Abstract / Summary
Abstract Background Diabetic kidney disease (DKD) is a leading cause of end-stage renal disease. Autophagy maintains cellular homeostasis but is dysregulated in DKD. Both DKD and autophagy are closely linked to immune responses, yet the mechanistic crosstalk among them remains unclear. Methods We integrated multi‑cohort transcriptomic data and used machine learning to identify five autophagy‑associated feature genes (FOS, JUN, SOD1, CASP3, CCL2), which showed robust diagnostic performance in a training set and three external validation datasets (ROC analysis). Based on these five genes, we identified two autophagy‑related molecular subtypes of DKD with significant differences in immune response intensity and immune checkpoint expression. Single‑cell transcriptomics characterized the expression patterns and interactions of these features across renal cell populations. In DKD mouse models, we validated abnormal expression of the five genes and common autophagy markers by qPCR and Western blot, and demonstrated significant correlations between key features, autophagy indicators, and multiple renal function parameters. Results The five autophagy‑associated genes exhibited strong diagnostic value. The two DKD molecular subtypes differed markedly in immune response and checkpoint gene expression. Single‑cell analysis revealed distinct expression patterns across renal cell types. Animal studies confirmed dysregulation of the five genes and autophagy markers, as well as significant correlations with renal function parameters. Conclusions This study delineates the autophagy landscape in DKD and reveals its connection to immune microenvironment remodeling, providing potential biomarkers and strategic insights for stratified diagnosis and targeted intervention in DKD.