Functional drug fingerprinting with all-optical electrophysiology

Project Summary Phase II: Functional drug fingerprinting with all-optical electrophysiology Q-State Biosciences has developed a therapeutic discovery platform that uniquely integrates i) scalable human cellular models of disease states, ii) diverse readouts of neuronal function using proprietary all-optical electrophysiology called OptopatchTM, and iii) artificial intelligence/machine learning (AI/ML) analytics to establish unique, disease-relevant phenotypes for assessment of therapeutics. Optopatch simultaneously records >500 electrical activity parameters from hundreds of neurons with 1 millisecond temporal resolution, single-cell spatial resolution, and a throughput >20,000-fold higher than manual patch clamp. Leveraging this high-throughput, high-content data, Q-State has developed “drug fingerprinting” (DFP) algorithms to learn the archetypal patterns of changes induced by disease states and chemical compounds, building a map of the functional electrophysiology of human neurons. In Phase I, we populated this map with fingerprints of 400 diverse and selective tool compounds for a wide range of hand-curated molecular targets, which serve as landmarks for interpreting the functional effects of disease states and novel chemical compounds. We also demonstrated that this technology could identify a disease phenotype in neurons and predict drug rescue. In Phase II, we will advance the DFP platform in 3 ways: 1) Build tools to improve the reliability, interpretability, accessibility, and automation of the technology. Our current platform is capable of extracting data from Optopatch assays and fingerprinting the impacts of perturbations using representation-learning. To achieve a production-ready pipeline, “explainable AI” techniques will be used to track how the algorithms make choices, and unit tests will affirm the correctness of all modules. The pipeline will be automated and deployed for access by non-coders. Finally, a new set of modules will extend DFP to high-content imaging data for mapping biological states across form and function of neurons. 2) Expand DFP library to include clinically relevant compounds and additional neuronal subtypes. To enable therapeutic discovery applications, we will expand our reference database to include a new custom-assembled library of ~3,000 clinically relevant drugs, including a comprehensive set of approved compounds, as well as unapproved bioactive molecules with acceptable safety profiles and favorable drug-like properties. Scaling to a 384-well format assay, we will collect data for all compounds in human iPSC-derived excitatory and inhibitory neurons, measuring pharmacological impact on both neuronal excitability and cell morphology in a dose-dependent format to create unique multi-modal fingerprints. 3) Demonstrate pharmacological rescue of a monogenic epilepsy in vitro and in vivo. As a proof-of-concept demonstration of the DFP platform, we will use our new database to identify compounds with potential to rescue effects of epilepsy causing mutations. Putative in silico hits will be experimentally tested and pharmacological rescue of phenotypes will be quantified. The best compounds will be advanced to in vivo testing in a rodent model as an initial demonstration of the ability of the platform to rapidly identify new, clinically viable treatments for epilepsy.