Screening of potential oxidative stress-related biomarkers and therapeutic drugs in rheumatoid arthritis based on integrative bioinformatics, machine learning, and molecular dynamics simulations

Background Rheumatoid arthritis (RA) is a prevalent autoimmune condition. Increasing evidence reveals that oxidative stress exerts an important effect in the pathogenesis of RA. This research aimed to systematically screen oxidative stress-related biomarkers for RA and further examine promising therapeutic drugs for RA. Methods This research first obtained transcriptome data of RA from the GEO database and identified differentially expressed oxidative stress-related genes (DEOSGs). Subsequently, core biomarkers were identified by integrating weighted gene co-expression network analysis with three machine learning algorithms. Their diagnostic performance was assessed utilizing receiver operating characteristic curves, and a clinical predictive nomogram was established. Functional enrichment analysis was implemented to systematically elucidate the biological processes involving DEOSGs in RA, and immune infiltration analysis was conducted concurrently. Furthermore, a potential therapeutic small-molecule compound was screened leveraging the CMap database and validated through molecular docking and molecular dynamics simulation. Finally, the expression levels of the core genes were quantified and analyzed utilizing quantitative real-time polymerase chain reaction and Western blot in a primary human RA synovial fibroblast model. Results Totally, 281 DEOSGs were identified. These genes were significantly enriched in pathways including the MAPK, AMPK, TNF, and Toll-like receptor signaling pathways. Based on the machine learning algorithms, three core genes were ultimately determined. The diagnostic model established on the basis of these genes demonstrated good diagnostic efficacy. Immune infiltration analysis revealed significant differences in the distribution of immune cell subsets between RA patient samples and normal control samples. Molecular docking and molecular dynamics simulation indicated that narciclasine exhibited good binding affinity with the target protein, and the stability of the binding complex was acceptable. Furthermore, experimental results from the in vitro cell model confirmed that the expression patterns of the core genes were consistent with findings from the bioinformatics analysis. Conclusion This research preliminarily suggests that CXCL10, EDNRB, and MMP13 may serve as potential oxidative stress-related biomarkers for RA. Simultaneously, it predicts that narciclasine may be a promising candidate drug for RA treatment. These findings offer new insights into the pathogenesis, targeted intervention, and treatment development for RA.