Type 2 diabetes mellitus (T2DM) is a metabolic disorder characterized by impaired regulation of blood glucose. One treatment strategy to decrease postprandial glucose responses is the inhibition of carbohydrate-digesting enzymes, such as α-glucosidases and α-amylase, requiring specific targeting by small molecules. Here, we used modeling and simulation techniques to predict and test potential inhibitors for these enzymes. We applied the site identification by ligand competitive saturation (SILCS) method using both polarizable (Drude) and nonpolarizable (CHARMM) force fields to generate the FragMaps and grid-free energies (GFEs) for the N- and C-terminal subunits of maltase-glucoamylase (MGAM) and pancreatic α-amylase. FragMaps are used to show the spatial arrangement of functional groups surrounding the enzymes, while GFEs estimate their binding affinity. The results suggest that the Drude polarizable model better captures hydrogen bonding and hydrophobic interactions compared to CHARMM, showcasing the significance of polarization in protein-ligand modeling. We tested known inhibitors, acarbose and miglitol, using SILCS-Monte Carlo (MC) pose refinement, and the predicted ligand grid free energies correctly identified the stronger inhibitor for each enzyme. We then built SILCS-based pharmacophore models and performed virtual screening with the Pharmit web server against multiple drug libraries. We ranked candidate compounds using SILCS-MC pose refinement and chose the top hits for further evaluation. We conducted CHARMM and Drude parametrization for the potential drug candidates and ran molecular dynamics simulations to examine their stability and binding in the enzyme active sites. This study sheds light on the inhibition of key enzymes in T2DM and provides guidance for developing new compounds with antidiabetic potential.
Mohammadi et al. (Sun,) studied this question.