Abstract

Background Atherosclerosis (AS) remains a major cause of cardiovascular morbidity and mortality. Lactylation—a recently described post-translational modification linking cellular metabolism to gene regulation—has been implicated in vascular inflammation, yet its roles in AS are not fully defined. Method We modelled AS by exposing HUVECs to oxidised LDL and performed data-independent proteomics. Public AS transcriptomes were integrated and batch-corrected; lactylation-related genes (LRGs) were profiled; consensus clustering and WGCNA defined subtypes/modules. Candidate biomarkers were prioritised by intersecting proteomic, differential and network features, then evaluated using an ensemble of 12 machine-learning algorithms with cross-validation and external validation. Immune infiltration (CIBERSORT) and single-cell data characterised immune contexts. Connectivity Map, molecular docking and molecular dynamics (MD) explored therapeutics. Core genes were validated by RT-qPCR, Western blotting and immunofluorescence in Apoe −/− mice. Results Proteomics identified 472 differentially expressed proteins; GEO analyses yielded 2,544 DEGs and WGCNA 2,059 module genes, converging on 25 candidates. The top ensemble (LASSO + GBM) achieved a mean AUC 0.979 across training and external sets, nominating UAP1, NRP1, QPRT and NDST1 as hub genes. These genes associated with immune-cell infiltration and showed prominent single-cell expression in macrophages and smooth muscle cells. RT-qPCR in vivo showed NRP1/NDST1/QPRT upregulated and UAP1 downregulated versus controls (all P < 0.05); WB/IF confirmed higher NRP1 and lower UAP1 protein abundance. CMap analysis suggested several candidate compounds. Among them, rivaroxaban was prioritised for further in silico evaluation because NRP1 emerged as a validated hub gene, and docking and molecular dynamics simulations supported a stable predicted interaction between rivaroxaban and NRP1. Conclusion UAP1, NRP1, QPRT and NDST1 represe