Network pharmacology, single-cell transcriptomics, machine learning and experimental validation identifying CHEK2 and...

Diabetic nephropathy (DN) is a severe microvascular complication of diabetes with limited therapeutic options. Berberine exhibits renoprotective potential, yet its precise molecular targets and mechanisms of action in DN remain unclear. Berberine targets were retrieved from TCMSP, SEA, and SwissTargetPrediction databases. Bulk transcriptomic microarray datasets (GSE96804, GSE30122) and single-cell RNA sequencing (scRNA-seq) dataset (GSE131882) were obtained from the Gene Expression Omnibus (GEO). Differentially expressed genes (DEGs) and weighted gene co-expression network analysis (WGCNA) were applied to GSE96804, followed by single-cell analysis and high-dimensional WGCNA (hdWGCNA) on podocytes. Key targets were refined via protein-protein interaction (PPI) network topology, random forest, and SHAP analysis. Molecular docking, 100 ns molecular dynamics (MD) simulations, and MM/PBSA calculations were performed to elucidate binding mechanisms. In vitro validation used human podocytes exposed to high glucose (30 mM, 48 h) ± berberine, with knockdown and overexpression plasmids transfected for rescue experiments. Single-cell analysis identified a total of 11 renal cell types; in DN, the number of podocytes was significantly reduced, and their intercellular communication was the most intense. CHEK2 and HPGD were identified as key targets-CHEK2 expression is upregulated in DN, whilst HPGD expression is downregulated-and both targets demonstrated robust diagnostic performance using nomogram, receiver operating characteristic curves, and decision curve analysis. Molecular docking revealed high binding affinities (CHEK2: -8.8 kcal/mol; HPGD: -9.9 kcal/mol), supported by stable molecular dynamics trajectories and favourable MM/PBSA binding free energies. In vitro, berberine normalised CHEK2 and HPGD expression, improved cell viability, reduced apoptosis, and restored structural markers of podocytes (Nephrin, Podocin, WT1), inhibited fibrotic proteins (Collagen I, fibronectin, α-SMA), and alleviated oxidative stress (decreased ROS, decreased MDA, increased SOD, increased GSH-Px). Importantly, rescue experiments demonstrated that the renoprotective effects of berberine were functionally dependent on CHEK2 and HPGD modulation in this in vitro model: CHEK2 gene silencing or HPGD overexpression further enhanced berberine's protective effects, whereas CHEK2 overexpression or HPGD gene silencing abolished this protection and restored the cellular phenotype to that observed under hyperglycaemic conditions. This comprehensive multi-omics study, incorporating rescue experiments, identified CHEK2 and HPGD as functionally important candidate mediators of berberine's renoprotective effects in DN, providing preliminary mechanistic insights to inform precision diagnosis and targeted therapeutic strategies.