Mendelian randomization analysis of gut microbiota-driven immune modulation in rheumatoid arthritis: New mechanistic insights and therapeutic targets

Abstract

Objective

To explore the complex interactions between gut microbiota and immune cell phenotypes in rheumatoid arthritis development and identify potential therapeutic targets within the gut microbiota–immune cell axis.

Methods

We conducted a Mendelian randomization analysis to explore the causal relationship between gut microbiota and rheumatoid arthritis, including the role of immune cell mediators. Sensitivity analyses assessed pleiotropy and heterogeneity, while mediation analysis identified pathways through which immune cells mediate gut microbiota effects on rheumatoid arthritis development. Key microbial taxa and their effects on rheumatoid arthritis were quantified.

Results

Our analysis identified 27 gut microbiota taxa significantly associated with rheumatoid arthritis, with Provencibacterium massiliense showing the strongest protective effect (odds ratio = 0.807, 95% confidence interval: 0.700–0.911, P = 0.003). Additionally, 20 immune cell phenotypes with IgD+ CD38dim AC were significantly linked to rheumatoid arthritis (odds ratio = 1.064, 95% confidence interval: 1.027–1.102). Mediation analysis uncovered 13 significant gut microbiota–immune cell pathways, with the UBA8517–CCR2 monocyte pathway mediating 10.1% of the total effect (beta1 = −0.595, beta12 = 0.027, mediation proportion = 10.1%).

Conclusion

This study offers novel insights into the gut microbiota–immune cell axis in rheumatoid arthritis, identifying Provencibacterium massiliense, IgD+ CD38dim AC and the UBA8517—CCR2 monocyte pathway as potential therapeutic targets for rheumatoid arthritis treatment.

Introduction

Rheumatoid arthritis (RA) is a chronic autoimmune disorder characterized by persistent synovial inflammation and systemic manifestations, leading to joint damage, disability, and significant societal burden. ¹ The incidence of RA, which affects approximately 0.5%–1.0% of adults globally, increases with aging, impacting over 75% of older adults.^2,3 Although advances in RA pathogenesis research have identified numerous genetic and immunological factors, the precise mechanisms driving RA remain unclear, highlighting the need for further exploration. Genome-wide association studies have identified numerous loci associated with RA risk, primarily involving immune pathways. This suggests immune dysregulation as a key component of RA pathophysiology. ⁴ Additionally, environmental and microbial factors have been implicated in RA onset and progression, suggesting a multifactorial etiology.^4,5

Emerging evidence highlights the potential role of gut microbiota (GM) in RA pathogenesis. Studies in animal models have linked dysbiosis to RA onset, while human studies have found microbial alterations, particularly in early RA stages, that correlate with clinical markers such as C-reactive protein and anti-citrullinated protein antibodies.^6,7 Certain bacterial species such as Porphyromonas gingivalis (a periodontitis pathogen) may trigger RA through aberrant citrullination and potential breaches in immune tolerance. ⁸ Other pathogens, including Proteus mirabilis, Escherichia coli, and Epstein–Barr virus, have been implicated through mechanisms such as molecular mimicry, although these remain unverified. ⁹

Although these findings emphasize potential links, the exact mechanisms by which GM influences immune responses in RA development are not fully understood. Notably, GM–immune cell (GM–IM) pathways, which may modulate RA risk, ¹⁰ are yet to be comprehensively investigated. This study addresses these knowledge gaps by examining the causal relationships between specific GM taxa and immune cell phenotypes in RA using Mendelian randomization (MR).

Methods

Study design

This study comprised three main components: (1) evaluating GM’s causal effect on RA, (2) assessing the causal association of immune cell phenotypes with RA, and (3) exploring the potential mediation effects of GM–IM pathways on RA.

Single-nucleotide polymorphisms (SNPs) were used as instrumental variables (IVs) in the MR framework, adhering to three key assumptions: (1) strong exposure association, (2) independence from confounders, and (3) influence on RA only through the exposure. ¹¹

Data source

GM data were sourced from the FinRisk 2002 cohort (n = 5959) that provided genomic data on 473 microbial taxa (Supplementary Table 1). ¹² Immune cell data were obtained from GWAS entries GCST90001391 to GCST90002121 (Supplementary Table 2), ¹³ covering 731 phenotypes from 359,399 European participants (1385 RA cases, 358,014 controls). RA-specific data were obtained from the Finngen R11 database (453,733 participants comprising 14,818 RA cases and 287,796 controls) (https://r11.finngen.fi/). This study performed secondary analyses on the publicly available GWAS summary statistics. Ethical approvals had been obtained for the original studies. No individual-level data were accessed; therefore, no additional approval was required.

IV selection

SNPs were selected based on genome-wide significance (P < 5 × 10⁻⁸) to ensure robust associations. Linkage disequilibrium clumping was performed using PLINK (version 1.90) with an r² threshold of <0.001 within a 10,000-kb window. ¹⁴ SNPs associated with the outcome (P < 0.05) and palindromic SNPs were excluded. Instrument strength was assessed using the F-statistic (F ≥ 10). ¹⁵ MR-PRESSO was used for pleiotropy correction, and Steiger filtering was applied to ensure validity and eliminate reverse causality.

MR analysis

We conducted a two-sample MR analysis using the inverse variance weighting (IVW) approach, complemented by MR-Egger, weighted median, weighted mode, and MR-PRESSO methods for comprehensive analysis. Results were reported as odds ratios (ORs) with 95% confidence intervals (CIs), and P < 0.05 was considered to indicate statistical significance. For single IV exposures, the Wald ratio method was used. ¹⁶ Data analyses were performed via R software (version 4.4.1) using MR and two-sample MR packages. No corrections for multiple testing were applied, as this was an exploratory study.

Bidirectional causality analysis

We performed a bidirectional MR analysis to assess the causal effects between GM, immune cells, and RA. In the analysis, GM and immune cells were considered the “exposure” and RA was considered the “outcome” in one direction; in the reverse analysis, RA was considered the “exposure,” with GM and immune cells being the “outcome.” SNPs significantly associated with RA, GM, and immune cells (P < 5 ×10⁻⁸) were selected as IVs, and the IVW method was used for robust causal estimates.

Mediation analysis

In the mediation analysis, we included GM and immune cells with significant causal effects on RA from the two-sample analysis (steps B and C in Figure 1). We assessed whether GM causally affects immune cells (step A in Figure 1) and explored whether immune cells mediate the pathway from GM to RA.

Figure 1.

Study overview. Step A represents the GM–IM pathway analysis and mediation effects of immune cells in the GM–RA relationship. Step B represents the causal effects of immune cells on RA. Step C represents the causal effects of GM on RA. IM: immune cell; RA: rheumatoid arthritis; GM: gut microbiota.

Sensitivity analysis

To ensure robust findings, we performed several sensitivity analyses. Cochran’s Q test was used to assess heterogeneity among the selected IVs, with significant Q values indicating heterogeneity. ¹⁷ Scatter plots were used to visualize SNP–exposure and SNP–outcome associations. A leave-one-out analysis was used to assess the influence of individual SNPs on causal estimates. ¹⁸ We also used MR-Egger regression and MR-PRESSO to detect and correct for horizontal pleiotropy, ¹⁹ removing significant pleiotropic outliers for more accurate estimates.

Results

Effect of GM on RA

We identified 30 significant GM taxa associated with RA, with Provencibacterium massiliense (P = 0.003, OR = 0.807, 95% CI: 0.668–0.975) showing the strongest protective effect (Supplementary Table 3). Heterogeneity analysis revealed three taxa with variability (Gillisia, Ruminococcus D bicirculans, and Ruminococcus D), and further analyses of 27 taxa revealed that 11 taxa exerted protective effects (OR < 1) and 16 increased the RA risk (Figure 2, Supplementary Table 4). Excluding taxa with heterogeneity improved the consistency of causal effect estimates. Sensitivity analyses (leave-one-out, funnel plot, and scatter plot) confirmed the robustness of our findings. Given the exploratory nature of the study, no multiple testing correction was applied. ²⁰ A reverse MR analysis showed no significant effect of RA on GM, supporting the microbiota’s role in RA risk.

Figure 2.

Forest plot of causal effects of significant gut microbiota taxa on RA. RA: rheumatoid arthritis.

Association between immune cell phenotypes and RA

We assessed the causal effects of 731 immune cell phenotypes on RA using MR, identifying 59 traits significantly associated with RA risk. HLA-DR on CD14+ CD16− monocytes showed the strongest association (P < 0.001, OR = 1.018, 95% CI: 1.004–1.032), indicating its significant role in RA susceptibility (Supplementary Table 5). After excluding heterogeneous and pleiotropic phenotypes, 20 immune cell traits were retained, with 12 showing increased RA risk and 8 exerting protective effects (Supplementary Table 6). The phenotype IgD+ CD38dim AC showed the most significant association (OR = 1.064, 95% CI: 1.027–1.102), highlighting its potential role in RA pathogenesis (Figure 3).

Figure 3.

Forest plot of causal effects of significant immune cell phenotypes on RA. RA: rheumatoid arthritis.

GM–IM pathways and mediation effects in RA

We integrated 27 GM taxa and 20 immune cell phenotypes to identify 13 significant GM–IM pathways via MR, involving 8 microbial taxa and 10 immune phenotypes, suggesting mechanistic links between GM and immune regulation in RA (Figures 4 and 5, Supplementary Table 7).

Figure 4.

Forest plot of GM–IM pathways, showing the effect of gut microbiota on immune cells. IM: immune cell; GM: gut microbiota.

Figure 5.

Forest plot of GM–IM–RA pathways, illustrating the combined effects of gut microbiota and immune cells on rheumatoid arthritis. IM: immune cell; RA: rheumatoid arthritis; GM: gut microbiota.

A two-step MR mediation analysis revealed that the UBA8517–CCR2 monocyte pathway had the strongest mediating effect, with a direct effect (beta1 = −0.595), indirect immune-mediated effect (beta2 =−0.046), and combined mediation effect (beta12 = 0.027), accounting for 10.1% of the total effect (Figure 6).

Figure 6.

The mediating effect of immune cell phenotypes of gut microbiota on RA. RA: rheumatoid arthritis.

Discussion

RA is driven by chronic synovial inflammation and immune dysregulation, leading to joint destruction and systemic effects. ²¹ Key RA features include synoviocyte proliferation and adaptive immune cell infiltration into the synovium. ²² These immune processes drive joint damage and systemic inflammation, both of which are hallmarks of the disease. Despite extensive research, the underlying mechanisms contributing to RA pathogenesis remain incompletely understood, emphasizing the need for further exploration of environmental, genetic, and immunological factors.

Emerging evidence has demonstrated that RA patients exhibit altered gut microbial compositions compared with healthy individuals, with specific taxa such as Prevotella copri associated with early RA development. ²³ However, causal relationships remain elusive due to the cross-sectional nature of most microbiome studies. The GM plays a pivotal role in immune regulation, with dysbiosis implicated in autoimmune diseases such as RA. ²⁴ The gut, the largest mucosal barrier, harbors over 100 trillion microorganisms involved in essential processes such as digestion, metabolism, immune regulation, and disease prevention.^25,26 An imbalance in gut microbial composition, referred to as dysbiosis, can disrupt immune homeostasis and trigger inflammatory responses, which may play a role in the pathogenesis of RA.^27–30 RA patients exhibit significant alterations in GM composition compared with healthy individuals.^31,32 Preclinical studies have further underscored the role of GM in RA, with evidence showing that fecal transplants from RA patients can increase the frequency of proinflammatory Th17 cells and exacerbate arthritis in germ-free mice. ³³ Probiotics and short-chain fatty acids (SCFAs), key metabolites produced by GM, have been investigated for their anti-inflammatory and immunomodulatory properties. Probiotics have been shown to regulate both innate and adaptive immunity by influencing epithelial cells, macrophages, dendritic cells, T cells, and B cells. ³⁴ They reduce the expression of pattern recognition receptors on immune cells, inhibit antigen presentation, and suppress proinflammatory cytokine production, ultimately decreasing RA disease activity.^35,36 SCFAs, particularly butyrate and propionate, play a key role in modulating dendritic cell differentiation, cytokine expression, and regulatory T cells (Tregs), which are critical for maintaining immune tolerance and controlling inflammation.^37–40

In this study, we utilized MR to explore potential causal associations between GM and RA susceptibility. Notably, P. massiliense, a provisionally classified microbial taxon detected in our analyses and other metagenomic studies, was identified as a protective factor. Our findings revealed a novel protective role of P. massiliense (P = 0.003, OR = 0.807), which appears to reduce RA risk by modulating immune responses and inflammation. P. massiliense currently lacks formal taxonomic characterization. Although direct evidence is limited, based on its likely affiliation with Firmicutes, it may contribute to gut immune homeostasis through the production of SCFAs, particularly butyrate, which has been associated with Treg induction and anti-inflammatory effects in autoimmune diseases.^41,42

Our study identified the UBA8517–CCR2 monocyte pathway as a key mediator of the GM–IM axis in RA. Specifically, UBA8517 demonstrated a direct protective effect on RA (beta1 = −0.595), alongside an indirect effect mediated by CCR2+ monocytes (beta12 = 0.027), accounting for 10.1% of the total causal pathway. These findings suggest that GM may influence RA pathogenesis through modulation of CCR2+ immune cell trafficking and inflammatory responses. Notably, we also observed negative mediation in some GM–IM pathways, indicating that certain gut microbes may downregulate the proinflammatory activity of immune cells, potentially conferring protection against RA progression. Recent studies have highlighted the complex role of the CCL2/CCR2 axis in RA pathogenesis. ⁴³ CCR2+ monocytes are known to mediate synovial inflammation, promote proinflammatory cytokine release, and contribute to osteoclastogenesis and joint destruction. ⁴⁴ However, clinical trials targeting CCR2, such as MLN1202, have failed to demonstrate significant therapeutic benefits, ⁴⁵ likely due to chemokine redundancy and compensatory immune pathways. Moreover, evidence suggests that CCR2 may exhibit context-dependent protective roles, as CCR2 deficiency in certain models exacerbated arthritis progression. Our findings, identifying a 10.1% mediation effect via the GM–CCR2 monocyte pathway, suggest that GM may modulate RA pathogenesis through immune cell trafficking and inflammatory signaling, offering a novel perspective that is distinct from direct CCR2 inhibition mechanisms.

Despite the valuable insights provided by our study, certain limitations must be acknowledged. First, as an in silico MR analysis based on summary-level GWAS data, our results should be interpreted as hypothesis-generating rather than definitive evidence. No functional or clinical validation has been performed; therefore, the proposed GM–IM pathways require further experimental support. Second, we did not apply corrections for multiple hypothesis testing, which may increase the likelihood of false-positive associations. This decision was made to preserve exploratory sensitivity and identify potential targets for follow-up studies. Nevertheless, future studies with larger cohorts and rigorous statistical controls, including false discovery rate or Bonferroni corrections, are essential for validation. Third, the study population was predominantly of European descent, which may limit the generalizability of our findings. Ethnic differences in genetic architecture and microbiota composition necessitate the inclusion of diverse populations in future research. Finally, the taxonomic resolution of microbiota in the available datasets remains limited. Some microbial taxa, including P. massiliense, are not fully characterized, and their biological functions remain unclear. This underscores the need for deeper metagenomic sequencing and functional microbiology approaches to verify their role in RA pathogenesis.

In conclusion, our study provides preliminary evidence that specific GM taxa and immune phenotypes, particularly the UBA8517–CCR2 monocyte pathway, may play a role in the pathogenesis of RA. These findings offer novel perspectives on the gut–immune axis in RA. However, as an exploratory MR-based analysis, our study results should be interpreted with caution. Future research should prioritize experimental validation, mechanistic elucidation, and clinical evaluation of microbiota-driven immunomodulatory strategies aimed at restoring immune balance and mitigating RA progression.

Data availability statement

The original contributions presented in the study are included in the article/supplementary materials. Further inquiries can be directed to the corresponding author.

Declaration of conflicting interests

Huangsheng Tan, Xinhai Gao, Chengcai Dai, Pan Shen, Juyi Lai, Yuanfei Fu, and Shenghua He declare that they have no conflict of interest.

Supplementary material

Supplemental material for this article is available online.