Background
Rheumatoid arthritis (RA) is a common chronic autoimmune disease that leads to significant morbidity and mortality. The prevalence of this disease is between 0.5 and 1% in the USA, with greater risk of incidence among women [
1]. The molecular cues that trigger RA are complex and likely require a combination of genetic and environmental factors. Ultimately, the loss of self-tolerance to self-proteins, typically citrullinated-proteins, results in disease progression through inflammation, synovial hyperplasia, and osteoclast activation [
2]. These processes lead to the recognizable hallmarks of the disease, including joint swelling, pain, and dysfunction. Although joint involvement is the major clinical manifestation of the autoimmune process, RA patients suffer from a number of co-morbidities including increased risk of cardiovascular disease even when accounting for the increased inflammatory state [
3,
4]. Significant strides have been made in the treatment of RA. Standard therapy for RA includes methotrexate and one of several disease-modifying anti-rheumatic drugs (DMARDs) that target the key inflammatory molecules that promote RA, including TNFα (infliximab, etanercept) and IL6 (tocilizumab). These treatments allow many patients to achieve disease-free status, but their use is associated with significant side effects and these biologic agents do not confer sustained remission after drug cessation [
1,
5]. Thus, there remains a clinical need for improved understanding of novel targets for drugs in the process of RA development.
The Wnt family proteins are developmental regulators that are becoming appreciated for their diverse roles in adult physiology. Wnt5a is the archetypal non-canonical Wnt, the subset of Wnts whose functions are independent of beta-catenin [
6]. In post-natal tissue, the functions of Wnt5a are diverse and include modulation of cell fate decisions and inflammation. Previous work has established Wnt5a as a pro-inflammatory molecule in a number of settings including vascular dysfunction [
7,
8], peripheral arterial disease [
9], and cardiometabolic disease [
10‐
12]. Wnt5a has also been demonstrated to promote expression of various cytokines that may contribute to RA progression [
13‐
16]. However, studies on the role of Wnt5a in immune regulation can be confounded by the likely presence of TLR4 agonists in commercial Wnt5a protein preparations [
17]. Thus, analyses of Wnt5a function using mouse genetic models are warranted as they are not subject to this potential confounder.
Wnt5a is increased in the synovium in patients with RA and osteoarthritis (OA) [
18], and inhibition of Wnt5a is reported to block the activation of cultured synovial fibroblasts from RA patients [
19]. In addition, the delivery of a soluble form of receptor-tyrosine orphan receptor 2 (ROR2), a candidate decoy receptor for Wnt5a, leads to reduced radiographic severity in mice undergoing collagen-induced arthritis (CIA) [
20]. These data underscore the potential clinical value of Wnt5a in the pathogenesis of RA. However, there is as yet no direct genetic evidence to support that endogenous Wnt5a plays a contributory role in the development of RA.
In the present study, we used inducible Wnt5a knockout (Wnt5a cKO) and littermate control mice in the K/BxN serum transfer-induced arthritis (STIA) model. Wnt5a cKO mice were protected from development of RA, both in terms of ankle thickness and other markers of disease severity. Histologic examination demonstrated reduced inflammation in the Wnt5a cKO, including reduced PMN infiltrate into the inflamed joints. Less osteoclast activity was also observed in the arthritic Wnt5a cKO joints compared to the controls. Mechanistically, this may result from decreases in osteoclast fusion as Wnt5a promoted osteoclast fusion in an in vitro culture of bone marrow-derived macrophage (BMDM) differentiation. Wnt5a increased the expression of critical molecules for osteoclast differentiation and fusion, specifically dendrocyte-expressed seven transmembrane protein (DCSTAMP) and matrix metalloproteinase 9 (MMP9). These data suggest Wnt5a signaling as a modulator of inflammation and osteoclast fusion could represent a component of disease activity in RA patients.
Discussion
Wnt5a is elevated in the synovium of RA patients [
18], and enhanced Wnt5a secretion by fibroblast-like synoviocytes has been attributed to their persistent activation [
19]. Thus, we sought to provide genetic evidence for a causal link between Wnt5a and RA development. The present study utilized a mouse model in which ablation of Wnt5a is inducible, thereby avoiding the embryonic lethality of the conventional knockout strain [
21]. This inducible ablation model also permits the ablation of Wnt5a proximal to the induction of arthritis, thereby eliminating the confounding effects of Wnt5a in bone development [
20] (Additional file
1: Figure S1-S3). Collectively, this study shows that mice deficient for Wnt5a are resistant to development of RA-like disease in the STIA model.
The acute genetic ablation of Wnt5a in the STIA model led to reductions in some markers of inflammation. Based upon histologic analysis, Wnt5a cKO mice exhibited statistically significant reductions in overall inflammatory status, extra-articular inflammation and PMN cell infiltration (Fig.
2). In this regard, it is well-appreciated that neutrophils are important mediators in the development of arthritis [
31,
32]. Failure to recruit neutrophils correlates with reduction in the overall progression of disease severity in other models of arthritis [
33]. Further, since Wnt5a is described to be chemotactic to neutrophils [
34], it is possible that this is a significant component of the mechanism by which Wnt5a cKO mice are protected from the development of STIA.
Using the Wnt5a cKO mouse model, we recently reported that Wnt5a ablation protects against adipose tissue inflammation and systemic metabolic dysfunction that is associated with diminished levels of monocyte chemoattractant protein 1 (MCP-1), IL6 and TNFα in the adipose tissue [
10]. Although these cytokines are shared in the pathogenesis of metabolic dysfunction and RA [
1,
5], only trends towards reduction of
TNFα,
IL6, and
IL1β transcripts were observed in the arthritic paws from the Wnt5a cKO mice (Additional file
1: Figure S6) as was a trend toward diminished lymphocyte infiltration in the histological sections (Additional file
1: Figure S5). In total, these data are indicative that Wnt5a is one of many modulators of inflammation in the RA model, which can be compensated for by other inflammatory molecules. However, there appears to be a modest and appreciable protection from the inflammation in RA conferred by Wnt5a deficiency. Further, we acknowledge the possibility that Wnt5a has a more profound impact on inflammation in the STIA model at other time points.
In contrast to the increase in Wnt5a expression in the synovium of RA patients [
18], we found a reduction in the amount of
Wnt5a in the joints from mice undergoing STIA. Beyond species differences, there are a number of possible explanations for this discrepancy. First, there was a concurrent increase in the Wnt5a co-receptor
ROR2 (Additional file
1: Figure S1), indicating that reduced amounts of Wnt5a might be sufficient to drive RA pathology. In this regard, Wnt5a is highly glycosylated and is believed to exhibit its effects through short-range paracrine interactions [
35,
36]. Thus, localized changes in Wnt5a may be sufficient to modulate RA development, particularly when paired with a concomitant increase in
ROR2. It should be also noted that our analysis was performed using the entire joint; including fibroblast-like synoviocytes, bone, bone marrow, and extra-articular inflammatory tissue. Indeed, although Wnt5a is reported to be among the most highly expressed Wnt homologs expressed in bone, immunohistochemical analysis demonstrated that it is predominantly expressed by osteoblast/osteoclasts and not the osteocyte [
20]. Thus, further exploration of Wnt5a in the arthritic space, beyond the synovial tissue, is warranted.
Wnt5a has been functionally implicated in the development of RA by a study that delivered a soluble version of ROR2, a putative Wnt5a receptor, to mice in the collagen-induced arthritis (CIA) model [
20]. Although mice receiving the soluble ROR2 were not protected from development of ankle thickness or clinical score, they did exhibit reduced radiographic damage [
20], consistent with a reduction in osteoclast activity. Our work supports and extends these observations in several ways. First, using the Wnt5a cKO mouse model we provide direct mouse genetic evidence to support that Wnt5a plays a role in the pathogenesis of RA. Second, the inducible Wnt5a cKO mice were used to ablate Wnt5a immediately prior to the induction of the STIA model, which eliminates the confounding effects of Wnt5a-deficiency on long-term bone development [
20,
37‐
39]. Third, we document an effect of Wnt5a in the STIA model, which recapitulates the initiating events of the disease with a higher penetrance than the CIA model. Notably, the STIA model also avoids the use of adjuvant that can potentially act as a confounder. Finally, our study provides mechanistic detail about the role of Wnt5a in osteoclast fusion.
Macrophage cell fusion is required for the formation of multinucleated osteoclasts that function in the catabolism of bone matrix. Osteoclast formation from myeloid precursors is regulated by a sequence of molecular events involving the cell surface receptors including DCSTAMP, CD36, CD47, TREM2 and E-cadherin [
40]. Additionally, osteoclast-specific genes including MMP9, CTSK and TRAP [
41], are also required for cell fusion during osteoclastogenesis. Although it has been implicated in differentiation of multiple cell types, Wnt5a is not widely appreciated to promote cellular fusion. In the present study, we find that exogenous Wnt5a added to fusing BMDM enhanced their fusion into osteoclasts and upregulated
DCSTAMP (Fig.
4), a cell surface molecule essential for formation of osteoclasts in vivo [
30].
MMP9 was also found to be upregulated by Wnt5a in fusing osteoclasts (Fig.
4c). MMP9 has been shown to be critical for the development of arthritis as MMP9 KO mice develop less inflammation and joint destruction in the serum transfer model [
42]. MMP9 has also been shown to modulate several aspects of osteoclast function, including enhancement of cell migration, potentiation of cell fusion, and increased bone metabolism [
43‐
46]. Altogether, our data indicate that Wnt5a promotes osteoclast formation, and suggests a possible role of Wnt5a in development of other bone resorption disorders.
Rheumatoid arthritis is associated with cardiometabolic diseases [
3,
4], and a growing number of animal studies have investigated this linkage. Assessment of arthritis development in ApoE-deficient mice using the CIA model has led to conflicting reports on the consequences of hypercholesterolemia on promoting arthritis [
47,
48]. However, using a modified chronic STIA model, more recent evidence demonstrated that ApoE-deficient mice display aggravated development of arthritis but not an increased plaque burden [
49]. Conversely, when placed on an atherogenic diet, K/BxA
g7 mice display increased plaque burden [
50]. The present study suggests that Wnt5a might be a possible common mechanism linking both RA and cardiovascular disease. It has been shown that Wnt5a contributes to increased cardiac dysfunction following ischemia/reperfusion injury [
11] and to impaired revascularization in a model of peripheral arterial disease [
9]. Moreover, previous research demonstrated that Wnt5a impairs metabolic function at least in part through its role in exacerbating adipose tissue inflammation in the obese state [
10,
12]. Wnt5a is markedly upregulated in visceral adipose tissue (VAT) and correlates strongly with indices of inflammation and diabetes mellitus [
10,
51]. Despite the existence of “rheumatoid cachexia” in late-stage disease, large scale meta-analysis reveals that obesity is a risk factor for RA that is associated with disease severity [
52,
53]. It has been proposed that adipokines, cytokines secreted by adipose tissue, are molecular drivers of both diseases [
52,
54]. RA patients exhibit elevated adipokine levels, including leptin, adiponectin, and resistin, in serum or synovial fluid compared to healthy patients [
52,
55‐
57], and these factors frequently correlate with disease progression [
58,
59]. Collectively, these observations warrant further study to establish whether Wnt5a functions as a putative nexus target in RA and cardiometabolic diseases.
Recent advances in DMARDs have revolutionized the treatment landscape for RA patients. However, cessation of treatment inevitably leads to relapse. Furthermore, a subset of patients remains refractory to these therapeutics agents. Current treatment modalities focus on inhibition of the inflammatory process and do not directly target osteoclast activity. Based upon the multi-modal influence of Wnt5a on RA etiology, specifically as a modulator of osteoclast activity and inflammation, it may represent a novel target that has potential to modulate both arms of RA disease progression.