by constitutive expression of IP-10. Neuronal expression of IP-10/CXCR3 under basal conditions may have a specific role to play that is analogous to the expression and release of fractalkine by neurons [42, 55]. Following strong neuronal excitation in the peripheral nervous system (i.e. trauma or disease), IP-10 may be released within the DRG and/or from central terminals in the spinal cord dorsal horn resulting in both local and distant glial activation [26, 56, 57].
Focal nerve demyelination changes in SDF-1 signaling via the CXCR4 receptor show still another pattern. In this case, the chemokine receptor is not generally expressed in neurons, but in satellite glia and Schwann cells. Upregulated CXCR4 expression, however, is primarily restricted to neurons making them a potential target for the release of SDF-1 from glia. It is interesting to note that a role for Schwann cell release of SDF-1 and for neuronally-expressed CXCR4 receptors has also been suggested in recently proposed models of HIV-1 and NRTI related neuropathies [14, 51].
It is clear that focal nerve demyelination injury-induced behavioral changes are correlated with widespread changes in the neuronal expression of chemokine/receptors and that the pattern of expression of each chemokine and its receptor is unique, suggesting that the influence of chemokine signaling on rodent nociceptive behavior may be complex. It should also be noted that the upregulated expression of different chemokines and receptors that we have observed may occur as part of a cytokine "cascade", where the expression of one chemokine or its receptor is dependent on previous events. If that is the case it is also possible that drugs which block several upregulated chemokine receptors may also prove to be effective if they are upstream of CCR2 expression.
In the course of these studies, we also noted that the PWT to mechanical stimulation decreased bilaterally. The degree of threshold change in the hindpaw contralateral to the nerve injury was qualitatively similar but smaller in magnitude, and briefer in time course, when compared with the hindpaw ipsilateral to the lesion. This type of bilateral hyperalgesia has previously been described in other rodent models of neuropathic pain [58–67]. As such, it is of interest to compare the LPC-induced peripheral nerve pain model with previously described rodent inflammatory pain models which demonstrated bilateral tactile pain behavior  and contralateral changes in the DRG [65, 68]. Milligan et al.  suggested that this phenomenon is likely due to changes in the spinal dorsal horn. This type of spinal mechanism may drive both bilatateral pain sensitivity and contralateral DRG changes in chemokines/receptors following unilateral sciatic nerve demyelination by releasing cytokines from activated microglia in the spinal cord dorsal horn following chronic activity in injured DRG afferent neurons . Spinal cord-derived cytokines or growth factors such as TNF-α, IL-1β, IL-6 and/or BDNF, may also directly signal contralateral lumbar DRG neurons through receptors present on primary afferent central terminations [71, 72]. Perhaps not surprisingly, blockade of the action of TNF-α or inhibition of glial metabolic activity can attenuate bilateral nociceptive pain behavior [69, 73], while the low dose administration of a gap junction protein decoupler only extinguishes only contralateral pain behavior . Although current trends in pain research favor bilateral spinal cord glial activation as a mechanism of central activation in the spinal cord, it does not appear to be central to all pain conditions . It is also interesting to note in the context of the present set of investigations that TNF-α can upregulate MCP-1 expression by sensory neurons , further supporting the possibility that it may function as an upstream regulator of chemokine signaling in the DRG.
Alternatively, aspects of bilateral tactile hyperalgesia may be due to spontaneous ectopic activity in A, but not C-fibers. This type of ongoing ectopic hyperexcitability in primary sensory neurons post-injury can occur in both injured neurons and adjacent, uninjured neurons [20, 37, 76, 77]. Changes necessary for this type of mechanism implicate modification of the electrical properties of the neurons [78–80]. Nerve demyelination in the mid-thigh may effectively trigger just this type of change in the primary sensory neuron (i.e. neurochemistry and physiology of primary afferent neurons), which then contribute to central sensitization and higher levels of nociceptive sensory processing. Taken together, the evidence of chronic changes in chemokine/receptor protein expression and the ability of certain chemokines to excite neuronal subpopulations [6, 8] is suggestive of a potential scenario.
Another behavioral component commonly observed with rodent pain models, especially those accompanied by robust inflammatory responses, is the presence of thermal hyperalgesia. Despite a presence of thermal hyperalgesia in the mouse focal nerve demyelination model , the rat nerve demyelination model does not exhibit changes in response to temperature. Thermal hyperalgesia is largely thought to be a pain-related symptom caused by peripheral sensitization. The absence of thermal hyperalgesia would suggest that there is a lack of ongoing inflammatory mediator-initiated sensory neuron signaling. Lack of thermal hyperalgesia in peripheral nerve injury models of pain, although not common, include perineural gp-120 administration ; Bhangoo and White, unpublished observations) acidic saline-induced hyperalgesia  and chronic constriction injury performed in a 5-HT transporter knockout mouse . Lack of thermal hyperalgesia in the model of muscle pain and CCI implicate central descending mechanisms for the display of bilateral hyperalgesia. It is possible that similar mechanisms are operating within the rat nerve demyelination injury model.
Taken together the data suggest that upregulation of chemokine signaling by sensory neurons may help to integrate several phenomena that account for changes in the properties of peripheral nerves resulting in bilateral pain hypersensitivity. The present results, together with previous studies , suggest that the mode of injury may determine which particular chemokines play a central role in maintaining the neuropathic pain state. Chemokine receptors may then represent novel targets for therapeutic intervention in demyelination associated neuropathic pain as well as other chronic pain states.