Autoimmunity to Citrullinated Proteins and the Initiation of Rheumatoid Arthritis

Abstract

Clinical manifestations of rheumatoid arthritis (RA), the second most common human autoimmune disease, are primarily focused on the joints, causing disability and requiring life-long treatment to ameliorate signs and symptoms. The etiology of RA is unknown; however, important discoveries in two areas have been made which provide hope that the causal mechanisms can be identified. First, the most severe form of this disease is associated with the presence of humoral and cellular autoimmunity to citrullinated proteins and peptides. Second, in the natural history of RA, autoimmunity to citrullinated antigens appears years prior to the onset of clinically apparent disease. Herein is described a model in which to consider how these two features are linked during very early disease development.

Rheumatoid Arthritis: Clinical and Autoimmune Features

Rheumatoid arthritis (RA) is the second most prevalent autoimmune disorder and presents clinically as a systemic, inflammatory autoimmune disease that most typically affects women (reviewed in [1]). RA exhibits a prevalence of 0.5-0.8 per 100 individuals in the general population, and an ~3-5-fold increase in first-degree relatives (FDRs) (reviewed in [2]). The term RA encompasses two major subsets of disease, seropositive and seronegative (reviewed in [3]). Based on clinical comparisons as well as studies of environmental and genetic associations, it is considered that the two forms of RA likely exhibit overlapping but individually distinctive pathogenic mechanisms [4;5].

RA is considered a prototypic polygenic disorder in which genetic risk factors interact with environmental exposures and, in the context of increasingly understood immune alterations, lead to the development of a severe autoimmune attack against self-tissues [6]. Consistent with this concept are observations that there are many genes which associate with the presence of clinically apparent RA, including high risk HLA-DRB1 alleles containing the shared epitope (SE) [7], and PTPN22 [8], as well as many additional non-HLA genes, a substantial number of whose products are components of immune-related signaling pathways (reviewed in [9]). In addition, there are several well-accepted environmental exposures which increase or decrease the risk of developing RA [10;11]. However, despite these insights into pathogenesis and recent improvements in therapy, only a relatively small number of patients achieve a durable drug free disease remission in standard clinical practice even with the best of current therapies [12].

Because of this situation, in order to achieve more dramatic changes in long term outcomes and prognosis, new approaches to the understanding and treatment of this disorder are needed. With that goal, this overview reviews the emerging evidence that RA does not begin as an articular disease but rather is initiated outside of the joints at sites of local inflammation, likely at mucosal surfaces. The underlying premise of this line of reasoning is that characterizing the relevant causal immune alterations at these sites should lead to the ability to treat this disease very early using therapies designed to beneficially modify these extra-articular processes, and perhaps even prevent the evolution of disease as it progresses from early asymptomatic stages through to the onset of clinically apparent deforming arthritis.

Key Features of Humoral and Cellular Autoimmunity in RA

Several types of autoantibodies are found in patients with clinically apparent RA. One is rheumatoid factor (RF), which is an autoantibody that recognizes the Fc domain of IgG (reviewed in [13]). Of note, in patients IgM, IgG and IgA RF isotypes can all be readily detected, the presence of which indicate that there are ongoing related but immunologically distinct processes [14]. The second type of RA-related autoantibody demonstrates reactivity to citrullinated proteins and peptides, and thus are broadly designated anti-citrullinated protein antibodies (ACPA) (reviewed in [15-17]). Peptidylcitrulline is generated through the conversion of arginine to citrulline by one of a family of enzymes designated peptidyl arginine deiminases (PADs) (reviewed in [18]). This process occurs as a part of normal intracellular homeostasis, but extracellular citrullination of proteins in tissues is a characteristic of many if not all inflammatory conditions. Notably, PADs exhibit different subcellular localizations and target specificities for RA-related antigens [19], and PAD4 is also itself a target of the citrullination process in a manner which changes its functional characteristics [20].

Thus, what is unique to RA is not the presence of citrullinated proteins but rather the development of an autoimmune response to this post-translational modification [6]. In patients with RA, reactivity against many citrullinated proteins has been detected, although much of the research focus has been on modifications of fibrinogen, vimentin, enolase, and Type II collagen. From these complex proteins, specific immunoreactive citrullinated peptides have been identified and applied to multiplex screening, demonstrating substantial heterogeneity across the RA population [21]. In one recent study 17 different subsets within RA patients were identified based on fine specificity profiles, with substantial differences across the subsets in the relationships to genetic and environmental factors [22]. Additionally, one particular reactivity to citrullinated alpha-enolase demonstrated a linkage to both smoking and DR4 SE status [23].

Importantly, a high proportion of B cells in the inflamed joint from patients with RA produce IgG ACPA at high levels in an apparently clonally restricted manner [24]. Finally, it is not only the specific epitope reactivity of ACPA that is important, as there are alterations in avidity [25;26] and the presence of G0 carbohydrates [27;28] as well as differential potential engagement of several pathogenic effector functions, including relative complement-fixing capacity [29], and the ability to engage Fc receptors [30] (reviewed in [31]).

The primary commercially-available tests for ACPAs are in ELISA format with cyclic citrullinated protein/peptide (CCP) or other similarly modified antigens as the substrate. Anti-CCP antibodies were first described as demonstrating a sensitivity of 60-70% with a specificity of ~98% [17;32]; however, the specificity falls modestly when other diseases with arthritis or pulmonary manifestations are evaluated (reviewed in [33]). Recently, a related type of antibody to carbamylated proteins has been described as highly specific for RA but of modestly lower sensitivity, wherein reactivity is found against the post-translational homocitrulline modification of lysine that is similar in structure to citrulline but with one additional carbon [34]. Although there is substantial overlap in reactivity to both targets within individual patients, there are individuals who lack anti-CCP antibodies but exhibit anti-carbamylated protein antibodies, and vice versa [34].

In addition to humoral autoimmunity to citrullinated antigens, patients with RA also demonstrate evidence of cellular immunity to citrullinated autoantigens. In one study, enhanced reactivity to a citrullinated vimentin peptide was detected in HLA-DRB1*0401 SE positive patients through the use of specific peptide:T cell receptor (TCR) tetramers and proliferative responses [35], and in another through enhanced proliferative responses [36]. In another report, T cell proliferative cytokine responses were studied, and secretion of IL-6, IL-17 and tumor necrosis factor (TNF)-alpha by CD4+ cells was found to be present in normal controls and patients with RA, while interferon (IFN) and IL10 were specific for RA, with citrullinated aggrecan generating the most robust responses [37].

A Detectable ACPA-Positive Preclinical Period Exists Prior to Onset of Arthritis in the Natural History of Rheumatoid Arthritis

Although one might consider that the onset of RA occurs around the time that clinically apparent arthritis appears, manifest by joint pain and swelling, it has been found instead that there is a prolonged period of highly specific RA-related autoimmunity in patients that occurs for 3-5, and in some instances more than 10, years prior to the onset of clinically apparent disease [reviewed in [38]]. This period is defined as the “preclinical” period of RA in subjects who eventually develop the disease [39], and is now considered to be an intrinsic part of the disease process itself (Figure 1). As described herein the study of this period of time is opening up new opportunities to understand the natural history and pathogenesis of this disease.

Figure 1.

Phases in the development of RA, with known genetic, clinical and biomarker alterations in patients below, and potential environmental exposures that influence the disease course at the top.

Many studies of such early RA pathogenesis have been performed using retrospective serum samples and data bases [reviewed in [38]]. However, because there are not yet sufficient studies of subjects who have transitioned in real time through these periods while under study, much of the current research is being performed on individuals who are at high risk for the future development of disease because they exhibit a pattern of autoantibodies and biomarkers that indicate they are at high-risk for the future development of clinically apparent disease (reviewed in [38]).

With regard to characteristics of subjects as they progress, there are several methods utilized and observations that have been made. First, one can assess individuals who exhibit only a family-based elevated risk of disease because they are FDRs, regardless of the presence or absence of autoantibodies in these at-risk family members. In this instance, there are increases of anti-CCP antibodies as well as multiple cytokines in this population [40]; additionally, ACPA of IgG, IgA and IgM isotypes, as well as IgM and IgA isotype RF are increased in unaffected members of multicase rheumatoid arthritis families [41], although the pattern of reactivity against individual antigens is more restricted than patients with clinically apparent RA [42;43]. When stratifying for the presence or absence of RA-related autoantibodies within FDR populations, a positive autoantibody status shows further preferential associations of cytokine elevations that are typically found in patients with active RA [44], and an increase in the number of individual ACPA peptide reactivities [43].

Finally, one can utilize samples serially collected in the past for other purposes to retrospectively evaluate biomarker evolution during the preclinical period. Using this approach, what is observed prior to the onset of clinically apparent arthritis is an increase in titer and expansion of epitope reactivity to stably encompass a wider variety of specific targets [45;46], an increase in the number of elevated cytokines [47-50], and avidity maturation of ACPA [26].

Notably, the finding that preclinical asymptomatic disease evolves into clinically apparent disease after a prolonged period of highly specific autoantibody positivity is a common feature of many autoimmune diseases. A similar type of progression occurs during the evolution of systemic lupus erythematosus [51], type 1 diabetes mellitus (T1DM) [52], celiac disease [53] and Hashimoto's thyroiditis [54]. What is unusual about RA, though, is that as opposed to other conditions in which disease-associated autoantibodies occur as an apparent manifestation of ongoing subclinical inflammation and organ-specific damage, such as in beta cells of the islets, is that the preclinical period in RA is apparently not characterized by detectable arthritis [55].

Are ACPA Pathogenic or Just Disease-Associated Biomarkers?

Observational studies in humans by their nature can only detect associations, and similarly to other diseases, there is the question of whether ACPA are pathogenic or simply markers of disease. Indeed, the finding that ACPA are present for years in the absence of arthritis suggests they alone are not alone sufficient to cause disease, although there are changes as noted above in their characteristics over time which may exceed a threshold of pathogenic potential. The question of pathogenic potential has been addressed primarily through murine models, wherein several studies have demonstrated that citrullinated autoantigen responses by both autoreactive B and T cells can lead to or amplify arthritis [56-60].

In addition, recently it has been shown that citrullination of mouse type II collagen eliminates the need to utilize adjuvant in the model in order to initiate arthritis in the collagen-induced arthritis model [61]. Finally, using purified human IgG reactive with mutated citrullinated vimentin, induction of osteopenia and bone resorption was demonstrated in vivo, and both binding to osteoclasts and osteoclastogenesis was seen in in vitro [62]. Thus, it appears that ACPA at least have the potential to be pathogenic, especially in the presence of pre-existing inflammation in the joint that leads to the elaboration of citrullinated proteins as “targets” of ACPA.

Where and How Do ACPA Originate at Extra-Articular Sites?

Even though the great majority of subjects with RA-related autoantibodies alone are asymptomatic, one might wonder whether there is subclinical inflammation. This subject has been addressed in several ways, but perhaps most importantly by performing synovial biopsy and magnetic resonance imaging in subjects who exhibit antibodies in the absence of clinically detectable arthritis. In that study, no histologic or MRI-based evidence of synovial inflammation was found in the knee [55].

In contrast to the lack of evidence of synovial inflammation, other studies strongly suggest that the RA-related autoimmunity and biomarker alterations may originate from a mucosal site. As one point in favor of this hypothesis, a substantial number of subjects manifest IgA ACPA and RF in the preclinical state [63-65]. Perhaps more importantly, though, alterations at specific mucosal sites, including the oral cavity, the lungs and intestines, suggest that microbial factors at these sites may affect the mucosal immune response and play an important role in the early pathogenesis of RA.

With regard to the oral cavity and periodontal region, there is an elevated occurrence of severe periodontitis in patients with RA, and patients with severe periodontitis exhibit higher disease activity scores [66]. Notably, patients with severe periodontitis manifest higher titers of IgG and IgM antibodies to Porphyromonas gingivalis, which is important because this organism exhibits PAD activity and citrullinates human fibrinogen and alpha-enolase in the presence of bacterial gingipains [67]. In addition, in subjects without RA but who exhibit RA-related autoantibodies in a pattern that is high-risk for future disease, there is a specific elevation of antibodies to Porphyromonas gingivalis but not to related strains [68]. In contrast, other studies have found that Prevotella and Leptotrichia, but not Porphyromonas gingivalis, are specifically expanded in the subgingival microbiota in patients with recent onset untreated RA [69].

Perhaps an even stronger case can be made for changes in the lung that promote the local development of ACPA. For instance, cigarette smoke exposure is highly associated with risk for the development of RA, especially in the presence of HLA expressing the DR4 SE [4;11]. In addition, although standard chest x-rays are typically normal, when assessed by sensitive high-resolution computed tomography, patients with early RA demonstrate a substantial number of pulmonary abnormalities [70]. The latter finding is particularly important because when individuals who do not have RA but exhibit high-risk RA-related autoantibodies, they similarly demonstrate a high rate of inflammatory airway disease that is not associated with smoke or other similar exposures [71]. Supporting the idea that there is local production of RA-related autoantibodies in the lung are studies of the sputa from subjects who exhibit family-based risk for the future development of RA, where RF and ACPA can be readily demonstrated in the sputa but not peripheral blood of a subset of individuals [72]. This is also notable because prior studies of subjects with RA and pulmonary disease had demonstrated local production of ACPA and RF by inducible bronchus associated lymphoid tissue [73].

Finally, the third mucosal site that has garnered attention is the intestine, where gut microbes are well known to be able to influence immune responses, and where experimental arthritis has been modulated by the exposure in the gut to a single organism [74]. Early studies have also suggested that change in the gut microbiota are associated with concurrent RA [75], and studies are being performed to evaluate whether similar changes might exist in subject manifesting high-risk autoantibodies. These studies, plus ongoing single cell analyses of human lymph node biopsies in patients and at-risk subjects [76], should continue to be informative with regard to the architecture of the ACPA response.

How Does Mucosal Autoimmunity Develop and then Transition to Local Inflammatory Arthritis?

With regard to the immune mechanisms that drive the loss of tolerance at mucosal sites, and also allow the evolution of disease from there to the joints, very little is known in patients; therefore, one can only speculate based on other similar situations and the little that is known about human at-risk subjects and very early arthritis. In the mucosal site, possible roles for citrullinated microbial antigens and molecular mimicry [67], Toll-like receptor (TLR) signals [reviewed in [77], and other innate immune activators and danger signals [78] exist. In addition, recent studies suggest that neutrophil extracellular traps (NETs) may play an important role in RA, as this type of neutrophil is found in tissue of patients with RA, circulates at levels that correlate with disease activity in patients, externalize citrullinated autoantigens, and are generated from neutrophils following activation by anti-citrullinated vimentin antibodies [79].

With regard to the transition from an autoimmune and inflammatory reaction at a mucosal site associated with circulating autoantibodies to then attack the joints, similarly little is known in human disease. Nevertheless, several potential mechanisms exist and have been explored either in patients or animal models (reviewed in [6;80]). One possibility, reflecting the presence of circulating immune complexes containing citrullinated fibrinogen in patients with RA [81], which notably can more effectively activate macrophages through the synergistic engagement of TLR4 and Fc receptors [82], is that immune complexes can enter the joint and deposit on cartilage and in synovial tissue. Another possibility is that transient expression of citrullinated antigens allows the ability of circulating autoantibodies [58;60] or T cells [83] to recognize and initiate or amplify inflammation. A third, potentially explaining the link between the early involvement of the lung with later arthritis, may be shared expression of citrullinated vimentin that is recognized by pathogenic autoantibodies at both sites [84].

Can One Use RA-Related Autoantibody Screening to Identify At-Risk Status and Halt Progression to Arthritis?

Because the pathway described out above strongly suggests that one might be able to interrupt the process, and there is a prolonged period of autoantibody positivity, one naturally could consider whether it is possible to screen for risk of developing RA and initiate a prevention therapy, perhaps focused on the mucosal site itself. Several recent publications have addressed this important topic [85;86], and interested readers are referred to these for further information.

Conclusion

RA is a potentially devastating chronic autoimmune disease that is treatable but is only moderately improved in most patients with current therapies. Substantial evidence from many investigators has been accumulated which demonstrates that the disease process begins years prior to the development of clinically apparent arthritis. The mechanisms by which autoantibodies and inflammatory biomarkers develop, in the absence of clinically and radiographically detectable arthritis, are under active investigation. This overview has summarized the evidence supporting the hypothesis that in a substantial number of patients, a dysbiosis at one or more mucosal sites leads to immune alterations and the initial break in self-tolerance to citrullinated autoantigens. This autoimmune process progresses to involve the joints secondarily. Although much work remains to further evaluate this concept, if it is validated there would be many important ramifications with regard to treatment approaches, therapeutic targets, and screening for future risk of disease.

Highlights.

• RA exhibits a long preclinical period with autoantibodies and other biomarkers

• Highly specific preclinical autoimmunity to citrullinated antigens develops early

• RA appears to be initiated by processes outside of the joints

• Inflammation at mucosal sites may drive the early phases of RA development