Integrative analysis of transcriptomics, single-cell RNA sequencing, and GraphBAN identifies de novo...

Diabetic retinopathy (DR) represents a key microvascular disorder resulting from prolonged diabetes mellitus. Hyperglycemia, a common driver of diabetes-related vision loss, can pathologically activate de novo lipogenesis (DNL); however, the underlying mechanisms remain incompletely understood. This work sought to profile signature genes correlated with DNL and DR, and offer novel insights into the disease's underlying pathogenesis.

DR-related transcripts and DNL-associated genes were obtained from public databases. Genes with DNL-related signatures associated with DR were identified through integrated bioinformatics analyses. These genes were incorporated into a nomogram, and their functional characteristics were further analyzed. The GraphBAN model was applied to predict potential gene-compound interactions, evaluate binding affinity, and assess docking stability. Single-cell RNA sequencing data were used to characterize the cellular landscape and to map gene expression patterns to specific cell populations within DR fibrous membranes, providing cellular context for the transcriptomic findings. Finally, RT-qPCR was performed to measure whole-blood expression of key genes between DR patients and control subjects.

AHR and SLC1A5 were identified as genes potentially associated with DR. RT-qPCR confirmed significant upregulation of both genes in the DR group. A nomogram established using these genes suggested that the predicted risk of DR increased with higher total scores. Functional enrichment analysis revealed that these genes were implicated in multiple biological pathways closely associated with DR pathogenesis. Integration of the GraphBAN model with compound screening predicted that two and fourteen compounds may interact with AHR and SLC1A5, respectively, and molecular docking suggested potential binding between these genes and several candidate compounds. At the single-cell level, conventional dendritic cell type 2 (cDC2) and classical monocytes showed notable transcriptional activity related to DR-associated pathways. cDC2 displayed relatively high activity across multiple metabolic pathways. During the differentiation of these cell populations, the selected genes exhibited dynamic expression patterns.

This exploratory study identified two DNL-associated candidate genes and explored their potential roles in DR-related molecular processes. These preliminary findings offer novel insights into the pathogenesis of DR and lay a foundational basis for future investigations into DNL-associated pathways and the development of therapeutic strategies targeting DR.