Significant joint pain is usually widespread beyond the affected joint, which results from the sensitization of nociceptive neurons in the central nervous system (central sensitization). This study was undertaken to explore whether the proinflammatory cytokine interleukin-6 (IL-6) in the joint induces central sensitization, whether joint inflammation causes the release of IL-6 from the spinal cord, and whether spinal IL-6 contributes to central sensitization.
In anesthetized rats, electrophysiologic recordings of spinal cord neurons with sensory input from the knee joint were made. Neuronal responses to mechanical stimulation of the rat knee and leg were monitored. IL-6 and soluble IL-6 receptor (sIL-6R) were applied to the knee joint or the spinal cord. Spinal release of IL-6 was measured by enzyme-linked immunosorbent assay. Soluble gp130, which neutralizes IL-6/sIL-6R, was spinally applied during the development of joint inflammation or during established inflammation.
A single injection of IL-6/sIL-6R into the rat knee joint as well as application of IL-6/sIL-6R to the rat spinal cord significantly increased the responses of spinal neurons to mechanical stimulation of the knee and ankle joint, i.e., induced central sensitization. Application of soluble gp130 to the rat spinal cord attenuated this effect of IL-6. The development of knee inflammation in the rat caused spinal release of IL-6. Spinal application of soluble gp130 attenuated the development of inflammation-evoked central sensitization but did not reverse it.
Our findings indicate that the generation of joint pain in the rat involves not only IL-6 in the joint but also IL-6 released from the spinal cord. Spinal IL-6 contributes to central sensitization and thus promotes the widespread hyperalgesia observed in the course of joint disease.
Rheumatoid arthritis (RA) and osteoarthritis (OA) are among the most frequent causes of chronic pain (1). It is now apparent that RA and OA pain involve substantial changes in the nociceptive (pain) system at all levels of the neuraxis, including the peripheral neurons and the central nervous system (2, 3). As a result, patients with advanced OA, for example, often report widespread pain beyond the affected joint (4, 5) and exhibit lower pressure pain thresholds in cutaneous and subcutaneous structures of the leg (4, 6). The brains of RA patients show strong activation of the so-called pain matrix (7), but atrophy of pain-related regions in the thalamus (8) and cortex (9) has also been reported, indicating severe alterations.
Widespread pain beyond the site of disease is attributed to “central sensitization,” a state of hyperexcitability of nociceptive neurons in the spinal cord (10). Indeed, patients with knee OA exhibit larger areas of referred pain and higher pain summation scores upon repetitive pressure on the knee than do controls, which indicates central sensitization (11). Spinal cord recordings in animals identified neuronal changes which are thought to underlie these findings in patients. Upon the development of inflammation in the joint, spinal cord neurons become more responsive to mechanical stimulation of the joint and, importantly, show enhanced responses to mechanical stimuli applied to healthy regions adjacent to and even remote from the inflamed joint (3). This spinal hyperexcitability is induced by sensory input from the inflamed joint. In addition, descending inhibitory systems from the brainstem may be less effective (12) and/or descending excitatory systems from the brainstem may be overactive (13, 14) in animals and humans with painful joint diseases. Spinal hyperexcitability also plays a role in the regulation of joint inflammation (15, 16).
Spinal hyperexcitability is elicited by changes in synaptic transmission mediated by the transmitter glutamate and the neuropeptides substance P and calcitonin gene-related peptide (3). More recently it has been recognized, mainly in studies on neuropathic pain (17), that mediators of the immune system, including interleukin-6 (IL-6) (18), modify the nociceptive processing in the spinal cord. Because the immune system plays a dominant role in painful RA, it is conceivable that the immune system contributes significantly to the generation of inflammatory joint pain. Indeed, both tumor necrosis factor (TNF) (19) and IL-6 (20) potently sensitize joint nociceptors. Because cytokines can also be produced in the spinal cord, e.g., by glial cells (17), the question of whether peripheral joint inflammation causes an increase in these mediators in the spinal cord and whether they contribute to central sensitization is intriguing. In fact, rats with articular Freund's complete adjuvant–induced inflammation and rats with generalized adjuvant-induced arthritis showed increased spinal levels of IL-6 (21, 22).
In the present study, we focused on the putative role of IL-6 signaling in the process of central sensitization in the rat spinal cord. In general, IL-6 can bind to the membrane-bound IL-6 receptor (IL-6R), which acts in cooperation with the transmembrane signal-transducing subunit glycoprotein 130 (gp130). Alternatively, IL-6 can bind to a soluble IL-6R (sIL-6R), and the IL-6/sIL-6R complex can bind to gp130 in cells that do not express the membrane-bound IL-6R, thus leading to IL-6 transsignaling (23, 24). Because the coadministration of IL-6 and sIL-6R caused stronger peripheral sensitization than IL-6 alone (20), in the present study we tested the effects of IL-6/sIL-6R on central sensitization. We recorded the responses of neurons in the exposed spinal cord in anesthetized rats and investigated first whether the injection of IL-6/sIL-6R into the knee joint is sufficient to induce central sensitization. We then examined whether application of IL-6/sIL-6R to the rat spinal cord changes the responsiveness of the neurons to stimulation of the joint. Finally, we investigated whether IL-6 is released from the spinal cord during the development of inflammation in the rat knee and whether the generation of central sensitization is attenuated by neutralizing spinal IL-6/sIL-6R.
MATERIALS AND METHODS
Established standard procedures were used for electrophysiologic recordings. Fifty-nine male Wistar rats (200–350 gm; University of Jena) were anesthetized by intraperitoneal injection of sodium thiopental (Trapanal; BYK Gulden) at an initial dose of 90–120 mg/kg. Further intraperitoneal injections of thiopental (20 mg/kg) were administered to maintain deep anesthesia, as assessed by the absence of corneal or leg withdrawal reflexes. The animals breathed spontaneously. Mean arterial blood pressure was measured with a catheter in the common carotid artery (most rats had normal values of 90–120 mm Hg). Body temperature was maintained at 37°C with a feedback-controlled system. Laminectomy was performed to expose spinal cord segments L1–L4, and the dura mater was opened. To apply compounds to the spinal cord, a trough with ∼50 μl capacity (or 120 μl capacity for release experiments) was formed over the region in which the recordings were to be performed by sealing an elliptic rubber ring (∼3 × 5 mm) onto the spinal surface with silicone gel. The area around the trough was sealed with 3% agar in Tyrode's solution.
Induction of inflammation in the rat knee joint.
We used the kaolin/carrageenan inflammation model, which allows direct monitoring of the generation of spinal hyperexcitability during the development of inflammation (25, 26). A kaolin suspension (4% in 70 μl; Sigma) was injected into the left knee joint, and then the joint was extended/flexed slowly for 15 minutes. Carrageenan (2% in 70 μl; Sigma) was then injected, and the joint was extended/flexed for 5 minutes.
Electrophysiologic recording and stimulation.
Using glass-insulated carbon filament electrodes, action potentials (APs) were recorded extracellularly from single dorsal horn neurons in the rat in vivo. Individual neurons were identified by their AP shape monitored on an oscilloscope. APs were stored on a hard disk for offline discrimination. Neurons that responded to pressure on the ipsilateral knee but not to brushing or squeezing of the skin over the knee were selected. The total receptive field was mapped using mechanical stimulation of the skin and deep tissue. For the testing protocol, innocuous and noxious pressure were each applied to the rat knee, ankle, and paw (Figure 1A). Each of the 6 stimuli lasted for 15 seconds, followed by an interval of 15 seconds without stimulation. This entire stimulation cycle was repeated every 5 minutes. A mechanical device (Correx; Haag-Streit) was used for compression of the rat knee joint in the mediolateral axis at innocuous intensity (1.9N/40 mm2) and noxious intensity (7.8N or 5.9N/40 mm2; felt painful when applied to the experimenter's finger). Modified crocodile clips were used to apply innocuous pressure (1.1N/20 mm2) or noxious pressure (5.8N/20 mm2) to the ankle and paw.
Release of IL-6 from the rat spinal cord.
Buffer solution (120 μl) was poured into the spinal trough, withdrawn after 20 minutes, and immediately frozen at −20°C. The samples were processed with a commercial enzyme immunoassay set for IL-6 (BDset OptEIA; eBiosciences).
The following questions were addressed.
Does the injection of recombinant human IL-6 plus recombinant human sIL-6R into the rat knee joint alter spinal responses to peripheral stimulation?
Recombinant human IL-6 was obtained from Bachem, and recombinant human sIL-6R was obtained from R&D Systems. IL-6 and sIL-6R were coinjected into the rat knee joint, each at a dose of 20 ng in a total volume of 100 μl of saline containing 0.25% bovine serum albumin (BSA) (20). Responses were assessed for 30 minutes before and 175 minutes after IL-6/sIL-6R injection. Control animals received an injection of 100 μl of vehicle.
Does topical application of IL-6/sIL-6R to the rat spinal cord alter the responses of nociceptive neurons to mechanical stimulation?
The neuron responses were initially assessed for 30 minutes with vehicle (Tyrode's solution with 1% BSA) in the spinal trough and then during the application of increasing concentrations of IL-6/sIL-6R (0.5 ng, 5 ng, and 50 ng in 50 μl of vehicle) to the rat spinal cord. Responses were monitored for 50 minutes after the administration of each dose.
Does coapplication of recombinant human soluble gp130 and IL-6/sIL-6R to the rat spinal cord prevent the effect of the spinal application of IL-6/sIL-6R alone?
Recombinant human soluble gp130 was obtained from R&D Systems. Soluble gp130 binds IL-6/sIL-6R complexes and thus prevents transsignaling (23, 24). IL-6/sIL-6R and soluble gp130 were applied to the rat spinal cord at a dose of 50 ng each in a total volume of 50 μl.
Is IL-6 released from the rat spinal cord during the development of joint inflammation?
The release of IL-6 was measured in 8 of the 13 rats in which the development of hyperexcitability was monitored with vehicle in the spinal trough (see below). Samples were obtained from the trough before and after the injection of kaolin/carrageenan.
Does application of soluble gp130 to the rat spinal cord alter the generation of inflammation-evoked spinal hyperexcitability?
Baseline responses were monitored for 30 minutes, soluble gp130 was applied to the rat spinal cord in an excessive concentration (50 ng in 50 μl), and recordings were continued for another 60 minutes. Inflammation was then induced, and recordings were performed for another 4 hours in the presence of soluble gp130. Vehicle was used in the spinal trough in 13 control rats.
Does application of soluble gp130 to the rat spinal cord during established inflammation and hyperexcitability influence spinal responses to mechanical stimulation?
Recordings were started 7–11 hours after kaolin/carrageenan injection. Baseline responses were monitored for 30 minutes with vehicle on the spinal cord, and then soluble gp130 (50 ng in 50 μl) was applied to the spinal cord and responses were monitored for 100 minutes.
Labeling for IL-6–like immunoreactivity in rat dorsal root ganglia (DRGs) and spinal cord sections.
Five adult rats were killed with CO2, and the DRGs and the dorsal roots from all spinal levels and the lumbar spinal cords were prepared and fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) for at least 24 hours at 4°C. The tissue specimens were watered, dehydrated, transferred to methyl benzoate (100%), embedded in paraffin (Histosec; Merck), cut into 5-μm sections, and mounted onto slides (Superfrost Plus; Menzel-Gläser). The sections were dewaxed in xylol, hydrated, and autoclaved for 15 minutes (120°C; 1 bar) in 0.1M citrate buffer (pH 6.0). After cooling, the sections were washed in PBS and incubated for 30 minutes in PBS plus 1% Triton X-100 plus 2% goat serum (Rockland).
To label for IL-6–like immunoreactivity, we used a polyclonal antibody raised in rabbits against recombinant human IL-6 produced in Escherichia coli that also recognized IL-6 in the rat (Abcam). The antibody was diluted 1:200 in a solution of PBS plus 1% Triton X-100 containing 1% gelatin from coldwater fish skin, in which all of the sections were incubated overnight at 4°C. As secondary antibodies for DRG and spinal cord sections, we used a Cy3-labeled goat anti-rabbit antibody (Dianova), diluted 1:100 in PBS plus 1% Triton X-100 containing 1% gelatin from coldwater fish skin, in which sections were incubated for 4 hours at 20°C. For IL-6 labeling in dorsal roots, we used a biotinylated goat anti-rabbit antibody (1:100; Dako) for 4 hours at 20°C, followed by application of avidin–biotin–peroxidase complex (Vectastain Elite ABC kit; Vector) for 40 minutes, and then visualization with Jenchrom px blue (MoBiTec). Labeling of spinal cord sections for IL-6–like immunoreactivity was combined with labeling for neurons with a monoclonal anti-NeuN antibody (against purified cell nuclei from the mouse brain, developed in mice; Millipore) diluted 1:100 and added overnight. The secondary antibody was a Cy2-labeled goat anti-mouse antibody (Dianova), diluted 1:100 in PBS plus 1% Triton X-100 containing 1% gelatin from coldwater fish skin. Dehydrated sections were embedded in Entellan (Merck). Control experiments were carried out with the omission of the primary antibodies, and no clear fluorescence signals were detectable.
In every second DRG section (of 20 sections per rat), we determined the proportion of neuronal profiles with IL-6–like immunoreactivity, using a light microscope (Axioplan 2; Zeiss) coupled to a CCD video camera and an image analyzing system (KS 300; Zeiss). We measured the diameter and fluorescence intensity of each DRG neuron. Neurons showing higher intensities than control neurons not treated with the anti–IL-6 antibody were considered to be positively labeled for IL-6. Using standard fluorescence filters, spinal cord sections were inspected for double labeling of NeuN-like and IL-6–like immunoreactivity.
To evaluate whether changes within experimental groups were significant, baseline responses and responses during time intervals following treatment were compared using Wilcoxon's matched pairs signed rank test. Differences between groups within the same time interval were compared by Mann-Whitney U test. Only raw data were used, and P values less than 0.05 were considered significant.
Rat spinal neuron samples.
Spinal cord neurons with sensory input from the rat knee were recorded in the gray matter of the deep dorsal horn (at a mean ± SEM depth of 743 ± 171 μm [range 404–1,150 μm]) of segments L1–L4 (1 neuron per animal). APs in these neurons were elicited by compression of the ipsilateral knee joint but not by brushing or squeezing the skin overlying the knee. Usually, these neurons also responded to compression of the muscles adjacent to the knee joint and to compression of the ankle joint, i.e., they had large receptive fields. Most of them were “wide dynamic range neurons” that showed small responses to moderate compression of the leg but much stronger responses to noxious (painful) compression of the tissue.
Effect of intraarticular injection of IL-6/sIL-6R on spinal hyperexcitability in the rat.
The coinjection of IL-6 and sIL-6R (20 ng each) into the normal knee joint sensitizes nociceptive C fibers of the knee joint so that their responses to mechanical stimulation increase within 2 hours (20). In this study, we tested whether such IL-6/sIL-6R coinjection into the rat knee joint induces spinal hyperexcitability. We monitored the responses of spinal neurons to innocuous and noxious pressure applied to the rat knee, ankle, and paw before and after the injection of IL-6/sIL-6R (Figure 1A). Approximately 1 hour after the injection of IL-6/sIL-6R, the responses to knee stimulation increased substantially, and the responses to ankle stimulation increased almost in parallel (Figure 1B). In Figure 1B, each symbol indicates the average of the responses of all neurons to the particular test stimulus, and the responses before the injection were set to 0. There was a significant increase in response, as measured by the average number of APs per stimulus, from before IL-6/sIL-6R injection to 2 hours after IL-6/sIL-6R injection (P < 0.05 by Wilcoxon's matched pairs signed rank test) (Figure 1C, top). Injection of vehicle into the rat knee joint did not change the responses (Figure 1C, bottom and Figure 1D). The difference between responses at baseline and in the second hour after injection into the knee was significantly higher in the IL-6/sIL-6R–treated rats than in the vehicle-treated rats (P < 0.05 by Mann-Whitney U test). Thus, injection of IL-6/sIL-R6 into the knee rendered spinal cord neurons more responsive to mechanical stimulation of both the injected knee and the ankle.
Effect of spinal application of IL-6/sIL-6R on spinal hyperexcitability in the rat.
We tested whether the application of IL-6/sIL-6R to the exposed spinal cord increased the responses to mechanical stimulation in the rat (Figure 2A). After the baseline response to stimulation was established, vehicle was replaced by increasing doses of IL-6/sIL-6R, and the response to each dose was monitored for 50 minutes (Figure 2B). Following IL-6/sIL-6R application, the responses to innocuous and noxious pressure applied to the rat knee and ankle gradually increased above baseline. Figure 2C shows the mean number of APs per stimulus after administration of different doses of IL-6/sIL-6R. The responses were already significantly increased after the addition of the lowest dose (P < 0.05 by Wilcoxon's matched pairs signed rank test). As shown in Figure 2D, the increase in responses to knee and ankle stimulation after IL-6/sIL-6R application was prevented by coapplication of soluble gp130 (each 50 ng in 50 μl), which neutralizes IL-6/sIL-6R complexes and thus prevents interaction with the transducing membrane complex (23, 24).
We further observed that 2 of 11 neurons did not respond to innocuous stimulation of the rat knee initially and started to respond to this stimulus after the application of IL-6/sIL-6R. One neuron was not activated by stimulation of the ankle initially but responded to such stimulation after the application of IL-6/sIL-6R to the rat spinal cord, and 3 neurons showed an expansion of their receptive fields to include the paw. Thus, spinally applied IL-6/sIL-6R enhanced the responses of spinal cord neurons to peripheral stimulation.
Spinal release of IL-6 during the development of inflammation in the rat knee joint.
After kaolin/carrageenan injection in the rat, inflammation reaches a plateau within 1–6 hours (25, 26). In order to explore whether knee inflammation causes the release of IL-6 from the spinal cord, we measured the content of IL-6 in the supernatant over the exposed spinal cord. Figure 3A shows the protocol. Supernatants were collected in 8 of the 13 rats in which we recorded the development of inflammation-evoked hyperexcitability with vehicle on the spinal cord (see below). Superfusates were obtained before any mechanical stimulation of the knee in 3 of the rats, and superfusates were obtained during the preinflammatory period when mechanical stimuli were applied, as well as in the second and fourth hour after the induction of inflammation, in all 8 rats. Before inflammation, the mean ± SEM IL-6 concentrations were 1.7 ± 0.5 ng/ml (prior to mechanical stimulation) and 1.7 ± 0.4 ng/ml (when mechanical stimuli were applied), indicating that IL-6 levels were not altered by the mechanical stimulation of normal rat joints. During the development of inflammation, the spinal release of IL-6 increased significantly, to 282% within the second hour and to 511% within the fourth hour, compared to the values during the period when mechanical stimuli were applied (P < 0.05 by Wilcoxon's matched pairs signed rank test) (Figure 3B).
Effect of the application of soluble gp130 to the rat spinal cord on inflammation-evoked spinal hyperexcitability.
Using neuronal recordings, we tested whether spinal release of IL-6 contributes to the development of spinal hyperexcitability. We recorded the responses of spinal cord neurons before and during the development of inflammation (Figure 3A), in rats treated with vehicle (n = 13) or soluble gp130 (50 ng in 50 μl; n = 7) in the spinal trough. Figure 3C shows the increases in the responses of the neurons above baseline in the presence of vehicle or soluble gp130. In both groups of rats, the responses to innocuous and noxious knee and ankle stimulation increased significantly (determined by Wilcoxon's matched pairs signed rank test), but the increase in the responses to innocuous stimulation was delayed in soluble gp130–treated rats, and the responses to noxious pressure reached lower values in soluble gp130–treated rats than in vehicle-treated rats. Within the first 3 hours of inflammation, the increase in the response to noxious pressure on the knee was significantly lower in the soluble gp130–treated group than in the vehicle-treated group (P < 0.05 by Mann-Whitney U test) (Figure 3C), whereas the difference between groups in the response to noxious stimulation of the ankle did not reach significance.
The effect of spinal application of soluble gp130 (50 ng in 50 μl) on neuronal responses was also tested in rats in which the joint was inflamed for 7–11 hours before the recordings started. In these rats, soluble gp130 did not affect the neuronal responses to stimulation of the leg (Figure 4). Thus, soluble gp130 did not reverse established spinal hyperexcitability.
IL-6–like immunoreactivity in primary afferent fibers and the spinal cord.
In order to identify the potential sources of endogenous spinal IL-6 in the rat, we labeled sections of DRGs, dorsal roots, and the spinal cord for IL-6–like immunoreactivity. Figure 5A shows that a proportion of DRG neurons exhibited labeling for IL-6–like immunoreactivity. Figure 5B displays a control section with omission of the primary antibody. IL-6–like immunoreactivity was also detected in rat dorsal roots (Figure 5A, right). For unbiased quantification, the fluorescence intensities of single neurons in antibody-labeled sections (Figure 5C) and in control sections without primary antibody (Figure 5D) were plotted. Figure 5C shows neurons with fluorescence above background (which is displayed in Figure 5D). A mean ± SD of 17.8 ± 6.1% of the DRG neurons (n = 5 rats) exhibited IL-6–like immunoreactivity, and most of these neurons were small and medium sized. Thus, primary afferent fibers are a source of spinal IL-6.
Figures 5E and F display a rat dorsal horn section labeled for NeuN-like immunoreactivity, a neuronal marker (Figure 5E), and for IL-6–like immunoreactivity (Figure 5F). IL-6–like immunoreactivity appeared as small spots (arrows in Figure 5F), but double labeling for NeuN-like immunoreactivity showed that these spots were not neurons. Thus, IL-6 in the spinal cord of the normal, healthy rat is expressed in nonneuronal structures, presumably glial cells (17, 18, 21).
This study addressed the role of IL-6 in the generation of inflammation-evoked pain in the rat. The administration of IL-6/sIL-6R either into the rat knee joint or topically to the spinal cord caused an increase in the responses of spinal neurons to mechanical stimulation of the knee and other parts of the leg, including an expansion of the size of the receptive field of the neurons. Knee inflammation in the rat induced significant spinal release of IL-6. Spinal application of soluble gp130 prevented the increase in responses that occurred after spinal application of IL-6/sIL-6R and attenuated the generation of spinal hyperexcitability during the development of inflammation. However, the application of soluble gp130 to the rat spinal cord did not reverse established hyperexcitability after inflammation was fully developed.
IL-6 and sIL-6R are key players in systemic inflammation and arthritis (27, 28). RA patients exhibit enhanced levels of IL-6 (29, 30) and sIL-6R (31) in serum and synovial fluid, which facilitates inflammation (32, 33) and joint destruction (34). The neutralization of IL-6 with tocilizumab has become an option in RA therapy (35). Recent experiments showed that IL-6 is also a potent sensitizer of nociceptive sensory fibers (20). IL-6 is likely to have, at least in part, a direct effect on the sensory neurons because peripheral nerve fibers express the transmembrane signal-transducing subunit gp130 (36, 37). In behavioral experiments, soluble gp130 injection into the knee joint at the time of the induction of antigen-induced arthritis significantly reduced inflammation-evoked mechanical hyperalgesia, while joint swelling was not reduced at this stage (38).
In the present study, the injection of IL-6/sIL-6R into the rat knee joint was sufficient to elicit a typical pattern of inflammation-evoked spinal hyperexcitability consisting of enhanced responses to both knee and ankle stimulation (25, 26) (Figure 1C). Such a pattern most likely results from the increased responsiveness of spinal neurons generated by intraspinal mechanisms (3, 10). However, the increase in responses was smaller than that usually seen during the development of inflammation (compare Figures 1C and 3C), presumably because IL-6/sIL-6R sensitizes only C fibers and not Aδ fibers (20) and because other mediators, such as TNF (19), contribute to inflammatory sensitization.
An increase in the responses to peripheral stimulation in the rat, including an expansion of the receptive fields of the spinal neurons toward the ankle or the paw, was also generated by the spinal application of exogenous IL-6/sIL-6R, showing that IL-6 also acts centrally. In fact, gp130, the signal-transducing complex and target of IL-6/sIL-6R, is abundant in neurons and glial cells of the spinal cord, suggesting different target sites of IL-6 (39, 40). The membrane receptor IL-6R (also called gp80) is mainly found in glial and endothelial cells and is sparse in neurons of the central nervous system (41). In this study, sensitization by application of IL-6/sIL-6R to the rat spinal cord was prevented by concomitant application of soluble gp130, which abolishes transsignaling (23, 24). It is likely that the application of both IL-6 and sIL-6R is necessary, since in a previous study the spinal application of IL-6 alone did not enhance spinal hyperexcitability (42). The precise mechanism of the sensitization is not known. The effects of IL-6 may depend on a decrease in inhibitory currents (43) and/or changes in transmitter release.
This is the first study to show that IL-6 is released from the spinal cord in the initial hours of peripheral inflammation in the rat. We cannot exclude the possibility that the high initial level of IL-6 was partly due to the preceding surgery (but see also ref.44). While microglia and astroglia are known sources of IL-6 (21), we also found IL-6–like immunoreactivity in a substantial proportion of DRG neurons in the rat, suggesting that significant amounts of IL-6 may actually be released from primary afferent fibers. In fact, cultured DRG neurons release IL-6 upon stimulation (Ebersberger A, et al: unpublished observations). The attenuation of spinal hyperexcitability by soluble gp130 strongly suggests that IL-6 release is important for the generation of full spinal hyperexcitability. Although a number of other cytokines share the transmembrane receptor gp130 with IL-6 (40), the effect of soluble gp130 is quite specific for IL-6/sIL-6R transsignaling (23).
The early release of IL-6 after induction of inflammation and the effect of soluble gp130 are strongly coincident. However, soluble gp130 did not fully prevent the development of hyperexcitability. This incomplete effect may have several causes. First, IL-6 is a modulator of synaptic transmission, whereas glutamate is the key mediator of synaptic transmission (10, 25), and neuropeptides and spinal prostaglandins (3, 26, 45) also contribute to spinal hyperexcitability. Second, soluble gp130 prevents transsignaling (implying that endogenous sIL-6R must be present) but not the direct effects of IL-6 on membrane-bound IL-6R. Substantial levels of IL-6R have been found in the DRGs and in cortical tissue under pathophysiologic conditions (41, 46, 47), indicating that the expression of IL-6R may limit the ability of soluble gp130 to block IL-6 signaling as a whole.
The application of soluble gp130 to the rat spinal cord 7–11 hours after the induction of inflammation and spinal hyperexcitability did not reduce the responses to stimulation of the inflamed knee joint and the adjacent and remote areas. An attractive hypothesis is that IL-6 triggers persistent effects and that after the induction of hyperexcitability, the continuous presence of IL-6 is no longer necessary to maintain it. The findings of previous studies support this idea. In peripheral nociceptors, soluble gp130 prevented the sensitization caused by IL-6/sIL-6R but did not reverse established mechanical sensitization (20). A single application of IL-6 primed muscle nociceptors for prostaglandin effects for at least 24 hours (48). The preventive application of soluble gp130 during the induction of antigen-induced arthritis reduced mechanical hyperalgesia, but systemic treatment with soluble gp130 after the onset of antigen-induced arthritis did not reduce mechanical hyperalgesia in the first 2 weeks (38). We propose, therefore, that IL-6 may be a key player in the induction of chronic pain in the deep somatic tissue, and quick pain relief may not be expected from the neutralization of IL-6 during established disease.
In summary, IL-6–deficient mice have been shown to exhibit less hyperalgesia in response to mechanical and thermal stimulation after subcutaneous injection of carrageenan than wild-type mice (49), and the up-regulation of spinal IL-6 in models of inflammation (21, 22) and models of neuropathic pain (50) suggested a role of spinal IL-6 in pain. The present study demonstrated a significant role of peripheral and spinal IL-6 signaling in the generation of inflammation-evoked central sensitization and hyperalgesia in the rat.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Ebersberger had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Vazquez, Kahlenbach, Segond von Banchet, König, Schaible, Ebersberger.
Acquisition of data. Vazquez, Kahlenbach, Segond von Banchet, König, Schaible, Ebersberger.
Analysis and interpretation of data. Vazquez, Kahlenbach, Segond von Banchet, König, Schaible, Ebersberger.
The authors thank Mrs. Konstanze Ernst and Mrs. Gabriele Weigand for excellent technical assistance.