A lactylation-related gene signature for the diagnosis of diabetic foot ulcers identified by machine learning and single-cell analyses

Diabetic foot ulcers (DFUs) are serious diabetes complications. Although histone lactylation has been linked to diabetic conditions, the pattern and potential diagnostic value of lactylation-related genes in DFU remain largely unexplored. Bulk RNA sequencing datasets and 376 lactylation-related genes were analyzed to identify DFU-associated lactylation-related genes (LRGs). Machine learning built a diagnostic model. Consensus clustering defined DFU subtypes. The cellular landscapes of DFUs were explored by single-cell RNA-sequencing (scRNA-seq) datasets. The gain- and loss-of-function experiments of the hub gene PFKFB2 were validated using a high-glucose-induced HUVEC model. A total of 10 differentially expressed LRGs were identified between the DFU and NC groups. Consensus clustering stratified DFU samples into two lactylation-related subtypes with distinct immune infiltration patterns and pathway enrichment. Based on machine learning methods, five LRGs (ADH1B, ARTN, KY, PFKFB2, and PFKFB4) were identified to construct the diagnostic model for DFU. A nomogram constructed from these five LRGs achieved high diagnostic accuracy in a bulk RNA set. Increased lactylation signature across all nine DFU-associated cell types, particularly in myeloid cells, fibroblasts, and epithelial cells. PFKFB2 was downregulated in vitro HUVECs model. The overexpression of PFKFB2 reduced the levels of lactate and histone lactylation and inhibited endothelial cell migration and tube formation, whereas its knockdown promoted these processes. This study established a lactylation-related diagnostic model for DFU based on five LRGs, highlighting the potential diagnostic and therapeutic target. Furthermore, PFKFB2 was identified as a key regulator, revealing a novel lactylation-associated metabolic-epigenetic mechanism in DFU.