[
International Worm Meeting,
2021]
Parkinson's disease (PD) is a movement disorder and the mechanisms that induce neurodegeneration in PD are still poorly understood. We recently found that RNAi-mediated knockdown of neuronal branched-chain amino acid transferase 1 (BCAT-1) in nematode C. elegans cause age-dependent spasm-like 'curling' phenotype mirroring PD clinical symptoms. Manual quantification of curling is labor-intensive making large-scale drug and genetic screens challenging. Here, we report the development of a machine learning-based automated workflow for C. elegans image analysis, and its application to the discovery of potential compounds that may be repurposed as late-in-life interventions for PD. This high-throughput workflow is 40X faster than the manual assay and constitutes a major advance in the efficiency and precision with which Parkinson's-like curling behavior in C. elegans can be quantified. In a screen of 50 FDA-approved drugs, we have identified four drugs (enasidenib, ethosuximide, metformin, and nitisinone) that reduced curling to <50% of vehicle-treated levels. These findings point to the utility of our high-throughput platform for screening for modifiers of the disease phenotypes, including but not limited to chemicals or genetic manipulations that may ameliorate or worsen motor function in the worms.