The influence of R and S configurations of a series of amphetamine derivatives on quantitative structure–activity relationship models

Chiral molecules need special attention in drug design. In this sense, the R and S configurations of a series of thirty-four amphetamines were evaluated by quantitative structure–activity relationship (QSAR). This class of compounds has antidepressant, anti-Parkinson and anti-Alzheimer effects against the enzyme monoamine oxidase A (MAO A). A set of thirty-eight descriptors, including electronic, steric and hydrophobic ones, were calculated. Variable selection was performed through the correlation coefficients followed by the ordered predictor selection (OPS) algorithm. Six descriptors (CHELPG atomic charges C3, C4 and C5, electrophilicity, molecular surface area and log P) were selected for both configurations and a satisfactory model was obtained by PLS regression with three latent variables with R2 = 0.73 and Q2 = 0.60, with external predictability Q2 = 0.68, and R2 = 0.76 and Q2 = 0.67 with external predictability Q2 = 0.50, for R and S configurations, respectively. To confirm the robustness of each model, leave-N-out cross validation (LNO) was carried out and the y-randomization test was used to check if these models present chance correlation. Moreover, both automated or a manual molecular docking indicate that the reaction of ligands with the enzyme occurs via pi–pi stacking interaction with Tyr407, inclined face-to-face interaction with Tyr444, while aromatic hydrogen–hydrogen interactions with Tyr197 are preferable for R instead of S configurations.

Figure optionsDownload as PowerPoint slideHighlights
► The QSAR model is not dependent of ligand conformation.
► Amphetamines were analyzed by quantum chemical, steric and hydrophobic descriptors.
► CHELPG atomic charges on the benzene ring are one of the most important descriptors.
► The PLS models built were extensively validated.
► Manual docking supports the QSAR results by pi–pi stacking interactions.