Design of cyclic peptides that bind protein surfaces with antibody-like affinity

There is a pressing need for new molecular tools to target protein surfaces with high affinity and specificity. Here, we describe cyclic messenger RNA display with a trillion-member covalent peptide macrocycle library. Using this library, we have designed a number of high-affinity, redox-insensitive, cyclic peptides that target the signaling protein Gαi1. In addition to cyclization, our library construction took advantage of an expanded genetic code, utilizing nonsense suppression to insert N-methylphenylalanine as a 21st amino acid. The designed macrocycles exhibit several intriguing features. First, the core motif seen in all of the selected variants is the same and shares an identical context with respect to the macrocyclic scaffold, consistent with the idea that selection simultaneously optimizes both the cyclization chemistry and the structural placement of the binding epitope. Second, detailed characterization of one molecule, cyclic Gαi binding peptide (cycGiBP), demonstrates substantially enhanced proteolytic stability relative to that of the parent linear molecule. Third and perhaps most important, the cycGiBP peptide binds the target with very high affinity (K₁ ≈ 2.1 nM), similar to those of many of the best monoclonal antibodies and higher than that of the βγ heterodimer, an endogenous Gαi1 ligand. Overall the work provides a general route to design novel, low-molecular-weight, high-affinity ligands that target protein surfaces.