Identification of lactylation-related genes as novel biomarkers for sepsis and in vitro evidence linking NDUFB3 K93 lactylation to mitochondrial dysfunction

Purpose Sepsis is defined as life-threatening organ dysfunction caused by a dysregulated host response to infection. Given the growing emphasis on identification and individualized treatment, novel diagnostic methods and reliable biomarkers for sepsis are urgently needed. Methods Our study integrated machine learning to identify novel biomarkers and construct a diagnostic model, while also providing a preliminary exploration into the potential involvement of NDUFB3 lactylation in sepsis progression. Using three Gene Expression Omnibus (GEO) datasets (GSE28750, GSE57065, GSE95233), we identified NDUFB3 and CA4 as sepsis hub genes. A diagnostic model was subsequently constructed by machine learning using NDUFB3 and CA4, and further validation through an observational cohort study (2022NL-158-02, Aug 26, 2022). To further investigate mitochondrial changes induced by the glycolysis inhibitor 2-Deoxy-D-Glucose (2-DG), we used transmission electron microscopy (TEM), ROS fluorescence staining, and MitoTracker. MitoCarta 3.0 screening identified NDUFB3. Liquid chromatography-mass spectrometry (LC-MS) revealed specific lactylation sites on NDUFB3. We generated site-specific mutant cell lines and performed co-immunoprecipitation (Co-IP), mitochondrial membrane potential assay, and reactive oxygen species (ROS) fluorescence staining. Results By intersecting sepsis differentially expressed genes (DEGs) with lactylation-related genes (LRGs), we identified NDUFB3 and CA4 as core hub genes. Sepsis 2 cluster displays upregulated hub genes, enhanced metabolic levels, and reduced inflammatory activity. Subsequently, a diagnostic model based on these two hub genes were constructed and validated across external and clinical cohorts. Functionally, our findings suggest that lactylation of NDUFB3 at the K93 site is closely associated with mitochondrial impairment; notably, a K93 mutation reduced lactylation levels and was accompanied by a restoration of mitochondrial function. Conclusion Our study establishes a novel diagnostic signature and shows that the NDUFB3 K93 mutation alleviates sepsis-induced mitochondrial dysfunction. These findings highlight NDUFB3 as a potential biomarker and a candidate target for further investigation into mitigating mitochondrial damage in sepsis.