Immune‐bone interplay in the structural damage in rheumatoid arthritis

Summary

The immune and bone systems maintain homeostasis by interacting closely with each other. Rheumatoid arthritis is a pathological consequence of their interplay, as activated T cell immune responses result in osteoclast‐mediated bone erosion. An imbalance between forkhead box protein 3 (Foxp3)⁺ regulatory T (T_reg) cells and T helper type 17 (Th17) cells is often linked with autoimmune diseases, including arthritis. Th17 cells contribute to the bone destruction in arthritis by up‐regulating receptor activator of nuclear factor kappa‐Β ligand (RANKL) on synovial fibroblasts as well as inducing local inflammation. Studies on the origin of Th17 cells in inflammation have shed light on the pathogenic conversion of Foxp3⁺ T cells. Th17 cells converted from Foxp3⁺ T cells (exFoxp3 Th17 cells) comprise the most potent osteoclastogenic T cell subset in inflammatory bone loss. It has been suggested that osteoclastogenic T cells may have developed originally to stop local infection in periodontitis by inducing tooth loss. In addition, Th17 cells also contribute to the pathogenesis of arthritis by modulating antibody function. Antibodies and immune complexes have attracted considerable attention for their direct role in osteoclastogenesis, and a specific T cell subset in joints was shown to be involved in B cell antibody production. Here we summarize the recent advances in our understanding of the immune‐bone interplay in the context of the bone destruction in arthritis.

Introduction

Rheumatoid arthritis (RA) is one of the most common autoimmune diseases, and is characterized by chronic joint inflammation along with local and systemic bone loss 1. Both environmental and genetic factors are involved in the etiology of RA, but the precise mechanisms underlying its onset and ensuing bone destruction remain elusive. The presence of autoantibodies, efficacy of T or B cell‐targeted therapies, genome‐wide association study (GWAS) analyses and animal models all support the importance of T and B cells in the pathogenesis of RA. In this study, we provide an overview of the current understanding of how the immune system contributes to bone destruction in RA, with a focus on the role of acquired immunity.

Osteoclasts in the bone destruction in RA

Bone loss in RA results from an excess of osteoclast‐mediated bone resorption and inhibition of osteoblast‐mediated bone formation 2. Osteoclasts can be formed from cultured rheumatoid synovial cells, indicating that both osteoclast precursors and osteoclastogenesis‐supporting cells are present in synovial cells 3. The molecular mechanism of osteoclast differentiation has been studied extensively ever since the identification of receptor activator of nuclear factor kappa‐Β ligand (RANKL), which was shown to be expressed on synovial cells 4, 5. Osteoclasts can be differentiated from macrophage/monocyte‐derived bone marrow cells in the presence of RANKL and macrophage colony‐stimulating factor (M‐CSF) in vitro.

A critical role for osteoclasts in bone destruction was demonstrated using mice which lack osteoclasts. Bone destruction did not occur in RANKL‐deficient mice lacking osteoclasts in a serum transfer model. Similarly, Fos‐deficient mice lacking osteoclasts did not exhibit bone erosion in a human tumour necrosis factor (TNF) transgenic (Tg) model. In both cases, inflammation was observed at a similar level to the controls 6, 7. Osteoprotegerin (OPG) is a soluble decoy RANKL receptor that inhibits RANKL function by interfering with RANK–RANKL binding. OPG‐deficient mice develop severe osteoporosis due to excessive osteoclastogenesis. Treatment with OPG ameliorated bone destruction, but not inflammation, in animal models of arthritis 8, 9. Importantly, even in the presence of joint inflammation, bone loss was scarcely detected in a RA patient with osteopetrosis caused by reduced osteoclast activity 10. This study, together with mouse studies, indicated that osteoclasts are indispensable for bone destruction in RA. Several clinical studies have shown that anti‐RANKL antibodies protect against bone erosion and provide increased bone mineral density in RA 11, 12, 13, 14. It is notable that anti‐RANKL antibodies have been approved for the treatment of bone destruction in RA in Japan.

Certain reports have suggested that osteoclasts are able to differentiate independently of RANKL 15, 16, 17, 18. However, most of these studies have not provided sufficient evidence for this claim to be accepted, as none of these experiments were performed under completely RANKL‐ or RANK‐deficient conditions in vivo. For example, it was reported that lysyl oxidase (LOX) induces osteoclast differentiation in vitro without any addition of RANKL 18. However, LOX failed to induce in vitro osteoclastogenesis from the bone marrow cells of RANKL‐deficient mice or rescue the osteopetrotic phenotype of RANKL‐deficient mice 19. Taken together, osteoclastogenesis is RANKL‐dependent in mice. In humans, RANKL mutation results in osteopetrosis, indicating that RANKL is essential for osteoclastogenesis under physiological conditions. The important role of RANKL in osteoclastogenesis in RA is supported by the efficacy of anti‐RANKL antibodies in the suppression of bone erosion. It will be necessary to examine the bone destruction in osteopetrotic patients with RANKL or RANK mutation to determine whether or not RANKL is absolutely required for osteoclastogenesis in RA.

The RANKL source in local bone destruction in RA

Under physiological conditions, the major sources of RANKL for bone remodelling are osteoblasts and osteocytes. Which cells induce osteoclast differentiation by expressing RANKL in arthritis? RANKL is expressed in the RA synovium mainly by synovial fibroblasts and T cells 4, 5, 8. Fixed activated T cells directly induce in vitro osteoclastogenesis by expressing RANKL 8. However, live activated T cells were shown to be unable to do so, because they also express cytokines such as IFN‐γ that inhibit osteoclastogenesis, suggesting that the osteoclastogenic activity of T cells is dependent upon the balance of cytokines they produce 20. Type 6a collagen (Col6a), a marker of mesenchymal cells, is expressed specifically on synovial fibroblasts in joints 21. Mice in which RANKL was specifically deleted in synovial fibroblasts and T cells were established using Col6a‐Cre and Lck‐Cre mice, respectively 22, to allow investigation of which cells are the primary RANKL‐expressing cells in vivo. It was demonstrated that osteoclast formation and bone destruction in the arthritic joints of Tnfsf11^flox/Δ Col6a‐Cre mice were inhibited, even when they displayed comparable joint inflammation as Lck‐Cre mice. In contrast, these activities were not inhibited in the arthritic joints of Tnfsf11^flox/Δ Lck‐Cre mice. Thus, synovial fibroblasts and not T cells in arthritic joints are considered to be the major RANKL source that induces osteoclast formation in mice 22. Recently, B cells were also reported to express RANKL in the RA synovium, but further studies need to establish the pathogenic role of B cell‐derived RANKL 23, 24, 25.

Bone destruction in arthritis takes place both locally and systemically. Systemic bone loss raises the risk of fracture in RA patients. It is possible that an increase in soluble RANKL in serum (or the RANKL/OPG ratio) may influence systemic bone loss. However, the contribution of RANKL to the osteoclatogenesis and cellular source of RANKL in systemic bone loss awaits elucidation in the future.

Th17 cells and T_reg cells in bone destruction in arthritis

Various immune cells, including T cells, B cells, neutrophils, macrophages and dendritic cells, infiltrate the arthritic synovium. Which immune cells up‐regulate RANKL expression and thus contribute to the bone destruction in arthritis? The contribution of CD4⁺ T cells in RA is supported by the presence of autoantibodies and the T cell‐genes associated with RA 26, 27, 28, 29, 30. It is also supported by the efficacy of cytotoxic T lymphocyte‐associated protein 4‐immunoglobulin (CTLA4‐Ig), a selective inhibitor of T cell activation 31, and studies from animal models 32, 33. Notably, a recent GWAS suggested that epigenetic changes in regulatory T (T_reg) cells may contribute to the pathogenesis of RA 34. Expression quantitative trait loci (eQTL) analysis revealed that activation of the TNF‐α pathway in CD4⁺ T cells is a causal cytokine event in RA 35.

The presentation of autoantigens by dendritic cells leads to the generation of various T helper (Th) cells, including Th1, Th2, Th17 and T follicular helper (Tfh) cells. In addition to RANKL, activated T cells express effector cytokines which either stimulate or inhibit osteoclastogenesis 2. Th1 and Th2 cells inhibit osteoclastogenesis by producing IFN‐γ and interleukin (IL)‐4, respectively. Th17 cells are the exclusive osteoclastogenic Th subset. Th17 cells express RANKL at the highest level of any of the Th cells. IL‐17, a hallmark factor of Th17 cells, stimulates osteoclast formation by up‐regulating RANKL on osteoclast‐supporting mesenchymal cells 36. In arthritis, IL‐17 induces osteoclastogenesis by up‐regulating RANKL expression on synovial fibroblasts and inducing innate immune cells to secrete inflammatory cytokines such as TNF‐α, IL‐1β and IL‐6. These cytokines further promote osteoclastogenesis by inducing osteoclast precursor cells to become highly responsive to RANKL and up‐regulating RANKL expression on synovial fibroblasts. Thus, the Th17 cell–synovial fibroblast axis promotes osteoclastic bone resorption in the inflamed synovium. Synovial fibroblasts not only induce osteoclastogenesis, but also suppress osteoblastic bone formation. TNF‐α up‐regulates Dickkopf‐related protein 1 (Dkk‐1) expression in synovial fibroblasts, leading to the suppression of the Wnt signalling that is required for osteoblastic bone formation 37.

T_reg cells play a central role in the prevention of autoimmune diseases 38, and are also able to suppress RANKL‐induced osteoclastogenesis. Certain studies have reported that the suppressive function is mediated by TGF‐β/IL‐4 or TGF‐β/IL‐10 in a cell–cell contact‐independent manner 39, 40, whereas others have suggested instead a cell–cell contact‐independent pathway via CTLA‐4 41, 42. CTLA‐4 binds to CD80/CD86 on osteoclast precursors and induces apoptosis in an indoleamine 2,3‐dioxygenase (IDO)‐dependent manner 42. T_reg cell transfer experiments have demonstrated that T_reg cells suppress osteoclast formation and increase bone volume in vivo 43, 44. Thus, it is suggested that the balance between T_reg cells and Th17 cells may be a critical factor in regulating the bone destruction that takes place in RA.

exFoxp3Th17 cells – the most osteoclastogenic T cells in both arthritis and periodontitis

The dynamic T_reg/Th17 balance is a critical determinant of the immune response to autoimmune diseases, including arthritis. In RA, anti‐IL‐6 treatments up‐regulate the T_reg/Th17 ratio 45. Interestingly, an increase in T_reg cells in the peripheral blood correlates with the clinical response elicited by anti‐IL‐6 therapeutics 46.

Forkhead box protein 3 (Foxp3) is a key transcriptional factor indispensable for T_reg cell development and function 38. The plasticity of Foxp3⁺ T_reg cells under inflammatory conditions has garnered attention because T cells which recognize self‐peptides are enriched in T_reg cells 47, 48, 49. It was demonstrated by tracking the fate of Foxp3⁺ T_reg cells in vivo using Foxp3‐Cre‐GFP ROSA‐YFP mice that, while CD25^hiFoxp3⁺ T cells maintain Foxp3 expression, CD25^loFoxp3⁺ T cells lose Foxp3 expression and are converted into a novel Th17 cell subset (termed exFoxp3Th17 cells) under arthritic conditions 50. This conversion was shown to be promoted by the IL‐6 produced by arthritic synovial fibroblasts. Considering the heterogeneity of arthritic synovial fibroblasts 51, it is a subject of interest to identify an arthritogenic synovial fibroblast subpopulation in future studies. Notably, exFoxp3Th17 cells in synergy with synovial fibroblasts induced osteoclastogenesis to a much greater extent than did naive CD4⁺ T‐derived Th17 cells, indicating that exFoxp3Th17 cells are the most osteoclastogenic T cell subset. FoxP3⁺IL‐17⁺ cells, which are considered to be in the transition state of the conversion, are observed frequently with high but not low activity in the RA synovium or in OA patients, suggesting the contribution of plastic FoxP3⁺ T cells to RA pathogenesis. Taken together, exFoxp3Th17 cells play a key role in the inflammation and bone resorption in autoimmune arthritis 50. The pathogenic role of exFoxp3T cells has been reported in a variety of diseases, such as diabetes, multiple sclerosis and asthma, suggesting that the conversion of Foxp3⁺ T cells into effector T cells may be a general underlying mechanism 47, 52, 53.

Why does this pathogenic T cell subset continue to exist? Foxp3⁺ T_reg cells protect against allergy and autoimmunity, but promote cancer and infectious disease. Thus, it is considered that the plasticity of Foxp3⁺T cells may be needed to adapt to changes in the surrounding environment. In line with this idea, a beneficial role of exFoxp3Th17 cells was reported recently in a mouse model of periodontitis, the most common infectious disease 54. Oral bacteria induce the production of IL‐6 which, in turn, promotes the conversion of Foxp3⁺ T cells into Th17 cells. exFoxp3Th17 cells were found to be the most potent bone‐damaging T cells in periodontitis as well as in autoimmune arthritis. Interestingly, these bone‐damaging exFoxp3Th17 cells protect against oral bacteria in two different ways. First, they elicit mucosal immune responses, including the production of anti‐bacterial products. Secondly, they inhibit bacterial dissemination by removing the tooth in which the bacteria reside. Thus, exFoxp3Th17 cells play a beneficial role in the host defence against oral bacteria, while they are exclusively harmful in the context of arthritis 54. This investigation highlighted the notion that T cell‐mediated bone damage, which was considered to be an adverse consequence of inflammation, may have developed to contribute to host defence against oral bacteria (Fig. 1).

Figure 1.

ex‐Forkhead box protein 3‐T helper type 17 (exFoxp3Th17) cells, the bone‐damaging T cells in rheumatoid arthritis (RA) and periodontitis. Foxp3⁺ T cells convert into exFoxp3Th17 cells, which are the most potent osteoclastogenic T cells in both arthritis and periodontitis. This conversion is promoted by mesenchymal cell‐derived interleukin (IL)‐6. exFoxp3Th17 cells induce osteoclastogenesis mainly by up‐regulating receptor activator of nuclear factor kappa‐Β ligand (RANKL) on mesenchymal cells. They evoke structural damage and are entirely pathogenic in the context of RA. While exFoxp3Th17 cells induce bone damage in periodontitis, they play a beneficial role in host defence against oral bacteria by ejecting the infected tooth and producing anti‐bacterial peptides.

The role of antibodies in bone destruction in RA

Antibodies against citrullianated protein (ACPAs) and rheumatoid factors are autoantibodies that are detected specifically in the peripheral blood as well as in the arthritic joints of RA patients, and are used for RA diagnosis 55. The immune complexes (ICs), conjugates of antibodies and their antigens, activate innate immune cells via Fcγ receptors and promote arthritic inflammation. The involvement of ACPAs in the bone loss in RA is suggested by several clinical observations. First, more severe local and systemic bone loss was observed in ACPA‐positive than ACPA‐negative RA 56, 57. Secondly, bone loss was detected in certain ACPA‐positive subjects with no clinical signs of arthritis 58. These findings support that ACPAs not only induce bone loss indirectly via an exacerbation of inflammation, but also induce bone loss directly in RA.

Recent studies have shown that ACPAs promote osteoclast differentiation directly in vitro and in vivo 59, 60, 61, 62. In the presence of RANKL, ACPAs promote osteoclastogenesis by binding to the antigens expressed on osteoclast precursors and up‐regulating TNF‐α and IL‐8 production 59, 60. As TNF‐α activates various cells and IL‐8 attracts neutrophils to the site of inflammation, it has been suggested that osteoclasts are a mediator of inflammation. Although this concept is intriguing, in vivo evidence needs to be provided in further studies.

ICs also induce bone loss by binding to Fcγ receptors on osteoclast precursors 61, 62. Osteoclast precursors express positive and negative FcγRs (FcγRI/III/IV and IIB, respectively). The IgG2 ICs increase and activate osteoclastogenesis by binding to FcγRI and FcγRIV, the expression of which is also up‐regulated under inflammatory conditions. Thus, the level of osteoclastogenesis is determined by the strength of FcγR signalling which, in turn, is dependent upon the relative expression of positive and negative FcγRs, as well as the availability of ICs 62. Recent studies have revealed that the IgG glycosylation pattern has an impact upon the pathogenesis of RA. Under arthritic conditions, a decreased level of IgG sialylation results in enhanced osteoclastogenesis 61. As mentioned below, increased IgG sialylation inhibits the inflammatory response in mouse models of arthritis, supporting the importance of antibodies in RA pathogenesis 63.

Taken together, antibodies induce systemic and local bone loss both directly and indirectly by evoking inflammation. It is worth noting that the anti‐CD20 antibody treatment that reduces both the B cell number and antibody production protects effectively against bone erosion 64, 65. Thus, antibodies constitute the link between adaptive immunity and bone in arthritis. However, the extent to which antibodies contribute to such bone loss at present remains unclear. Elucidation of the relative contribution of autoantibodies to the local and systemic bone loss in association with RANKL will help to clarify further their pathogenic significance.

T cell subsets that help B cell antibody production in RA

B cell differentiation and antibody production require the help of T cells. Tfh cells are a specialized T cell subset responsible for B cell differentiation into plasma cells and memory B cells in lymph node follicles 66. In RA synovium, T–B cell aggregates are observed frequently. Using mass cytometry of RA synovial tissues, a recent study identified a new pathogenic T cell subset in RA synovium, named peripheral helper T (Tph) cells, that helps B cells to produce antibodies 67. Tph cells are CXCR5^–programmed cell death‐1 (PD‐1)^hi T cells and are observed more frequently in arthritis synovium than Tfh cells, which are CXCR5⁺ PD‐1^hi cells. Tph cells help with plasma cell differentiation by IL‐21 production and signalling lymphocyte activation molecule 5 (SLAM5) interaction 67. Further studies are needed to clarify how and where Tph cells are differentiated in arthritis.

IL‐21 and IL‐22 are produced not only by Tfh and Tph cells, but also Th17 cells. Indeed, a recent report showed that Th17 cells promote the inflammatory activity of autoantibodies and trigger the clinical onset of autoimmune arthritis in an IL‐21 and IL‐22 manner 63. IL‐21 and IL‐22 down‐regulate sialyltransferase activity in plasma cells and IgG sialylation is thus reduced, enhancing the activity of IgG. A decrease in the sialyltransferase activity and IgG sialylation parallels the clinical onset of RA, suggesting a key role of glycosylation and thus the activity of antibodies in the onset of RA. Anti‐IL‐17A antibodies have been shown to be less effective in established RA compared to other biologicals 68, 69, 70. It is possible that Th17 cells may contribute to RA pathogenesis in the early phase of arthritis rather than the chronic phase through the production of various Th17‐related cytokines.

Taken together, these studies indicate the importance of T cell B cell interaction and the production of antibodies in the pathogenesis of RA. Further characterization of the arthritogenic T cell subsets and elucidation of the molecular mechanism that governs their pathogenicity will be important for the understanding of and establishing the strategy for the treatment of RA in the future.

Conclusions and perspective

It is well established that T cells are important in the pathogenesis of RA. Recent studies have emphasized that the Th17–synovial fibroblast–RANKL axis primarily underlies the inflammatory bone loss in arthritis. Th17 cells converted from Foxp3⁺ T cells comprise the osteoclastogenic T cell subset, not only in arthritis but also in other pathological settings, such as periodontitis. These bone‐damaging T cells are entirely harmful in arthritis, while they may have developed in order to play an important role in host defence against oral bacteria, suggesting a beneficial side of T cell‐mediated bone loss.

The effects of antibodies on bone destruction have been determined recently, and it has been shown that T cells contribute to bone destruction by regulating both the quality and quantity of antibodies. Thus, the T cell B cell–antibody axis has emerged as an alternative important driver in bone destruction in arthritis. Further studies are needed to clarify the relative contribution of these two axes as well as their interaction in the structural damage in RA (Fig. 2).

Figure 2.

Mechanism of bone destruction in rheumatoid arthritis (RA). The aetiology of RA is not elucidated completely, but it is assumed that both genetic and environmental factors are involved. Autoantigens are presented by dendritic cells to T cells, which undergo differentiation into T helper (Th) cells that mediate autoimmune inflammation. Presentation of autoantigens by dendritic cells leads to the generation of various Th cells. Th17 cells up‐regulate receptor activator of nuclear factor kappa‐Β ligand (RANKL) expression on synovial fibroblasts and activate innate immune cells to produce proinflammatory cytokines [interleukin (IL)‐6, tumour necrosis factor (TNF)‐α, IL‐1] that activate osteoclast precursors and up‐regulate RANKL further on fibroblasts, resulting in the induction of osteoclastogeneisis. exFoxp3Th17 cells comprise a pathogenic Th17 cell subset. Tfh and Tph help B cells produce antibodies that promote RANKL‐dependent osteoclastogenesis both directly and indirectly by activating innate immune cells. Thus, the Th17–synovial fibroblast–RANKL axis and the T cell–B cell–antibody axis co‐operatively induce structural damage in RA.

The studies performed to date have advanced our understanding of the mechanism by which T cells induce structural damage in arthritis, but it remains unclear how autoantigen‐specific T cells are generated and contribute to local joint damage. Although we focused upon T cells and B cells in this review, other cell subsets residing in arthritic synovium, such as dendritic cells, innate lymphoid cells and synovial fibroblasts, also play a distinct role in the course of arthritis and may contribute to bone damage 71, 72, 73. However, it is not fully understood how these cells constitute the local intercellular network to induce RA pathogenesis. It will be important to address these questions for the elucidation of new links between autoimmunity and the structural damage in RA.

Disclosures

The authors declare no competing financial interests.