A predictive acetylcholinesterase inhibition model: an integrated computational approach on alkaloids and synthetic derivatives

Abstract

Computational techniques have become powerful tools for studying biological systems, including receptor–ligand (R–L) complexes. In medicinal chemistry, these in silico approaches are widely used for modeling and predicting molecular interactions, as well as for designing new ligands with biological activity. However, obtaining a direct correlation between the structure and activity of a set of active compounds is a challenging task. This study aims to develop a computational pipeline to find a direct correlation between structure and acetylcholinesterase (AChE) inhibitory activity across a structurally diverse set of 224 Amaryllidaceae alkaloids and synthetic derivatives. Standard docking protocols failed to generate reliable correlations with experimental data, and although the inclusion of molecular dynamics (MD) simulations improved performance, the results remained insufficient for robust prediction. Incorporation of quantum theory of atoms in molecules (QTAIM) analyses on MD-refined geometries was essential to capture key R–L interactions, yielding a strong correlation with relative IC50 values (R = – 0.9131). This approach not only explained differences in activity among structurally related compounds but also distinguished active, moderately active, and inactive ligands across multiple alkaloid families. For the first time, a QTAIM analysis is reported providing detailed insights into the molecular interactions stabilizing AChE–ligand complexes, including natural alkaloids, as well as synthetic dual-site inhibitors designed to engage both the catalytic active site and the peripheral anionic site of the enzyme. These findings suggest that simple appropriately combined computational methodologies can yield predictive and explanatory models applicable to chemically diverse scaffolds, supporting the rational design of novel AChE inhibitors.