Generalized predictive model of estimation of inhibition of muscarinic receptors M1-M5

A general predictive model for assessing the inhibition constant (K_i) value of human acetylcholine muscarinic receptors M1-M5 by potential ligands has been constructed. We used information on the three-dimensional structure of human M1, M2, M4, and M5 receptors, as well as a model of the M3 receptor constructed according to homology based on the structure of the rat M3 receptor. A set of complexes of known inhibitors with the target receptor constructed by means of molecular docking, was selected using an additional option: the coincidence of the spatial position of 4 pharmacophore points of a tested inhibitor and tiotropium, for which the position in the crystal structure was known. For five types of M receptors 199 complexes with known K_i values were selected. Based on the data obtained during molecular dynamics simulation of these complexes by means of the MM-PBSA/MM-GBSA methods, their energy characteristics were calculated. They were used as independent variables in linear regression equations for pK_i value prediction. The R² prediction for the generalized equation was 0.7, and the mean prediction error was 0.55 logarithmic units with a range for pK_i=4.7.