Per-Arnt-Sim (PAS) domains are ubiquitous protein modules that enable cells to detect and respond to environmental signals. For instance, circadian rhythm regulators leverage PAS domains to sense stimuli and initiate protein-protein interactions critical for maintaining biological oscillations. Structurally, the sensory region of PAS domains detects environmental cues—such as fluctuations in phosphorylation levels—while the effector domain converts these signals into cellular responses, including altered gene expression or protein interactions. Inspired by this natural framework, our project aims to design de novo sensory domains that selectively recognize tyrosine phosphorylation, a key post-translational modification in cellular signaling, through association/dissociation between bound and unbound states regulated by the phosphorylation/dephosphorylation cycles. During the design phase, we prioritized synthetic peptide targets for initial proof of principle and systematically deployed computational pipelines: (1) Rosetta introduced phosphotyrosine modifications into pre-designed protein-peptide heterodimer scaffolds; (2) iterative LigandMPNN with Rosetta FastRelax optimized binding interfaces to accommodate the phosphotyrosine modifications; (3) RFdiffusion Partial Diffusion enhanced the structural diversity around promising designs with the aim of improving affinity and specificity; and (4) Chai-1 and AlphaFold enabled in silico folding and structure-based filtering of final candidates. High-confidence designs will be expressed and purified from E. coli, and then undergo in vivo characterization via size exclusion chromatography (SEC) binding assays and enzyme-linked immunosorbent arrays (ELISA) to quantify their binding affinity, specificity, and the function of phosphorylation-dependent switching. Validated scaffolds will then be integrated with pre-designed effector domains to assemble fully de novo PAS domains. This modular platform establishes a foundation for designing phosphorylation-sensitive biosensors. Future adaptation to natural phosphorylation sites could yield programmable tools for interrogating signaling networks, advancing synthetic biology, and enabling precise manipulation of cellular communication pathways.

In nature, Per-Arnt-Sim (PAS) domains comprise a sensor that undergoes conformational changes upon signal recognition which either activates or deactivates an effector domain. Natural PAS domains detect environmental cues, such as oxygen, light, and small ligands; however, they do not sense phosphorylation, a key post-translational modification. Here, we present a designed de novo phosphorylation-inducible heterodimer that serves as a sensor domain. This system toggles between association and dissociation states in response to phosphorylation and dephosphorylation events. To engineer reversible association and dissociation, we designed phosphorylated peptides and their corresponding binders. Starting from a library of previously designed peptide-binder complexes, mutations were introduced into the peptide sidechains, replacing selected residues with phosphorylated tyrosine, serine, or threonine. Next, we ran iterative cycles of LigandMPNN-FastRelax to generate binder sequence candidates. Finally, we used AlphaFold2 and Chai1 to predict the folded structures of our input sequences and selected those that were predicted with high confidence. For experimental validation, the designed proteins will be overexpressed in Escherichia coli and purified using affinity and size exclusion chromatography. Phosphorylation-dependent binding specificity and affinity will be assessed through enzyme-linked immunosorbent assays (ELISA), surface plasmon resonance (SPR), and fluorescence polarization (FP). Subsequently, we will fuse these sensor domain designs to a collection of previously designed hinge proteins—which can bind/release an effector protein—to produce de novo PAS domains, thereby linking the sensing event to downstream functional responses. This adaptable system offers broad applications in biomaterials and synthetic biology, including the development of responsive scaffolds for biosensors and synthetic protein motors with controlled conformational cycles.