AI-Integrated Multi-Target Validation of Coreopsis tinctoria Polyphenols as a Functional Food Ingredient Against Diabetic Nephropathy

Diabetic nephropathy (DN) is a severe diabetic complication with substantial clinical burden. The complex pathogenesis of DN has hindered the development of targeted therapies, creating an urgent need to develop novel strategies that directly address its underlying inflammatory and fibrotic mechanisms. Coreopsis tinctoria (CE) is an edible plant rich in polyphenols, but its mechanism against DN remains understood. An integrated framework combining network pharmacology and machine learning was developed to prioritize active polyphenols and their targets. A multi-layer perceptron classifier, trained on 3.16 million compound–target pairs from Binding DB, predicted interactions between 36 CE polyphenols and 12,030 DN-associated genes. The top 100 targets were subjected to KEGG enrichment analysis, and the identified pathways were validated in a high-fat diet/STZ-induced DN rat model. The MLP model achieved superior performance (AUC-ROC = 0.9219, AP = 0.9592). Five lead polyphenols (flavonoids/chalcones) showed high predicted activity. KEGG analysis revealed enrichment in PI3K-Akt, calcium signaling, metabolic pathways, and cellular senescence. In vivo, CE treatment (150–600 mg/kg/day) dose-dependently improved glucose/lipid metabolism and renal function, and ameliorated histopathological damage, including glomerular hypertrophy, fibrosis, and mesangial expansion. Mechanistically, CE suppressed NFκB/TGFβ/Smad signaling, restored PPARγ and Nrf2/HO-1/FoxO1 antioxidant defenses, and inhibited apoptosis via Bcl-2/Bax regulation. CE exerts multi-target renoprotective effects through coordinated modulation of metabolic, inflammatory, fibrotic, and antioxidant pathways, supporting its potential as a functional food ingredient for DN management.