A novel palmitoylation-based molecular signature reveals COX6A1 as a key regulator in metabolic dysfunction-associated steatotic liver disease.

BACKGROUND: Metabolic dysfunction–associated steatotic liver disease (MASLD) is highly prevalent, but the molecular links between palmitoylation, mitochondrial function, and immune remodeling remain unclear. METHODS: We integrated four GEO bulk RNA-seq cohorts (GSE126848, GSE130970, GSE135251, GSE213621) and one single-cell RNA-seq cohort (GSE136103). Expression matrices were normalized to TPM, log-transformed, and batch-corrected with ComBat. A curated set of palmitoylation-related genes (PRGs) was assembled from GeneCards and used for consensus clustering (ConsensusClusterPlus) to define MASLD subtypes. Differential expression (limma; |log2FC|>0.5, P < 0.05), functional enrichment (clusterProfiler for GO/KEGG), and GSEA were performed. WGCNA identified PRG-associated modules; hub genes were prioritized using a machine-learning pipeline (12 algorithms; 113 model combinations) with 10-fold cross-validation in GSE213621 and external validation in GSE126848, GSE130970, and GSE135251. Immune infiltration was inferred using xCell, EPIC, MCP-counter, QUANTISEQ, CIBERSORT, and TIMER. In vitro, HepG2 cells exposed to palmitic acid were transfected with COX6A1 siRNA; lipid accumulation (Oil Red O/Nile Red), mitochondrial membrane potential (JC-1), mitochondrial ROS (MitoSOX), apoptosis (Annexin V/PI flow cytometry), and COX6A1/NDUFA4 expression (qRT-PCR/Western blot) were assessed. Performance was evaluated by ROC AUC and decision-curve analysis. RESULTS: Two MASLD subtypes, Cluster 1 and Cluster 2, were identified, with Cluster 1 enriched in metabolic pathways and Cluster 2 showing immune activation and mitochondrial metabolism pathways. Hub genes, such as COX6A1, COX7A2, and NDUFA4, were identified with diagnostic potential (AUC > 0.75). Single-cell analysis revealed differential immune cell infiltration, with Cluster 2 showing increased immune activity. COX6A1 knockdown in vitro alleviated palmitic acid-induced mitochondrial dysfunction and apoptosis, suggesting its potential as a therapeutic target. CONCLUSION: Our study found that COX6A1 is closely related to MASLD status and molecular subtypes, and it shows strong diagnostic value across independent cohorts. COX6A1 is associated with abnormal mitochondrial redox metabolism and immune regulation—particularly macrophage signaling—and its modulation directly affects lipotoxic injury (mtROS, ΔΨm, apoptosis) in hepatocyte models. Together, these findings nominate COX6A1 as a mechanistically grounded biomarker and a potential therapeutic target for MASLD. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12967-025-07253-0.