3D-QSAR Study of Human DHFR Inhibitors Based on GRIND Descriptors and Designing of New Inhibitors

Dihydrofolate reductase (DHFR), an indispensable enzyme in the folate metabolism, converts dihydrofolate (DHF) to tetrahydrofolate (THF) in the presence of NADPH found in all organisms and regulates the level of THF in the cell. DHFR has been a validated drug target for the development of therapeutics for human cancer for decades [1]. Hence, there is an urgent need to identify and develop new inhibitors with high potency and selectivity relative to a specific species of DHFR. In this regard, 3D-QSAR modeling is one of the powerful computational methods employed in medicinal chemistry for deciding about the structure-activity relationship of new chemical compounds. In the present study, we studied a data set including 2, 4-diaminopyrimidine scaffold with the acetylenic linker to another substituted aryl ring with human DHFR receptor for 3D-QSAR modeling.Moreover, we implemented a 3D-QSAR model based on GRIND descriptors on a series of 36 dihydrofolate reductase inhibitors. Genetic algorithm (GA) variable selection technique was applied to reduce the number of the molecular descriptors. The correlation between molecular descriptors and biological activity was established using PLS regression method. To assess the quality of the model, internal and external validations were implemented by different statistical procedures [2]. The results show that all the calculated external/internal validation parameters are in good agreement with acceptance criteria. New ligands were designed based on results of 3D-QSAR studies. Designed compounds were evaluated by docking and their biological activities were predicted by the 3D-QSAR model. Finally, ADME/Tox parameters of new ligands were predicted.