Abstract
Molecules in immune cells plays a vital role in the pathogenesis of ischemic stroke (IS). The aim of this study is to profile the landscape of molecules on the basis of immune cells in IS peripheral blood and construct an immunoregulatory competing endogenous RNA (ceRNA) network. We collected and combined multiple public transcriptome datasets from the peripheral blood of IS patients and healthy controls. CIBERSORT deconvolution revealed that the proportions of CD8 and CD4 naive T cells, monocytes, and neutrophils changed significantly in the IS group. Intersecting the immune cell-related genes identified by weighted gene co-expression network analysis (WGCNA) and differential expression analysis, 38 overlapping candidate biomarkers were selected. Three machine learning algorithms, including least absolute shrinkage and selection operator (LASSO), support vector machine-recursive feature elimination (SVM-RFE), and random forest were applied, and 11 distinct immune cell-related genes were identified. We obtained the mRNA-miRNA and miRNA-lncRNA interactions from StarBase v3.0, and constructed a ceRNA network based on the differentially expressed mRNAs, miRNAs, and lncRNAs. The aberrant expression of HECW2-centered ceRNAs in the peripheral blood of in-house patients was validated using quantitative PCR. We also revealed that the expression of HECW2 was positively correlated with lncRNAs LINC02593 through miRNAs miR-130a-3p, miR-130b-3p and miR-148b-3p in cells. These results show that there are distinct immune features between IS patients and healthy controls. The ceRNA network may help elucidate the mechanism of immune cell-related genes in IS and may serve as a therapeutic target.