Uncover Microbiome's Role in RA Progression

Emerging evidence suggests that changes in microbiota begin prior to clinical disease and further change with treatment of RA. Although studies identify numerous microbial candidates at oral and intestinal sites, some patterns emerge (Figure 1). However, data to date focus primarily on bacteria, with little information regarding viruses and fungi; thus, we focus our Review on the bacterial microbiome in RA, with brief mention of data associating Candida species and viruses.

Microbial and immune changes across the stages of rheumatoid arthritis. Changes in the composition of gut and oral microbiota as well as systemic responses observed in three stages of rheumatoid arthritis (RA): preclinical (Pre-RA), new-onset and early RA, and established RA. Pg, P. gingivalis; IPA, indole-3-propionic acid; IAA, indole-3-acetic acid; DMARD, disease-modifying antirheumatic drug. Black text indicates increased markers, blue text indicates decreased markers, and green text indicates markers that show both increases and decreases.

Bacterial dynamics in individuals at risk for RA. Most studies define “at risk for RA” as individuals who are serum ACPA+ but without typical clinical features of RA such as synovitis on exam. In such individuals, Prevotella spp., particularly Segatella (formerly Prevotella) copri, are consistently enriched in the stool (11, 12). Host genetics further influence gut microbiota composition in this context, as supported by the finding that a polygenic risk score for RA positively associated with the presence of Prevotella spp., in the absence of clinical symptoms (13). This finding suggests a bidirectional relationship in which RA-related genetic variants may shape the gut microbiome, while microbial dysbiosis, in turn, may contribute to immune activation and disease progression. Supporting this conclusion, a recent prospective analysis of serum ACPA+ individuals with musculoskeletal symptoms who ultimately developed RA revealed significant fluctuations in the microbial community composition, especially Prevotella spp., that occurred approximately 10 months before RA onset. This finding is in comparison with a control population of individuals with serum ACPA+ and musculoskeletal symptoms who did not develop RA during a 12-month follow-up period (12). This instability may signify a late microbial shift that contributes to RA progression.

Beyond gut microbiota, numerous studies have reported changes in the oral microbiomes of individuals at risk for RA. An increased abundance of Prevotella and Veillonella has been observed (14, 15). However, the presence of oral P. gingivalis is inconsistent, with some studies reporting an increase (16) and others a decrease (15, 17) in prevalence, although these discrepancies may be influenced by variations in study cohorts, collection methods, geography, genetic predisposition, and other contributing factors.

Bacterial changes in individuals with established RA. Cross-sectional studies evaluating the oral and gut microbiomes from individuals with RA have associated numerous bacteria with disease. Machine learning approaches have demonstrated that fecal, dental, and oral microbiomes from individuals with RA can distinguish disease with high accuracy (AUCs of 0.93, 0.87, and 0.81, respectively) (18), although these findings require validation in independent cohorts.

Like individuals at risk for RA, Prevotella spp., particularly S. copri, is consistently detected in the stool from individuals with established disease (12, 19, 20). Notably, it is also found in synovial tissue, but its role in pathophysiology at this site is unknown (20, 21). The pathogenic potential of S. copri is associated with genomic adaptations, particularly the acquisition of conjugative transposons that enhance its ability to modulate host innate immune responses (22).

Similarly, products from Fusobacterium nucleatum and Eggerthella lenta, enriched in the stool from patients with RA, have been identified in the synovial fluid of patients with RA and associated with markers of inflammation and measures of disease severity (20, 23–25). However, whether and how the presence of bacteria in the synovium and synovial fluid affect RA is unknown. Expansions of Collinsella (stool), Escherichia coli (stool), and Lactobacillus spp., particularly L. salivarius (stool, oral) also associate with RA and clinical factors such as autoantibodies and systemic inflammatory cytokines (18, 20, 26–28). Conversely, some Lactobacillus spp. like L. salivarius strain UCC118 and L. plantarum WCFS1 (stool), as well as Parabacteroides distasonis (stool), are reduced in abundance and may be protective in RA through associated antiinflammatory mechanisms such as inducing IL-10 or suppressing Th17 cells (29, 30). The abundance of periodontal and oral bacteria, like Porphyromonas gingivalis, varies across studies, with some reporting an increase (31), others a decrease (17, 18), and some showing no significant change (32–34). The abundance of P. gingivalis may be more strongly linked to periodontitis severity rather than being specific to RA (33).

The tonsillar microbiota in individuals with RA is enriched for pathogenic species such as S. pyogenes, S. dysgalactiae, and S. agalactiae, with experimental models demonstrating their role in exacerbating arthritis severity by promoting immune cell activation and inflammatory responses. Conversely, the depletion of protective species such as S. salivarius in patients with RA contributes to microbial imbalance and immune dysregulation (35). The loss of S. salivarius is particularly significant, as this species produces salivaricins, which are lantibiotic peptides with immunomodulatory properties (36, 37) that inhibit T follicular helper (Tfh) cell differentiation and IL-21 production, thereby reducing antibody production and systemic autoimmunity (37). Notably, supplementation with S. salivarius or its salivaricins in murine models effectively attenuates arthritis progression (35, 37).

Bacterial changes in response to treatment. RA treatments also actively reshape the gut microbiome, leading to microbial shifts that may influence disease progression and treatment outcomes (10, 38). Immunomodulatory therapies, including DMARDs and biologics, have been shown to partially restore gut dysbiosis in patients with RA, with microbial shifts correlating with improved clinical outcomes and reduced disease activity. However, these effects are often incomplete, as patients with longstanding RA retain a distinct microbiota composition compared to healthy individuals despite prolonged treatment (18).

Among conventional DMARDs, methotrexate (MTX) treatment did not significantly alter overall gut microbiota composition (39, 40) but induced significant shifts in gene family abundance, particularly in pathways related to pyrimidine synthesis, protein synthesis, and ABC transporters (40). Hydroxychloroquine may similarly contribute to microbiota modulation by suppressing proinflammatory bacterial overgrowth (41), although data on its role in RA microbiome effects are scarce. Biologic therapies, including TNF inhibitors such as etanercept, also influence gut microbial composition. TNF inhibitor treatment in patients with RA partially restores gut microbiota composition by increasing beneficial bacterial taxa and reducing dysbiosis-associated changes, with notable modulation of Euryarchaeota, which correlates with disease severity (39).

Fungal and viral microbiomes in RA. The fungal and viral microbiomes in RA have been studied less extensively compared with bacteria. Fungi constitute only a minor fraction of the intestinal microbiota — typically accounting for 0.1% to 1.0% (42, 43). Among gut fungi, Candida albicans is a predominant member of the intestinal mycobiome, recognized both as a commensal organism and an opportunistic pathogen (42, 43). Relative abundances of Candida species are increased in the fecal microbiota of individuals with RA (44, 45), and colonization of mice by C. albicans can worsen disease in murine arthritis (46, 47). Notably, β-glucan, a structural component of the C. albicans cell wall, acts as an immunological adjuvant capable of promoting autoimmune arthritis in mice (48–50).

The intestinal virome in individuals at risk for developing RA compared with controls is enriched with Streptococcaceae, Bacteroidaceae, and Lachnospiraceae phages, which associated with cyclic citrullinated peptide (CCP) positivity (51). Interestingly, a phage-encoded phosphonate phosphodiesterase that associated with CCP-positive at-risk individuals (51), is an ortholog of a gene that associated with predicted response to MTX therapy in another study (52). Within individuals with established RA, crAss-like phages are significantly reduced compared with controls (53), although when examined by treatment status, family Phycodnaviridae were significantly decreased in treated patients (54).

There are likely multiple mucosal sites, as indicated by the multiple sites of microbial dysbiosis described above, and pathways by which the microbiota can contribute to the development of RA. Among the pathways are the generation of neoantigens through citrullination, molecular mimicry, epithelial barrier permeability, microbial translocation, and microbial education of immune responses (Figure 2). These mechanisms, along with the mucosal sites where they occur, are not mutually exclusive. Multiple mechanisms and mucosal sites likely converge to lead to autoimmunity and ultimately clinical RA.

Potential mechanistic effects of the microbiome in RA pathogenesis. Bacterial dysbiosis in the gut and oral cavity promotes barrier disruption, bacterial translocation, and altered microbial metabolite production. Increased gut permeability facilitates dissemination of microbial components such as LPS and polysaccharide A (PSA), priming innate immune responses and promoting T cell polarization toward Th1, Th17, and Tfh subsets. Microbial metabolites, including short-chain fatty acids (SCFAs), tryptophan derivatives (e.g., indole, IPA, IAA), and bile acids (e.g., LCA, DCA), modulate local and systemic immune responses. Specific bacteria contribute to RA via molecular mimicry (S. didolesgii, P. copri, E. lenta) or by promoting antigen citrullination (P. gingivalis, A. actinomycetemcomitans, S. parasanguinis), leading to the generation of autoantibodies. Neutrophil activation and NETosis further expose citrullinated microbial and host antigens. The combined effects of microbial translocation, antigenic stimulation, molecular mimicry, and citrullination establish a link between mucosal microbiota and systemic autoimmunity in RA.

Generation of citrullinated antigens. Given the central role of ACPAs in the diagnosis and pathophysiology of RA (55), identifying the source of citrullinated antigens that drive ACPA production remains a key focus of investigation. Citrullination of arginine residues in proteins occurs as a posttranslational modification, catalyzed by the family of peptidyl arginine deiminases (PADs). Microbial factors, particularly bacterial PADs and microbe-induced host PAD activation, contribute to the generation of citrullinated antigens in mucosal sites. The only bacterium known to express PAD is the periodontal pathogen P. gingivalis, which can generate citrullinated fibrinogen and enolase, antigens often targeted by ACPAs (56, 57). Experimental models demonstrated that PAD-expressing strains of P. gingivalis exacerbated arthritis, whereas PAD-deficient strains failed to do so (58, 59). Conversely, some periodontal bacteria and pathogens, such as Aggregatibacter actinomycetemcomitans, indirectly induce citrullination by triggering host PAD activation to subvert host immunity. The resulting hypercitrullination of host proteins closely resembles the citrullinated antigens found in RA joints (60, 61). Additional bacteria like Staphylococcus aureus and viruses like rhinovirus and cytomegalovirus similarly induce host hypercitrullination that may result in loss of tolerance leading to RA (61). However, the importance of bacterial species, the processes of antigen citrullination, and the temporal relationship between microbial exposure and citrullination in the generation of ACPAs remain unresolved.

Molecular mimicry. For decades, researchers have sought to identify the elusive “arthritogenic” antigen, citrullinated or not. A leading hypothesis proposes that microbial antigens resemble host proteins that may provoke cross-reactive immune responses, ultimately breaking immune tolerance and promoting systemic autoimmunity (7). S. copri, Subdoligranulum didolesgii, and Streptococcus spp. each have promising data supporting molecular mimicry that leads to RA.

Initial studies to identify potential microbes that could serve as molecular mimics of self-antigens utilized HLA-DR peptidomics. Peptides presented by HLA-DR on cells from the peripheral blood, synovial tissue, and synovial fluid from individuals with RA revealed sequence homology between the S. copri protein Pc-p27 and self-antigens, including N-acetylglucosamine-6-sulfatase (GNS) and filamin A (FLNA), both of which are expressed in RA synovial tissues. Pc-p27 elicited robust T cell responses that cross-reacted with their human counterparts, producing a pronounced Th1 response in patients with RA, while IgA antibodies targeting Pc-p27 associated with increased levels of ACPA and Th17 cytokines (62–64). In arthritis-prone SKG mice, S. copri colonization promoted Th17 expansion, joint inflammation, and autoantibodies. Furthermore, S. copri–educated T cells from SKG mice could trigger arthritis when transferred to naive T cell–deficient mice (65), supporting a causal role for cross-reactive immune responses in driving autoimmune arthritis.

In another approach to identify possible molecular mimics, autoreactive monoclonal antibodies (mAbs) derived from patients at risk for and with RA were used to identify possible cross-reactive bacteria in a pool of fecal samples. Through 16S sequencing of bacteria bound to the mAbs, followed by culturing bacteria from the primary fecal samples, Subdoligranulum didolesgii emerged as a candidate molecular mimic. In addition to being targeted by the autoreactive mAbs, S. didolesgii specifically activated T cells with a Th17 phenotype in a MHC class II–dependent manner from individuals with RA and not controls. To demonstrate causality, colonization of germ-free mice with S. didolesgii induced joint inflammation, Th17 activation, and pathogenic autoantibody production as evidenced by the ability of serum from S. didolesgii colonized mice to transfer arthritis (66). Although the specific T and B cell antigens are yet to be identified, these findings highly suggest that S. didolesgii may be another trigger for molecular mimicry in RA.

As a third approach to identify molecular mimics, paired bacterial and human transcriptomics of blood from individuals with concomitant periodontitis and RA demonstrated systemic translocation of citrullinated oral Streptococcus spp. from the oral mucosa preceding an RA flare. Inflammatory ISG15+HLA-DRhi and S100A12+ monocytes and antibody effector response transcripts also associated with RA flare, suggesting that the citrullinated Streptococcus could trigger autoantibody responses. Indeed, ACPA mAbs derived from human RA plasmablasts cross-reacted with citrullinated bacteria including Streptococcus, but not uncitrullinated bacteria (67), implicating citrullinated Streptococcus spp. as another potential molecular mimic.

These findings support molecular mimicry as one possible mechanism in RA, yet critical gaps remain. Although microbial peptides that mimic self-antigens have been suggested, their precise role in breaking immune tolerance and driving disease progression is not fully elucidated. Microbial antigens may act as initial triggers or exacerbate preexisting autoreactivity, and these antigens may act individually, sequentially, or in concert. Finally, the diversity of microbial epitopes capable of eliciting cross-reactive immune responses complicates efforts to define causative agents in RA.