Toward resistance-compromised DHFR inhibitors part 1: Combined structure/ligand-based virtual screenings and ADME-Tox profiling

• Published in: Journal of chemometrics 2016-08, Vol.30 (8), p.462-481
• Main Authors: Gahtori, Prashant, Pandey, Rahul, Kumar, Vinod, Ghosh, Surajit Kumar, Das, Aparoop, Kalita, Jun Moni, Sahu, Supriya, Prakash, Anil, Bhattacharyya, Divya Ranjan
• Format: Article
• Language: English
• Subjects: 1,3,5‐triazines; 5-triazines; ADMET; Antimalarial; Binding sites; CBScore; Computation; Correlation analysis; CScore; Dihydrofolate reductase; Drugs; GARD; Hydrogen bonds; LibDock; Ligands; Mathematical models; Metabolism; Pharmacology; pharmacophore; Scoring; scoring functions; Screening
• ISSN: 0886-9383; 1099-128X
• DOI: 10.1002/cem.2814

The front‐line antimalarial drugs, for example, chloroquine, mefloquine, sulfadoxine, pyrimethamine, atovaquone, and artemether, are often failing because of the worldwide spread of drug‐resistant parasites. There has been significant recent interest in virtual screening to drive innovative drug discovery and to combat resistance efforts for a wide range of diseases. In fact, virtual screening has become the “gold standard” for major pharmaceutical industries and some university groups. Therefore, we present herein a structure‐based LibDock/CHARMM modeling and a set of appropriate scoring function evaluation criteria: correlation, consensus score, correlation‐based score, generally applicable replacement for root‐mean‐square deviation using a training set of 38 phenylthiazolyl‐1,3,5‐triazines from our previous reports and followed by a ligand‐based model to identify molecular features like hydrogen‐bond acceptor, hydrogen‐bond donor, hydrophobicity, and ring aromatic (RA) using Catalyst HipHop/HypoGen module. Next, TOPKAT module was applied to predict ADME‐Tox properties. The combined structure/ligand‐based approaches inadvertently arrived at a conserved Arg122 binding site from reliable LigScore1_dreiding top scoring function and are subsequently attributed to reserve important interactions and combat mutational drug resistance. The best pharmacophore model suggested that 1 hydrogen‐bond acceptor, 2 hydrophobicities, and 1 ring aromatic feature with good sensitivity at 0.50, specificity at 0.66, enrichment at 1.60, and accuracy at 0.50. Finally, good pharmacokinetics, metabolic stability, and toxicity endpoints were predicted in the comparison of proguanil and cycloguanil. These druggability insights are useful for researchers to deliver more effective, safer, both wild‐type and resistance‐compromised, and more economical dihydrofolate reductase inhibitors in the near future. Scientists and academicians need to adopt a faster, more economical, and user‐friendly computational methods to drive innovative drug discovery projects. As part of our work on computational methods, we introduce herein a generalized procedure for a structure‐based docking and scoring, a ligand‐based pharmacophore, and pharmacokinetic‐metabolic stability‐toxicity profiling for lead optimization.