AI-assisted behavioral profiling of an invertebrate model to identify new psychedelic therapeutics for stimulant use disorders

项目摘要

Project Abstract Substance use disorders are neurobehavioral disorders in which afflicted individuals continue to use a substance despite its adverse effects on their lives. Opioids and cocaine use were responsible for 70% and 22%, respectively, of drug overdose deaths in the US alone in 2019. While there are FDA-approved treatments for opioid use disorders, no such therapeutics are available for cocaine addiction. Opioids have a well-characterized mechanism of action in the brain. However, cocaine targets multiple neurotransmitter pathways, and the molecular basis of addiction has not been elucidated. A significant challenge in discovering new therapeutics for cocaine use disorders (CUD) is the need to perform behavioral assays on animals to determine if a candidate drug reduces or abolishes addictive behavior. The need for behavioral assays in rodents is not compatible with high-throughput drug discovery screens. NemaLife, Inc. has developed a microfluidic platform that enables high-throughput in vivo screening using the nematode Caenorhabditis elegans. Video data recorded by our platform is automatically processed by our AI analysis pipeline NemaStudio.ai for metrics such as live/dead and animal activity. This Phase 1 project will implement additional machine learning algorithms into NemaStudio.ai to classify worm behavior. This new functionality will enable us to perform high-throughput behavioral profiling of animals. This form of behavioral profiling using worm models of CUD is expected to rapidly identify new treatments for cocaine addiction. Aim 1 - Develop high-throughput AI-assisted behavioral profiling assays for cocaine use disorders. In this aim, we will take existing worm CUD paradigms and optimize them for our microfluidic platform. Specifically, we will create microfluidic chip-based versions of cue-conditioned place preference and self-administration CUD paradigms. Once these assays are developed, we will collect large datasets of animal behavior either on or off cocaine. This data will then be used to train NemaStudio.ai to classify and quantify addictive behaviors in animals exposed to cocaine and form our behavioral profiling pipeline. Aim 2 - Identify potential CUD therapeutics from a targeted library of psychedelic compounds. With our screening pipeline in place, we will examine whether psychedelic compounds, which are known to have anti- addiction properties, might represent a new source of treatments for CUD. We will screen a variety of psychedelic compounds, select the top candidates, and perform RNA-seq on psychedelic-treated and untreated CUD animals. This data will help identify possible mechanisms of action by which these candidates reduce/abolish the effects of cocaine addiction in the worm. Our data-driven insights will enable targeted rodent studies for pre- clinical validation of the hit compounds. In sum, this work leverages AI tools to build a high throughput in vivo drug discovery platform to accelerate the development of new therapeutics for CUD.