Atomic 3D-QSAR study based on pharmacophore modeling of resveratrol derivatives as selective COX-2 inhibitors

This work, based on pharmacophore modeling, aimed to elucidate the three-dimensional structural features of resveratrol derivatives by generating the three-dimensional common pharmacophore responsible for the binding to the inflammation-inducible target. A five points pharmacophore, with three hydrogen bond acceptors and two aromatic rings as pharmacophoric features, was developed. The fit values of the generated pharmacophore are high, except for the Resv34 compound, which could be due to its lowest activity toward cyclooxygenase-2. This pharmacophore yielded a statistically significant atom-based three-dimensional-quantitative structure-activity relationships model with values of R (2), F and standard deviation of 0.9735, 165.4, and 0.2058, respectively, for a training set of twenty-three molecules. Obtained data for a test set of nine compounds with 0.9343, 0.3676, and 0.9694 values for correlation coefficient Q (2), root-mean-square error and Pearson R, respectively, suggested an excellent predictive power of the model. Further, the visualization of the three-dimensional-quantitative structure-activity relationships model provided information about the structure-activity relationships by revealing the importance of electron withdrawing and hydrophobic features, on the chemical structure of compounds for the inhibition of cyclooxygenase-2 enzyme activity. The model therefore provides explicit advances for the design of better resveratrol analogs as cyclooxygenase-2 inhibitors. Thus, the obtained three-dimensional-quantitative structure-activity relationships model can predict novel compounds derived from the low active resveratrol analogs, with improved structures, allowing them to have a better inhibitory effect toward cyclooxygenase-2.