Computational molecular modelling of N-cinnamoyl and hydroxycinnamoyl amides as potential α-glucosidase inhibitors

Abstract

In the present work, a molecular modelling has been performed on a total of 17 cinnamic acids, derived with different amino acids. These have recently been evaluated as potential α-glucosidase inhibitors. Consequently, we proposed a quantitative 3D structure–activity relationships in order to obtain predictive models of the biological activity. Two types of descriptors were used for the modelling: quantum chemical and multi-linear algebraic maps. The first one was obtained by using Density Functional Theory (DFT) at the M06/6-311++G(d,p) level, while for the second, attribute sets were obtained by using 3D molecular topology, atom weighed by physicochemical properties including the polar surface area, hardness, softness, van der Waals volume and charges. A genetic algorithm scheme was employed to determine the best subset of attributes for the application of the multiple linear regression. The most robust model (M4R_TD) was obtained with four multi-linear algebraic map descriptors (R2 = 0.946, Q2 = 0.901, F = 85.70), which indicate a good correlation between experimental and calculated results. Seven new simple structural modifications in the molecular core were suggested specifically for the three most active compounds (1, 2 and 5). The modifications to compound 1 appear to lead to an increase in the predicted activity from a pIC50 value of 9.872 to 10.656 in compound 1.3. The modifications to compound 2 do not affect an increase in the biological activity, and finally the modifications to compound 5 seem to increase the activity with a predicted pIC50= log[10/(10−6×IC50)]IC50 value of 10.094 for compound 5.1.