Deep Sequencing of a Systematic Peptide Library Reveals Conformationally‐Constrained Protein Interface Peptides that Disrupt a Protein‐Protein Interaction

Abstract

Disrupting protein‐protein interactions is difficult due to the large and flat interaction surfaces of the binding partners. The BLIP and BLIP‐II proteins are unrelated in sequence and structure and yet each potently inhibit β‐lactamases. High‐throughput oligonucleotide synthesis was used to construct a 12,470‐member library containing overlapping linear and cyclic peptides ranging in size from 6 to 21 amino acids that scan through the sequences of BLIP and BLIP‐II. Phage display affinity selections and deep sequencing revealed that, despite the differences in interaction surfaces with β‐lactamases, rapid enrichment of consensus peptide regions originating from both BLIP and BLIP‐II contact residues in the binding interface occurred. BLIP and BLIP‐II peptides that were enriched by affinity selection were shown to bind β‐lactamases and disrupt the BLIP/β‐lactamase interaction. The results suggest that peptides that bind at and disrupt PPI interfaces can be identified through systematic peptide library construction, affinity selection, and deep sequencing.

Targeting peptide‐mediated interactions in omics

“…For this previous studies have been primarily been performed at peptide sequence level by addressing residue mutation on the natural SIPs. The mutation can be conducted by random high‐throughput strategy such as phage‐displayed or chemically synthetic peptide library [54–24], where the core SLiMs or hotspot residues of natural SIPs can be defined as the invariable motif of the library to randomly generate numerous short peptides containing such motif, from which those with increased interaction strength with protein B were screened and identified. The molecular optimization of direct readout can also be guided by rational strategy such as machine learning‐guided evolution [56], empirical peptide docking/scoring [57], rigorous QM/MM hybrid calculation [58], and dynamics simulation/energetics analysis [59].…”

“…The former is only applicable for helical peptides; it introduces a full‐hydrocarbon linker across specific residue spanning sites such as [ i ,i+3], [ i , i +4] and [ i , i +7] to “staple″ the native helical conformation of peptides [69], while the latter can be classified into disulfide‐bonded cyclization”, which cyclizes a double‐stranded peptide by introducing one or more disulfide bridges across its two strands [70], and head‐to‐tail cyclization [71], which cyclizes loop peptide by linking its N‐ and C‐termini. Previously, the hydrocarbon stapling and cyclization technique have been successfully employed to treat a variety of SIPs with improved competitive potency to disrupt diverse dPMI complexes by reducing indirect readout, such as PD1–PDL1 (for tumor immune) [54], ACE2–spike protein (for SARS‐CoV‐2) [72], MMP13–TIMP1 (for osteoarthritis) [73] and Keap1–Nrf2 (for sepsis) [74]. For example, the YAP (protein A) employs its two hotspot regions of α‐helix and Ω‐loop to interact with its cognate partner TEAD (protein B) to form a PMI system.…”