Predicting Cross-Reactive Epitopes in the Gut Microbiome of Lupus Patients

Abstract

Systemic Lupus Erythematosus (SLE) is an inflammatory chronic autoimmune disease that affects various systems throughout the body. Although several autoantibodies are known to be closely associated with the pathogenesis of SLE, the precise etiology remains elusive. Previous studies have suggested that the gut microbiota may harbor specific cross-reactive epitopes capable of triggering host immune responses, potentially promoting the production of host autoantibodies. In this study, we collected gut microbiome samples from SLE patients before and after treatment and analyzed the gut microbiome of eight SLE patients using next-generation sequencing (NGS). To identify antigenic determinants on bacterial proteins that may elicit autoimmune responses, we employed the immunoinformatics pipeline CRESSP, which predicts similar cross-reactive epitopes between two proteomes based on structural properties. Consequently, we identified 1,156 potential cross-reactive B cell epitope candidates. To validate the predicted cross-reactive epitopes, we analyzed autoantibodies in serum samples from the same patients using phage immunoprecipitation sequencing (PhIP-Seq). A total of 1,013 autoantibody levels decreased in patients post-treatment, and 153 of these autoantibodies overlapped with those predicted by CRESSP to have cross-reactivity with bacterial pathogens. This group included known lupus-specific autoantibodies such as TRIM21 (Ro52), SSB (Ro60), XRCC5, and SNRPA. Upon further examination of which epitopes were predicted for the identified autoantibodies by CRESSP, we found that the selected autoantibodies exhibited a high average BLOSUM62 matrix score of at least 2 per amino acid residue, indicating strong similarity. Additionally, structural analysis confirmed considerable structural similarities among the predicted epitopes in three-dimensional space. This study demonstrates that the microbiome in SLE patients exhibits molecular mimicry between bacterial antigens and human autoantibodies, and that these epitopes can be predicted using informatics tools. These findings provide insights into the underlying causes of SLE and may contribute to the development of therapeutic strategies.