Novel AChE and BChE inhibitors using combined virtual screening, text mining and in vitro binding assays

Abstract

In the current work, we developed a computational pipeline method for predicting the binding affinities of studied compounds at the specific target sites. Since many approved therapeutic compounds involve indole or indole-derivative rings, in the current study we focused compounds including these fingerprints. Initially, 212520 compounds were retrieved from Specs-SC library and after the conversion of IUPAC text file format, compounds that include ‘indol’ keyword (5194 compounds) were used in binary QSAR-based models to screen against a defined therapeutic activity “Alzheimer’s disease” (AD). The molecules that have higher AD therapeutic activity values (>0.5) were then used in the 26 different toxicity-QSAR models. Binary QSAR models resulted 89 hits that have high AD therapeutic activity and no toxicity. Selected 89 molecules were then screened against acetylcholinesterase (AChE) targets using molecular docking and top-docking poses of compounds were used in initially short (10 ns) molecular dynamics (MD) simulations. Molecular Mechanics/Generalized Born Surface Area (MM/GBSA) binding free energy calculations were performed for 89 ligands and tightly bound 17 ligands based on average MM/GBSA scores were selected for long (100 ns) MD simulations. The same protocol was also applied for the known 4 AChE inhibitors. Selected hits were also docked to the binding pocket of butyrylcholinesterase (BChE). Finally, based on MM/GBSA scores, as well as their corresponding docking scores and metabolite production profiles, 7 compounds were selected and their in vitro tests were performed. Out of 7 compounds, 6 of them showed μM-level inhibition for both AChE and BChE targets.Communicated by Ramaswamy H. Sarma