Seminar by Dr. Jana Shen

University of Maryland School of Pharmacy, Baltimore (USA)

Using simulations and AI to understand ion channels, drug permeation, and discover druggable sites across the proteome

Fueled by computing power, algorithm development, and the rise of AI, we are witnessing a transformation in the way we understand the cellular machines and pharmacology of drugs. In this talk, I will discuss three projects from my lab that harness this transformation: the mechanistic elucidation of 1) ion channels and 2) membrane-dependent pharmacology of drugs; 3) the development of ML models for proteomewide drug discovery. 1) Human Hv1 is a voltageand pH-gated channel exclusively selective for protons. Despite its physiological importance and prototypical role in our understanding of voltage- and pH-gated channels, mechanistic knowledge is very limited due to the lack of experimental structures for the inactive or the active state. I will discuss how we illuminated the channel activation mechanism of Hv1 by leveraging state-of-the-art AI tools and the all-atom particle-mesh Ewald continuous constant pH (CpH) molecular dynamics (MD) method in Amber24. 2) Many drugs, including opioids, are weakly basic; it is poorly understood whether and how they permeate the membrane at physiological pH. For example, fentanyl and morphine both have a pKa around 8.5, and yet experimental evidence suggests that fentanyl (but not morphine) partitions into the membrane and uses it as a reservoir to activate the mu-opioid receptor. I will discuss the weighted-ensemble CpHMD simulations that yield membrane permeation profiles and rates of fentanyl and morphine. 3) Finally, I will discuss our development of ML models leveraging Large Language Models (LLMs) and chemoproteomic data to map the cysteine-directed covalently druggable sites across the human proteome.