In Silico Evaluation of Withanolides from Withania somnifera Against Acetylcholinesterase: An Integrated Molecular Docking, ADMET, DFT and Molecular-Dynamics Study for Alzheimer's Disease

Abstract

Background: Alzheimer's disease (AD) is a progressive neurodegenerative disorder in which the cholinergic deficit remains a validated therapeutic target, and acetylcholinesterase (AChE) inhibition is the mechanistic basis of most symptomatic drugs. Withania somnifera (ashwagandha), a nootropic botanical of the South Indian and broader Ayurvedic tradition, is rich in withanolides with reported neuroprotective activity, making it an attractive source of candidate AChE ligands. Objective: To computationally prioritise twelve W. somnifera constituents against human AChE using an integrated workflow, benchmarked against donepezil. Methods: Twelve withanolides, withanosides and glycowithanolides plus donepezil were docked (five poses per ligand; 65 records including the reference) into the AChE catalytic gorge. Poses were interaction-profiled and compounds evaluated for drug-likeness and ADMET, density-functional-theory (DFT) electronic descriptors, and, for the top three, 100 ns molecular dynamics (MD) with MM-GBSA free energies. Results: Donepezil scored the strongest affinity (−10.04 kcal/mol). Among the phytochemicals, the tri-glycosylated withanosides VI (−9.57) and IV (−9.29 kcal/mol) topped the docking table but carried three Lipinski violations, low gastrointestinal absorption and no predicted blood–brain-barrier (BBB) penetration. Withaferin A (−9.15 kcal/mol) combined strong docking with zero Lipinski violations, high absorption, predicted BBB penetration, favourable ADMET, the most reactive DFT profile (HOMO–LUMO gap 3.007 eV) and a dynamically stable MD complex (protein RMSD 1.90 Å; MM-GBSA ΔG −49.61 kcal/mol). Ligands engaged catalytic-triad (Ser203, His447), anionic (Trp86) and peripheral (Trp286) sites in a donepezil-like dual-site mode (Phe338 contacted in 43/65 poses). Conclusion: Withaferin A emerges as the most drug-like, BBB-penetrant lead, whereas the strong-docking withanosides are computationally unsuitable CNS candidates. These purely computational predictions are hypothesis-generating only and must be regenerated with validated software and confirmed by enzyme-inhibition (Ellman) assays and cell/animal models.