Engineering and analysis of peptide-recognition domain specificities by phage display and deep sequencing

Protein interaction networks depend in part on the specific recognition of unstructured peptides by folded domains. Understanding how members of a domain family use a similar fold to recognize different peptide sequences selectively is a fundamental question. One way to advance our understanding of peptide recognition is to apply an existing model of peptide recognition for a particular domain toward engineering synthetic domain variants with desired properties. Successes, failures, and unintended outcomes can help refine the model and can illuminate more general principles of peptide recognition. Using the PDZ domain fold as an example, we describe methods for (1) structure-based combinatorial library design and directed evolution of domain variants and (2) specificity profiling of large repertoires of synthetic variants using multiplexed deep sequencing. Peptide-binding preferences for hundreds of variants can be decoded in parallel, enabling comparisons between different library designs and selection pressures. The tremendous depth of coverage of the binding peptide profiles also permits robust computational analysis. This approach to studying peptide recognition can be applied to other domains and to a variety of structural and functional models by tailoring the combinatorial library design and selection pressures accordingly.