ABSTRACT: Dihydrofolate reductase (DHFR) is an essential cellular enzyme and thereby catalyzes thereduction of dihydrofolate to tetrahydrofolate (THF). In cancer medication, inhibition of humanDHFR (hDHFR) remains a promising strategy, as it depletes THF and slows DNA synthesis and cellproliferation. In the current study, ligand-based pharmacophore modeling identified and evaluatedthe critical chemical features of hDHFR inhibitors. A pharmacophore model (Hypo1) was generatedfrom known inhibitors of DHFR with a correlation coefficient (0.94), root mean square (RMS)deviation (0.99), and total cost value (125.28). Hypo1 was comprised of four chemical features,including two hydrogen bond donors (HDB), one hydrogen bond acceptor (HBA), and onehydrophobic (HYP). Hypo1 was validated using Fischer's randomization, test set, and decoy setvalidations, employed as a 3D query in a virtual screening at Maybridge, Chembridge, Asinex,National Cancer Institute (NCI), and Zinc databases. Hypo1-retrieved compounds were filtered byan absorption, distribution, metabolism, excretion, and toxicity (ADMET) assessment test andLipinski's rule of five, where the drug-like hit compounds were identified. The hit compounds weredocked in the active site of hDHFR and compounds with Goldfitness score was greater than 44.67(docking score for the reference compound), clustering analysis, and hydrogen bond interactionswere identified. Furthermore, molecular dynamics (MD) simulation identified three compounds asthe best inhibitors of hDHFR with the lowest root mean square deviation (1.2 Å to 1.8 Å), hydrogenbond interactions with hDHFR, and low binding free energy (-127 kJ/mol to -178 kJ/mol). Finally,the toxicity prediction by computer (TOPKAT) affirmed the safety of the novel inhibitors of hDHFRin human body. Overall, we recommend novel hit compounds of hDHFR for cancer and rheumatoidarthritis chemotherapeutics.
