Steve Finkbeiner, M.D., Ph.D.

Finding Causes and Treatments for Neurodegenerative Disease with Stem Cells, Genomics, Robotics, and Artificial Intelligence

Abstract: Despite numerous clinical trials, it remains the case in 2020 that there are no disease-modifying therapies for any of the major adult-onset neurodegenerative diseases such as Alzheimer’s disease (AD), Frontotemporal dementia (FTD), Parkinson’s disease (PD), Huntington’s disease (HD) or amyotrophic lateral sclerosis (ALS). The preclinical pipeline to identify causes and find treatments that translate into effective therapies has largely failed to serve its purpose.  In this talk we will describe our work to address this major unmet medical need. In an effort to create more relevant disease models, we developed human brain cell models of AD, FTD, PD, HD and ALS from patient-derived stem cells. To exploit these models, we invented an automated imaging platform, called robotic microscopy (RM), capable of high throughput longitudinal single cell analysis. RM and longitudinal single cell analysis is 2–3 orders of magnitude at detecting phenotypic differences than conventional snap shot HTS approaches. Moreover, multivariate longitudinal single cell data can be integrated using Bayesian analysis methods to develop models that predict fate at a single cell level and used to create blueprints for interventions. Recently, artificial intelligence / deep learning methods have been developed to uncover insights in imaging data that are not evident to the human eye. Observational studies from different models and genomics suggested that deficits in protein homeostasis were predictors of whether and when cells would undergo neurodegeneration. Based on these observations, we developed a high throughput assay for autophagy flux based on optical pulse labeling to search for compounds that could stimulate autophagy. We developed nanomolar potent orally available compounds that stimulate autophagy in neurons in vitro and in vivo, promote the clearance of tau, synuclein, TDP43, and huntingtin, and mitigate cytopathology and neurodegeneration phenotypes in our patient-derived models of HD and ALS. We propose that deficits in the ALP may be a common thread that cuts across multiple neurodegenerative diseases and could be a useful therapeutic target.