Deciphering mitochondrial dysfunction in keratoconus: Insights into ACSL4 from machine learning-based bulk and single-cell transcriptome analyses and experimental validation.

Keratoconus (KC) is a prevalent ectatic corneal disease and the leading cause of corneal transplantation globally. Despite evidence of mitochondrial abnormalities in KC, the underlying mechanisms remain unclear. Our aim was to investigate the role of mitochondrial dysfunction in this pathological condition. Based on transcriptomics datasets of KC, mitochondria-related differentially expressed genes (mDEGs) were identified and analyzed for potential functional pathways, protein-protein interaction (PPI), and gene regulatory networks. Hub genes were further screened and validated by multiple machine learning (ML) algorithms, followed by a comprehensive visualization of single-cell atlas and immune landscape. Additionally, bioinformatic results were validated through quantitative PCR, Western blot, and transcriptomics analysis in an in vitro KC model based on matrix stiffness using human stromal keratocytes. In total, 104 mDEGs were identified, enriched in pathways related to oxidative stress, apoptotic mitochondrial changes, ferroptosis, and inflammatory responses. Nine characteristic genes (CYP24A1, ACSL4, ACADL, HELZ2, AMT, DEPTOR, TUBA1A, TYMS, and ACSL5) were selected and validated using multiple ML models. Single cell sequencing data highlighted ACSL4 as the most promising biomarker, primarily expressed in corneal stromal cells (CSCs). Immune infiltration analysis revealed that ACSL4 was positively associated with monocytes and negatively correlated with eosinophils in KC. In cellular experiments, ACSL4 expression was significantly upregulated in response to decreased substrate stiffness, suggesting its critical role in KC development. These findings suggest a mitochondrial-related molecular mechanism implicated in KC pathogenesis. The identified pivotal biomarker ACSL4 provides a novel framework for future mechanistic and therapeutic studies of KC.