Multi-Omics and Clinical Validation Identify Key Glycolysis- and Immune-Related Genes in Sepsis

Background: Sepsis is characterized by profound immune and metabolic perturbations, with glycolysis serving as a pivotal modulator of immune responses. However, the molecular mechanisms linking glycolytic reprogramming to immune dysfunction remain poorly defined. Methods: Transcriptomic profiles of sepsis were obtained from the Gene Expression Omnibus. Differentially expressed genes (DEGs) related to glycolysis were identified through a combination of ssGSEA, WGCNA and differential expression analysis. Hub genes were prioritized using Mendelian randomization and machine learning algorithms (LASSO, SVM-RFE, and Boruta), and validated in an independent dataset and by RT-qPCR in a clinical sepsis cohort. Immune cell infiltration was assessed using CIBERSORT to profile the immune landscape, and single-cell RNA sequencing (scRNA-seq) was employed to delineate the cell type-specific transcriptional profiles. Results: The ssGSEA scores derived from the glycolysis signature indicated a marked reduction in glycolytic activity associated with sepsis. By employing an integrative framework that includes WGCNA, differential expression analysis, Mendelian randomization, and machine learning algorithms, this study successfully identified five pivotal genes associated with glycolysis: DDX18, EIF3L, MAK16, THUMPD1, and ZNF260. The diminished expression of these genes was significantly correlated with immune remodeling, characterized by an increase in neutrophils and a decrease in lymphocytes. In a clinical sepsis cohort, RT–qPCR of peripheral blood, in conjunction with routine hematological profiling, validated their expression pattern and immune associations. Moreover, scRNA-seq facilitated a comprehensive characterization of these transcriptional alterations within distinct subsets of immune cells. Conclusion: This study identifies five glycolysis-related genes linked to immune remodeling in sepsis, revealing a metabolic–immune axis that may drives disease pathogenesis and offers promising targets for therapeutic intervention. Keywords: sepsis, glycolysis, Mendelian randomization, machine learning, single-cell RNA sequencing