Molecular Modeling

Compounds from 2 CNS-targeted databases (Enamine CNS Library and ChemDiv CNS-MPO Library) were subjected to the LigPrep module of the Schrödinger molecular modeling suite 2017 (Schrödinger, Inc, LLC) to generate 3D structures including all possible stereoisomers and tautomers.²⁵

The crystallographic structure of human BChE was downloaded from Protein Data Bank (http://www.rcsb.org, PDB: 5NN0).²⁰ All solvent molecules except HOH724, HOH734, and HOH805 were eliminated. The raw PDB protein structure was then prepared by giving preliminary treatment including adding hydrogen atoms, refining the loop region, and finally energy minimization by using the Protein Preparation Wizard of the Schrödinger modeling suite.²⁶

The generated 3D structures were subjected to a hierarchical docking-based virtual screening workflow using the docking module of the Schrödinger suite, Glide (Grid-based ligand docking with energetics).²⁵ At the beginning, the prepared libraries were screened using glide high-throughput virtual screening mode. Compounds with better performance (Docking score ≤ −7) were screened using standard precision (SP) docking mode. Next, potential hit compounds (Docking score ≤ −8) from SP docking were processed through extra-precision (XP) docking mode. The output of XP docking was filtered for an XP Gscore of ≤ −9.

The MM-GBSA method is widely exploited to investigate the binding of small ligands to the macromolecules and rank affinities of ligands bound to the same macromolecule.²⁷ The binding free energies of the filtered output of XP docking were estimated by using the Prime MM-GBSA Panel of the Schrödinger modeling suite. Distance from ligand was set to 5 Å. Other parameters were maintained as the defaults. The threshold for the binding free energy (dG) was set to a value of −90 kcal/mol. Finally, 9 hits were identified and 8 available hits were purchased from Topscience (www.tsbiochem.com), with purity >95% (liquid chromatography-mass spectrometry).