Аннотация:Positive allosteric modulators (PAMs) of the AMPA-sensitive glutamate receptors are expected to be helpful in treatment of learning and memory deficits, which could be caused, e.g., by the Alzheimer’s disease. In this work we tried to create and tune a system for the estimation of PAMs activity using MM-PB(GB)SA approach and CoMFA. This system can be used in the refinement of the virtual screening results to identify novel PAM chemotypes.
Crystal structures of PAM complexes with GluA2 ligand binding domains dimers were used for ligand alignment for CoMFA. The best obtained model (Q2 = 0.57) was created using MMFF94 charges. As one can note the increase in activity can be achived by filling the two symmetrical binding pockets which are formed by Ile502, Pro515, Ser750 and Lys751 (GluA2).
Calculation of the binding free energy for a set of AMPA receptor PAMs was undertaken and a reasonable correlation was found for the obtained values and the available data for pEC50 of PAMs. Combining various calculated parameters we were able to construct a linear model for the pEC50 with the correlation coefficient (R2) of 0.61 (MM-GBSA) on a large set of PAMs with highly diverse structures. In the considered case the overall performance of the MM-GBSA is comparable with the MM-PBSA method for solvation energy calculation. It was found for the MM-PBSA method that the default value of the dielectric constant is inappropriate in this case giving positive value for the binding energy. Variation of the internal dielectric constant led to achievement of the relatively high correlation coefficient (R2 = 0.62). It should be noted that a usage of frames from the beginning of the trajectories leads to the noticeable decrease in the correlation coefficient which can be associated with the conformational changes in the structures of complexes. The inclusion of the calculated conformational entropy term did not improve the correlation both for MM-GBSA and MM- PBSA.