MTDL-GAN: De novo Design of Multi-Target Directed Ligands for Alzheimer’s Disease from Unpaired Sets of Target-Focused Chemical Library

BACKGROUND. Alzheimer’s disease (AD), the most common form of dementia, causes memory loss, cognitive decline, and behavioural changes, affecting over 32 million people globally. Current AD treatments that focus on single-target intervention often fail to slow disease progression and may not be effective for all patients. Given AD's complex nature, a more effective approach may involve targeting multiple pathways simultaneously. METHODS. This study proposes the use of a cycle-consistent adversarial network to design multi-target directed ligands (MTDL-GAN), drugs designed to inhibit two primary AD target enzymes simultaneously; our targets of interest are acetylcholinesterase (AChE), beta-secretase 1 (BACE1), and glycogen synthase kinase 3 beta (GSK3), each contributing to AD through distinct pathological mechanisms. Inhibitor libraries of the targets were curated from ChEMBL database and further characterised to represent each inhibitor domain, resulting in 69 AChE, 572 BACE1, and 246 GSK3 inhibitors. MTDL-GAN was trained on these unpaired datasets to generate MTDL-like molecules, with all generated molecular structures stored after each training step to construct large \textit{in silico} MTDL libraries. Subsequently, 300 molecules were sampled from each library to evaluate their \textit{in silico} binding affinities using molecular docking simulation. RESULTS. The proposed method effectively transformed molecules from the input inhibitor domains into MTDL-like molecules that exhibited dual-target binding affinity and structural novelty; several MTDLs surpassed \textit{in silico} binding affinities of investigational drugs in phase 2 and 3 clinical trials and less than 0.15\% of the generated molecules have Tanimoto similarity greater than 0.85 to the known bioactive molecules in ChEMBL and ExCAPE database.