Post-translational modifications (PTMs) critically impact cellular decision-making in many contexts, including plant’s responses to changing environments. However, innovative analytical and experimental strategies are required to reveal how PTM networks integrate to control cellular logic at the systems level. Protein phosphorylation is the most common PTM, and this modification is catalyzed by over 1000 different protein kinases in the model plant Arabidopsis.
We seek to illuminate the phosphorylation networks that drive plant development and resilience to environmental stress by connecting these protein kinases to their tens of thousands of cognate substrates. We will construct a genetically encoded library of over 33,000 Arabidopsis (phospho)peptide sequences that mimic experimentally supported phosphorylation sites. Using a cutting-edge genetic code expansion system that can conditionally encode serine or phosphoserine in target amino acid sequences, this library will enable high-throughput identification of kinase-substrate relationships for key stress-responsive kinases and phosphorylation-dependent protein-protein interactions.
In parallel, we will adaptively train Phosphormer, a large language model designed to predict human protein kinase-substrate interactions, on Arabidopsis protein kinase-substrate networks that are not represented in human genomes, enabling predictive modeling of phosphorylation networks in the green lineage.
Deliverables include a publicly available (phospho)peptide library for community distribution, validated protein kinase-substrate networks, phosphorylation-dependent interaction datasets, and an enhanced version of Phosphormer tailored to plant biology. This integrative approach combines synthetic biology, proteomics, and machine learning to transform how we understand and engineer the signal transduction networks underlying plant resilience by offering scalable tools to advance our grasp of cellular regulatory complexity.
