Modeling the Binding Affinity of p38α MAP Kinase Inhibitors by Partial Least Squares Regression

Dibenzepinones, dibenzoxepines and benzosuberones based p38α MAP kinase inhibitors: Their pharmacophore modelling, 3D-QSAR and docking studies

In the present study, a series of dibenzepinones, dibenzoxepines, and benzosuberones targeting p38α MAP kinase were subjected to pharmacophore modelling, 3D-QSAR and molecular docking studies. The IC₅₀ values for these 67 compounds ranged between 0.003 and 6.80 μM. A five-point model (DDHHR.8) was generated using these compounds. This model was found to be statistically significant and was found to have high correlation (R² = 0.98), cross-validation coefficient (Q² = 0.95) and F (330) values at six component PLS factor. Tests were performed to ascertain the efficacy of the generated model. These tests included external validation, Tropsha's test for predictive ability, Y-randomisation test and domain of applicability (APD). In order to check the restrictivity of the model, enrichment studies were performed with inactive compounds by using decoy set molecules. To evaluate the effectiveness of the docking protocol, the co-crystallised ligand was extracted from the ligand-binding domain of the protein and was re-docked into the same position. Both the conformers were then superimposed, suggesting satisfactory docking parameters with an RMSD value of less than 1.0 Å (0.853 Å). A 10 ns molecular dynamics simulation confirmed the docking results of the 3UVP-ligand complex and the presumed active conformation. The outcome of the present study provides insight into the molecular features that promote bioactivity and can be exploited for the prediction of novel potent p38α MAP kinase inhibitors before carrying out their synthesis and anticancer evaluation.