Project Summary/Abstract Our understanding of biology and human diseases is taking a significant leap forward due to access to whole genome sequences and detailed protein structural information. The number of high-resolution protein structures determined by single particle cryogenic electron microscopy (cryo-EM) is growing exponentially. The AlphaFold neural network may soon provide high-resolution structures for the entire human proteome. Starting from a three- dimensional protein structure, physics-based molecular dynamics (MD) simulation offers an atomic-level view of protein’s motion. Fueled by the exponential growth of computing power, MD is becoming a powerful tool for structure-function studies and assisting target-based drug discovery. Despite the aforementioned progress, molecular mechanisms of pH-driven and proton-coupled dynamic pro- cesses remain poorly understood. This is because proton positions are not resolved in most experimental struc- tures and conventional MD does not describe proton-coupled dynamics or explicitly account for solution pH. One such example is the human ATP-binding cassette subfamily G member 2 protein (ABCG2), which contributes to cancer drug resistance as well as renal excretion of urate. While high-resolution structures for the entire proteome may soon become available, a large fraction of the pro- teome is currently considered undruggable, i.e., intractable to traditional drug discovery efforts. The development of chemical proteomics platforms for discovery of reactive and ligandable cysteines and lysines in human cell lines holds the promise to significantly expand the druggable space. Nonetheless, the covalent ligandability of a large fraction of the proteome remains unexplored, and a systematic knowledge is lacking. In the previous R01 grant period, the Shen group has made significant progress in the development of GPU- accelerated continuous constant pH MD (CpHMD) methods and application to elucidate proton-coupled structure- dynamics-function relationships of various aspartyl proteases, cysteine proteases, kinases, as well as the mul- tidrug efflux pump AcrB, sodium-proton antiporter NhaA, and µ-opioid receptor. The Shen group has also de- veloped and applied a CpHMD method to predict reactive cysteine and lysine sites in a large number of kinases and other proteins. Building on the progress and taking advantage of the vast data from structural genomics and chemical proteomics, this R35 project seeks to fill the aforementioned gaps in tool development and knowledge. We will tackle the remaining challenges in the development of the all-atom CpHMD method to enable routine studies of proton-coupled dynamic processes. We will apply the all-atom CpHMD and other state-of-the-art MD tools to illuminate the mechanism of the multidrug transporter and urate exporter ABCG2. Finally, we will evalu- ate the entire proteome for covalent inhibition by integrating CpHMD, machine learning, and structure as well as chemo...