Abstract
BACKGROUND: Growing evidence suggests a link between systemic lipopolysaccharide (LPS) exposure and worsening diabetic retinopathy (DR). This study aims to investigate DR's pathogenesis by analyzing LPS-related genes (LRGs) through bioinformatics. METHODS: The CTD database was utilized to identify LRGs. The datasets associated with DR were acquired from the GEO database. The Venn diagram was used to identify the differentially expressed LRGs (DLRGs), and the putative molecular mechanism of these DLRGs was investigated through functional enrichment analysis. We used WGCNA, Lasso regression, and RF to identify hub DLRGs. The expression levels of these hub DLRGs were validated in an independent dataset (GSE102485) and cell experiments. Employing the CIBERSORT algorithm, we examined the infiltration of 22 distinct immune cell types in DR and assessed the association between key DLRGs and immune infiltrates through correlation analysis. RESULTS: A total of 71 DLRGs were detected. These genes exhibited significant enrichment in pathways associated with inflammation. In addition, the in-depth analysis uncovered that five hub DLRGs (STK33 and EPHX2) linked to bacterial LPS displayed noteworthy diagnostic potential for individuals diagnosed with DR. The hub DLRGs expression in the high glucose-induced DR model was confirmed by qRT-PCR analysis. Furthermore, examination of immune infiltration indicated a significant association between these five genes and the extent of immune cell infiltration. CONCLUSION: STK33 and EPHX2 serve as biomarkers related to bacterial LPS. Exploring these genes in-depth could provide innovative ideas and a foundation for comprehending the progression of the disease and developing targeted treatments for DR.