Abstract
BACKGROUND: Atherosclerosis (AS) is a chronic inflammatory disease with a poor prognosis, and 5-methylcytosine (m5C) RNA modification plays a significant role in AS-induced cardio-cerebrovascular diseases (CCVDs). However, the effects of m5C modification and related genes in AS remain unclear. METHODS: We analyzed the correlations between m5C RNA modification and its regulatory genes in AS using microarray databases. Specifically, microarray datasets of AS (GSE90074, GSE27034, GSE59421, and GSE159677) were obtained from the Gene Expression Omnibus (GEO) database. Following differential expression and Spearman correlation analyses using GSE90074, m5C regulator-related genes (MRRGs) were screened to construct a protein-protein interaction (PPI) network. A least absolute shrinkage and selection operator (LASSO) logistic regression model was constructed and validated using receiver operating characteristic (ROC) curves in GSE90074 and GSE27034 to identify feature genes. Consensus clustering analysis was then performed to classify AS samples into distinct clusters. In addition, Spearman correlation analysis was used to explore the associations between differentially expressed m5C regulators (DE-MRs) and immune cells based on the identified clusters. Weighted gene co-expression network analysis (WGCNA) was applied to identify AS-related module genes. Subsequently, intersecting genes common to module genes and differentially expressed genes (DEGs) across AS-related clusters were considered candidate biomarkers and were validated by quantitative real-time polymerase chain reaction (qRT-PCR) in human myeloid leukemia mononuclear cells (THP-1). Single-cell RNA sequencing (scRNA-seq) analysis was performed to characterize the immune microenvironment of AS. In vitro experiments and genetic interventions were conducted to investigate the effects of the m5C regulator NSUN3 on macrophage function. Finally, a competitive endogenous RNA (ceRNA) network targeting the identified biomarkers was predicted using the miRNet database. RESULTS: Based on two differentially expressed m5C regulators (NSUN3 and NSUN5), 546 of 2247 DEGs between AS and control samples were identified as MRRGs for PPI network construction. Twenty hub MRRGs were further incorporated into the LASSO logistic regression model, yielding nine feature genes. Based on these feature genes, AS samples were classified into two clusters, with five immune cell types showing significant differences between clusters. Both NSUN3 and NSUN5 showed the strongest correlations with M0 macrophages. A total of 643 module genes were identified and overlapped with DEGs from the two AS-related clusters, resulting in five biomarkers-MCL1, F13A1, RGS2, Toll-like receptor 8 (TLR8), and TAGAP. The expression patterns of these five biomarkers in the foam cell model were consistent with those observed in public datasets. Furthermore, NSUN3 regulated the production of proinflammatory cytokines in macrophages. Finally, a ceRNA regulatory network was constructed. CONCLUSION: Five potential diagnostic biomarkers for AS-MCL1, F13A1, RGS2, TLR8, and TAGAP-were identified. In addition, the m5C regulator NSUN3 plays a critical role in macrophage function, providing experimental evidence that may support the diagnosis and treatment of AS.