Background
Synovitis is a hallmark of rheumatoid arthritis (RA) and the majority of osteoarthritis (OA) patients. Both diseases have a huge socio-economical impact on society [
1]. Pain, the major clinical symptom for the RA and OA patients, is of unknown origin and correlates poorly with radiographic disease scores, and very often, treatment responses are low. Therefore, there is an urgent need to clarify the mechanisms behind pain perception in rheumatic diseases. It is thought that part of the pain perception in RA and OA is related to mediators of synovitis [
2‐
4].
In several studies, we demonstrated that S100A8/9 are involved in inflammation and subsequent structural joint pathology, in humans and experimental RA and OA. S100A8 and S100A9 are proteins that form heterodimers and are abundantly expressed and released by neutrophils, monocytes, and activated macrophages. S100A8/9 heterodimers have been shown to signal via Toll-like receptor 4 (TLR4) and are important regulators of the innate inflammatory response [
5]. We demonstrated significantly higher levels of S100A8/9 levels in early OA patients that show progression of disease regarding Kellgren and Lawrence score [
6], and pathology in experimental RA and OA strongly depends on S100A8/9 [
6,
7].
Since S100A8/9 can mediate synovitis in both RA and OA models, it is difficult to investigate whether S100A8/9 plays a direct role in pain perception. TLR4 has been implicated in pain perception, is expressed by afferent sensory neurons, and may play a role in local peripheral nociception [
8,
9]. A role for S100A8 was demonstrated in the activation of these neurons, leading to the production of MCP-1 and the local influx of monocytes that produce S100A8 in the dorsal root ganglion (DRG) [
10]. Therefore, S100A8/9 may not only be involved in local nociception but also in sensitization by monocyte influx or local MCP-1 production in the DRG. Sensitization leads to increased pain perception and is thought to be involved in chronic pain that occurs in OA [
11].
Activation of the peripheral C- and Aδ-fibers leads to pain sensation [
12]. Upon excitation of the peripheral pain fiber, the neuron, with the cell body residing in the DRG, is activated, and this may result in the upregulation of several marker genes, among which substance P, CGRP, NPY, galanin, NAV1.7, P2RX3, α2δ1, ATF3, and GAP43, depending on the cause and nature of the pain [
13].
The hypothesis of this study is that S100A8/9 induces pain as by direct activation of afferent nerve endings and by sensitization of the peripheral nervous system locally in the DRG, where it is produced by infiltrating monocytes.
Synovitis is studied in streptococcal cell wall (SCW) arthritis, which is based on local activation of the synovium by intra-articular deposition of SCW. SCW bind to TLR2 rather than to TLR4 [
14], which makes this a good candidate to study the involvement of S100A8/9, a known TLR4 ligand, in pain perception.
Several methods to measure different aspects of pain have been described [
15]. In this study, we used gait analysis and the incapacitance tester to study pain upon static and dynamic loading and von Frey testing to determine allodynia, as a measure for pain sensitization.
Discussion
Synovitis and bone marrow lesions have been proposed to determine OA pain [
23]. In RA, it is well accepted that synovitis is an important source of pain, but also here, mechanisms are poorly understood [
24]. An intriguing question is whether key mediators can be identified in the joint that are particularly involved in the generation and maintenance of pain caused by synovitis. S100A8/A9 is a potent mediator of synovitis and joint destruction during mouse and human OA and these alarmins are produced within the joint throughout the course of the disease. In the present study, we set out to determine if and to what extent the S100A8/9 heterodimer is involved in pain that is mediated by synovitis. We clearly demonstrate that the heterodimer S100A8/9 is indeed involved in pain behavior in an experimental model for synovitis. S100A8/9 seem particularly important in nociception, since we found a clear normalization of joint loading and gait in
S100a9−/−.
SCW synovitis is based on the activation of TLR2, rather than TLR4, which would disturb the measurement of isolated effects of the TLR4 ligand S100A8/9 [
14]. Interestingly, in contrast to other models [
6,
7,
25], the inflammation in the SCW model was not mediated by S100A8/A9. This allowed us to study the direct effect of S100A8/A9 in pain, independent of the differences in inflammation. The SCW arthritis model is a generalized model for acute synovitis. The infiltrate is characterized by cells of the innate immune system, like monocytes, macrophages, and PMN, rather than lymphocytes, cell types relevant for both RA and OA
. Inflammation is an important source of pain in many processes, including arthritic diseases [
10,
26]. It has been recently shown that a 32-mer aggrecan fragment is a potent activator of TLR2 and is involved in inflammation and pain in models for OA [
27,
28].
Much is still unknown about the involvement of the DAMP S100A8/9, in pain perception. We studied S100A8/9 involvement in 2 different aspects of pain: (1) acute inflammatory or nociceptive pain, caused by direct excitation by S100A8/9 of the relevant afferent nerve fibers, likely via TLR4 or RAGE, directly resulting in pain [
8,
29]; (2) sensitization, induced by binding of TLR4 or RAGE, and resulting in increased excitability of the afferent nerve fiber, via cell influx into the DRG or via regulation of nociceptors or ion channels.
By comparing the differential effects of S100A8/9 on the specific parameters, we deducted the involvement of S100A8/9 in these different aspects of pain. The first indication that pain caused by synovitis is mediated by S100A8/A9 came from our finding that static weight bearing, determined with the incapacitance tester [
30,
31], was unchanged in
S100a9−/− but not in WT mice. Gait was analyzed to measure the dynamic loading of the paws and “limping” [
32]. These parameters were quantified as (1) the mean stand phase of all four paws and (2) the TDS. Both parameters have been described before to reflect pain behavior [
33,
34]. We found an increase in stand phase in the unaffected paws in WT mice, rather than a decrease of the stand phase of the arthritic paw. It is assumed that the loading of unaffected limbs is increased to spare the affected limb. This effect has been described before in a rat model for arthritis [
35]. TDS of the right hind paw is considered an indicator of pain, and changes represent “limping” behavior. This clear demonstration of lack of pain behavior when S100A8/9 is absent affirms the importance of S100A8/9 in general pain behavior during synovitis.
To study the second aspect of pain, sensitization, we determined the pressure-pain threshold (tactile allodynia) using von Frey filaments [
36,
37]. Although clearly present in WT and
S100A9−/− mice, allodynia was equal between the two groups, excluding a role for S100A8/9 in sensitization in acute synovitis.
Together, these data strongly suggest a role for the TLR4 ligand S100A8/9 in acute nociception rather than modulation of pain sensitivity. This is in line with a recent finding, in which TLR4 deficiency did not alter sensitization [
10]. Although S100A8 stimulation caused excitation of afferent neurons, when a model for OA, destabilization of the medial meniscus (DMM), was induced in the knee joints of
Tlr4−/− mice, no effect on sensitization was demonstrated. In contrast to these findings, a study in which LPS was injected in the paw of WT and
Tlr4−/− mice did demonstrate a role for TLR4 in this process [
38]. This indicates that endogenous TLR4 ligands as formed in (osteo) arthritis models may play a redundant role. S100A8/9 can bind to RAGE, but most literature suggests a role for RAGE in neuropathic pain and central sensitization, for which the current model is not suitable [
39]. We therefore cannot claim a role for this receptor. Future experiments could include i.a. injection of S100A8 or S100A9 homodimers to test the hypothesis that these compounds induce an acute pain response, and using specific knockout mice, this could shed light on the receptors that are involved in this. However, it would be difficult to determine whether this would be a direct effect on pain behavior or indirect through the mediation of inflammation.
To shed more light on the cellular and molecular processes that may underlie the differences in pain behavior, we studied the DRG, where the cell bodies of the Aδ- and C-fibers reside [
40]. Ipsilateral DRG from L3–L5 were pooled, since the neuronal bodies of the afferents from the knee joint reside at these levels, using retrograde fluorogold staining [
41]. We tested the expression of markers for neuronal inflammation, activation, and neuropathy. No induction of markers for inflammation was found, which is in contrast to previous studies which were performed in a model for OA, the DMM [
10,
42]. Also in contrast to these studies, we did not observe an influx of monocytes in DRG, likely explained by the lack of upregulation of MCP-1 by the DRG. The lack of DRG inflammation in synovitis during acute SCW-induced arthritis refutes our hypothesis, derived from the DMM-data, that influx of cells or production of MCP-1 by the DRG would be responsible for the sensitization in both WT and
S100a9−/− mice [
10,
43,
44]. An obvious difference between both models is the acute nature of the SCW arthritis compared to the DMM, with long-lasting mild synovitis. Whether the DRG inflammation during DMM is S100A8/A9-dependent is currently under investigation. Possibly in the chronic variant of the SCW model, DRG inflammation is comparable to DMM, which will be subject to future experiments [
45]. The relevance of an acute flare of inflammation during OA for joint pathology was demonstrated elegantly in a study in rats where an acute inflammation was induced during an induced instability model for OA by intra-articular carrageenan injection. This indeed caused acute pain and also led to more pronounced end-stage pathology [
46].
Stimulation of TLR4 by LPS leads to the expression of IL-1 and TNFα in the DRG [
47]. In addition, TLR4 may play a role in the conversion of acute to chronic pain, which was demonstrated in a model for chronic arthritis, where TLR4 was involved in sustaining sensitization after the inflammation waned [
48,
49].
Cytokines that are produced locally in the joint during arthritis, like IL-1, TNFα, IL-6, and IL-17, have been shown to modulate neurons [
50]. However, in osteoarthritis, cytokine levels are considerably lower and therefore may have less impact. In contrast, levels of S100A8/9 are high [
6], and therefore, S100A8/9 may play a particularly relevant role in OA pain. In vitro stimulation of DRG with S100A8 resulted both in the excitation of the neurons and the production of inflammatory mediators [
10]. In a study in rats, macrophages were stimulated with LPS in co-culture with DRG neurons, and this led to apoptosis of neurons, suggesting a role for TLR4 in neuropathy [
51].
Male mice showed TLR4-mediated hyperalgesia when challenged with LPS, whereas the role of TLR4 in allodynia was comparable between male and female mice [
52]. This, and other studies, indicate sex as an important factor in pain. In the present study, experiments were performed in males only, which is a limitation of our approach.
The neuronal markers that were differentially expressed in the DRG were NAV1.7, ATF3, and GAP43, which were increased in WT DRG and not
S100a9−/−. These markers appear related to both nerve injury and inflammatory pain since they are also expressed in collagen antibody-induced arthritis and signify activation of the Aδ- and C-afferents [
13]. The fact that they are differentially expressed in
S100a9−/− is in line with the demonstrated difference in pain behavior. They are related to nerve injury and this suggests that local joint inflammation during SCW causes damage to peripheral afferent fibers [
53]. However, we did not find clear signs of nerve injury at this early time point, like prolonged changes in pain behavior in this acute synovitis. Unfortunately, due to the experimental setup, we were not able to further study local processes. Possibly, the neuronal markers for tissue inflammation, CGRP, substance P, and α2δ1, but also TrkA, would have been differentially expressed [
51,
54]. In a study in which the effect of local macrophage activation in the DRG was studied, activated macrophages induced CGRP production by afferent neurons. Our lack of monocyte influx may at least explain the lack of CGRP regulation [
51]. Nevertheless, we were able to confirm the effects of S100A8/9 on pain behavior on a molecular level.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.