ProP-PD for proteome-wide motif-mediated interaction discovery

Abstract

Many protein–protein interactions are mediated by short linear motifs (SLiMs; 3–10 amino acid stretches; typically found in intrinsically disordered regions; low-to-mid micromolar affinities). The human proteome is expected to contain tens of thousands of SLiMs. However, to date only ~3000 SLiM instances have been discovered. The field required a scalable and accurate experimental approach to tackle SLiM discovery. Proteomic peptide-phage display (ProP-PD) is based on classical phage display updated with a combination of computational library design, custom oligonucleotide library synthesis, next-generation sequencing (NGS), and integrative data analysis. ProP-PD can discover hundreds of SLiMs in a single experiment and has transitioned the field into a high-throughput era. ProP-PD can systematically map unexplored parts of the human proteome and illuminate how viruses exploit SLiM-based interactions to hijack the host cell, which may uncover leads for innovative inhibitor design. Based on mature technologies: M13 peptide-phage display, oligonucleotide library synthesis, and NGS. Monovalent or multivalent display, depending on phage coat protein used. Hundreds of thousands of peptides screened for binding in a single experiment, permitting proteome-wide screens. Scalability in terms of bait proteins from the parallelization in 96-well plates, barcoding, and NGS. No need for specialized equipment. Experimental data has strong discriminatory attributes for biophysical binders: replicated or overlapping peptides, high sequencing counts, and presence of shared consensus motifs. Two distinct data types produced: direct interfaces and binding pocket specificity determinants. SLiMs are discovered with amino acid resolution. ProP-PD requires purified bait proteins. Selections may fail even when bait proteins are of good quality; optimized immobilization protocols may be needed. High recall requires several replicate selections. Some peptides are promiscuous; care needs to be taken in the data analysis. There is no strict correlation between affinity and sequencing counts. Not applicable to interactions that are dependent on post-translational modifications. Screens lack cellular spatiotemporal separation and results require post-processing to deconvolute biological and biophysical binders. Putative biologically relevant interactions require additional validation. This work was supported by the Swedish Foundation for Strategic research (Y.I.: SB16-0039), the Swedish Research Council (Y.I.: 2020-03380), and a Cancer Research UK Senior Cancer Research Fellowship (N.D.: C68484/A28159). No interests are declared.

Topics & Concepts