BATF regulates collagen-induced arthritis by regulating T helper cell differentiation

Background We recently demonstrated that BATF, a member of the activator protein-1 (AP-1) family, regulates osteoarthritic cartilage destruction. Here, we explored the roles and regulatory mechanisms of BATF in collagen-induced arthritis (CIA) in mice. Methods CIA and K/BxN serum transfer were used to generate inflammatory arthritis models in wild-type (WT) and Batf−/− mice. RA manifestations were determined by examining CIA incidence, clinical score, synovitis, synovial hyperplasia, angiogenesis in inflamed synovium, pannus formation, bone erosion, and cartilage destruction. Immune features in RA were analyzed by examining immune cell populations and cytokine production. Results BATF was upregulated in the synovial tissues of joints in which inflammatory arthritis had been caused by CIA or K/BxN serum transfer. The increases in CIA incidence, clinical score, and autoantibody production in CIA-induced WT mice were completely abrogated in the corresponding Batf−/− DBA/1 J mice. Genetic ablation of Batf also inhibited CIA-induced synovitis, synovial hyperplasia, angiogenesis in synovial tissues, pannus formation, bone erosion, and cartilage destruction. Batf knockout inhibited the differentiation of T helper (Th)17 cells and the conversion of CD4+Foxp3+ cells to CD4+IL-17+ cells. However, BATF did not modulate the functions of fibroblast-like synoviocytes (FLS), including the expressions of chemokines, matrix-degrading enzymes, vascular endothelial growth factor, and receptor activator of NF-κB ligand (RANKL). Conclusion Our findings indicate that BATF crucially mediates CIA by regulating Th cell differentiation without directly affecting the functions of FLS.

Background BATF (basic leucine zipper transcription factor, ATF-like) is a member of the activator protein-1 (AP-1) family whose members regulate various biological functions [1][2][3]. BATF, which lacks a transactivation domain, heterodimerizes with JUN to bind the AP-1 site for transcriptional regulation [3,4]. We recently demonstrated that BATF regulates osteoarthritis (OA) in mice by modulating anabolic and catabolic gene expression in chondrocytes [4]. We found that overexpression of BATF upregulated matrix-degrading enzymes and downregulated cartilage matrix molecules; we also found that BATF expression in mouse joint tissues promoted OA cartilage destruction and that, conversely, knockout of Batf in mice (Batf −/− ) suppressed experimental OA [4].
OA and rheumatoid arthritis (RA), which are the most common types of joint arthritis, share certain phenotypic features, such as cartilage destruction [5]. However, these diseases clearly differ in their etiologies, pathogenic mechanisms, and the cell types associated with each pathogenesis. OA is a degenerative joint disease that begins with the destruction of surface articular cartilage [6,7]. Mechanical stresses (i.e., joint instability) and factors that predispose toward OA (i.e., aging) are important causes of OA pathogenesis [6,7]. In contrast, RA is an inflammatory autoimmune disease that mainly targets the synovium, resulting in destruction of the joint architecture. Various cell types of joint tissues are associated with RA pathogenesis, including T cells, B cells, macrophages, synoviocytes, chondrocytes, and osteoclasts [8][9][10]. T cell-mediated autoimmune responses play a critical role in the RA pathogenesis, in which interleukin (IL)-17-producing T helper (Th) cells act as crucial effectors [8,11,12]. RA is characterized by synovitis with infiltration of immune cells, synovial hyperplasia that arises via proliferation of synovial cells, such as macrophage-like synoviocytes (MLS) and fibroblast-like synoviocytes (FLS), and angiogenesis in the hyperplastic synovium [8][9][10]. Synovial cells express numerous cytokines that have been implicated in many of the immune processes involved in RA [13]. RA manifestations also include erosion of the bone and cartilage which is caused by the formation of the pannus, which is an aggressive front of hyperplastic synovium. The pannus invades and destroys mineralized cartilage and bone through the action of osteoclasts [8][9][10].
BATF is known to regulate OA cartilage destruction, and we previously showed that BATF overexpression in joint tissues causes synovial inflammation [4] suggesting that BATF could contribute to inflammatory arthritis. This notion is supported by reports that proinflammatory cytokines, such as IL-1β and IL-6, increase BATF expression in naive CD4 + T cells [14,15], BATF directly regulates IL-17 expression, and Batf-deficient mice show resistance to experimental autoimmune encephalomyelitis [16]. BATF also controls the development of follicular Th cells (Tfh) and class-switch recombination in B cells [17]. Additionally, inhibition of the transcriptional activity of c-Fos/AP-1 suppresses arthritic joint destruction in a mouse RA model [18]. However, while these previous reports suggest that BATF may be involved in RA pathogenesis, the role of BATF and its regulatory mechanisms are not yet well understood.
In this study, we examined whether BATF is required for collagen-induced arthritis (CIA), which is a commonly used experimental model of inflammatory arthritis caused by a T cell-dependent, antibody-mediated autoimmune response directed against cartilage type II collagen [19]. Here, we show that genetic ablation of Batf in mice suppresses the manifestations of CIA, including synovitis, synovial hyperplasia, angiogenesis in the inflamed synovium, and cartilage/bone erosion in the joint tissues. We also reveal that BATF regulates CIA by regulating Th cell differentiation without directly affecting the functions of FLS.

Mice and experimental RA
Male wild-type (WT) and Batf −/− DBA/1 J mice were used to generate the CIA models. C57BL/6-background Batf −/− mice [4] were backcrossed with DBA/1 J mice to generate Batf −/− DBA/1 J mice. All mice were used in accordance with protocols approved by the Animal Care and Ethics Committees of the Gwangju Institute of Science and Technology. CIA was induced by a standard protocol [19,20]. Mice were intradermally injected with incomplete Freund's adjuvant alone (nonimmunized; NI) or Freund's adjuvant containing 100 μg collagen type II (CIA). A booster injection was given 21 days later. The incidence and severity of arthritis were evaluated on the indicated days after the first immunization. Severity was evaluated using a clinical score (grade 0−4) of paw swelling [19,20]. Joint tissues were fixed, decalcified with 0.5 M EDTA, embedded in paraffin, and sectioned at 5-μm thickness. Synovitis was evaluated by hematoxylin and eosin (H&E) staining, and synovial inflammation (grade 0−4) was scored as previously described [19,20]. The pannus was visualized by H&E staining and quantified by scoring (grade 0−4) [19,20]. Cartilage destruction was examined by safranin-O staining and scored using the OARSI (Osteoarthritis Research Society International) grading system [4,20]. Inflammatory arthritis was also induced by K/BxN serum transfer [21] in WT and Batf −/− C57BL/6 mice. Arthritic transgenic mice (K/BxN) and nontransgenic littermates (BxN) were generated by crossing KRN T cell receptor (TCR)-transgenic (K/B) mice with nonobese diabetic (NOD) mice. K/BxN and control sera were collected from K/BxN and BxN mice, respectively, and administered intraperitoneally to recipient mice on days 0 and 2. Mice were sacrificed on day 14 after serum transfer.
FLS culture and proliferation assays FLS were isolated from WT or Batf knockout (KO) mice and cultured as described by Zhao et al. [23]. FLS of passages 4-8 were used for further analysis. Pure FLS (> 90% CD90 + /< 1% CD14 + ) were identified by flow cytometry using antibodies against CD90 and CD14 (Abcam). FLS proliferation in culture was quantified by measuring bromodeoxyuridine (BrdU) incorporation [20]. Briefly, FLS cultured in a 96-well plate were treated with or without tumor necrosis factor (TNF)-α (100 ng/ml), and BrdU labeling was detected using the cell proliferation enzyme-linked immunosorbent assay (ELISA) BrdU kit (Roche). Proliferating cells in synovial sections were identified by detecting Ki67 using an antibody obtained from Abcam. The empty adenovirus (Ad-C) and BATF-expressing adenovirus (Ad-Batf ) were as previously described [4]. FLS were infected with Ad-Batf and Ad-C at the indicated multiplicities of infection (MOI) for 2 h, washed, and maintained for 48 h before analysis.

ELISA of autoantibody production
Collagen type II-specific antibodies were measured by ELISA [20]. Sera from NI and CIA mice were loaded to type II collagen-coated 96-well plates, incubated overnight at 4°C, washed, and incubated for 1 h with alkaline phosphatase-labeled monoclonal antibodies against mouse IgG1, IgG2a, or IgG2b (Immunology Consultants Lab). p-Nitrophenyl phosphate was used as a substrate for chromogenic reactions, and the resulting color reaction was quantified using an ELISA plate reader.

Statistical analysis
The nonparametric Mann-Whitney U test was used for the analysis of data based on an ordinal grading system, such as the synovitis, pannus, and OARSI grades. For results obtained from ELISA and analyses of joint thickness, TRAP-positive cells, and BrdU incorporation, the data were first tested for conformation to a normal distribution using the Shapiro-Wilk test. The data were analyzed by the Student's t test (pair-wise comparisons) or analysis of variance (ANOVA) with post-hoc tests (multi-comparisons) as appropriate. Significance was accepted at the 0.05 level of probability (P < 0.05).

BATF is upregulated in synovial tissues of arthritic joints
To explore the possible functions of BATF in inflammatory arthritis, we first examined the expression patterns of BATF in the arthritic joints of DBA/1 J mice treated with CIA. Immunostaining revealed that BATF expression was markedly increased in the inflamed synovial tissues of CIA (Fig. 1a). In contrast, BATF was not detected (i.e., there was no obvious synovitis or synovial hyperplasia) in the synovial tissues of Batf −/− mice subjected to CIA conditions (Fig. 1a). As expected, TNF-α and IL-6 were also markedly increased in the CIA synovial tissues (Fig. 1b). To identify cell types expressing BATF in synovial tissues, we performed double immunofluorescence staining of BATF with markers of various synovial cell types, including vimentin for FLS, CD4 for T cells, B220 for B cells, and CD11b for macrophages. BATF was detected in subsets of CD4 + T cells (> 80%), CD11b + macrophages (4 0%), and vimentin-expressing FLS (< 50%), but not in B220-expressing B cells (Fig. 1c).

Genetic ablation of Batf inhibits CIA
We next examined whether BATF regulates CIA. The CIA manifestations observed in WT mice, including CIA incidence, clinical scores, and paw swelling, were completely abrogated in Batf −/− DBA/1 J littermates (Fig. 2a, b). The production of IgG autoantibodies against type II collagen is a key pathological change in RA pathogenesis, and the IgG2a autoantibody is particularly predominant in the CIA model [25]. Consistent with the above results, IgG2a production was markedly increased in the sera of CIA-induced WT mice, but this was completely suppressed in Batf −/− littermates (Fig. 2c). These results collectively indicate that a b c Fig. 1 Upregulation of BATF in arthritic synovial tissues. a Representative images (n = 10) of BATF immunostaining in synovial tissue sections of nonimmunized (NI) and collagen-induced arthritis (CIA) wild-type (WT) DBA/1 J mice and Batf −/− littermates (knockout (KO)). b Representative immunostaining images (n = 8) of tumor necrosis factor (TNF)-α and interleukin (IL)-6 in CIA synovial tissues determined by immunohistochemical staining and immunofluorescence microscopy, respectively. c Representative triple-immunofluorescence microscopic images of mouse CIA synovial sections (n = 12) immunostained for BATF, cell type-specific markers, and DAPI. Scale bars = 50 μm BATF is required for the pathogenesis of CIA in DBA/1 J mice.
Genetic ablation of Batf inhibits CIA-induced synovitis, synovial hyperplasia, and angiogenesis in synovial tissues CIA is characterized by a synovial hyperplasia that arises through the proliferation of synoviocytes, synovitis with infiltration of immune cells, and angiogenesis in synovial tissues [8][9][10][11]. Here, we found that CIA-induced synovitis was significantly blocked in Batf −/− mice, as determined by H&E staining and inflammation scoring (Fig. 3a, b). To examine synovial hyperplasia, we determined synovial cell proliferation by Ki67 staining [26]. Ki67 was highly expressed in the synovial tissues of WT mice subjected to CIA-inducing conditions, whereas it was completely absent from those of the corresponding Batf −/− mice (Fig.  3c). The inhibition of synovial cell proliferation in Batf −/− mice was further confirmed by BrdU incorporation assays performed using dissociated FLS obtained from WT and Batf −/− mice. Compared with WT FLS, Batf-deficient FLS showed significantly less proliferation in response to TNF-α (Fig. 3d).

Genetic ablation of Batf inhibits CIA-induced bone and cartilage erosion
As CIA involves the formation of pannus, which invades and destroys mineralized cartilage and bone [8][9][10][11], we next examined the possible role of BATF in pannus formation and subsequent bone and cartilage destruction. Indeed, we found that the CIA-induced pannus formation found in WT mice was completely abrogated in Batf −/− mice (Fig. 4a). Next, we examined the expression of RANKL, which promotes osteoclast differentiation to stimulate bone erosion [29,30]. The induction of CIA in WT mice increased the expression of RANKL at the bone/pannus interface, but this was completely abrogated in Batf −/− littermates (Fig. 4b). Although FLS are known to produce RANKL [20], BATF overexpression in FLS did not trigger any upregulation of RANKL (Fig. 3f ), suggesting that the upregulation of BATF in the FLS of CIA synovium is not directly associated with RANKL expression. In addition, TRAP-positive osteoclasts, which were highly increased at the bone/pannus interface of WT mice, were not observed in Batf −/− littermates (Fig. 4c). These results collectively suggest that Batf KO inhibits bone erosion under CIA-inducing conditions by blocking RANKL expression and osteoclast differentiation. . c Type II collagen-specific autoantibody production in NI and CIA WT and Batf −/− DBA/1 J mice (n = 10 mice per group). Values are means ± SEM; *P < 0.01, **P < 0.001, ***P < 0.0001 CIA in WT mice caused cartilage destruction, as determined by safranin-O staining and OARSI grading (Fig.  4d). Similar to the case of bone erosion, cartilage destruction was completely abrogated in Batf −/− littermates (Fig.  4d). Cartilage destruction during CIA can be caused by pannus, which invades and destroys mineralized cartilage, or matrix-degrading enzymes produced by synoviocytes and chondrocytes [31,32]. Although the matrix-degrading enzymes in inflammatory joint tissues are primarily produced by synoviocytes, such as FLS [31,32], BATF overexpression in FLS did not cause any upregulation of the tested matrix-degrading enzymes (MMP3 and MMP13) or chemokines and chemokine receptors (CCL2, CCL5, CXCL1, CXCL5, and CXCL10) (Fig. 3f). Notably, however, we previously showed that BATF overexpression causes chondrocytes to produce MMP3 and MMP13 [4]. These results collectively suggest that matrix-degrading enzymes produced by chondrocytes, not FLS, may contribute to cartilage destruction during CIA.

Genetic ablation of Batf does not affect thymic T cell development
Given that T cells, such as Th17 cells, play crucial roles in inflammatory arthritis [33], we examined whether Batf deficiency affected T cell development in DBA/1 J mice. Flow cytometric analysis revealed that thymic T cell development was not affected by Batf deletion in DBA/1 J mice (Fig. 5a). However, consistent with a previous report [34], Batf −/− mice exhibited altered T cell populations in peripheral organs, such as the spleen and lymph nodes (Fig. 5b, c). Compared with WT mice, T cells were slightly reduced in the spleen (Fig. 5b) and lymph nodes (Fig. 5c) of Batf −/− mice. CD4 + effector/memory T cells were also slightly reduced in the spleen (Fig. 5b) and lymph nodes (Fig. 5c) of Batf −/− mice. Overall, although the populations of T cell subsets in the periphery were altered by Batf deletion, the degree of alteration was small (i.e., less than 10%) (Fig. 5b, c). To further elucidate the functions of BATF, we examined whether Batf deficiency affects the in-vitro differentiation of CD4 + T cells into Th cells, including Th1 (CD4 + IFNγ + ), Th2 (CD4 + IL-4 + ), Th17 (CD4 + IL-17A + ), and Treg cells (CD4 + Foxp3 + ). Interestingly, Batf deletion significantly increased Th2 cell differentiation in Batf −/− mice, while in-vitro Th1 and Treg cell differentiations were not affected by Batf gene deletion (Fig. 6a). Batf gene deletion also significantly reduced Th17 cell differentiation in Batf −/− DBA/ 1 J mice compared with WT control mice (Fig. 6a). Consistent with this finding, the Th17 cell population was significantly reduced in the lymph nodes of Batf −/− mice subjected to CIA-inducing conditions (Fig. 6b). Moreover, the CD4 + Foxp3 + IL-17 + cell population was dramatically reduced in the peripheral lymph nodes of CIA-induced Batf −/− mice (Fig. 6c). As the transconversion of Treg cells to Th17 cells can reportedly exacerbate Th17-mediated inflammation [35], we examined whether BATF affects the conversion of Treg cells to Th17 cells in vitro. Indeed, BATF deficiency dramatically abrogated the transconversion of Treg cells to Th17 cells (Fig. 6d), indicating that BATF regulates this conversion. Finally, we analyzed cytokine production in the lymph nodes of Batf −/− mice and WT littermates under CIA-inducing conditions. Consistent with the increase in Th2 cell differentiation, the production of the Th2 cytokines IL-4, IL-10, and IL-13 were increased in CIA-induced Batf −/− mice compared with WT littermates (Fig. 6e). Additionally, IL-6 was also increased in CIA-induced Batf −/− mice (Fig. 6e). In contrast, the production of the inflammatory cytokines IL-2, IL-17, and TNF-α were reduced in Batf −/− mice under CIA-inducing conditions (Fig. 6e).
Genetic ablation of Batf does not affect inflammatory arthritis caused by K/BxN serum transfer Finally, we examined a possible role of BATF in T cell-independent inflammatory arthritis using C57BL/6 mice subjected to K/BxN serum transfer [21]. Immunostaining revealed that BATF expression was markedly increased in the inflamed synovial tissues caused by K/BxN serum transfer (Fig. 7a). However, all examined manifestations of inflammatory arthritis caused by K/BxN serum transfer in WT mice (i.e., paw thickness, synovial inflammation, and cartilage erosion) were not markedly inhibited in Batf KO mice (Fig. 7b, c). Our results suggest that BATF is not essential for T cell-independent inflammatory arthritis.

Discussion
We herein show that BATF is required for CIA since Batf −/ − mice exhibit complete suppression of the manifestations of CIA. We further demonstrate that BATF regulates Th cell differentiation during CIA. The pathogenesis of RA is associated with enrichment of Th17 cells in the inflamed synovium and enhancement of IL-17 production [8,9,36,37]. A recent report indicated that the conversion of Treg cells to Th17 cells contributes to bone destruction in CIA mice [35], suggesting that an imbalance between these cell populations contributes to CIA. BATF is known to regulate Th17 cell differentiation by modulating the expression of the transcription factor RORγt [16], and to control class-switch recombination in T and B cells [17]. Thus, BATF was previously known to function at multiple hierarchical levels in two cell types to globally regulate switched-antibody responses. Here, we demonstrated that BATF deficiency decreases Th17 cell differentiation both in vivo and in vitro. It was previously reported that a portion of Foxp3-positive cells also expresses IL-17, and that CD4 + Foxp3 + IL-17 + cells can be found in CIA mice [35]. However, we found here that Batf deficiency completely abolished CD4 + Foxp3 + IL-17 + cell generation under CIA-inducing conditions. An additional interesting finding of our current study is that BATF regulates Th2 cell differentiation; we demonstrated that IL-4-producing CD4 + T cells were increased in CIA-induced Batf −/− mice compared with WT control mice. This is consistent with a previous report that IL-4 negatively regulates the induction and progression of CIA [38]. In addition, BATF is not detected in the infiltrated B cells of CIA synovial tissue. Although this does not rule out expression and the possible role of BATF in B cells of peripheral lymphatic tissues, our results indicate that upregulation of BATF in the infiltrated B cells is not essential for pathogenesis of CIA. Furthermore, we found that T cell-independent inflammatory arthritis caused by K/BxN serum transfer was not affected by Batf −/− mice. Based on these findings, we propose that BATF deficiency alters Th cell differentiation, enabling Batf −/− mice to resist CIA. Thus, BATF inhibition could be a useful strategy for the treatment of RA.
We also report that FLS are not directly associated with the BATF-mediated regulation of CIA, although BATF appears to regulate the proliferation of FLS in Values are presented as means ± SD; *P < 0.01, **P < 0.001, ***P < 0.0001. ns not significant response to TNF-α. Accumulating evidence indicates that FLS are key players in RA pathogenesis [10,28]. Cytokines and chemokines produced by FLS attract T cells to RA synovium, and the interaction of FLS with T cells results in the activation of both cell types. FLS also produce matrix-degrading enzymes involved in cartilage destruction, angiogenic factors associated with neovascularization, and RANKL [10,28]. The latter factor regulates osteoclastogenesis, which requires physical contact of precursor cells with RANKL-expressing FLS or T cells [29,30,39]. We showed previously that HIF-2α modulates various functions of FLS, including proliferation, the expressions of RANKL and various catabolic factors, and osteoclastogenic potential [20]. Here, we found that, unlike HIF-2α, TNF-α and IL-6 do not cause BATF expression in FLS, and BATF overexpression in FLS does not modulate the expressions of various matrix-degrading enzymes or chemokines. These results collectively support our notion that the ability of BATF to regulate CIA is not due to a direct BATF-mediated modulation of FLS functions. It has proven difficult to elucidate the role of chondrocytes in cartilage destruction during RA pathogenesis [5]. Such destruction occurs primarily at the interface of the pannus and calcified cartilage [5,40]. There is evidence that proteoglycans are lost from the superficial zone, where cartilage contacts with synovial fluid, but not with the pannus [5]. Because RA synovium produces various matrix-degrading enzymes [5,8,9], cartilage destruction at the superficial zone may be due to synovial cell functions. However, proteoglycans can also be lost from the middle and deep zones of the cartilage [5], suggesting that a chondrocyte may help degrade its own matrix by releasing matrix-degrading enzymes. Indeed, we recently demonstrated that BATF upregulates MMP3 and MMP13 in chondrocytes, leading to cartilage destruction during OA pathogenesis [4]. Therefore, our current and previous [4] results suggest that the BATF-mediated regulation of MMP3 and MMP13 expression in chondrocytes is associated with cartilage destruction during CIA.
The results of the present study collectively indicate that BATF regulates CIA in mice, and our previous work showed that BATF functions as a catabolic regulator of OA cartilage destruction by upregulating catabolic enzymes (e.g., MMP3 and MMP13) in chondrocytes [4]. Thus, despite their different etiologies and pathogeneses, both RA and OA are regulated by BATF. However, different mechanisms are involved; BATF regulates OA pathogenesis by upregulating matrix-degrading enzymes in chondrocytes, whereas it appears to regulate RA pathogenesis by regulating Th cell differentiation. Thus, BATF could be a useful target for the regulation of RA.

Conclusions
In summary, we demonstrated here that BATF regulates CIA, including synovitis, synovial hyperplasia, angiogenesis in the inflamed synovium, cartilage destruction, and bone erosion in the joint tissues. We also reveal that BATF regulates CIA by regulating Th cell differentiation without directly affecting the functions of FLS.

Availability of data and materials
The data supporting the conclusions of this article are included within the article.
Authors' contributions SGP and JSC contributed equally as corresponding authors by conceiving and designing the experiments. SHP, JR, and SKK performed the experiments. JAK, JSK, YS, and WSC analyzed the data. SHP, JR, SGP, and JSC wrote the manuscript. All authors read and approved the final manuscript.
Ethics approval and consent to participate Not applicable.