The receptor-dependent LQTA-QSAR: application to a set of trypanothione reductase inhibitors

Abstract

A new Receptor-Dependent LQTA-QSAR approach, RD-LQTA-QSAR, is proposed as a new 4D-QSAR method. It is an evolution of receptor independent LQTA-QSAR. This approach uses the free GROMACS package to carry out molecular dynamics simulations and generates a conformational ensemble profile for each compound. Such an ensemble is used to build molecular interaction field-based QSAR models, as in CoMFA. To show the potential of this methodology, a set of 38 phenothiazine derivatives that are specific competitive T. cruzi trypanothione reductase inhibitors, was chosen. Using a combination of molecular docking and molecular dynamics simulations, the binding mode of the phenotiazine derivatives was evaluated in a simulated induced fit approach. The ligands alignments were performed using both ligand and binding site atoms, enabling unbiased alignment. The models obtained were extensively validated by leave-N-out cross-validation and y-randomization techniques to test for their robustness and absence of chance correlation. The final model presented Q(2) LOO of 0.87 and R² of 0.92 and a suitable external prediction of [Formula: see text]= 0.78. The adapted binding site obtained is useful to perform virtual screening and ligand structure-based design and the descriptors in the final model can aid in the design new inhibitors.