Abstract:AIM: To identify biomarkers and potential therapeutic targets related to angiogenesis and RNA modification in diabetic retinopathy (DR). METHODS: The Gene Expression Omnibus (GEO) datasets GSE12610, GSE87433, and GSE111465, which contain retinal samples from diabetic and normal mice, were analyzed to identify differentially expressed genes (DEGs). The DEGs were then intersected with angiogenesis- and RNA modification-related genes (A&RMRGs) obtained from GeneCards and other databases to identify differentially expressed A&RMRGs in DR. Hub genes were subsequently identified from the protein-protein interaction (PPI) network built on the enrichment results. Their diagnostic value was then estimated by receiver operating characteristic (ROC) analysis, and their expression levels were tested for associations with immune cell infiltration. Regulatory networks of hub genes were constructed using MicroRNA Target Prediction Database (miRDB), ChIPBase, and the Comparative Toxicogenomics Database (CTD). To validate the bioinformatic findings, hub gene transcript levels were quantified with reverse transcription quantitative polymerase chain reaction (RT-qPCR) in high-glucose-treated rat retinal microvascular endothelial cells (rRMECs). RESULTS: Forty-two A&RMRGs showed differential expression in the diabetic retina, with Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment implicating RNA modification and immune-inflammatory regulation, alongside Gene Set Enrichment Analysis (GSEA) evidence of activated RNA-silencing and fatty-acid-transport programs and suppressed neuronal and retinoid-cycle programs. Seven hub genes (Wdr3, Crebbp, Sdad1, Stat3, Gnl2, Nhp2, and Hsp90aa1) were selected, all of them were upregulated in DR. The DR retina showed higher levels of central memory CD8+ T cells and lower levels of monocytes, Tregs, and CD56bright natural killer (NK) cells. Monocyte infiltration was inversely associated with Stat3 expression (r=-0.604, P<0.001), and central memory CD8+ T cell levels were positively associated with Nhp2 expression (r=0.539, P=0.003). A total of 60 miRNAs, 43 transcription factors (TFs), and 47 drugs may regulate RNA modification of angiogenesis-associated genes in DR. Gnl2 [area under the ROC curve (AUC)=0.914, 95% confidence interval (CI): 0.810–1.000) and Stat3 (AUC=0.964, 95%CI: 0.892–1.000) showed high diagnostic value. Exposure of rRMECs to high glucose led to a marked rise in Stat3 expression and a fall in Sdad1, leaving the transcript levels of the remaining five hub genes unchanged. CONCLUSION: These findings identify Stat3 as a point of convergence between RNA modification, immune dysregulation, and angiogenesis, nominating the hub genes as candidate biomarkers and the associated drug–gene interactions as repurposing leads.
