In silico Identification of Potential serotonin 5-HT6 Receptor Antagonists as Cognitive Enhancers for Alzheimer's disease: An Integrated Computational Study

In this investigation, we employed a quantitative Read-Across Structure-Activity Relationship (q-RASAR) approach to develop a statistically robust machine learning (ML)-based model for predicting the antagonistic activity of compounds targeting the serotonin 5-HT6 receptor (5-HT6R), a pharmacologically relevant target associated with cognitive function and Alzheimer’s disease (AD). A strictly validated univariate q-RASAR linear regression (LR) model was developed using a structurally diverse dataset of 3,102 5-HT6R antagonists curated from the publicly available BindingDB database. The model was subsequently applied for predictions of the Mcule and InterBioscreen (IBS) databases, comprising approximately 10.18 million chemical compounds, to identify potential 5-HT6R antagonist candidates with cognitive-enhancing potential. Predicted compounds were ranked according to their estimated antagonistic activity, and prioritized candidates were further evaluated using molecular docking, ADMET profiling, molecular dynamics (MD) simulations, and density functional theory (DFT) calculations. The top-ranked compounds, MCULE-9410489047, STOCK1N-76057, and STOCK6S-93236, were investigated for their binding modes and key receptor-ligand interactions, predicted pharmacokinetic and toxicity profiles, and electronic and structural properties. Furthermore, 100-ns MD simulations were performed to assess the stability of the receptor–ligand complexes and persistence of key interactions. DFT analyses provided additional insights into frontier molecular orbital and molecular electrostatic potential characteristics. Overall, the integrated q-RASAR, molecular docking, ADMET, MD, and DFT framework enabled systematic prioritization of potential 5-HT6R antagonists for further experimental evaluation as therapeutic leads for AD-related cognitive impairment.