AI-guided design of leucopelargonidin-derived ligands targeting glucokinase regulatory protein: Quantum descriptors, binding dynamics, and network pharmacology

Type 2 diabetes mellitus is a multifactorial metabolic disorder that requires therapeutic strategies extending beyond glucose lowering alone. In this study, leucopelargonidin was used as a natural-product-inspired scaffold for AI-guided design of glucokinase regulatory protein (GKRP) ligands. Pharmacodia CyberAIDD generated a focused library of 2000 derivatives, which was prioritized by docking, drug-likeness, and ADMET screening. Docking against GKRP (PDB ID: 4OP1 ) identified three principal leads, P281AM7755, P281AM8215, and P281AM13980, each showing a Glide score of approximately -12.8 kcal/mol and reproducing the key interaction pattern of the crystallographic pocket. These ligands were advanced to 200 ns molecular dynamics simulations, where P281AM7755 and P281AM13980 showed better pocket retention, restrained fluctuations, and more coherent residue-coupling behaviour than P281AM8215. Principal component analysis, dynamic cross-correlation matrix analysis, and free energy landscape mapping further indicated the conformational stability trend P281AM7755 > P281AM13980 > P281AM8215. Density functional theory showed that the conserved aminopyridyl-sulfonamide-piperazine core was the principal electronically active region, with the HOMO-LUMO gap trend P281AM8215 < P281AM13980 < co-crystal < P281AM7755, supporting greater electronic stability for P281AM7755. Network pharmacology linked the prioritized derivatives to PI3K-Akt, insulin signalling, insulin resistance, AMPK, mTOR, FoxO, and cAMP pathways. Overall, P281AM7755 emerged as the most promising computational GKRP-directed lead, with P281AM13980 as a credible secondary candidate for experimental validation.