In silico evaluation of 5-hydroxypyrazoles as LSD1 inhibitors based on molecular docking derived descriptors

Abstract

The lysine specific demethylase 1 (LSD1) overexpression is linked to a lung, breast and prostate cancers. In order to explore significant biochemical features affecting LSD1 inhibition mechanism and help design new potent inhibitors, molecular docking, classification and quantitative structure-activity relationship (QSAR) were performed based on 5-hydroxypyrazole analogs. The molecular docking studies indicate residues of Arg316, Thr810 and Tyr761 with total binding energy more than 75 kcal mol-1 located in LSD1 receptor play main rules in inhibitory mechanism by making strong polar interactions. Furthermore, there are two backbone donor hydrogen bonds (Ser760 and Val333), two backbone acceptor hydrogen bonds (Val288 and Gly287) and three nonpolar interactions (Ala331, Trp751 and Val811) which have key roles in inhibition. The actual inhibitors' conformations in the main binding pocket of LSD1 with the minimum binding free energies were used to generate molecular descriptors and performing further QSAR models. As a simple to use graphical classification tool, the PLS-DA (partial least squares-discriminant analysis) was effectively implemented to classify the inactive and active LSD1 inhibitors. The robust and valid QSAR analysis were done through the stepwise MLR (Multiple Linear Regression) based on the breakpoint procedure and SVM (Support Vector Machine) methods as variable selection and mapping tools, respectively. The QSAR model (R2p = 0.837, Q2LOO = 0.871, R2LSO = 0.790 and r2m = 0.758) shows that the descriptors of JGI1, MATS3e and Mor20v are major inhibitors’ features affecting inhibition activities. The performed molecular docking, PLS-DA classification and breakpoint stepwise MLR/SVM analysis may provide a guideline to design new potent 5-hydroxypyrazole analogs as LSD1 inhibitors.