Programmed cell death-driven molecular subtyping identifies ARPC1B as a potential immune-fibrotic biomarkers in diabetic nephropathy via multi-model machine learning

Abstract Background Programmed cell death (PCD) is closely associated with the occurrence and progression of diabetic nephropathy (DN). This study aims to explore the diagnostic biomarkers of PCD in the DN and its underlying mechanisms. Methods DN-related datasets were downloaded from the GEO database, and PCD-related genes were collected. PCD-related subtypes were identified through cluster analysis, and differentially expressed genes between subtypes were determined. Subtype-related genes were identified using WGCNA, and key genes were identified through multiple machine learning algorithms. The expression levels of the key genes were validated in several datasets, and their diagnostic values were assessed using ROC curves. The biological processes involving these genes were analyzed through KEGG, GO, and GSEA, and their associations with immune cells were examined. The expression levels of ARPC1B and IGFBP6 in the HK-2 cell line cultured in high-glucose medium were verified by PCR. Functional knockdown experiments were performed for ARPC1B to examine its association with epithelial-mesenchymal transition (EMT) related changes. Results By acquiring DN-related datasets from the GEO database and integrating multiple PCD-associated genes, we identified PCD subtypes (P1 and P2) through cluster analysis. Results revealed significant differences in PCD scores and immune cell infiltration levels between the two subtypes. WGCNA analysis revealed that the yellow module was closely associated with PCD subtypes. Further integration of multiple machine learning algorithms identified two potential genes, ARPC1B and IGFBP6. Additionally, functional analysis indicated that both genes participate in diverse biological processes and are associated with multiple immune cell types. PCR experiments confirmed that the mRNA expression levels of both ARPC1B and IGFBP6 were elevated in the HK-2 cell line cultured in conditions with high glucose. ARPC1B knockdown was associated with changes in EMT-related markers, providing preliminary functional evidence for its involvement in EMT-related processes. Conclusion Integrative transcriptomic analyses prioritized ARPC1B and IGFBP6 as PCD-associated candidate biomarkers for DN. Preliminary functional evidence supports an association between ARPC1B and EMT-related changes.