QSAR studies combined with DFT-calculations and Molecular docking of polyamine-sensitive inhibitors of the NMDA receptor

Abstract

A quantitative structure-activity relationship (QSAR) was carried out to analyze inhibitory activity of 35 compounds, new polyamine-sensitive inhibitors of the NMDA receptor, using multiple linear regression (MLR), artificial neural networks (NN), and the molecular descriptors were calculated using DFT method. This study shows that the compounds' activity correlates reasonably well with six selected descriptors by MLR method. The correlation coefficients calculated by MLR and after that by NN, R =0.878 and R =0.978 respectively, are relatively kind to evaluate the proposed quantitative model, and to predict activity for new polyamine-sensitive inhibitors of the NMDA receptor. The test of the performance of the NN model, using a cross-validation method with a leave-one-out procedure (LOO) shows that the predictive power of this model is relevant (R=0.966). The constitutional molecular descriptors (nN and nHBD) have the most significant impact in the formulation of the QSAR model. The molecular docking investigations exploring the influence of the structural differences in the interaction potency demonstrate that the number of N atoms expressed by multiple hydrogen bonds helps the ligand to be fixed to NR2B subtype of NMDA receptor.