Machine Learning-Driven Bioprospecting of Cholinergic Medicinal Plants for Putative Leads Against...

Butyrylcholinesterase (BChE) plays a key role in preserving appropriate cholinergic neurotransmission that is essentially altered in the brains of advanced Alzheimer's disease (AD), hence a therapeutic target. This study employed a machine learning (ML) bioactivity predictive model to explore the chemical space of potential BChE inhibitors. A cheminformatics pipeline was explored to create a ML model for BChE inhibition using structural insights complemented with a comprehensive variance importance plot (VIP) and correlation matrix analysis. Specifically, a compiled library of 2179 secondary metabolites (SMs) from 50 Nigerian medicinal plants with reported cholinergic activity was investigated using the ML model. After which, molecular modelling was used to further screen the active SMs (827). The final predicted models demonstrated significant robustness, with a correlation coefficient of 0.8981. Molecular docking investigation of the 827 SMs identified the top five candidates based on their scores: three triterpenoids (adipedatol, lupenone and β-amyrin), one steroid (9 (11)-dehydroergosterol benzoate) and one flavonoid (tiliroside). These leads exhibited favourable ADMET properties and promising safety profiles. Among these five, lupenone (-48.78 kcal/mol), β-amyrin (-49.19 kcal/mol) and adipedatol (-48.87 kcal/mol), from Peltophorum pterocarpum, Bryophyllum pinnatum and Alchornea laxiflora, respectively, were the most promising leads with significant binding free energy compared to decamethonium (-17.64 kcal/mol) and more favourable van der Waals, electrostatics and nonpolar solvation energetics. Furthermore, the binding of these leads resulted in optimised interaction profiles that preserved the structural integrity of the BChE and aligned well with desirable drug-like characteristics. These findings position the leads as promising candidates for therapeutic applications targeting BChE for AD management, subject to further in vitro and in vivo validation investigations.