Identification of extracellular matrix stiffness-related genes in human scleral fibroblasts through bioinformatics analysis and machine learning.

BACKGROUND

Scleral extracellular matrix (ECM) remodeling plays a key role in myopia development. This study aims to identify ECM stiffness-related genes in human scleral fibroblasts (HSFs) using RNA sequencing, bioinformatics analysis, and machine learning.

METHODS

HSFs were cultured on soft and stiff polyacrylamide hydrogels (mimics myopic and normal extracellular matrix). RNA sequencing was performed on HSFs, and differentially expressed genes (DEGs) were identified. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis, gene set enrichment analysis (GSEA), protein-protein interaction (PPI) network, and machine learning algorithms were used to explore the ECM stiffness-related key genes. Nomogram and receiver operating characteristic curve were constructed to assess the diagnostic value in HSF sequencing dataset from myopia patients. Differential microRNAs (miRNAs) between myopia and normal samples in GSE131831 were analyzed, and miRNA-messenger RNA network was also developed.

RESULTS

The dataset included 1368 upregulated and 1481 downregulated DEGs, which were primarily associated with change of scleral ECM stiffness. Five key genes (LDLR, DNM1, DUSP15, NES, and PPL) were identified. DNM1 was the best marker for myopia diagnosis.

CONCLUSION

LDLR, DNM1, DUSP15, NES, and PPL were identified as ECM stiffness-related genes in HSFs, of which DNM1 was the optimal diagnostic marker for myopia. The association of these five genes with myopia and scleral remodeling warrants further investigation.