Using Shape-Focused Pharmacophore Modeling to Improve Docking Screening with Acetylcholinesterase

Abstract

Acetylcholinesterase is a serine hydrolase whose main role is the degradation of the neurotransmitter acetylcholine in the cholinergic synapse. The activity of this enzyme is associated with neurodegenerative disorders, such as Alzheimer's disease and Parkinson’s disease. The emergence of the protein’s 3D structure has enabled structure-based drug discovery for acetylcholinesterase using affordable in silico methods. This study applies a novel computer-aided drug discovery method, O-LAP modeling, to enhance docking-based virtual screening yield, and human acetylcholinesterase is used as a test case. In the O-LAP modeling, shape-focused pharmacophore models are built by merging overlapping atomic content filling the target protein’s ligand-binding cavity. The cavity-filling input can originate for example from flexible molecular docking of active small-molecule ligands or from cavity detection/filling done based on the ligand-binding cavity geometry. During docking rescoring, the shape complementarity of the O-LAP models with the flexibly docked ligand poses is compared using an established similarity comparison algorithm ShaEP. O-LAP modeling significantly improves docking yield for acetylcholinesterase: the best results are achieved by systematically exploring alternative clustering settings, carefully curating the cavity-filling atomic input, and optimizing the O-LAP model’s atomic composition using a greedy search. This in silico study, relying on demanding and random training/test set division for the benchmarking set, concludes that O-LAP modeling should work excellently in docking-based virtual screening for acetylcholinesterase, and, thus, it is ready for practical work involving in vitro verification.