Identification of Hub Gene Characteristics and Immune Landscapes Across Aging Subtypes in Atherosclerosis: A Machine Learning-Based Multi-Omics Study With Experimental Verification

Abstract

Aging is a major risk factor for atherosclerosis (AS), but the aging-associated molecular characteristics and immune heterogeneity of AS remain incompletely understood. This study aimed to identify aging-related hub genes and characterize immune landscapes across aging subtypes of AS using integrated bioinformatics and experimental validation. AS-related candidate genes were found by overlapping DEGs identified by limma and key module genes identified by Weighted Correlation Network Analysis (WGCNA) based on the GSE100927 dataset. Consensus clustering based on aging-related DEGs (differential expression analysis based on 125 aging-related genes between AS and control) was performed to identify aging subtypes and characterize immune landscapes. Typical genes were picked out using four machine learning models: Support Vector Machine (SVM), Random Forest (RF), eXtreme Gradient Boosting (XGB), and Generalized Linear Model (GLM). The expression of central genes was verified through single—cell data analysis. Patients with AS were enrolled (n = 30), and atherosclerotic plaque and adjacent normal arterial tissues were collected. Quantitative Polymerase Chain Reaction (qPCR) validation was performed in atherosclerotic plaques and oxidized Low-Density Lipoprotein (ox-LDL)-treated THP-1 macrophage. The effect of crucial genes on ox-LDL-induced THP-1 macrophage inflammatory response and foaming were validated by qPCR, ELISA and Oil Red O Staining. We identified 76 aging-related DEGs and classified AS samples into two aging-related subtypes (C1 and C2) with distinct immune infiltration characteristics. Machine learning analysis based on 43 candidate genes identified 5 hub genes: heat shock protein family B (small) member 7 (HSPB7), myelin expression factor 2 (MYEF2), dual specificity phosphatase 26 (DUSP26), tandem C2 domains, nuclear (TC2N), and phospholamban (PLN), whose expression patterns were further validated by single-cell RNA sequencing. Moreover, TC2N and PLN were significantly downregulated in atherosclerotic plaque tissues. TC2N and PLN expressions in atherosclerotic plaque tissue of AS patient were significantly reduced. TC2N was significantly decreased in ox-LDL-treated THP-1 macrophage. TC2N overexpression alleviated ox-LDL-induced THP-1 macrophage inflammatory response but also alleviated cell foaming. This integrated bioinformatics and experimental study identified five hub genes (HSPB7, MYEF2, DUSP26, TC2N, and PLN) associated with AS. Experimental validation confirmed that TC2N is significantly downregulated in human atherosclerotic plaques and ox-LDL-treated THP-1 macrophages, TC2N overexpression alleviates inflammatory responses and foam cell formation. These findings provide insights into the molecular mechanisms linking aging and immune dysregulation in AS and highlight TC2N as a potential regulator of AS progression.