Identification of angiogenesis-related hub genes in proliferative diabetic retinopathy via integrated transcriptomics and machine learning

Le résumé fourni par la source

BACKGROUND: Proliferative diabetic retinopathy (PDR), characterized by pathological neovascularization, is one of the leading causes of blindness in diabetic patients. A better understanding of its molecular mechanisms is crucial for identifying novel biomarkers and therapeutic targets. METHODS: The dataset GSE102485 was retrieved from the GEO database to identify differentially expressed genes (DEGs). Weighted gene co-expression network analysis (WGCNA) was applied to detect PDR-associated gene modules. GO and KEGG enrichment analyses were performed using SangerBox. A protein-protein interaction (PPI) network was constructed, and core genes were identified using MCODE and cytoHubba plugins within Cytoscape. Key genes were further refined through LASSO regression. Immune cell infiltration was assessed using the MCPcounter algorithm. Receiver operating characteristic (ROC) curves and gene expression levels were evaluated in both training and external validation datasets. Western blotting, transwell, tube formation, reactive oxygen species (ROS) detection, and ELISA were performed to verify the role of the hub gene laminin subunit beta 1 (LAMB1) in high glucose (HG)-induced human retinal endothelial cells (HRECs). In addition, LAMB1 expression was examined under hypoxic conditions to assess its responsiveness to microenvironmental stimuli relevant to PDR. RESULTS: A total of 5,275 DEGs were identified, including 3,453 upregulated and 1,822 downregulated genes. Among these, 468 were retained in PDR-associated WGCNA modules. A total of 105 overlapping genes were obtained by intersection of DEGs, WGCNA modules and angiogenesis-related targets. PPI network analysis and LASSO algorithm identified six hub genes: LAMB1, COL15A1, ITGA1, THBS1, POSTN, and COL5A1. These genes correlated with altered immune infiltration. ROC analysis confirmed their predictive values in both the training and validation datasets. Functional assays demonstrated that silencing LAMB1 suppressed angiogenesis in HG-treated HRECs. CONCLUSION: This study identified angiogenesis-related hub genes in PDR using transcriptomics and machine learning, and demonstrated that LAMB1 promoted angiogenic activity in HRECs, offering a potential therapeutic target for PDR.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.