Lactylation modulates immune infiltration in sepsis-induced acute respiratory distress syndrome: a multi-omics and machine learning study with experimental confirmation

OBJECTIVES: To investigate lactylation-driven mechanisms in the pathogenesis of sepsis-induced acute respiratory distress syndrome (ARDS). METHODS: Multi-cohort transcriptomic data sets (GSE10474, GSE32707, and GSE66890) were integrated with machine learning algorithms (LASSO, support vector machine, random forest) to identify differentially expressed lactylation-related genes (LRGs). Five hub genes (ALDH1A1, CALM1, CCNA2, HIST1H2BN, SH3GL1) were prioritized. Orthogonal experimental validation was performed using qRT-PCR and Western blotting. Subsequent analyses explored immune cell correlations (focusing on ALDH1A1), regulatory networks (transcription factors and miRNAs), and potential therapeutic drug candidates. RESULTS: Integration of bioinformatics analyses identified 25 differentially expressed LRGs and prioritized 5 hub genes. Experimental validation (qRT-PCR/Western blot) consistently demonstrated downregulation of all five hub gene proteins. Notably, this contradicted the bioinformatically predicted upregulation of CCNA2, HIST1H2BN, and SH3GL1, revealing a significant transcriptional-translational discordance. Further analysis revealed ALDH1A1-associated myeloid-derived suppressor cell and neutrophil correlations, a regulatory network comprising 68 transcription factors and 79 miRNAs, and 26 prioritized drug candidates. CONCLUSIONS: This study establishes a protein-verified lactylation-related gene signature with diagnostic relevance for sepsis-induced ARDS. Crucially, it highlights post-transcriptional regulation, evidenced by the discordance between mRNA levels and protein expression of key hub genes, as a key mechanistic feature in the pathogenesis of this condition. The identified regulatory networks and drug candidates provide potential avenues for further research and therapeutic development.