Associations Between the Gut Microbiota and Its Related Metabolic Pathways and Uveitis: A Bidirectional Two-Sample Mendelian Randomization Study

Abstract

Purpose: Some experimental reports have proposed an interaction between gut microbiota (GM) and uveitis. However, the exact association between GM and its metabolic pathways and uveitis remains unknown. This study was conducted to explore the bidirectional causal relationship between GM and its metabolic pathways and uveitis.

Methods: Summary data of the GM and its metabolic pathways and uveitis were leveraged from the Dutch Microbiome Project and the Genome-Wide Association Studies (GWAS) Catalog, respectively. We then conducted Mendelian randomization (MR) analysis to explore whether the GM and its metabolic pathways have a corresponding causal relationship with uveitis. To confirm the credibility of the findings, we utilized MR Egger, the MR-PRESSO global test, and the Cochran Q test to detect pleiotropy and heterogeneity.

Results: According to the inverse variance weighting method, the species Bacteroides faecis (odds ratio [OR] = 0.598, 95% confidence interval [CI] = 0.390–0.919, P = 0.019) and the superpathway of sulfate assimilation and cysteine biosynthesis (OR = 0.179, 95% CI = 0.038–0.843, P = 0.029) had beneficial effects on uveitis. In contrast, the genus Sutterellaceae (OR = 3.493, 95% CI = 1.121–10.879, P = 0.030); the species Parabacteroides distasonis (OR = 5.932, 95% CI = 1.321–26.635, P = 0.020), Faecalibacterium prausnitzii (OR = 4.838, 95% CI = 1.067–21.936, P = 0.040), and Bacteroides caccae (OR = 3.818, 95% CI = 1.010–14.437, P = 0.048); and the L1,2–propanediol degradation (OR = 2.084, 95% CI = 1.098–3.954, P = 0.024), galactose degradation I (Leloir pathway; OR = 3.815, 95% CI = 1.108–13.135, P = 0.033), TCA cycle VI (obligate autotrophs; OR = 2.955, 95% CI = 1.015–8.606, P = 0.046) and UMP biosynthesis (OR = 4.979, 95% CI = 1.000–24.782, P = 0.049) pathways had adverse effects on uveitis. No pleiotropy or heterogeneity was found. Leave-one-out analysis showed the reliability of the above findings.

Conclusions: Our analysis revealed a causality between certain GM species and metabolic pathways and uveitis via genetic prediction, which may provide new perspectives into the etiology and therapies of uveitis.

Translational Relevance: This study provides evidence that modulating the intestinal flora and its metabolic pathways is effective in treating uveitis.

Introduction

Uveitis is an inflammatory disease involving the iris, ciliary body, choroid, and surrounding tissues. Uveitis can occur in patients of all ages but is more common in young adults. Recurrent episodes of uveitis can affect the patients’ quality of life. If left untreated, uveitis can lead to blindness.¹ Uveitis can be classified as infectious or noninfectious depending on the etiology, with infections accounting for 30% to 60% of cases in developing nations and noninfectious uveitis being more common in developed nations, with the noninfectious type believed to be mediated mainly by immune factors.^1,2 Noninfectious uveitis includes Vogt‒Koyanagi‒Harada (VKH) disease, Behçet's disease (BD), and acute anterior uveitis (AAU).³ AAU is the most common anatomic type of uveitis.⁴ 

The gut microbiota (GM) is widely distributed in the human digestive tract and is dominated by bacterial species.⁵ Researchers have shown that these microorganisms take an increasingly important part in keeping immune and metabolic homeostasis, resisting pathogenic agents, and promoting the occurrence and evolution of tumors, and many immune-related diseases are closely related to dysregulation of the GM.^6,7 The GM is linked to host genotype, dietary structure, lifestyle, and immune response, etc. Meanwhile, recent studies have revealed that alterations in the composition of the GM is correlated with various ocular diseases,⁸ such as those with glaucoma,^9,10 uveitis,^11,12 high myopia,¹³ and diabetes retinopathy.^14,15 One of the metabolites of GM is short-chain fatty acids, which maintain intestinal barrier function and homeostasis. Oral administration of short chain fatty acids to uveitis model mice resulted in a decrease in intestinal chemokines, which inhibited the migration of effector T-cells to the eye and ameliorated the symptoms of uveitis.¹⁶ 

HLA-B27 is a susceptibility factor for acute uveitis. The GM of mice with genetically modified HLA-B27 and HLA-B27-knockout mice differed from that of normal mice, suggesting that the GM perhaps contribute to the occurrence of autoimmune uveitis through inflammatory response.¹⁷ Fecal transplantation from patients suffering from VKH to rats with experimentally induced autoimmune uveitis exacerbated endophthalmitis in these mice. In patients with VKH, the altered GM was partially restored after the administration of immunosuppressive therapy.¹⁸ An increase in specific Gram-negative bacteria may trigger ocular inflammation and aberrant immune responses.¹⁹ However, our understanding of GM is still limited, and the causal relationship between GM and uveitis needs further clarification. 

Mendelian randomization (MR) is a statistical method for uncovering causal relationships using whole-genome sequencing data that effectively minimizes bias and potential confounders. The present research was conducted to explore the bidirectional causal relationship between GM and its metabolic pathways and uveitis from a genetic perspective. 

Methods

Study Design

Framework of this study is shown in Figure 1. Single-nucleotide polymorphisms (SNPs) strongly associated with GM had been chosen for instrumental variables (IVs), which must satisfy three fundamental hypotheses: IVs should be closely related to the exposure, IVs should be independent of confounders, and IVs should affect the outcome solely through the exposure. To eliminate the interference of reverse causality, we conducted a reverse analysis. 

Figure 1.

 

Framework of the study.

Data Sources

The genetic data of GM were extracted from the Dutch Microbiome Project,²⁰ which incorporated 7738 subjects of European descent with a mean age of 48.5 years. The data encompassed 207 taxa (5 phyla, 10 classes, 13 orders, 26 families, 48 genera, and 105 species) and 205 bacterial pathways. Genetic data of patients with uveitis were sourced from the Genome-Wide Association Studies (GWAS Catalog²¹) and comprised 134 patients and 456,214 controls of European ancestry (Table 1). All the data used were acquired from publicly available data bank and did not need further ethical approval. 

Table 1.

 

Date Sources

 

IV Selection

SNPs significantly related to the GM were first screened as IVs. The genome-wide statistical significance threshold was set at P < 1 × 10⁻⁸. Unfortunately, few SNPs met this threshold. Therefore, to explore more comprehensive results, the threshold was set at P < 1 × 10⁻⁶, and this threshold has been widely applied in previous studies. The threshold parameters (R² = 0.001, kb = 10,000) were selected to avert the effects of linkage disequilibrium, and duplicate SNPs and palindromic SNPs were removed.²² Confounding factors were further excluded by the FastTraitR package in the R software (version 4.3.3). SNPs with F values < 10 were regarded as weak instrument and were eliminated. 

MR Analysis

Two-sample MR and MR-PRESSO packages in R software (version 4.3.3) were applied for statistical analysis. The inverse variance-weighted (IVW), MR–Egger regression, weighted median estimator (WME), simple mode (SM), and weighted mode (WM) methods were utilized to evaluate the causal relationship between the GM and its metabolic pathways and uveitis. Estimates from IVW method were derived from a pooled analysis of the Wald ratios for all genetic variants, and IVW was selected as the primary analytic method. The results are expressed as odds ratios (ORs) with 95% confidence intervals (95% CIs). However, IVW suffers from a certain multiplicity bias, so sensitivity tests are needed to determine the validity of the results. Differences were considered statistically significant at P < 0.05. 

Sensitivity Analysis

The heterogeneity test was conducted via Cochrane’s Q method, with P < 0.05 suggesting the existence of heterogeneity.²³ The Egger regression method was used to explore the horizontal pleiotropy of IVs,²⁴ with P < 0.05 indicating horizontal pleiotropy. Sensitivity analyses were performed via the leave-one-out method to appraise the reliability of the results.²⁵ 

Reverse MR Analysis

To explore latent reverse causation, reverse MR analysis was carried out, with uveitis as the exposure and GM and its metabolic pathways as the outcome. The analysis process was consistent with that of the forward MR analysis. The direction of causality between GM and its metabolic pathways and uveitis can be further verified if the forward MR analysis is significant but the reverse is not. 

Results

Causal Effects of the GM on Uveitis

The F values of the SNPs for GM and its metabolic pathways ranged from 20.83 to 61.11, all of which were > 10, indicating the absence of weak IVs and the dependability of the results. According to the IVW results, there was a causal relationship among five different GM species and five bacterial pathways and uveitis (Fig. 2). The species Bacteroides faecis was negatively correlated with uveitis (OR = 0.598, 95% CI = 0.390–0.919, P = 0.019), indicating that it is a protective bacterium. The genus Sutterellaceae (OR = 3.493, 95% CI = 1.121–10.879, P = 0.030) and species Parabacteroides distasonis (OR = 5.932, 95% CI = 1.321–26.635, P = 0.020), Faecalibacterium prausnitzii (OR = 4.838, 95% CI = 1.067–21.936, P = 0.040), and B. caccae (OR = 3.818, 95% CI = 1.010–14.437, P = 0.048) were positively associated with uveitis, implying that these bacteria are risk factors. The superpathway of sulfate assimilation and cysteine biosynthesis (OR = 0.179, 95% CI = 0.038–0.843, P = 0.029) was negatively related to uveitis, signifying that an increase in the activity of this pathway reduces the risk of developing uveitis. L 1,2-Propanediol degradation (OR = 2.084, 95% CI = 1.098–3.954, P = 0.024), galactose degradation I (Leloir pathway; OR = 3.815, 95% CI = 1.108–13.135, P = 0.033), TCA cycle VI (obligate autotrophs; OR = 2.955, 95% CI = 1.015–8.606, P = 0.046), and UMP biosynthesis (OR = 4.979, 95% CI = 1.000–24.782, P = 0.049) were positively linked with uveitis, denoting that the increased activity of these pathways exacerbates the risk of uveitis. A scatterplot visualizing the associations of each SNP with the GM and uveitis is shown in Figure 3. 

Figure 2.

 

Causal relationships between gut microbiota and its metabolic pathways and uveitis. Forest plot of IVW analyses.

Figure 3.

 

The scatter plots demonstrated the effect of gut microbiota on uveitis. (A) Genus_Sutterellaceae_unclassified; (B) species_Bacteroides_faecis; (C) species_Parabacteroides_distasonis; (D) species_Faecalibacterium_prausnitzii; (E) species_Bacteroides_caccae; (F) L 1,2-propanediol degradation; (G) superpathway of sulfate assimilation and cysteine biosynthesis; (H) galactose degradation I (Leloir pathway); (I) TCA cycle VI (obligate autotrophs); (J) UMP biosynthesis.

Sensitivity Analyses

Sensitivity analyses revealed P values > 0.05 for Cochran’s Q test, MR Egger regression, and MR PRESSO for the five pathways and five GM species, suggesting that causality was not influenced by heterogeneity or horizontal pleiotropy (Table 2). Results of the leave-one-out method did not identify any notable outliers, confirming the reliability of the MR research results (Fig. 4). 

Table 2.

 

Heterogeneity and Horizontal Pleiotropy of Gut Microbiota and its Pathways

 

Figure 4.

 

Plots of leave-one-out analysis of the gut microbiota and its metabolic pathways in relation to uveitis. (A) Genus_Sutterellaceae_unclassified;B:species_Bacteroides_faecis; (C) species_Parabacteroides_distasonis; (D) species_Faecalibacterium_prausnitzii; (E) species_Bacteroides_caccae; (F) L 1,2-propanediol degradation; (G) superpathway of sulfate assimilation and cysteine biosynthesis; (H) galactose degradation I (Leloir pathway); (I) TCA cycle VI (obligate autotrophs); (J) UMP biosynthesis.

Reverse MR Analysis to Evaluate the Effect of Uveitis on the GM

Through reverse MR analysis, a total of 11 GM species and 2 bacterial pathways were identified to have a reverse causal relationship with uveitis. The sensitivity analysis of the reverse MR results revealed no statistical heterogeneity or horizontal pleiotropy, which indicated that there was no reverse causal relationship (Table 3; Fig. 5). 

Table 3.

 

Heterogeneity and Horizontal Pleiotropy of Uveitis

 

Figure 5.

 

Mendelian randomization analyses of uveitis effects on the gut microbiota and its metabolic pathways. Forest plot of IVW analyses.

Discussion

The results of this study revealed that five GM species and five bacterial pathways are causally associated with uveitis, and we performed rigorous quality control of research results to avoid confounding factors and reverse causality. Specifically, Bacteroides faecis and the superpathway of sulfate assimilation and cysteine biosynthesis exhibit protective effects, whereas the genus Sutterellaceae; the species Parabacteroides_distasonis (P. distasonis), Faecalibacterium prausnitzii (F. prausnitzii), and Bacteroides caccae (B. caccae); and the L 1,2-propanediol degradation, galactose degradation I (Leloir pathway), TCA cycle VI (obligate autotrophs), and UMP biosynthesis pathways are risk factors for uveitis. However, uveitis does not have a reverse causal regulatory effect on these GM species and bacterial pathways. 

Currently, academics have made some progress in the study of mechanisms related to the pathogenesis of ocular diseases caused by intestinal microorganisms. The existence of an intestinal-ocular axis has been demonstrated, and changes in the intestinal microflora are closely related to the physiopathologic changes in ophthalmopathy. Our eyes have an immune privilege mechanism that can regulate inflammation but limit the ability to renew and repair. Therefore, changes in GM are more likely to develop ocular inflammation, and one possible mechanism is the transfer of bacteria to eyes through blood or mesenteric lymphatic vessels.²⁶ Approximately 4% to 6% of complications in inflammatory bowel disease are uveitis, indicating a potential link between dysbiosis of GM and uveitis.²⁷ The abundance of intestinal microorganisms in the human body varies, but high-abundant microbiota may be involved in a disease together with low-abundant microbiota. Therefore, exploring the metabolic pathways of gut microbiota is also important. Scholars have found that GM affects the occurrence of colorectal cancer by mediating the host’s urea cycle metabolic pathway.²⁸ 

Sutterella has been linked to a variety of conditions, such as autism,²⁹ ulcerative colitis,³⁰ and diabetes.^31,32 Compared with that in healthy adults, the composition of intestinal microorganisms in patients with BD was greatly different, accompanied with the reduction of the abundance of Sutterella, suggesting that Sutterella may be involved in the progression of BD.³³ Fecal transplantation from patients with BD to rats triggers a significant increase in uveitis activity and boosts the production of inflammatory factors.³⁴ 

P. distasonis is an obligate anaerobic, inactive Gram-negative bacterium that grows on bile-rich media.³⁵ Ankylosing spondylitis (AS) is an autoimmune disorder with a clear association with HLA-B27. The most common type of ocular damage in patients with AS is uveitis, with approximately 50% of patients experiencing acute, unilateral anterior uveitis.³⁶ The abundance of P. distasonis in feces is considerably increased among patients with AS, suggesting that P. distasonis perhaps takes part in autoimmune disorders such as AS.^37,38 Microbial dysbiosis during infancy may induce juvenile idiopathic arthritis (JIA) or accelerate its progression. Uveitis is the most common extra-articular manifestation of JIA. P. Distasonis can increase the probability of JIA occurrence, and its interaction with Prevotella copri can increase the production of IFN - γ cells, leading to autoimmune reactions.³⁹ 

F. prausnitzii is a Gram-negative, extremely oxygen-sensitive microbe that is highly important in the GM. F. prausnitzii can generate numerous substances that suppress inflammatory effects caused by prevention of NF-κB activation and IL-8 production.^40,41 Depletion of F. prausnitzii is associated with microbial dysbiosis accompanied by a wide range of metabolic and/or immune-mediated chronic diseases. The present study has found that abundance of F. prausnitzii increases in patients with AS. 

Among mothers and their infants, B.caccae has been recognized as the most widespread strain.⁴² The abundance of B.caccae is increased in HLA-B27-positive controls compared with HLA-B27 related uveitis patients, suggesting that B.caccae likely takes a protective role in the progression of uveitis.¹⁹ 

The tricarboxylic acid (TCA) cycle provides energy for the body and is a common pathway for the oxidation and decomposition of primary nutrients. Succinic acid is one of the major intermediate metabolic products of the TCA cycle, and it can induce inflammation by binding specifically to receptors in cells and prompting the production of reactive oxygen species, even causing autoimmune diseases. Heat-induced abnormal succinate accumulation has been linked to an increased occurrence of uveitis and aggravation of intraocular inflammation.⁴³ Intraperitoneal injection of succinic acid in mice with experimental autoimmune uveitis exacerbates intraocular inflammation, and elevated levels of succinic acid have been demonstrated in patients with BD and AAU.⁴⁴ 

Because noninfectious uveitis is mainly caused by autoimmune reactions, it may be affected by intestinal dysbiosis. Mice with autoimmune uveitis given oral antibiotics presented significant variations in the gastrointestinal flora and a reduction in uveitis severity,⁴⁵ suggesting the possible presence of uveitis-causing bacteria and inspiring new strategies for future treatments. 

This study also has some limitations: (1) the selected uveitis samples were all of European ancestry, which does not allow for the generalization of the findings to other ethnic populations; (2) the data obtained lack more detailed cohort data, such as age and gender, making it impossible to conduct subgroup analysis; (3) the HLA allele can affect the composition of gut microbiota and may consequently contribute to the development of uveitis. We lack an analysis of this relationship. More research is needed to understand the pathogenesis of uveitis; and (4) this study is a preliminary exploration of the effects of GM and its metabolic pathways on uveitis, and future expansion of the sample size is needed to clarify this relationship. 

Conclusions

In summary, we demonstrated that five GM species and five bacterial pathways are causally associated with uveitis. Owing to the recurrent and blinding properties of uveitis, exploring the protective and risk factors for uveitis is important. The protective or pathogenic mechanism of the GM and its metabolic pathways in uveitis requires further investigation. 

Acknowledgments

The authors thank all the researchers who have generously shared the data publicly, which greatly facilitated the conduct of our research. 

Supported by the Sichuan Provincial Department of Science and Technology (No. 2024ZYD0114) and Sichuan Medical Association (No. 202407100226). 

Author Contributions: Maomei Luo, Investigation, validation, visualization, writing–original draft, writing–review & editing. Zhen Xing, Investigation, validation, writing–review & editing. Yanhao Gou, Formal analysis, visualization. Xinlin Yang, Formal analysis. Xinran Zhang, Investigation. Wei Yu, Investigation. Hongbin Lv, Formal analysis, data curation, supervision. 

Availability of Data and Materials: All data are publicly available. 

Disclosure: M. Luo, None; Z. Xing, None; Y. Gou, None; X. Yang, None; X. Zhang, None; W. Yu, None; H. Lv, None