Current therapies for rheumatoid arthritis (RA) have substantially improved the outcome of the disease and the quality of life of these patients. However, achieving and maintaining long-term remission is still challenging. Moreover, a significant proportion of RA patients do not adequately respond to current therapies (1). One of the reasons for the imperfect management of RA is that a critical cell type, fibroblasts, is not adequately targeted.

RA is primarily inflammation of the synovium, cartilage degradation, and bone erosion. In a normal joint, synovium is a thin loosely organized relatively acellular connective tissue without a basal membrane. Instead, synovium is bordered by a lining layer which comprises cells of monocyte in origin (type A synoviocytes) and resident fibroblasts (previously referred as fibroblast-like synoviocytes, type B synoviocytes) (2, 3). The hallmark of RA pathology is hyperplasia of the synovium. In RA, the synovium grows enormously into a mass-like tissue containing a large number of immune and inflammatory cells. In several aspects, RA synovium is considered an analog of a tumor (4). Thus, RA synovium displays neovasculature forming pannus and invades into adjacent cartilage and bone leading to joint destruction (5). The tumor like feature of RA synovium is largely contributed by fibroblasts (4, 6). RA fibroblasts proliferate and are resistant to apoptosis. They build stromal network which harbors immune and inflammatory cells. Moreover, fibroblasts actively interact with immune and inflammatory cells leading to persistent inflammation of the synovium; support the formation of ectopic lymphoid follicles (7). In addition, fibroblasts are effector cells producing inflammatory cytokines participating inflammatory process and matrix metalloproteinase (MMP) and directly invading articular cartilage and subchondral bone. It is well-established that infiltration of immune and inflammatory cells in RA synovium is histologically heterogeneous between individual patients ranging from fully organized lymphoid structures to diffusely distributed lymphoid and myeloid cells throughout the synovium, and to scares of immune and inflammatory cells (pauci-immune). In contrast to the highly variable presence of immune and inflammatory cells, fibroblasts are invariably present in all pathotypes of synovitis (8). Thus, fibroblasts are indispensable in formation of RA synovitis and actively contribute to cartilage and bone destruction. Thereby, a therapeutic strategy directed at modulation of fibroblast function or ablation of fibroblasts has long been proposed (9–12). However, fibroblast targeted therapies have not yet been developed for clinical practice owing to poorly understood molecular mechanisms that drive synovial fibroblast behavior in RA. Recently, more insightful understanding of RA fibroblasts coupled by advanced technology enable us to explore fibroblast targeted therapies for RA. Readers are directed to excellent comprehensive reviews for molecular and cell biology of fibroblasts, their interplay with immune and inflammatory cells in the synovium, and discussion on restoration of synovial homeostasis in RA (12–16). In this article, I shall highlight several different strategies selectively targeting fibroblasts, which have been explored in preclinical arthritis models and/or in early phase of clinical investigations.

The number of fibroblasts in RA synovium increases substantially in both lining and sublining regions. The expansion of fibroblasts is the result of increased proliferation and decreased cell death (17–19). Earlier studies suggest that fibroblasts arise from local epithelial to mesenchymal transition and differentiation from pluripotent mesenchymal stem cells (20–22). There is little evidence indicating in situ proliferation of fibroblasts in the lining region. In contrast, recent studies have demonstrated proliferation of sublining fibroblasts (23–25). The gradient reduction of Thy1 (CD90) and augmentation of proteoglycan (PRG)-4 from sublining to lining fibroblasts suggest that divided fibroblasts migrate from sublining to lining region (24, 26). The origin of sublining fibroblasts is not clear. A recent study suggests that circulating preinflammatory mesenchymal (PRIME) cells, which bear hallmarks of synovial sublining fibroblasts, migrate into synovium during RA flare (27). It is conceivable that PRIME cells further differentiate into sublining fibroblasts upon interaction with endothelial cells (15, 26, 27). Several studies have shown that RA fibroblasts display aberrant apoptosis at several levels (28) and also probably due to increased autophagy (17, 19).

Several subpopulations of RA synovial fibroblasts have been described based on their distinct profiles of gene and protein expression. It is likely that these subpopulations of fibroblasts function differently, which are related to their locations in the compartments and interaction with other cells in the synovium (Figure 1) (23, 25, 29, 30). The strategy at ablation of fibroblasts has been hampered due to the lack of specific cell surface targets. Recently, several surface proteins have been described for identifying RA synovial fibroblast populations. Namely, CD55, podoplanin, and protein tyrosine phosphatase receptor sigma (PTPRS) are expressed by lining fibroblasts, while Thy1 (CD90), and CD248 are expressed on sublining fibroblasts (Figure 1).

The ideal cell surface targets should be expressed by all subpopulations of fibroblasts. CD34, cadherin-11, and fibroblast activation protein (FAP) can be detected in both lining and sublining fibroblasts (Figure 1) (23–25, 29–32). CD34, expressed by hematopoietic stem cells, is not a specific marker for fibroblasts. CDH11 is critical for lining layer formation and CDH11 deficient mice are resistant to arthritis induction (33); and has been explored as a therapeutic target. However, in a phase II clinical trial, an anti-CDH11 monoclonal antibody failed to show efficacy in RA patients with inadequate response to tumor necrosis factor (TNF) inhibitors (34).

In RA synovium, the expanded fibroblasts in the lining and sublining layers reside in different anatomical regions and execute related but different activities (29), and phenotypically are distinct subpopulations. However, they are likely derived from same origin. Perivascular fibroblasts receive signals from vascular endothelial cells to proliferate and differentiate and migrate to lining layer. This process is evident by gradient increase of PRG4 but decrease of Thy1 expression from sublining to lining fibroblasts (24). Among the signals conduced between endothelial cells and fibroblasts, NOTCH signaling, particularly NOTCH3 is critical. NOTCH3 is highly expressed by perivascular fibroblasts. In an organoid culture system, silencing NOTCH3 gene expression by siRNA diminished endothelium induced fibroblast expansion and gradient cell surface Thy1 and PRG4 expression. The critical role of NOTCH signaling has been tested in animal models of arthritis. In a rat CIA model, a NOTCH1 and NOTCH3 inhibitor (LY411575) showed therapeutic effect of arthritis, presumably by inducing fibroblast death since LY411575 is able to induce death of human synovium derived fibroblast cell line, MH7A in vitro (69). Using monoclonal antibodies, Wei et al. (24) demonstrated that anti-NOTCH3 monoclonal antibody has profound therapeutic effect in arthritis compared with moderate effect by monoclonal antibody to NOTCH1 (Figure 2B). Furthermore, NOTCH3 knockout mice had diminished arthritis induced by KBxN serum transfer compared with wild type animals (24). NOTCH3 gene deleted mice develop a normal joint structure suggests that NOTCH3 targeted therapy is safe for treating arthritis (24, 26). Further development for RA therapy by targeting NOTCH3 signaling is warrant.

Several modalities with various mechanisms of action are able to inhibit proliferation, migration and/or invasion of RA synovial fibroblasts (Figure 2C) and have been investigated for development of fibroblast targeted therapy.

Promotion of cellular senescence has been proposed and actively explored as an approach to cancer therapy (89, 90). RA synovial fibroblasts display some features of tumor cells (4), thereby, pro-senescence in these fibroblasts is potentially therapeutic for RA (Figure 2D). Indeed, Montero-Melendez et al. (91) demonstrated that induction of synovial fibroblast senescence via activation of a G-protein coupled receptor, melanocortin type 1 receptor (MC1R) can ameliorate an experimental arthritis. MC1R is highly expressed by RA fibroblasts. Activation of MC1R using a selective agonist, BMS-470539, but not a non-selective agonist, induced senescence in RA fibroblasts. The specific role of MC1R in this senescence was further confirmed in primary mouse synovial fibroblasts that MC1R knockout abrogate the effect of BMS-470539. Furthermore, this MC1R activation induced fibroblast senescence involves phosphorylation of extracellular signal-regulated protein kinase (ERK) 1/2, a mitogen-activated protein kinase (MAPK) family protein, which have been shown to promote cellular senescence (92). Importantly, BMS-470539 was able to suppress KBxN serum transfer induced arthritis in mice and the therapeutic effect was associated with in vivo synovial fibroblast senescence. The therapeutic effect of BMS-470539 was countered by senolytic drugs (91).

These results are encouraging although further confirmatory evaluation in chronic arthritis models would be required for long-term efficacy and possible adverse effects associated with pro-senescence. Senescence of joint cells, especially chondrocytes is increased in osteoarthritis (OA) (93). It is to be determined if therapy with pro-senescence in fibroblasts and off-targeted senescence of other joint cell types will result in OA of the joint. Another concern associated with pro-senescence as a therapy is the removal of senescent cells from the tissue. Insufficient elimination of the senescent cells after induction of pro-senescence might be detrimental and senolysis may be required. Interestingly, local clearance of senescent cells is beneficial in post-traumatic OA model (94).

RA synovial fibroblasts show reduced rate of apoptosis. Thereby, inducing cell death is an attractive approach to ablate synovial fibroblasts for RA therapy (Figure 2E).

The tumor-like behaviors of RA fibroblasts such as activation, migration and invasion are associated with increased cell metabolism. There is ample evidence to indicate that glucose metabolism plays an important role. Therefore, interruption of glucose metabolism of fibroblasts is a plausible approach to novel therapy of RA. As has been demonstrated in several murine models of RA, inhibition of glycolysis can significantly reduce severity of arthritis (104, 105). However, global inhibition of glycolysis is not desirable as a therapy. Fibroblast specific inhibitor of glucose metabolism is required. Among all the glycolytic enzymes involved, hexokinase 2 (HK2) may be a relatively selective target (Figure 2F). HK2 is an inducible isoform that is selectively highly expressed in skeletal and cardiac muscles and adipose tissue (106), and is highly expressed in RA but little in OA synovial tissue (107). Silencing HK2 resulted in reduction of migration and invasion of RA fibroblasts in vitro. Ablation of HK2 significantly reduced severity of arthritis and bone and cartilage damage in KBxN serum transfer induced arthritis (107).

Currently, a HK2 specific inhibitor is not available, but a glucose analog, 2-deoxy-D-glucose (2-DG) has been extensively evaluated as an agent for inducing cancer cell death and cancer therapy (108). Administration of 2-DG reduced the severity of spontaneous arthritis in KBxN mice. However, the therapeutic effect of 2-DG was attributed to inhibition of glucose metabolism in T follicular helper (Tfh) cells. The effect of 2-DG on fibroblast activity remains to be possible, but unfortunately this was not investigated in this work (109). This result on T cells is in consistence with the previous findings that loss of HK2 mildly reduced colitis in interleukin-10 deficient mice and ovalbumin induced airway inflammation (110). Furthermore, HK2 is not essentially required for T cell glucose metabolism in vitro and loss of HK2 did not impair clearance of lymphocytic choriomeningitis virus infection, suggesting inhibiting HK2 activity would have less immune compromise (110). Further studies are required to clarify the targeted cell types of 2-DG in RA models and a fibroblast specific HK2 inhibition strategy is required to develop.

Janus kinase (JAK) inhibitors are approved therapy for RA. JAK-signal transducer and activator of transcription (STAT) pathway is active in RA synovial fibroblasts in response to cytokine stimulation (111). As expected, JAK inhibitors are able to block cytokine mediated inflammatory production by fibroblasts; these include stimulation by cytokines such as oncostatin M, IL-6, TNF, IL-1, and interferon-γ (111, 112).

Current JAK inhibitors are not fibroblast specific. Among the JAK inhibitors, peficitinib was able to suppress synovial fibroblast migration in vitro and may induce fibroblast apoptosis suggesting that peficitinib may have advantage over other JAK inhibitors on targeting fibroblasts (Figure 2G) (113, 114). Since high dose of peficitinib is tolerated, it may in particular be useful for treating RA patients with pauci-immune pathotype which is poorly responsive to conventional DMARDs or TNF inhibitors (115). Selective JAK inhibitors for JAK-STAT pathway in fibroblasts are to be uncovered for this purpose. On the other hand, synovial fibroblasts are major source of IL-6 production in RA joint (25, 116). Moreover, IL-6 acts in an autocrine amplification mechanism on activation of fibroblasts (116). Therefore, JAK inhibitors with profound activity to block IL-6 signaling may have profound impact on fibroblast activation in RA.

Pirfenidone [5-methyl-1-phenyl-2-(1H)-pyridone], one of the heterocycle pyridones, is approved for treating idiopathic pulmonary fibrosis. Pirfenidone was first developed for anti-pyretic and analgesic use, but was found to have anti-fibrotic effect. Pirfenidone has pleiotropic effects including inhibition of transforming growth factor (TGF)-β-mediated proliferation of fibroblasts. In cultured synovial fibroblasts, pirfenidone can reduce the expression of intercellular adhesion molecule-1. Pirfenidone can also suppress TGF-β induced collagen type I production and inhibit myofibroblasts activity to reduce extracellular matrix deposition (117). In a rat CIA model, pirfenidone was shown to ameliorate arthritis and reduction of MMP3 and vascular endothelial growth factor (VEGF) expression in the joint (118). In Simian Vacuolating Virus 40 large T antigen transformed human synovial fibroblast cell line, MH7A cells, pirfenidone significantly reduced TNF-induced production of inflammatory cytokines, VEGF, and MMPs. Moreover, pirfenidone significantly inhibited VEGF by vascular endothelial cell line, EA.hy926 cells (118). Further studies are required to delineate whether these effects of pirfenidone fibroblast selective.

Another anti-fibrotic drug, nintedanib, a potent tyrosin kinase inhibitor that inhibits growth factor stimulated migration and proliferation of fibroblasts and TGF-β-induced transformation to myofibroblasts. In a zymosan-induced arthritis and pulmonary fibrosis model in SKG mice, treatment at early stage of disease, nintedanib was able to significantly reduce severity of arthritis but has no effect on progression of lung fibrosis. In contrast, treatment started when pulmonary fibrosis established, nintedanib was able to reduce pulmonary fibrosis but had no effect on established arthritis (119). Nintedanib is approved for treating idiopathic pulmonary fibrosis. There are case reports in the literature that nintedanib has been tried in RA related interstitial lung disease and showed benefit (120). It would be interesting to investigate further how ninetedanib will affect synovial fibroblasts in RA patients and whether ninetedanib will show anti-arthritic effect.

RA synovial fibroblasts are valid therapeutic target candidates and current strategies for intervention have been explored (Table 1 and Figure 2). Especially, fibroblasts are not professional immune cells for host defense although they display features of inflammatory cells and can present self-antigens during the disease process of RA. Therapies directed at suppression of the function or ablation of synovial fibroblasts will be less likely to cause immunosuppression. Therefore, development of fibroblast targeted therapy will be particularly practical for combination therapy with currently available immune component directed therapies. Another concern with regard to fibroblast ablation therapy is impairment of wound healing since fibroblasts are critical for tissue repair. However, in preclinical tumor models, ablation of fibroblasts did not negatively affect wound healing although these need to be further evaluated in human studies (51, 52, 59).

In addition to the strategies discussed above, mitogen-activated protein kinase (MAPK) are highly activated in RA synovial fibroblasts. However, targeting MAPK pathway for therapy has not be successful in clinical trials (discussed in details by Nygaard and Firestein (12). Also, targeting the imprinted signature of RA synovial fibroblasts is another attractive approach. For example, modulation of histone-modifying enzymes may lead to remodeling and restoration of the homeostasis of RA synovial fibroblasts [reviewed by (12)].

Clinically, RA patients with pauci-immune synovial pathotypes respond poorly to current DMARDs which are mainly directed to immune suppression. Presumably fibroblasts are the predominant effector cell types mediating the disease process in these patients (8, 115). This RA subpopulation will better serve for testing efficacy of fibroblast targeted therapies in clinical studies. Peficitinib, one of the newer JAK inhibitors was shown to preferentially act on JAK-STAT pathway in RA synovial fibroblasts and hence would be a good candidate to try on this RA subpopulation (113, 114).

The author confirms being the sole contributor of this work and has approved it for publication.

C-QC's work was supported by an Innovative Award from American College of Rheumatology Research Foundation and by a VA Merit Review grant (I01BX005195).

The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.