Small cyclic peptides combine excellent binding properties with the potential to passively cross cell membranes, making them attractive for developing orally available drugs to challenging disease targets or for addressing intracellular proteins. However, the generation of membrane-permeable cyclic peptides against new targets has been challenging, mostly due to the lack of methods for generating and screening large libraries of small cyclic peptides. Such peptide libraries with high structural diversity may be best produced by chemical synthesis which allows using hundreds of commercially available Fmoc-amino acids and other chemical building blocks, but the number of peptides that can be chemically synthesized and tested in parallel has so far been limited to the lower ten-thousands.

In my PhD thesis, I aimed at developing methods for generating and screening much larger libraries of small, chemically synthesized peptides. Towards this end, I proposed a strategy in which a large number of m short, chemically synthesized peptides are combinatorially ligated with another large number of n short peptides, to generate m × n different peptide pairs. The peptides are paired in wells of microtiter plates by chemical reactions and directly screened using functional assays for assessing target engagement.

In a first project, I generated libraries of linear peptide pairs by connecting two short peptides, each containing a thiol group, over a disulfide bridge. I showed that thiol-bearing peptides, synthesized in 384-well plates, can efficiently be paired by disulfide formation in aqueous buffer containing dimethyl sulfoxide. Subsequently, large libraries of peptide pairs were generated in 1,536-well plates by transferring the peptides with acoustic dispensing followed by screening the reaction mixture directly in the same plate. In a proof-of-concept study, I generated and screened a library with one peptide pair per well and identified several strong hits of the disease target thrombin. In a follow-up study, I generated and screened multiple peptide pairs per well to increase throughput and showed that active pairs can be rapidly identified by testing the peptide pairs of a hit well individually. This strategy allowed screening over 800,000 peptide pairs and led to the identification of nanomolar thrombin inhibitors.

In my second project, I developed a combinatorial peptide pairing strategy for the synthesis and screening of cyclic peptides. I proposed to synthesize short peptides containing thiol-reactive groups at both ends, and to then dimerize these 'reactive' peptides with short dithiol peptides. To produce the 'reactive' peptides, I used a strategy in which an excess of volatile divinyl sulfone is incubated with dithiol peptides that have a thiol on their N- and C-terminus thereby installing a thiol-reactive vinyl sulfone group at each end of the peptide. The excess of divinyl sulfone is conveniently removed by evaporation. I showed that the peptide pairing strategy works with crude peptides and that cyclic peptides are efficiently generated in a combinatorial fashion.

The methods developed in my PhD thesis allow generating libraries that contain over a million chemically synthesized, structurally highly diverse peptides. I hope that in the future these libraries will allow identifying binders of challenging disease targets to generate membrane-permeable peptide drugs for oral application and to address intracellular targets.