Cannabis sativa Phytochemicals as Potent Regulators of Aquaporin-4 in Alzheimer’s Disease: A Cheminformatics and Machine Learning Approach

Purpose
Aquaporin-4 (AQP4) plays a crucial role in regulating brain water homeostasis and facilitating amyloid-beta clearance. Impaired AQP4 function is implicated in the pathogenesis of Alzheimer’s disease (AD). However, developing AQP4-targeting therapies is limited by challenges related to selectivity, toxicity, and brain delivery. This study aimed to investigate phytochemicals from Cannabis sativa (CS) as potential AQP4 modulators, addressing the need for novel, safe, and effective interventions for AD.

Methods
A ligand- and structure-based in silico approach was employed. A curated dataset of 81 known AQP4 modulators was compiled through comprehensive literature mining. Six classification models were developed, with a radial basis function (RBF) kernel-based machine learning model yielding the best predictive performance. The model was validated through tenfold cross-validation. CS-derived phytochemicals were screened using the optimized model, followed by molecular docking and ADMETOx profiling of selected hits.

Results
The RBF kernel model demonstrated high accuracy (0.941), AUC (0.933), and F1 score (0.889), with cross-validation confirming model robustness (accuracy: 0.844 ± 0.12). Screening identified two promising compounds: cannabinolic acid and an ecdysterone derivative, both exhibiting docking energies comparable to a reference AQP4 binder. ADMETOx predictions indicated favorable drug-like and metabolic properties, including hydroxylation, carboxylation, and glucuronidation pathways.

Conclusion
This study introduces the first QSAR model for AQP4 modulators and highlights the potential of CS-derived phytochemicals in AD therapy. Further in vivo validation is required to assess their therapeutic efficacy and safety profiles.