Immunometabolic reprogramming and glycolysis-associated signatures in sepsis: insights from single-cell RNA sequencing and machine learning

Background: Immunometabolic remodeling is central to sepsis, yet robust glycolysis-associated biomarkers and their cell-type context remain unclear. Methods: We integrated peripheral blood scRNA-seq (GSE175453) and a whole-blood microarray cohort (GSE100159). Glycolysis activity was scored by AUCell (HALLMARK_GLYCOLYSIS), hub genes were prioritized by LASSO/random forest/Boruta, communication was inferred by ligand-receptor analysis, and qRT-PCR was performed in CLP vs control mice. Results: Sepsis showed myeloid predominance and increased glycolysis scores, most evident in monocytes and plasma cells. Five candidates (GLRX, MDH1, MDH2, TGFBI, COPB2) displayed good discriminatory performance in bulk data; TGFBI was monocyte-enriched and centrally positioned in a dense communication network with B cells/plasma cells/neutrophils. qRT-PCR confirmed a significant between-group difference for TGFBI in the CLP model. Conclusions: These findings link enhanced glycolysis-associated programs to a monocyte-centered TGFBI communication pattern and prioritize TGFBI as a candidate biomarker for further validation in sepsis.