Sensitization of unmyelinated sensory fibers of the joint nerve to mechanical stimuli by interleukin-6 in the rat: An inflammatory mechanism of joint pain




Pain during mechanical stimulation of the joint and spontaneous pain are major symptoms of arthritis. An important neuronal process of mechanical hypersensitivity of the joint is the sensitization of thin myelinated Aδ fibers and unmyelinated C fibers innervating the joint. Because interleukin-6 (IL-6) is a major inflammatory mediator, we investigated whether this cytokine has the potential to sensitize joint afferents to mechanical stimuli.


In electrophysiologic experiments conducted on anesthetized rats, action potentials were recorded from afferent fibers supplying the knee joint. Responses to innocuous and noxious rotation of the tibia against the femur in the knee joint were monitored before and 1–2 hours after injection of test compounds into the joint cavity.


Injection of IL-6 and coinjection of IL-6 plus soluble IL-6 receptor (sIL-6R) caused a gradual increase in the responses of C fibers to innocuous and noxious rotation within 1 hour. The increase in responses to IL-6 and IL-6 plus sIL-6R was prevented by coadministration of soluble glycoprotein 130 (sgp130), but sgp130 did not reverse established mechanical hyperexcitability. Responses of Aδ fibers were not altered by the compounds. While injection of sIL-6R alone into the normal knee joint did not influence responses to mechanical stimulation, injection of sIL-6R into the acutely inflamed knee joint caused an increase in responses.


IL-6 has the potential to sensitize C fibers in the joint to mechanical stimulation. Thus, IL-6 contributes to mechanical hypersensitivity, most likely due to an action of IL-6 on nerve fibers themselves.

The proinflammatory cytokine interleukin-6 (IL-6) plays an important role in the pathogenesis of rheumatoid arthritis (RA). The concentration of IL-6 is elevated in the serum and synovial fluid of patients with arthritis (1, 2), corresponding to disease activity (3, 4). IL-6 acts on target cells in a 2-step process. First, IL-6 binds to a specific IL-6–binding protein, IL-6 receptor (IL-6R), which is located either on the cell membrane or is soluble (sIL-6R) in the extracellular space. Second, the IL-6/IL-6R complex binds to the transmembrane signal-transducing subunit glycoprotein 130 (gp130), which confers the signal to intracellular cascades (5–8). By binding to sIL-6R, IL-6 can also stimulate cells that express only gp130 but lack membrane-bound IL-6R. The sIL-6R concentration is also elevated in synovial tissue in RA and correlates with the degree of leukocyte infiltration (9–11). Experimentally, actions of IL-6 can be antagonized by application of soluble gp130 (sgp130), which binds IL-6 and prevents it from activating gp130 (12, 13).

An important symptom of arthritis is pain at rest and/or during normal movements, and the inflamed joint is painful during palpation (mechanical allodynia). Joints are densely innervated. Sensory endings of afferent fibers are located in the fibrous capsule, ligaments, and in the synovial layer (14). An important neuronal mechanism of inflammatory joint pain is the sensitization of slowly conducting thin myelinated sensory Aδ fibers and unmyelinated C fibers to mechanical stimuli. Essentially, sensitization has 2 consequences. It lowers the mechanical activation threshold of high-threshold nociceptors such that they are activated by normally innocuous stimuli, and it increases the responses of these fibers to suprathreshold stimuli. In addition, sensitization increases the responsiveness of low-threshold Aδ fibers and C fibers that respond with low discharge rates at innocuous stimulus intensities and with higher discharge rates at noxious intensities (15).

Direct activation and sensitization of articular Aδ fibers and C fibers to mechanical stimuli is produced by inflammatory mediators such as bradykinin, prostaglandins, and others (for review, see ref.15). Recently, evidence was provided that proportions of sensory neurons also express receptors for cytokines, and indeed, peripheral nociceptive effects of cytokines have been reported in behavioral studies (16, 17). IL-6 plus sIL-6R induces thermal hyperalgesia on rat skin (18), and sIL-6R alone or in combination with IL-6 was shown to sensitize cutaneous nociceptors to noxious heat (19). In patch clamp experiments, bath application of IL-6 plus sIL-6R (Hyper–IL-6), a gp130-stimulating designer IL-6–sIL-6R fusion protein, potentiated heat-activated inward currents in isolated dorsal root ganglion (DRG) neurons (18), showing prompt action of IL-6 on sensory neurons. Long-term administration of IL-6 to cultured DRG neurons potentiated the expression of the neurokinin 1 receptor (20), which also suggests direct, long-lasting effects on sensory neurons.

In the present electrophysiologic study, we investigated whether IL-6 has a pronociceptive effect in the joint in situ by influencing the response properties of joint afferents. Specifically, we tested whether application of IL-6 alone or a combination of IL-6 and sIL-6R (Super–IL-6) can enhance the responsiveness of Aδ fibers and C fibers in the rat knee joint to innocuous and noxious joint movements. Furthermore, we tested the effect of the antagonist sgp130. Preliminary data have been previously reported (21).


Animal preparation.

The experiments were approved by the animal committee of the regional government of Thüringen, Germany (Reg. No. 02-07/04). Male Wistar rats (weight 300–460 gm) were anesthetized using intraperitoneal (IP) injections of 100 mg/kg thiopentone sodium (Altana, Konstanz, Germany). Supplemental doses (20 mg/kg IP) were administered as necessary to maintain areflexia. The trachea was cannulated, and the animals breathed spontaneously during surgery and the recording of data. A gentle jet of oxygen was blown toward the opening of the tracheal cannula. Mean arterial blood pressure was continuously monitored by means of a catheter in the right carotid artery. The absence of changes in blood pressure due to noxious stimuli confirmed the depth of anesthesia. A catheter in the right external jugular vein allowed volume substitution (5% glucose solution) if necessary. Body temperature was kept at 37°C using a feedback-controlled temperature controller (L/M-80; List, Darmstadt, Germany).

In 16 rats, acute inflammation (for review, see ref.22) was induced in the right knee joint 7–11 hours before recording of action potentials (APs). Through the patellar ligament, 0.1 ml of a 4% kaolin suspension (Sigma, Deisenhofen, Germany), dissolved in distilled water, was slowly injected into the articular cavity. Then the joint was slowly flexed and extended for 15 minutes. Thereafter, 0.1 ml of a 2% carrageenan solution (Sigma), dissolved in distilled water, was injected, and the joint was moved for another 5 minutes.

For exposure of the right knee and recording of APs, the rat was lying on its back. The skin was incised from the medial side of the lower leg to the belly. The skin flaps were sewn to an oval metal ring fixed over the leg to form a pool, which was filled with paraffin oil to prevent drying of the tissue. For stimulation of joint afferents, the knee was rotated outward and inward (for review, see ref.23). Briefly, a special fastener fixed the right femur. The right hind paw was fixed in a shoe-like holder that was connected to a force transducer and torque meter.

Figure 1A illustrates the recording setup. The anteromedial aspect of the knee is innervated by the medial articular nerve, which usually joins the saphenous nerve (or sometimes the femoral nerve). Medial articular nerve afferents were recorded as closely as possible to the inguinal region because a long distance from the knee reduces the mechanical instability of the recording conditions. The femoral nerve was cut, and thin bundles were placed on a small Perspex plate for further dissection with ultrafine watchmaker tweezers. To eliminate fiber activity from the lower leg, the saphenous nerve was transected distally to the medial articular nerve.

Figure 1.

Recording of neuronal activity from sensory fibers innervating the knee joint. A, Experimental setup showing the innervation of the anteromedial portion of the knee joint by the medial articular nerve and the approximate position of the recording electrode. coll. lig. = collateral ligament. B, Nerve activity in a filament that contained a fiber with action potentials (APs) of large amplitude and additional fibers with small amplitudes in the background. APs were elicited by outward rotation at a noxious torque. C, Peristimulus time histogram showing the sequence of stimuli in 1 test block and the evoked responses displayed as the number of APs/second (Hz). Each stimulus was applied for 15 seconds, followed by a 15-second break. RA = resting activity; Inn. OR = innocuous outward rotation; Inn. IR = innocuous inward rotation; Nox. OR = noxious outward rotation; Nox. IR = noxious inward rotation. D, Centers of mechanical receptive fields of C fibers. E, Centers of mechanical receptive fields of Aδ fibers.

Recording of afferent fiber responses.

Thin nerve filaments were placed on a platinum wire electrode. The reference platinum wire electrode was attached to connective tissue. Amplified APs were continuously monitored with a digital oscilloscope to observe their shape and size. A filament was accepted when it contained either just 1 fiber or 2–3 fibers with discernible APs (see Figure 1B). The signals were also entered into a personal computer (interface card DAB 1200; Microstar Laboratories, Bellevue, WA) for offline analysis of APs, using the SPIKI/SPIDI software package (24), and construction of peristimulus time histograms (Figure 1C).

Medial articular nerve afferents were searched by probing the anteromedial side of the knee with a blunt glass rod at moderate intensity. If APs were elicited, the exact location and size of the receptive field of this fiber were characterized using calibrated von Frey hairs. Only fibers that had a receptive field in the knee and responded to outward and/or inward rotation of the knee were included. Filaments that exhibited regular spontaneous activity (typical of muscle spindles) were discarded. The conduction velocity of the nerve fibers was determined by stimulating the mechanical receptive field using a bipolar electrode (1–10V, 0.5 msec pulse width) and dividing the distance between the receptive field and the electrode by the latency between stimulus artifact and evoked AP. Nerve fibers conducting at ≤1.25 meters/second were classified as C fibers, those conducting at 1.25–10 meters/second were classified as Aδ fibers, and those conducting at ≥10 meters/second were classified as Aβ fibers (25).

Experimental protocol.

Fiber responses were tested with a sequence of mechanical stimuli (Figure 1C). In each test block, resting activity was recorded for 1 minute, then innocuous outward and inward torque (20 mNm, for 15 seconds each) was applied, followed by noxious outward and inward torque (40 mNm for 15 seconds each). Rotation to 20 mNm was considered innocuous because at this intensity, the lower leg could be rotated to the end of the normal movement range. Rotation to 40 mNm was considered noxious because this rotation was performed against the resistance of the joint structures. The interval between test blocks was 3 minutes. The first 6–8 test blocks defined the baseline. Then 0.1 ml of the test substance was injected into the knee joint cavity, and another 12–15 test blocks were applied. In some experiments, a second injection was performed. At the end of the experiments, the rats were killed by intravenous administration of thiopentone sodium.

To assess the effects of volume expansion of the joint space on fiber responses, Tyrode's solution was injected into 7 normal rats and 3 rats with acute knee inflammation. Vehicle effects (1% bovine serum albumin [BSA] in phosphate buffered saline [PBS]) were tested in 4 normal rats. The following specific compounds were injected in a volume of 100 μl: recombinant human IL-6 (Bachem, Weil am Rhein, Germany), dissolved in 1% BSA in PBS (10 ng in 5 rats, 20 ng in 6 rats, 100 ng in 11 rats, and 200 ng in 5 rats); recombinant human sIL-6R (R&D Systems, Minneapolis, MN), dissolved in 1% BSA in PBS (20 normal rats and 6 rats with knee inflammation). Six normal rats and 6 rats with acute knee inflammation received 125 ng sIL-6R alone. A combination of IL-6 and sIL-6R (Super–IL-6) was coapplied in another 14 normal rats (sIL-6R 10 ng plus IL-6 10 ng in 5 rats; sIL-6R 20 ng plus IL-6 20 ng in 6 rats; and sIL-6R 100 ng plus IL-6 100 ng in 3 rats). In 14 normal rats and 7 rats with inflammation, we tested the effect of recombinant human sgp130 (R&D Systems) dissolved in 1% BSA in PBS. Four normal rats and 7 rats with inflammation received only 100 ng sgp130. Coapplication of 100 ng sgp130 and 100 ng IL-6 was performed in 4 normal rats, and simultaneous coapplication of 20 ng sgp130 and 20 ng IL-6 plus sIL-6R was performed in 3 normal rats. In 3 normal rats, we coadministered sIL-6R and IL-6 (both 100 ng) first and injected 100 ng sgp130 90 minutes later.

Data evaluation and statistical analysis.

Data are expressed as the mean ± SD. APs were discriminated offline according to shape and size and were counted before and during each stimulus. For normalization, we averaged all responses to each type of stimulus preceding drug application (baseline) and subtracted these values from the responses to the mechanical stimuli after drug application. Furthermore, we averaged data from 3 blocks of the recording protocol to 1 data point in the diagrams. For statistical analysis of changes within groups, we used Wilcoxon's matched pairs signed rank test. We compared the mean of 3 subsequent blocks before injection with the mean of groups of 3 subsequent test blocks during the recording period after injection. P values less than 0.05 were considered significant.


In 70 normal rats, recordings were performed on 101 C fibers and 12 Aδ fibers. Typically, prior to application of compounds, the receptive fields in the joint were small. Receptive fields of C fibers (Figure 1D) and Aδ fibers (Figure 1E) were located in the capsule over the tibial and/or femoral portions of the knee joint, and, in a few cases, in the capsule over the joint cavity. None of the fibers was spontaneously active at the resting position of the knee except for a few single spikes in some cases. The response pattern of a typical fiber is shown in the peristimulus time histogram in Figure 1C. Innocuous outward rotation and inward rotation evoked a small response, whereas noxious outward rotation elicited a much stronger response. In this fiber, noxious inward rotation caused only a small response. Single fibers had their activation thresholds between 5 and 38 mNm. Approximately 50% of them had a response within the innocuous range (<20 mNm). The mean ± SD response to innocuous outward rotation at 20 mNm was 28.1 ± 38.5 APs/15 seconds in Aδ fibers and 72.4 ± 129.6 APs/15 seconds in C fibers, and the average response at 40 mNm (noxious torque) was 208.6 ± 124.6 APs/15 seconds in Aδ fibers and 174.6 ± 223.8 APs/15 seconds in C fibers. Because responses to inward rotation were generally smaller, responses to outward rotation were chosen for display of data.

Effects of different doses of IL-6 on responses of Aδ fibers and C fibers.

Figures 2 and 3 show recordings of groups of C fibers that were tested for the effects of IL-6 on mechanically evoked responses. In all groups, the mean basal responses (baseline) to innocuous and noxious outward rotation before injection were set at 0 and responses after injection were expressed as the change from baseline. Injection of Tyrode's solution into the joint cavity caused a small but insignificant enhancement of responses to outward rotation (Figure 2A). Similarly, the injection of 0.1 ml of 1% BSA in PBS (vehicle) into the knee joint had no significant effect (Figure 3A). Intraarticular injection of 10 ng IL-6 alone (Figure 2B) did not significantly change responses to innocuous outward rotation (baseline mean ± SD 46.2 ± 12.7 APs/15 seconds) and noxious outward rotation (baseline 133.6 ± 9.7 APs/15 seconds). A second injection of 10 ng IL-6 72 minutes later had no effect either.

Figure 2.

Effects of interleukin-6 (IL-6) and the combination of IL-6 plus soluble IL-6 receptor (Super–IL-6 [SIL-6]) on responses of C fibers to mechanical stimulation of the joint. Curves show the change in the responses of groups of fibers to innocuous and noxious outward rotation of the knee from baseline. The baseline was set at 0, and arrows indicate the time of injection of the compounds. Changes in responses after injections of A, Tyrode's solution, B, 10 ng IL-6, C, 20 ng IL-6, and D, 20 ng Super–IL-6 (20 ng IL-6 plus 20 ng soluble IL-6 receptor) are shown. Values are the mean ± SD; n = number of fibers. ∗ = first statistically significant difference from baseline (P < 0.05 by Wilcoxon's matched pairs signed rank test). Innoc. outward = innocuous outward rotation; nox. outward = noxious outward rotation; AP = action potential.

Figure 3.

Effects of high concentrations of IL-6 on responses of C fibers to mechanical stimulation of the joint with A, vehicle (1% bovine serum albumin in phosphate buffered saline), B, 100 ng IL-6, and C, 200 ng IL-6. Recording periods were shorter with these concentrations. Values are the mean ± SD; n = number of fibers. ∗ = first statistically significant difference from baseline (P < 0.05 by Wilcoxon's matched pairs signed rank test.) See Figure 2 for definitions.

In contrast, application of 20 ng of IL-6 caused a slow and continuous increase in the responses to noxious torque, from a mean ± SD of 195.3 ± 4.7 to 262 ± 15.8 APs/15 seconds in C fibers, that reached significance 102 minutes after injection (Figure 2C). The increase in the responses to innocuous torque was not significant. A slight transient increase in both innocuous and noxious torque immediately after the injection was probably due to volume expansion of the joint cavity. Injection of 100 ng of IL-6 caused a higher and faster increase in the responses of C fibers to both innocuous torque (from 11.6 ± 3.3 to 37.1 ± 3.5 APs/15 seconds) and noxious torque (from 75 ± 18.2 to 182 ± 3.5 APs/15 seconds) that reached significance 48 minutes after injection (Figure 3B). A second injection of 100 ng of IL-6 did not further increase the responses (data not shown). However, injection of 200 ng of IL-6 decreased the responses of C fibers to both innocuous and noxious torque below baseline levels (innocuous torque from 86.7 ± 19.7 to 86.4 ± 6.8 APs/15 seconds, noxious torque from 176.9 ± 7.8 to 157.1 ± 10.3 APs/15 seconds) within 66 minutes in all experiments (Figure 3C). This decrease, however, did not reach significance. Positive IL-6 effects were observed in both low- and high-threshold C fibers. Aδ fibers were not influenced by IL-6. Ongoing spontaneous activity was induced in none of the fibers.

Coadministration of IL-6 and sIL-6R (Super–IL-6).

The application of sIL-6R alone was tested in 12 C fibers and did not influence responses to innocuous and noxious torque. Coapplication of 10 ng sIL-6R plus 10 ng IL-6 (10 ng Super–IL-6) had no effect either (8 C fibers). But application of 20 ng Super–IL-6 caused an increase in responses to noxious torque from mean ± SD 136.3 ± 15.7 APs/15 seconds (baseline) to a maximum of 204 ± 15.8 APs/15 seconds (Figure 2D). Significance was reached 48 minutes after injection. A second injection of 20 ng Super–IL-6 caused a further rise in responses to noxious rotation, up to 227.5 ± 14.5 APs/15 seconds 48 minutes after the application. Responses to innocuous torque insignificantly decreased below baseline (from 42.2 ± 16.3 to 27.5 ± 5.6 APs/15 seconds) after the first injection of 20 ng Super–IL-6, but increased significantly after the second injection of Super–IL-6 to 75.9 ± 6.5 APs/15 seconds within 72 minutes. In contrast, the responses of Aδ fibers were not affected by 20 ng Super–IL-6, even when a second injection was added 90 minutes after the first one. The application of 100 ng Super–IL-6 to 7 C fibers caused only an insignificant increase in the responses to innocuous and noxious torque.

Influence of sgp130 on IL-6 and Super–IL-6 effects.

As mentioned above, sgp130 acts as an antagonist of IL-6. The application of 100 ng sgp130 alone did not change the responses to innocuous and noxious rotation within 84 minutes after injection (Figure 4A). However, sgp130 prevented the increase in responses due to IL-6 and Super–IL-6. Whereas IL-6 and Super–IL-6 substantially increased responses in C fibers to noxious outward rotation of the knee (Figures 2C and D, and 3B), coadministration with sgp130 prevented these effects (see Figure 4A for coadministration of 100 ng sgp130 and 100 ng IL-6). Rather, an insignificant decrease in responses below baseline was observed when 20 ng sgp130 was coadministered with 20 ng Super–IL-6 (Figure 4B).

Figure 4.

Effects of soluble gp130 (sgp130) on the responses of C fibers to outward rotation. A, No change in responses to innocuous and noxious outward rotation after injection of sgp130 either alone or with IL-6. B, Summary of the effects of sgp130. The baseline value before injection of sgp130 was set at 100%. The average change in the responses 54 minutes after injection when 100 ng sgp130 was applied alone, when it was coadministered with 100 ng IL-6 or 20 ng Super–IL-6 (SIL-6), and when it was applied after an enhanced response rate had been established by application of 100 ng Super–IL-6 is shown. Values are the mean ± SD; n = number of fibers. ∗ = P < 0.05 versus baseline, by Wilcoxon's matched pairs signed rank test. See Figure 2 for other definitions.

We also tested whether sgp130 reverses the established increase in the responses due to Super–IL-6. In these experiments, 100 ng Super–IL-6 significantly increased the responses of C fibers within 72 minutes after injection (innocuous torque from mean ± SD 79.6 ± 2.5 to 115.7 ± 5.5 APs/15 seconds, noxious torque from 142.9 ± 8.8 to 192.7 ± 6.1 APs/15 seconds, respectively), but injection of 100 ng sgp130 60 minutes after administration of Super–IL-6 did not reverse the responses (Figure 4, last column). Rather, the responses further increased and reached 151.4 ± 8.8 APs/15 seconds (innocuous torque) and 244.7 ± 6.4 APs/15 seconds (noxious torque) within 72 minutes after the injection of sgp130.

Effects of sgp130 and sIL-6R on joint afferents from the inflamed knee joint.

In 16 rats with inflammation of the right knee joint, APs were recorded from 29 C fibers that responded to both innocuous and noxious rotations (mean ± SD baseline innocuous torque 118.4 ± 340.6 APs/15 seconds; baseline noxious torque 261.3 ± 455 APs/15 seconds). Thirteen of these fibers showed some spontaneous activity, ranging between 0.3 Hz and 2 Hz.

As can be seen in Figure 5A, the response rate to noxious torque in inflamed knees was not influenced by intraarticular injection of 0.1 ml Tyrode's solution. Injection of sgp130 7 hours after the induction of inflammation did not alter the responses to noxious outward rotation either. Responses were insignificantly reduced by 11.5% within 126 minutes after sgp130 injection (baseline 405.6 ± 12.2 APs/15 seconds). Application of 125 ng sIL-6R into the knee joint caused a slow but continuous increase in responses to noxious torque, from a baseline level of 172.3 ± 2.9 APs/15 seconds to a maximum of 219.6 ± 2.5 APs/15 seconds at 72 minutes after injection. Then, the responses to noxious rotation slightly decreased but remained above baseline (Figure 5B). The responses to innocuous torque first declined to 63% of baseline, but then started to increase and reached a maximum value of 149.1% above baseline at 108 minutes after injection.

Figure 5.

Effects of Tyrode's solution, soluble gp130 (sgp130), and soluble IL-6 receptor (sIL-6R) on responses of C fibers from the acutely inflamed knee joint. A, Change from baseline in responses to innocuous and noxious outward rotation of the acutely inflamed knee after injection of 125 ng sIL-6R. Arrow indicates the time of injection of the compound. B, Average change in the responses of C fibers to noxious outward torque 72 minutes after injection of Tyrode's solution, 100 ng sgp130, or 125 ng sIL-6R. The baseline value before injection was set at 100%. Values are the mean ± SD; n = number of fibers. ∗ = P < 0.05 versus baseline, by Wilcoxon's matched pairs signed rank test. See Figure 2 for other definitions.


In the present study, we have shown for the first time that the cytokine IL-6 influences responses of unmyelinated joint afferents to mechanical stimulation of the knee joint in vivo. After local application of IL-6 and IL-6 plus sIL-6R (Super–IL-6) to the normal joint, responses of C fibers increased slowly, and this increase could be prevented by coapplication of sgp130. However, application of sgp130 after establishment of IL-6– or Super–IL-6–induced sensitization did not reverse enhanced responses. In the inflamed knee, local application of sIL-6R caused an increase in responses to mechanical stimuli, but local application of sgp130 did not significantly reduce the response rate. The increase in the responses after application of sIL-6R under inflammatory conditions suggests that IL-6 was already present due to the inflammatory process. Together, the data show a long-lasting IL-6–mediated mechanical sensitization of C fibers but not of Aδ fibers. Thus, IL-6 and its receptor signaling may be important factors in the generation of mechanical hypersensitivity under arthritic conditions.

The complexity of arthritis pain has recently been emphasized (26). Joint afferents play a key role in arthritis pain because they are the interface between the disease process and the nervous system that produces the conscious pain response. Joints of patients with RA show an even higher density of substance P–positive sensory nerve fibers (27). Recordings from afferent fibers allow us to investigate the mechanisms of mechanical sensitization, a key process in the generation of joint pain (for review, see ref.15), and they are particularly suitable for monitoring the peripheral effect of antinociceptive compounds (28–30). We therefore chose this experimental approach to study the role of IL-6 signaling in mechanical sensitization.

The IL-6 effect was specific because intraarticular injections of the same amount of Tyrode's solution or of the vehicles did not significantly change responses. The effect was dose-dependent, and it was blocked by coadministration of sgp130, which prevents IL-6R activation (13). Interestingly, higher doses of IL-6 inhibited mechanical sensitization of C fibers. Principally, Super–IL-6 had a more pronounced effect on C fibers than did IL-6 alone, but similar to IL-6 alone, it had no effect on the mechanosensitivity of Aδ fibers. Injection of sIL-6R alone into the normal joint had no effect on the mechanosensitivity of C fibers.

The present data do not necessarily imply that the effect of IL-6 was only produced by direct action of IL-6 at the nerve terminal itself. The slight temporary increase in response to mechanical stimulation after the injection of 20 ng IL-6 (see Figure 2C) could reflect an effect of the injected volume as well as a slight immediate effect of IL-6. However, experimental work on DRG neurons in culture has provided strong evidence of the direct effects of IL-6 on primary afferent neurons. First, the vast majority of cell bodies of primary afferent fibers express gp130 (18, 20, 31). Second, bath application of 20 ng/ml Super–IL-6 significantly increased heat-activated ionic currents at a short latency in isolated DRG neurons (18). Furthermore, application of IL-6 for 2 days to cultured DRG cells caused a dose-dependent increase in neurokinin 1 receptor, which is activated by substance P (20). Both the short-term effect on heat-activated currents and the long-term effect on the expression of neurokinin 1 receptors were prevented by compounds that interfered with intracellular signaling of IL-6, and in both cases, the IL-6 effect was not prevented by cyclooxygenase inhibitors (18, 20). Unfortunately, it is not possible to test the effect of IL-6 on the mechanosensitivity of isolated DRG neurons, in particular on responses to innocuous and noxious stimuli.

It is noteworthy that sensitization of joint afferents after a single application of IL-6 or Super–IL-6 in the dose range of 20–100 ng was long-lasting in most experiments, with no evidence of a tachyphylactic effect within the recording time. Thus, IL-6 may be a key mediator that causes a long-lasting effect on mechanosensitivity. This is of considerable interest because inflammation in the joint is usually associated with a persistent increase in mechanosensitivity of Aδ fibers and C fibers (14). Interestingly, the application of sgp130 after establishment of mechanical hyperexcitability did not reverse the increased mechanosensitivity, although pretreatment with sgp130 prevented the generation of mechanosensitivity induced by IL-6 or Super–IL-6. Therefore, an interesting possibility is that IL-6 is mainly required for the initial sensitizing process, and that the maintenance of sensitization does not require persistent action of IL-6 on the neurons.

Curiously, however, a dose of 200 ng did not evoke hypersensitivity of unmyelinated joint afferents, and when this dose was administered after sensitization by 20 or 100 ng, hypersensitivity to mechanical stimuli was decreased. Thus, the pattern of effect of IL-6 seems to be bell-shaped. A possibility is that different cellular pathways come into play when different doses of IL-6 are used. The increase in heat-evoked responses by IL-6R activation could be blocked by the inhibition of protein kinase C (18). In cortex slices, IL-6 reduced the glutamate release, and this IL-6 effect involved activation of the STAT-3 pathway together with inhibition of the MAPK/ERK pathway (32). The lower doses of IL-6 are closer to the range of IL-6 concentrations in the synovial fluid of patients with arthropathies (33), and therefore, the inhibitory IL-6 effect may not be relevant under pathophysiologic conditions. A 1-to-1 comparison is difficult because the net effect of IL-6 probably depends on a number of other factors, such as the presence of other mediators, including sIL-6R.

Another interesting finding is that none of the compounds changed the responses of Aδ fibers to mechanical stimulation, although articular Aδ fibers are sensitized to mechanical stimuli during peripheral inflammation (22). Because the vast majority of DRG neurons express gp130, it is unlikely that Aδ fibers cannot be activated by IL-6 or IL-6/sIL-6R complexes. Possibly, mechanical sensitization by IL-6 requires a cofactor that is only expressed in C fibers, or IL-6 acts on a site of C fibers that is covered by myelin in Aδ fibers, e.g., the distal axon, where APs are generated.

While the application of sIL-6R to normal joints had no effect on mechanosensitivity, the injection of sIL-6R into the acutely inflamed joint enhanced the responses of the fibers to mechanical stimulation. The most likely explanation for this difference is that in the inflamed joint, IL-6 is present, and thus, complexes are formed that act on the neurons. As in the normal knee, the application of sgp130 alone to the inflamed knee did not alter the responses to rotatory stimulation. This finding supports our conclusion from experiments with normal knee joints that sgp130 does not reverse established mechanical sensitization, at least not within a recording period of ∼1 hour. Whether sgp130 reduces mechanical sensitization in the long term needs to be explored. In a model of RA in mice, gp130 of the RA antigenic peptide–bearing soluble form reduced inflammatory symptoms in the long term, suggesting that sgp130 is able to stop the STAT-3 pathway in inflammatory diseases (33).

The present data suggest an important role of IL-6 in the mechanical sensitization of C fibers in the joint. However, they also show the problem of reversing mechanical sensitization by interfering with IL-6 signaling. Further research should elucidate the cellular mechanisms of mechanical sensitization by IL-6 and investigate possibilities to reverse IL-6–induced effects on neurons.


Dr. Brenn 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 design. Dr. Schaible.

Acquisition of data. Drs. Brenn and Richter.

Analysis and interpretation of data. Drs. Brenn, Richter, and Schaible.

Manuscript preparation. Drs. Richter and Schaible.

Statistical analysis. Drs. Brenn and Richter.


We thank Mrs. Konstanze Ernst for excellent technical assistance.