Development of New Bicyclic Peptide Formats and their Application in Directed Evolution

Cyclic peptides have a range of properties that make them attractive as therapeutics such as high binding affinity, good target selectivity and low toxicity. While most peptides in the clinic are derived from nature, various powerful in vitro evolution techniques allow now the de novo development of cyclic peptide ligands to targets of interest. Our group is specialized in the development of bicyclic peptide ligands by phage display. A current limitation of phage-selected bicyclic peptides is their conformational flexibility limiting the binding affinity due to entropic effects or even preventing the generation of binders to challenging targets. The main goal of my thesis was the development of new bicyclic peptide formats that are conformationally more rigid or even have a stable fold in solution. The proposed bicyclic peptide formats are based on a rigid chemical structure containing multiple polar groups and peptides that are anchored to the structure and fold tightly around by forming non-covalent interactions. I also envisioned to cyclise peptide phage display libraries with several such chemical structures in parallel to generate and screen structurally highly diverse bicyclic peptide libraries. In a first project, I designed and synthesized three new compounds that have i) three thiol-reactive groups to react with peptides and ii) several polar groups. Two of these compounds efficiently and selectively cyclised linear peptides containing three cysteine residues. Functional and structural analysis of the bicyclic peptides showed that the relatively small compounds at the centre of bicyclic peptides significantly influence the activity and conformation of the peptides. These results demonstrated the prominent structural role of chemical linkers in peptide macrocycles. In a second project, I applied the compounds developed in the first project in phage selections. The sequences of isolted bicyclic peptide ligands were strongly influenced by the small chemical structures, suggesting that the peptides fold closely around them. X-ray structure analysis revealed that one of the chemical structures indeed nucleated peptides by forming H-bond interactions. This result showed that small rigid chemicals can serve as nucleating scaffolds and directs the way towards the generation of peptide ligands being folded in solution. A third project is based on an unexpected observation in the second project, namely that panning peptide phage libraries without prior alkylation with compounds yielded mostly peptides with four cysteines. The peptides bound with good affinity and characterization revealed that they contain two disulphide bridges and thus formed bicyclic peptide structures. An interesting feature of disulfide-based bicyclic peptide libraries was the high topological diversity: oxidation of peptide libraries of the form Xm CXn CXo CXp yielded 3 × (m+n+o+p) different bicyclic peptide topologies because the fourth cysteine could appear in any of the (m+n+o+p) randomized amino acid positions (X) and the four peptides in such peptides could pair in three different ways. X-ray structure analysis revealed an important structural role of the disulfide bridges. The new method offers an easy and robust way to generate bicyclic peptide ligands.