This study aimed to generate a competitive endogenous RNA (ceRNA) network and identify novel biomarkers for diabetic kidney disease by combining single-cell RNA-seq (scRNA-seq) and bulk RNA-seq data analysis. Methods: Four datasets associated with diabetic nephropathy were downloaded, and their differentially expressed genes were identified. Enrichment analysis was conducted using Metascape. Next, we identified co-expressed mRNA-lncRNA pairs, analyzed subcellular lncRNA localization, and constructed a ceRNA network. Next, a specific protein-protein interaction was used to identify key biomarkers. Then, we determined the expression profiles of key biomarkers from scRNA-seq data and validated in an external independent dataset. Finally, we explored the correlations between key biomarkers and the clinical features and analyzed the drug perturbation. Results: Thirteen cell types were identified in the scRNA-seq data, and 106 common differentially expressed genes were identified in both scRNA-seq data and bulk RNA-seq data. These genes were mainly enriched in the focal adhesion pathway of the extracellular matrix. Next, we overlapped these genes with genes predicted by differentially expressed miRNAs, identified co-expressed mRNA-lncRNA pairs and generated a ceRNA network containing 60 mRNAs, 10 miRNAs, and 5 lncRNAs. All lncRNAs were localized in the cytoplasm or cytosol. Analyses identified five key biomarkers (VCAN, TIMP1, TNC, C3 and CP); these proteins were mostly differentially expressed in fibroblasts and renal tubular cells. Furthermore, most of these proteins are associated with glomerular filtration rate, serum creatinine, and proteinuria. In total, seven potential therapeutic molecular medicines were predicted for the treatment of diabetic nephropathy.
Conclusion:In summary, we identified a novel ceRNA-network and five key biomarkers in diabetic kidney disease, which may help to elucidate the mechanisms underlying diabetic nephropathy and facilitate new treatments.