The alarmins S100A8 and S100A9 mediate acute pain in experimental synovitis

Synovial activation was elicited in male C57BL/6 mice and in S100a9^−/− mice backcrossed to the C57BL/6 background for more than 12 generations. Myeloid cells of S100a9^−/− mice also lack S100A8 at the protein level [16]. Mice were used between the age of 14 and 16 weeks. All animal studies were conducted according to the Dutch law and approved by the Dutch Central Animal Experimentation Committee (project 2015-0014). A total of 80 mice were used, 6–8 per experimental group.

SCW were produced as described previously [17]. After randomization, unilateral arthritis was induced by intra-articular injection of 25 μg SCW (rhamnose content) in 6 μL PBS into the right knee joint of 20 WT mice and 10 S100a9^−/− mice. As a sham control, saline was injected into the right knee joint of 20 WT and 10 S100a9^−/− mice. At day 1 and day 7 after induction, serum was collected and S100A8/9 levels were determined using an ELISA that was designed in-house [18].

To quantify joint swelling as a measure for inflammation, the uptake of ^99mTc-pertechnetate was measured [19]. The signal was expressed as a ratio relative to the naive contralateral knee joint. Briefly, mice were injected subcutaneously with 12 μCi ^99mTc-pertechnetate and sedated. After 30 min, the amount of radioactivity was assessed by measuring the gamma radiation with the knee in a fixed position, using a collimated NaI scintillation crystal. Swelling was measured at baseline and at days 1 and 7 after injection.

On days 1 and 7 after induction of SCW arthritis, the knee joints were isolated and fixed in 4% formaldehyde, decalcified in formic acid, and subsequently dehydrated and embedded in paraffin. From a part of the murine joints, synovium was isolated for the preparation of washouts and RNA isolation. Briefly, the synovium was isolated from the murine knee joints using a standardized method comprising removal of the quadriceps and section of the patella ligament. Tissue was put in RPMI (Gibco, Invitrogen, Carlsbad, CA, USA), enriched with 0.1% BSA (Sigma–Aldrich) for 1 h at room temperature. Prior to incubation, two 3-mm biopsies were taken for RNA isolation. DRG were isolated at the L3–5 levels of the spinal column. Briefly, the spines were isolated and cut in half transversely, and the ipsilateral L3–5 were removed. DRG were fixed in 4% formaldehyde, subsequently dehydrated, and embedded in paraffin. Paraffin sections were cut at 7 μM or 5 μM and mounted on coated slides. Hematoxylin/eosin (H&E) staining was performed to study the inflammatory cells and tissue morphology. Of the knee joints, inflammation scored using an arbitrary scoring system. The severity was determined using an arbitrary score (0–3), performed blindly by two observer: 0, no influx; 1, mild cellularity; 2, higher cellularity; and 3, very high cellularity. Immunohistochemical (IHC) staining was performed on S100A8 and S100A9 in WT mice. As primary antibodies, anti S100A8 and S100A9 were used that were prepared as described previously [5]. On DRG, IHC staining was performed for F4/80, to detect monocytes/macrophages, and for NAV1.7 (Alomone Labs, asc-008). For all IHC detections, isotype-matched IgG was used as a negative control. For S100A8, S100A9, and NAV1.7, IHC staining was scored in a semi-quantitative manner by assessing the diaminobenzidine (DAB) intensity using ImageJ (NIH, Bethesda). Since DAB intensity is not linearly related to expression, the staining is qualified as semi-quantitative. Briefly, photomicrographs were corrected for background, area of interest was selected, and the mean intensity of the area was assessed using the “H DAB” plugin after color deconvolution. Quantification of F4/80 staining in the DRG was done by counting the amount of positive cells per field of view after deconvolution.

In order to reliably measure pain in animal models, using more than one way to measure pain behavior in experimental models is highly recommendable [15]. None of the available pain tests for rodents is specific only for pain, since in all cases, behavior is studied, and behavior can easily be influenced by other parameters. Therefore, in our experiments, extreme care was taken to include a proper adjustment period for acclimatization, sham controls were used, and baseline measurements were taken at 3 alternate days before the start of the experiments. By taking these precautions, and by combining several methods, pain measurements can be reliably performed, and several groups have shown that these methods reflect pain behavior since the use of analgesics normalizes behavior [20]. All measurements were performed by the same observers (AB, EBD, and EG) who were blinded for treatment and phenotype. Mice were randomized prior to the start of the experiments. The methods we used were (a) incapacitance testing, (b) gait analysis, and (c) von Frey testing.

Gene expression was determined using quantitative real-time PCR (qRT-PCR). DRG for RNA isolation were homogenized in RLT buffer using the MagNA Lyser Instrument (Roche). Total RNA was isolated using the RNeasy MinElute kit (Qiagen) with a Proteinase K step according to the manufacturer’s protocol. The RNA concentration was determined using a Nanodrop spectrophotometer and subsequently reverse transcribed into cDNA. qRT-PCR was performed using specific primers (Table 1) and the SYBR Green Master Mix in the Applied Biosystems StepOnePlus real-time PCR system (Applied Biosystems). Reactions were presented as minus delta threshold cycle Ct (−ΔCt) values, calculated by correcting the negative threshold cycle (−Ct) of the gene of interest to the −Ct of the reference gene GAPDH.

Statistical analyses were performed using Graphpad Prism version 5.03. Differences between the groups were tested using a one-way or two-way analysis of variance (ANOVA). Differences in histology were tested with a non-parametric Mann–Whitney U test. p values lower than 0.05 were considered significant. Results are expressed as mean values ± standard deviation (SD).

First, we determined whether in the acute synovitis model of SCW induced arthritis in C57Bl/6 mice, S100A8 or S100A9 genes and proteins were expressed. After a single i.a. SCW injection, the expression of both S100a8 and S100a9 mRNA was significantly upregulated locally in the synovium. The peak expression was found on day 1 (resp. 29- (ddCt 5.4 ± 2.1) and 34-fold increase (ddCt 5.8 ± 1.8) compared to sham), and levels decreased at day 2- to 8-fold increase for both (ddCt resp. 3.0 ± 1.6 and 3.0 ± 1.2) compared to sham and were back to baseline at day 7 (ddCt resp. 0.8 ± 1.1 and 0.9 ± 1.5). This upregulation in S100a8 and S100a9 mRNA was reflected by increased protein levels in serum and synovial washouts (Fig. 1a, b).

Using immunohistochemistry we demonstrated that levels of both S100A8 and –A9 peaked at day 1, decreased slightly at day 2 and were low, but still above baseline at day 7 after SCW injection (Fig. 1c, d).

To study the involvement of S100A8/9 in synovitis, we first determined swelling of the knee joint at day 1 and at day 7 after SCW injection. Swelling increased, as determined by measuring the ^99mTc-pertechnetate uptake R/L ratio, 1 day after injection of SCW in WT mice (Fig. 2a). At day 7, joint swelling was down to baseline levels. In S100a9^−/− mice, effectively lacking both S100A9 and S100A8, swelling was induced to comparable levels as in WT mice. Swelling at day 7 was not different from baseline levels. To determine whether S100A8/9 deficiency leads to changes in cell influx, we scored local inflammation using histology (Fig. 2b–d). Inflammation was high at day 1 and was decreased but still present at day 7 after induction. Interestingly, however, no differences in cell numbers and cell types could be observed between WT and S100a9^−/− mice (Fig. 2d). Mediators for inflammation, except TNFα and IL-1, were increased in synovial washouts during SCW arthritis, but none of the inflammatory mediators (KC, MCP-1, IL-6) were significantly different between WT and S100a9^−/− (Fig. 2e).

The lack of effect of S100A8/9 on inflammation presented us with the perfect conditions to determine the role of S100A8/9 in pain irrespective of inflammation. We first measured pain behavior defined by static loading of the right and left paws using an incapacitance tester. At day 1 after induction, the percentage of total weight on the right hind leg decreased from approximately 50% at baseline to just below 30%, p < .001 (Fig. 3a). In contrast, mice lacking S100A9 showed no shift. Seven days after induction, incapacitance normalized almost completely to 44% in the WT and 48% in the S100a9^−/− and was no longer significantly different from saline-injected controls.

Next, animals were allowed to walk freely, and gait was monitored at day 1 after injection for 2 parameters, the stand phase of each paw, and the terminal dual stance (TDS). Remarkably, the stand phase of the right hind leg was unchanged 1 day after SCW injection, both in WT and S100a9^−/− mice. However, in WT mice, the stand phase in the remaining legs, left hind and right and left front legs, increased significantly (respectively p < .001, p < .05, and p < .001). This increase was not observed in S100a9^−/− mice (Fig. 3b). Next, the TDS was determined from the Catwalk data. This parameter can be described as “limping.” The TDS significantly increased in wild-type mice after injection of SCW, resulting in “limping” behavior (Fig. 3c). In contrast, no change in TDS was observed in the S100a9^−/− mice with SCW arthritis, underlining the role of S100A8/9 in pain perception.

To determine whether S100A8/9 are contributing to pain sensitization, we measured hind paw mechanical allodynia using von Frey filaments. At day 1 after SCW injection, allodynia could be clearly demonstrated (p < .001) in WT mice (Fig. 4a). Saline-injected animals did not differ from baseline and had a 50% paw withdrawal threshold of approximately 3 g, whereas SCW-injected mice showed a dip of approximately 1 g at day 1 after injection, which returned to baseline levels at day 7. In contrast to the parameters for general pain perception, allodynia in S100a9^−/− was present (p < .001) and did not differ from WT (Fig. 4b).

To study the possible mechanisms that underlie differences in pain perception between WT and S100a9^−/−, we determined whether the inflammatory response in the relevant DRGs (L3-L5) was different between WT and S100a9^−/−. In the joint, S100A8/9 in WT may bind to the afferent neuronal ending, thus causing activation of the neuronal bodies in the DRG. We did not find increased expression of MCP-1 in the DRG 1 or 7 days after induction of synovitis (Fig. 5a). Also, other mediators that may signify cell influx or inflammation (S100A8, IL-1β, NGF, and TNFα, the latter not shown) were unchanged (Fig. 5b–d). Histological examination of the DRG proved a lack of cell influx upon induction of synovitis. This was confirmed by immunohistochemical detection of F4/80, which was not different in DRG of SCW versus saline-injected mice (Fig. 5e, f).

Now that we could not demonstrate regulation of inflammatory response genes in the DRG, we detected genes that are involved in neuron function in the DRG and may serve as markers for activation or damage: Substance P, CGRP, NPY, galanin, NAV1.7, P2RX3, α2δ1, ATF3, and GAP43. First, we determined levels of mRNA expression of these genes in WT mice injected i.a. with SCW and compared this to WT mice that received saline (Table 2). One day after induction of SCW arthritis, 3 genes of this panel were upregulated: NAV1.7 (1.7-fold), ATF3 (1.8-fold), and GAP43 (1.9-fold) (Fig. 6a–c). Subsequently, regulation of these genes in DRG of S100a9^−/− mice after induction of SCW arthritis was absent. In contrast, even a downregulation of ATF3 compared to WT mice was observed (2.1-fold). At day 7 after induction of SCW, none of the genes were differentially expressed, in WT nor in S100a9^−/−. The mRNA findings of NAV1.7 were confirmed by immunohistochemistry on DRG, where enhanced staining was found in WT compared to S100a9^−/−, although protein levels were not significantly different between WT and S100a9^−/− until day 7 (Fig. 6d, e).