Using AI to Repurpose Small Molecules to Target Amyloid-tau Interactions in AD

Project Accomplishments: This project developed an integrated AI‑driven drug discovery platform that combines optogenetic robotic microscopy with deep neural network analysis to identify small‑molecule compounds capable of disrupting the synergistic interaction between amyloid precursor protein (APP) pathology and tau aggregation in Alzheimer’s disease. Over the course of the project, the team engineered a light‑controlled tau aggregation system, generated human iPSC‑derived neuron models with familial AD mutations, and conducted high‑throughput longitudinal imaging that produced more than 18 million single‑cell images. Using this dataset, two complementary AI models—the Rescue Model and the Reversion Model—were trained to detect APP‑related disease phenotypes and tau‑aggregation dynamics with up to 90% accuracy. The models provided the first quantitative single‑cell evidence of bidirectional synergy between APP and tau pathologies, validatinglong‑standing hypotheses that cross‑talk between these pathways drives disease progression. 

The platform enabled a two‑dimensional screening approach that evaluated more than 120 compounds for dual action against both tau aggregation and APP‑associated pathology. Statistical analysis identified 15 high‑confidence “double‑hit” compounds that simultaneously prevented tau aggregation and normalized APP‑driven phenotypes, representingpromising therapeutic candidates that target the disease‑relevant interaction rather than individual pathways. The project not only achieved all of its original objectives but expanded its scope through methodological advances, generating one of the largest dynamic single‑cell Alzheimer’s datasets to date and establishing a validated screening system now poised for further development. 

Initial Proposal Abstract: Extracellular amyloid plaques and intraneuronal neurofibrillary tangles are two histopathological hallmarks of Alzheimer’s disease (AD) that interact during disease progression. However, the precise nature of their interaction and how they collectively drive AD pathogenesis remain unclear. Developing compounds and identifying therapeutic targets to block Tau-ABeta interactions could form the basis for therapies that halt AD progression, complementing current ABeta-focused therapeutics that show limited efficacy in slowing disease.

Progress in understanding the multifaceted dynamics of Tau-ABeta interactions has been hindered by the complexity of their spatial and temporal relationships. Operant BioPharma has developed technologies to resolve dynamic phenotypic changes over neurodegeneration’s slow timeline, which are critical for identifying pathways governing Tau-ABeta interplay. Using proprietary AI, we identify small-molecule compounds that uncouple the pathological synergy between Tau and APP (the precursor to ABeta), thereby blocking disease progression in human neurons and slowing AD-related neurodegeneration.

Our strategy prioritizes drug repurposing: identifying compounds already proven safe in clinical trials for non-AD indications, which could be rapidly repositioned for AD and other tauopathies. Concurrently, we generate a phenotypic embedding of small-molecule effects to elucidate structure-activity relationships (SARs) between efficacy, cytotoxicity, and Tau-ABeta modulation, enabling data-driven optimization of AD therapeutics.