Uncovering the molecular basis of kinase activity and substrate recognition with phospho-PCA

Protein kinases relay information to various cellular processes, and their dysregulation underlies nu- merous human diseases. Despite their importance, our understanding of how kinase domains and their variants impact protein stability, catalytic activity, and substrate recognition is incomplete. In this work, we develop the phosphorylation protein complementation assay (phospho-PCA), which enables quantitative measurements of kinase-substrate interactions by coupling them to the growth of bud- ding yeast, thereby enabling deep mutational scanning of kinase domains. When combined with deep mutational scans targeting folding stability and Bayesian modeling, phospho-PCA can disentangle the relative impact of mutations on kinase domain stability, catalytic activity, and substrate specificity. We demonstrate the accuracy and breadth of phospho-PCA, showing its applicability to both tyrosine and serine/threonine kinase domains. We then apply our method to three closely related protein kinases with distinct substrate preferences, evaluating over 15,000 kinase variants against a panel of three sub- strates for both catalytic activity and substrate specificity. The resulting dataset constitutes the largest and most detailed variant-to-function map assembled for this enzyme family to date, revealing numerous mutations that alter kinase activity and substrate specificity. Physics-based modeling reveals how these mutations operate through diverse mechanisms, including long-range allosteric communication, to alter both activity and substrate specificity. Given its scalability, we believe phospho-PCA can measure the functional impact of variants across the entire human kinome.