Comprehensive Analysis of Starvation Response-Related Genes in the Diagnosis of Ischemic Stroke: Based on Machine Learning and Single-Cell RNA Sequencing Data

Abstract

Background: The pathogenesis of ischemic stroke (IS) involves a starvation response (SR); however, the mechanistic link between the two remains unclear. Therefore, it is imperative to identify SR-related gene (SRRG) biomarkers involved in IS. Methods: IS datasets were obtained from the Gene Expression Omnibus, and SRRGs were acquired from the Molecular Signatures Database. Biomarkers were screened using differential expression analysis, machine learning, and expression validation. Subsequent analyses included immune infiltration, gene set enrichment analysis (GSEA), drug prediction, and single‑cell RNA sequencing (scRNA‑seq). The middle cerebral artery occlusion mouse model and oxygen-glucose deprivation/reoxygenation (OGD/R)-treated BV2 cells were used. PAK2/Iba1 co-expression was assessed using immunofluorescence double-staining. The effects of PAK2 knockdown on cell proliferation, apoptosis, and inflammation were assessed using functional assays. Results: Two biomarkers, PAK2 and lysosome-associated membrane protein 2 (LAMP2), were identified and used to construct a diagnostic model that demonstrated good performance. Immune infiltration analysis revealed changes in 33 immune cell types. GSEA revealed that biomarker‑related genes were enriched in different pathways, including NOD‑like receptor and T‑cell receptor signaling pathways. Drug prediction analysis identified 48 potential agents targeting PAK2 and 35 agents targeting LAMP2. Microglia were identified as a key cell subtype using scRNA‑seq analysis. PAK2 mRNA was significantly upregulated, and LAMP2 mRNA was downregulated in mouse ischemic penumbra; immunofluorescence confirmed PAK2/Iba1 co-localization. In vitro, PAK2 knockdown suppressed apoptosis, promoted proliferation, and reduced inflammatory cytokine levels in OGD/R-injured BV2 cells. Conclusion: In this study, two SR‑derived biomarkers were identified for IS, and a diagnostic model was established, providing new insights into IS diagnosis and therapeutic development.