Introduction
Rheumatoid arthritis (RA) is an autoimmune disease and characterized with chronic joint inflammation. Chronic joint inflammation causes cartilage damage and ultimately total joint destruction but is also associated with ongoing pain and increased pain [
1‐
3]. The magnitude of pain may not necessarily be associated with the severity of the underlying disease, and pain may persist even when disease exacerbations have apparently lessened. In RA, the neurophysiological mechanisms underlying pain remain unclear. Experimental animal models of inflammatory arthritis suggest that changes in neuronal sensitivity at both peripheral and central levels may be important [
4,
5]. The causes of RA-associated pain could also differ in early and late disease stages. The acute phase of pain could be associated with acute joint inflammation, but the chronic phase could be linked to inflammatory components of neuron–immune interactions and non-inflammatory components such as neuron–glia interactions or central mechanisms [
6,
7].
In RA, resident macrophages become activated in synovial tissues; they along with infiltrated macrophages secrete pro-inflammatory cytokines (e.g., tumor necrosis factor α [TNFα], interleukin 6 [IL-6]), mediators, and enzymes to regulate synovial inflammation and joint destruction [
8]. Activated macrophages also produce anti-inflammatory cytokines (e.g., IL-10) to promote the resolution of inflammation and tissue repair, thus ameliorating the disease. Two different macrophage phenotypes, M1 (classically activated) and M2 (alternatively activated), are responsible for producing pro- and anti-inflammatory cytokines, respectively [
9]. The imbalance between pro- and anti-inflammatory cytokines (M1/M2) could be a key mechanism of rheumatic disease progression. RA patients display a more M1 macrophage profile than do people with spondyloarthritis such as psoriatic arthritis [
10] or osteoarthritis [
11]. An acute hypoxia environment favors M2 macrophage polarization, but chronic hypoxia triggers M1 polarization [
12].
In a monoarthritis or collagenase arthritis model, arthritic rats showed satellite glial cell (SGC) proliferation and activation in dorsal root ganglia (DRG) [
13,
14]. Intrathecal injection of a glial inhibitor (fluorocitrate) blocked collagenase-induced nociception [
14], which suggests that SGCs play some roles in arthritic pain. IL-17A may act on glial cells to sensitize neuron function [
15]. Genome-wide association studies demonstrated that genetic variants in the T cell death-associated gene 8 (TDAG8) locus are associated with spondyloarthritis [
16], and T-helper 17 (Th17) cells in spondyloarthritis patients show high expression of TDAG8 gene [
17]. TDAG8 gene-deficient mice show reduced number of Th17 cells and secretion of IL-17A [
18]. Consistent with these data, RA disease severity and RA-evoked pain were attenuated in TDAG8 gene-deficient mice in the RA mouse model [
19]. However, how TDAG8 regulates RA and RA-evoked pain in the early and late pain phases remains unclear.
In this study, we used a previously established arthritis mouse model [
19] in TDAG8 gene-deficient mice to investigate whether TDAG8-modulated chronic pain and disease severity is related to immune cells or glial cells. TDAG8 gene deficiency reduced SGC number, and SGC inhibition attenuated the chronic phase of RA pain, which suggests that TDAG8 deficiency relieved the late phase of RA pain by regulating SGCs in part. Moreover, reduced M1 macrophage number but not synovial macrophage number in TDAG8–deficient mice may explain the less attenuation of acute-phase RA pain. Consistent with the results of TDAG8 deletion, long-term suppression of TDAG8 gene expression and function by using previously developed salicylanilide derivatives [
20,
21], CCL-2d, and LCC-09, reduced RA pain by modulating the number of SGCs and pro-inflammatory macrophages. Accordingly, TDAG8 gene deficiency relieved RA disease severity and pain by reducing SGC number and pro-inflammatory macrophage number.
Materials and methods
Agents
Complete Freund’s adjuvant (CFA) and DL-fluorocitric acid (FC) barium salt were from Sigma-Aldrich. Salicylanilide derivatives CCL-2d (3-(4-Chloro-2-fluorophenyl)-7-methoxy-2H-benzo[e][1,3]-oxazine-2,4(3H)-dione) and LCC-09 (N-(3-cyanophenyl)-20,40-difluoro-4-hydroxy-[1,10-biphenyl]-3-carboxamide) were synthesized as described [
20,
21]. Tofacitinib (commercial RA drug) [
22] was from Selleckchem. For animal experiments, all drugs or compounds were first solved in dimethylsulfoxide and then diluted in saline before injection.
Arthritis mouse model
Male or female ICR mice (8–12 weeks old) were purchased from BioLASCO Taiwan (Taipei) and housed 3–4 per cage under a 12-h light/dark cycle (lights on at 7:00 am) with food and water ad libitum in a temperature- and humidity-controlled environment at National Central University. TDAG8
−/− and TDAG8
+/+ mice on a B6 background were generated as described [
23]. The genotyping primer sequences for TDAG8
−/− were 5′-gaaccattagtttggctcatgtgactg/5′-cttgtgtcatgcacaaagtagatgtcc and for TDAG8
+/+, 5′-cgaactctagctggcttttatccaataat/5′-gaaccattagtttggctcatgtgactg. Care and use of mice conformed to the Guide for the Use of Laboratory Animals (US National Research Council) and the experimental procedures were approved by the local animal use committee (IACUC, National Central University, Taiwan).
All behavioral testing was performed between 9:00 am and 5:00 pm. Efforts were made to minimize the number of animals used and their suffering. Arthritis was induced as described [
19]. Briefly, TDAG8
+/+, TDAG8
−/− or wild-type (WT ICR) mice were injected with 5 μg CFA in the right ankle joint once a week for 4 weeks (CFA-ctrl). Compounds (CCL-2d [360, 3600 μg/kg], LCC-09 [39, 390 μg/kg], was intraperitoneally injected once at 4w after CFA injection, followed by mechanical tests. For long-term treatment, compounds (CCL-2d [360, 3600 μg/kg], LCC-09 [39, 390 μg/kg], 3 mg/kg tofacitinib) or saline (vehicle) was intraperitoneally or orally (using oral feeding needle, ST-F173 ψ0.9 mm × L 70 mm) administered weekly for 9 consecutive weeks after CFA injection. Some experiments were only intraperitoneally injected once with CCL-2d and LCC-09, followed by the rota-rod tests (by the Taiwan Mouse clinic, Taiwan). FC (0.01 mM) was intrathecally administered at week 3 after CFA injection. Behavioral tests for mechanical or thermal stimuli were performed before and after CFA injection. In some experiments, L4-6 DRG were excised for measuring gene expression, and joints were fixed for hematoxylin and eosin (H&E) staining or immunostaining.
Immunohistochemistry and immunostaining
The severity of the arthritis was scored from 0 to 5 as previously described [
19]. Each limb was graded and given a maximum possible score of 15; the maximum score for an animal was 60.
Histological staining was performed as previously described [
19]. Briefly, at 12 weeks after CFA injection, the tibiotarsal joint was excised, fixed, decalcified, embedded in paraffin and sectioned longitudinally at 10 μm with use of a microtome, then stained with HE (by the Taiwan Mouse Clinic, Taipei). Images were observed by light microscopy (Lecia, LAS EZ). Arthritic changes were scored on a scale of 0 to 5 as previously described [
19]. From each joint, 6 areas of 2 sections were used to provide a representative sample of the whole joint. Cell number counted is dependent on different type of macrophages, so cell density (cells/mm
2) is used to present the data. Mean scores were the average of all section scores for each animal. Three animals are used for each point.
Some joint sections were stained with the antibodies for CD80 (1:250, Biorbyt, UK), CD68 (1:100, Biorbyt, UK) or CD163 (1:100, Biorbyt, UK), followed by alkaline phosphatase-conjugated goat anti-rabbit IgG (1:5000, Jackson Immunoresearch). Signals were developed by nitro-blue tetrazolium chloride and 5-bromo-4-chloro-3′-Indolyphosphate p-toluidine (Millipore). Immunoreactivity-positive cells were counted.
Staining of SGCs was as previously described [
24]. Briefly, at 0, 1, 4, 8, and 12 weeks after CFA injection, DRG isolated from vehicle- or compound-treated mice were frozen in freezing solution and cut at 12 μm by using a cryostat (Leica microsystem 3510S, Bensheim, Germany). Sections were co-stained with the antibodies for peripherin (PERI, 1:500, Sigma) and glial fibrillary acidic protein (GFAP; 1:1000; Dako), followed by TRITC-conjugated goat–anti-mouse IgG antibody (1:250, Sigma) and FITC-conjugated goat-anti-rabbit-IgG antibody (1:250, Sigma), respectively. The digitized images were captured by using MetaVue. PERI-positive neurons surrounded by GFAP-positive SGCs in one-third or more of the PERI circumference were counted (PERI
GFAP+) and expressed as a percentage of total PERI-IR neurons (PERI
T) in the fields analyzed (PERI
GFAP+/PERI
T). Data for each treatment group were collected from 10 DRG sections. The distance between two sections at least 60 μm. Total 1000–2000 PERI-positive neurons were counted for each group.
Assessment of arthritic pain in mice
Pain behavioural tests were as described previously [
25]. Briefly, mice were tested for withdrawal thresholds to mechanical stimuli applied to the plantar aspect of the hindpaw. Mice were pre-trained for 2 h each day and for 3 days before the test. Before and after CFA injection, a series (ascending force) of von Frey fibers (Touch-Test, North Coast Medical, Morgan Hill, CA) was applied. A von Frey fiber was applied to each paw 5 times at 5-s intervals. The paw withdrawal threshold (PWT) was when paw withdrawal was observed in more than 3 of 5 applications.
For thermal nociceptive response to radiant heat applied to the plantar surface of the paw, before and after CFA injection, the plantar surface of mouse hindpaws was stimulated with a lit light bulb (30% intensity, 25 s for cut-off time). The latency to withdrawal of the paw (PWL) from radiant heat was measured. Measurements from three trials at 1-min intervals in each paw were averaged. The mean basal withdrawal latencies of 15~20 s was obtained in non-injected mice.
Measurement of cytokine levels in serum
Mice with or without compound treatments were sacrificed at 8 or 12 weeks. Blood was collected by cardiac puncture. For serum samples, blood was left to clot for 30min at 4 °C, followed by centrifugation for 20 min at 2000×g. Serum was aliquoted and stored at – 80 °C. TNF-α or IL-6 levels were measured with kits from R&D systems (Minneapolis, MN, USA).
Quantitative RT-PCR
Lumbar 1–5 (L1–5) DRG ipsilateral and contralateral to injected paws were removed at 0 or 12 weeks for RNA extraction, with DRG from 0 weeks as a control. RNA extraction was performed as described [
26]. Each DRG pool of each point contained at least 10 DRG from one side of 3 mice. RNA was extracted by using the RNeasy kit (Qiagen, Valencia, CA, USA). Each gene primer (100 nM), derived cDNA, and master mix (SYBR green I and AmpliTaq Gold DNA polymerase [Applied Biosystems, Foster City, CA, USA]) were mixed for PCR reactions and product detection by using the ABI Prism 7300 system. For each assay, preparations were run in triplicate. The thermal cycling conditions were 95 °C for 10 min, followed by 40 cycles of 95 °C for 15 s, and 60 °C for 1 min. The threshold cycle (Ct) values for both targets and the internal reference (mGAPDH) were measured from the same samples, and the expression of target genes relative to that of mGAPDH was calculated by the comparative Ct method.
The primer sequences for TDAG8 (197 bp) were 5′-atagtcagcgtcccagccaac (forward)/5′-cgcttcctttgcacaaggtg (reverse) and for mGAPDH (233 bp), 5′-ggagccaaacgggtcatcatctc (forward)/ 5′-gaggggccatccacagtcttct (reverse).
Calcium imaging
To detect TDAG8-mediated signaling, calcium imaging was performed as described [
24]. Briefly, human embryonic kidney, adenovirus type 5-transformed 293 cells (HEK293T, obtained from the Bioresource Collection and Research Center of Food Industry Research and Development Institute, Taiwan) were cultured on coverslips and transfected with 1.2 μg pIRES-GFP-TDAG8. Transfected cells were pre-incubated with 2.5 μM Fura-2 acetoxymethyl ester (Fura-2-AM, Molecular Probes) for 40 min in HEPES/MES buffer (125 mM NaCl, 1 mM KCl, 5 mM CaCl2, 1 mM MgCl2, 8 mM glucose, 10 mM HEPES, and 15 mM MES, pH7.6). After washing, cells were stimulated with HEPES/MES (pH 5.5) buffer, followed by [Ca
2+]i recording with a Leica DMI3000B fluorescence microscope and a Ca
2+ imaging system and analyzed by using MetaFluor software. The fluorescence ratio at two excitation wavelengths (340/380 nm, Ca
2+-bound Fura-2-AM/free Fura-2-AM) was recorded and analyzed. The pH-evoked calcium transients and number of cells responding to the indicated pH values were recorded. For compound treatments, cells were stimulated with HEPES/MES buffer (pH 5.5) containing different concentrations of compounds.
Statistical analysis
All data are presented as mean ± SEM. One-way or two-way ANOVA with post hoc Bonferroni correction was used to compare results from multiple groups. Non-parametric Mann-Whitney U test was used to compare results from two groups. For SGC number analysis, the z test for two proportions was used to test the level of significance, with 95% confidence intervals estimated. p < 0.05 was considered statistically significant. A statistical power analysis was performed for sample size estimation, based on data from pilot study, comparing TDAG8−/− to TDAG8+/+ group. The effect size (ES) for t test was 3.4 and for two-way ANOVA was 0.30. With an alpha = 0.05 and power = 0.80, the sample size needed with the effect size (GPower 3.1) was approximately N = 3/group for the t test or total N = 48 for two–way ANOVA. Thus, our sample size of N ≥ 3 or N ≥ 48 was adequate.
Discussion
We previously found TDAG8 involved in RA disease progression and associated pain [
19]. Here, we demonstrated that RA development increased the number of SGCs, and inhibition of glial activation blocked the chronic phase of RA-associated pain. TDAG8 gene deletion inhibited SGC activation and attenuated the chronic phase of RA-associated pain. Thus, TDAG8 may modulate activation of SGCs to attenuate the chronic phase of RA-associated pain. In addition, TDAG8 deletion reduced the number of pro-inflammatory M1 macrophages, which may explain the decrease in arthritis scores and associated pain in TDAG8-deficient mice. We examined the effects of two salicylanilide derivatives, CCL-2d and LCC-09, which were originally designed to inhibit receptor activator of NF-κB ligand (RANKL)-induced osteoclastogenesis [
20,
21] in RA disease progression and associated pain. Both compounds attenuated RA progression and associated pain. These two compounds inhibited TDAG8 gene expression in RA mice and blocked TDAG8 signalling, thereby reducing the number of SGCs and pro-inflammatory M1 macrophages. Thus, TDAG8 may control the number of SGCs and proinflammatory macrophages to modulate RA disease severity and associated pain.
TDAG8 gene deletion greatly attenuated chronic hyperalgesia (> 7 weeks) but less effectively acute hyperalgesia in our mouse model. TDAG8 may regulate different components contributing to the early and late RA pain stages. The acute phase of pain could be largely attributed to acute joint inflammation because of an increase in synovial macrophage number. TDAG8 deficiency reduced the number of pro-inflammatory M1 macrophages (CD80
+ cells) but not synovial macrophages (CD68
+ cells). Therefore, the acute hyperalgesia was less attenuated in TDAG8
−/− mice, which may explain why arthritis scores were not reduced in the early stage of RA in TDAG8
−/− mice. The hypoxia condition may favor M1 polarization [
12]. Given that TDAG8 is expressed in macrophages [
27‐
29] and TDAG8 responds to acid [
30,
31], mice lacking the TDAG8 gene may have reduced macrophage responsiveness to the hypoxia condition, which would not favor M1 polarization.
In the chronic phase, RA progression mainly depends on the increase in M1 macrophage number and cytokine levels (IL-17 and IL-6). Given that both IL-6 and IL-17 are involved in arthritis-induced hyperalgesia [
32‐
34] and administration of a macrophage blocker reduced hyperalgesia [
35], the reduced M1 macrophage number and IL-6 and IL-17 levels we showed may have synergistically contributed to a decline in chronic hyperalgesia and arthritis score. Moreover, neuron–glia interactions play a role in the chronic phase. Consistent with previous studies in monoarthritis or osteoarthritis [
13,
14], our RA mice showed increased number of SGCs. With TDAG8 deficiency, the increased number of SGCs was completely inhibited. Administration of an SGC inhibitor at week 3 attenuated the chronic hyperalgesia (from 6 weeks). Reduced chronic hyperalgesia caused by TDAG8 deletion could also be attributed to an inhibition of the increased SGC number. This suggestion could explain why hyperalgesia was greatly attenuated in the late phase in TDAG8
−/− mice. SGCs can be activated by substance P or calcitonin gene-related peptide (CGRP) [
36] or by IL-17 [
15]. Given that TDAG8 gene knockdown in a peripheral nerve reduces neuron activity [
23], TDAG8 deficiency may reduce neuron activity and release of substance P or CGRP, thereby blocking SGC activation. TDAG8 gene-deficient mice show reduced number of Th17 cells and secretion of IL-17A [
18]. Alternatively, TDAG8 deficiency could decrease IL-17 levels to directly attenuate RA-induced hyperalgesia or indirectly affect SGC activation thereby reducing hyperalgesia. TDAG8
−/− mice showed reduced IL-17 level from week 8, which corresponded to attenuated hyperalgesia in the chronic phase.
A previous study of TDAG8–deficient mice found that TDAG8 deficiency promotes arthritis using anti-collagen antibody/lipopolysaccharide model [
37], suggesting that TDAG8 is a negative regulator for RA. However, recent studies in human patients found that spondyloarthritis patients are associated with genetic variants of TDAG8 locus [
16], and their Th17 cells show high expression of TDAG8 gene [
17]. TDAG8 gene-deficient mice show reduced number of Th17 cells and secretion of IL-17A [
18]. These results suggest that TDAG8 is a positive regulator for RA progression, which are consistent with our RA animal results.
Similar to results from TDAG8-deficient mice, consecutive administration of the salicylanilide derivative CCL-2d or LCC-09 suppressed TDAG8 expression and function, further reducing the bilateral mechanical hyperalgesia induced by arthritis. Both CCL-2d and LCC-09 treatments also reduced the number of SGCs and M1 macrophages. These two compounds likely act on TDAG8 to reduce the proliferation of SGCs and M1 macrophage polarization, thereby affecting bilateral hyperalgesia.
Low doses of CCL-2d (360 μg/kg) or LCC-09 (390 μg/kg) had similar analgesic effects as 3 mg/kg tofacitinib injection, which suggests that CCL-2d or LCC-09 could be more effective analgesics than tofacitinib. The analgesic effect of CCL-2d or LCC-09 was transient (lasted 90–120 min) for a single-dose injection, but consecutive injection (once per week) seemed to have synergistic effects in later weeks. Mechanical hyperalgesia before each injection (CCL-2d 0 min or LCC-09 0 min) was gradually attenuated after week 5 (after the first injection). The reason for the synergistic effects is unclear. Given that both compounds inhibit TDAG8 expression and TDAG8-deficient mice also showed gradually attenuated mechanical hyperalgesia from week 7, the synergistic effects of CCL-2d or LCC-09 on hyperalgesia could be attributed to long-term suppression of TDAG8 gene expression and function.
Although TDAG8 gene deletion significantly reduced RA pain, arthritis scores, bone erosion and cartilage damage, it only slightly reduced pannus. Less reduction in synovial inflammation could be due to distinct effects on macrophages and synovial fibroblasts that are predominantly expressed in the synovial lining of the joint and play many roles in synovial inflammation [
38,
39]. Both synovial macrophages and intimal fibroblast-like synoviocytes are positive for CD68, whereas CD80 recognizes macrophages in synovial sublinings, bone, and other joint areas. Given that TDAG8 deficiency reduced the increased number of CD80
+ macrophages with no inhibition of CD68
+ macrophages, TDAG8 may regulate macrophages but not synoviocytes, which explains the slight reduction of pannus. Similarly, CCL-2d greatly decreased CD80
+ but not CD68
+ cell number, for only a slight reduction in pannus. In contrast, LCC-09 significantly decreased CD68
+ and CD80
+ cell number, which agrees with the reduced pannus.
The presence of osteoclasts, a form of macrophages in bone, seems essential for RA-induced bone erosion because articular bone remains well preserved in the absence of osteoclasts despite overexpression of proinflammatory cytokines [
40]. TDAG8 deficiency reduced CD80
+ cell number, so it probably also decreased osteoclasts to prevent bone erosion. Acidosis in joints induces chondrocyte apoptosis [
41‐
43], which may explain the prevention of cartilage damage in TDAG8-deficient mice.
In our previous study of TDAG8, gene expression suppressed in peripheral nerves of mice, only the initial phase of RA-induced pain was reduced with TDAG8 knockdown [
19]. Nervertheless, in the current study, both the initial and chronic phases of RA pain were reduced in TDAG8-deficient mice. Given that suppression of TDAG8 reduced CD68
+ and CD80
+ cell number [
19] but TDAG8 gene deficiency decreased only CD80
+ cell number, CD68
+ cells could be the major factor for the initial development of RA pain rather than the chronic phase, whereas CD80
+ could contribute to both the initial and chronic phase of RA pain. The finding also explains the attenuated initial arthritis scores in TDAG8-knockdown mice but not TDAG8 gene-deficient mice. Alternatively, we have previously found that ASIC3, TRPV1, and TDAG8 are involved in RA progression and associated pain [
19]. Therefore, we have examined ASIC3 and TRPV1 gene expression in TDAG8 knockout mice without RA induction and found no significant change in ASIC3 and TRPV1 gene expression. It suggests that no ASIC3 and TRPV1 influence on TDAG8 knockout mice before RA induction, although it cannot be excluded a possibility that TDAG8 knockout mice have declined expression levels of ASIC3 and TRPV1 genes after RA induction. In both ASIC3 and TRPV1 knockout mice with attenuated chronic hyperalgesia, expression levels of all three genes were eliminated or at low levels at 12 weeks after RA induction. It is likely that high expression levels of ASIC3, TRPV1, and TDAG8 at 12 weeks after RA induction are essential for maintenance of chronic hyperalgesia.
Accordingly, TDAG8 gene expression in peripheral nerves could be critical to regulate CD68+ and CD80+ cell number, thus contributing to the initial development of RA and associated pain, and TDAG8-mediated satellite glial activation and increased IL-6 and IL-17 levels could be essential for maintenance of RA pain.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.