In Silico Cavity Detection-Guided Docking Reveals Epigallocatechin Gallate as a Multitarget Phytochemical and Ursolic Acid as a Selective SGLT2/PPARγ Modulator in Type 2 Diabetes Mellitus Polypharmacology

Type 2 diabetes mellitus (T2DM) is a complicated metabolic disease that has been linked to dysfunction of several therapeutic targets, such as α-glucosidase, sodium–glucose cotransporter 2 (SGLT2), peroxisome proliferator-activated receptor gamma (PPARγ), and dipeptidyl peptidase-4 (DPP-4). Small compounds with the capacity to modify these proteins can be found in natural phytochemicals, and computational methods offer a quick and economical way to assess their potential. This study used CB-Dock, which combines cavity detection with AutoDock Vina-based docking, to evaluate the interaction of four phytochemicals: fisetin, baicalein, epigallocatechin gallate (EGCG), and ursolic acid. Cavity volume and coordinates were used to characterize the binding sites, and centroid–pocket distance measurements were used to confirm that the docking postures were correctly localized within the anticipated active sites. Baicalein showed positive interactions with DPP-4 (–9.0 kcal/mol), ursolic acid showed selective binding to SGLT2 (–9.4 kcal/mol), and EGCG was shown to be a strong multitarget binder with high affinity for PPARγ (–10.0 kcal/mol) and α-glucosidase (–9.2 kcal/mol) according to docking studies. Fisetin's function as a secondary multitarget chemical was supported by its moderate but steady binding across a variety of proteins. In addition to hydrophobic and π–π stabilizations mediated by PHE, residue-level interaction analysis identified conserved polar interactions, such as hydrogen bonding with HIS, ASP, and ARG. In Discovery Studio, protein–ligand complexes were further investigated. There, 2D interaction diagrams and 3D pocket visualizations verified that important contacts remained within validated cavities. Together, these results show that ursolic acid is a selective SGLT2 binder, baicalein is a promising DPP-4 inhibitor, and EGCG is a multitarget phytochemical candidate. They also show that docking combined with cavity detection, centroid validation, and interaction mapping offers a strong framework for describing natural product interactions with proteins linked to diabetes mellitus.