VIRTUAL SCREENING OF THE ZIMBABWE NATURAL PRODUCT DATABASE FOR GLUCOKINASE ACTIVATORS

Objective: This study aimed to identify potential glucokinase activators within Zimbabwean natural products using virtual screening techniques. Methods: Twenty-one compounds filtered from ChEMBL ID 3820 (pEC50 ≥ 8) were used to generate a pharmacophore model, validated with DUD-E data. The model screened the 6220 compounds in the Zimbabwe Natural Products Database (ZiNaPoD) using LigandScout. Hit compounds were docked with glucokinase (protein ID 4NO7) using AutoDock Vina and AutoDock 4 in PyRx, followed by adsorption, distribution, metabolism, and excretion (ADME) screening by SwissADME. Molecular dynamics simulations were conducted on the resulting complexes using the CHARMM36m force field on GROMACS. Results: The validated pharmacophore model (80% accuracy, 95% sensitivity, 80% specificity) produced 149 hits, 16 of which had binding energies ≤ -8 kcal/mol after the two rounds of molecular docking. The ADME analysis narrowed the selection to four compounds, with binding energies ranging from -8.35 to -9.82 kcal/mol. All four demonstrated stability in molecular dynamic simulations, with average root mean square deviation (RMSD) values ranging from 1.491 to 3.835 Å. The Sphenostylisin I and Dihydroxymethyl dihydroxybenzyl chromanone (DMDBC) complexes exhibited the highest stability with average RMSD values of 1.491±2.794 Å and 2.875±1.452 Å, respectively. They also exhibited low-binding free energies of -30.30±0.38 and -30.20±0.49 kcal/mol, making them promising targets. Conclusion: Four potential glucokinase activators were identified, with Sphenostylisin I and DMDBC showing promise as candidates for developing new diabetes treatments due to their stability, favorable binding, and absence of liver-toxic groups.