Integrated Virtual Screening and Molecular Electrostatic Surface Potential Descriptors for Predicting BACE1 Inhibitor Interactions in Alzheimer’s Disease

Abstract

Developing brain-penetrant BACE1 inhibitors remains a major challenge in Alzheimer’s disease. Unlike conventional studies focusing solely on binding affinity, this work presents an integrated computational pipeline that prioritizes both potency and pharmacokinetic developability from the outset. A robust quantitative structure–activity relationship (QSAR) model was built for 41 N-(pyridin-3-yl)picolinamide derivatives using Molecular Electrostatic Surface Potential (MESP) descriptors. Principal component analysis identified three key electrostatic features – mean MESP, variance of negative MESP, and number of local MESP maxima/minima as drivers of BACE1 inhibitory activity. A key novelty is the early integration of Absorption, Distribution, Metabolism, Excretion and Toxicity (ADMET) filtering, using ADMETlab 3.0, to ensure blood-brain barrier (BBB) permeability and low P-glycoprotein efflux properties whose absence contributed to the failure of several clinical BACE1 inhibitors. Only compound N-[3-[(1S,5S,6S)-3-amino-1-(difluoromethyl)-5-(fluoromethyl)-2-thia-4-azabicyclo[4.1.0]hept-3-en-5-yl]-4-fluorophenyl]-5-chloropyrazine-2-carboxamide (5) simultaneously satisfied the criteria for high predicted potency, favorable BBB penetration, low PgP efflux, and good synthetic accessibility. Molecular docking revealed that compound 5 binds strategically to the S1 and S3 substrate-recognition pockets through hydrogen bonds with THR280, GLN121, LYS155, as well as a halogen interaction with ARG283, while avoiding direct interaction with the catalytic dyad. This binding mode balances affinity and permeability, distinguishing compound 5 from clinically unsuccessful inhibitors. The practical implication is a rational framework that rapidly prioritizes candidates with high potency and favorable Central Nervous System (CNS) drug-like properties, thereby reducing the risk of late-stage attrition. All predictions are computational and require experimental validation (enzymatic, cellular, and in vivo) before any translational claims can be made. Thus, combining QSAR, ADMET, and molecular docking with an early emphasis on BBB penetration offers a time-efficient strategy for CNS drug discovery against Alzheimer’s disease.