Abstract by Fabian Hink

The potential to combine the benefits of small molecules, namely oral availability, and cell permeability with the advantages of antibodies such as potency and selectivity make peptides attractive drug modalities. The development of mRNA display offers the possibility to screen trillion of peptides simultaneously to find de novo binders for a target protein. The combination with the flexizyme technology, known as the RaPID system, enables genetic code reprogramming allowing the incorporation of non-canonical amino acids and the formation of cyclic peptides. The focus of this thesis was to explore the potential and limitations of the RaPID system with respect to target binding, with the aim to expand the druggable human proteome. For this, we focused on the discovery of cyclic peptides that target helical peptide-protein interactions, small protein domains, RNA-protein interactions, and intrinsically disordered proteins.

In chapter 3, we tested the ability of the RaPID system to target helical peptide-protein interactions. Peptides show promise as a class of molecules to inhibit disease-associated interactions between proteins. Macrocyclization strategies such as peptide stapling and capping can enhance the binding affinity to a target protein by constraining the peptide into a biological active  α-helical conformation. We utilized RaPID to find cyclic peptides targeting the model protein Mcl-1, which binds the  α-helix-forming BH3 domain. Intriguingly, a small macrocycle discovered among the top-enriched peptides was identified as a novel N-capping motif. We found that a thioether-connected 15-membered macrocycle between a cysteine at the 4^th position and the acetylated N-terminus stabilized helicity and improved binding of the peptides. Further exploration highlighted the pivotal role of the first amino acid's sidechain in the N-capping motif. Of the tested amino acids in the N-terminal position, D-phenylalanine exhibited the highest helix-inducing effect. Moreover, incorporation of an N-terminal D-phenylalanine in conjunction with macrocyclization showed the potential to induce helicity in naturally occurring BH3 peptides. Our results demonstrate the discovery of an N-capping motif that enhances helicity and binding of peptides. Compatibility with the ribosomal synthesis without protection groups enables future RaPID-based de novo screening attempts of helix-focused peptide libraries. Additionally, its ease of solid-phase peptide synthesis renders this N-capping motif a promising entity for future endeavors in rational design-based peptide drug discovery.

In chapter 4, in a first step the RaPID system was established and optimized by screening against the small folded protein domains PDZ2 of PSD95 and ubiquitin as part of M1-linked ubiquitin chains. Due to the availability of sequence data for peptide binders for PDZ2 of PSD95 and ubiquitin, the discovered peptide sequences could be compared with the reported motifs from literature. The overlap in the motifs supported the idea that the RaPID system can identify sequences that mimic natural motifs. Then, the newly established RaPID system was employed to evaluate the applicability for the screening against RNA binding domains. RNA-binding domains are highly abundant, often involved in pathological processes and challenging to target due to shallow surfaces. Here we targeted the two RNA recognition motifs (RRMs) of hnRNPA1, shown to be potential cancer targets and important in the modulation of phase-separating behavior of hnRNPA1. RaPID selection against RRM1 resulted in an enrichment based on nucleic acid binding. Efforts to minimize these interactions included the addition of a catalytically inactive RNase able to bind DNA-RNA duplexes aiming for a masking effect. These endeavors have not yet proved successful. In contrast, a RaPID selection against RRM2 resulted in the discovery of cyclic peptide binders. The tightest binder showed affinity in the mid-nM range. Furthermore, the peptide was able to inhibit RNA binding. Our results demonstrate that the RaPID system can be an option to find de novo peptide binders for RNA binding domains with the ability to prevent RNA binding. Future efforts should focus on the reduction of binding between the nucleic acids and the RNA binding domain to make this approach more generalizable.

In chapter 5, we investigated the suitability of the RaPID system to find de novo cyclic peptide binders for intrinsically disordered proteins. More than 30 % of the human proteome are not well structured. Some of these highly abundant intrinsically disordered proteins (IDPs) play crucial roles in various diseases. The discovery of molecules with sufficient binding properties to target IDPs has proven difficult in the past. In a first test, we employed the RaPID system to find cyclic peptides to target the monomeric  α-synuclein (aSN), an important protein in the development of Parkinson’s Disease. An initial screen resulted in the enrichment of false positive hits. Addition of the kosmotropic agent trimethyl N-oxide helped with the enrichment of aSN-specific peptides. Those peptides showed the N-cap described in chapter 3 and formed amphipathic helices. Protein NMR indicated weak C-terminal binding for one of the peptides. Our results demonstrate the possibility to discover peptides binding to the IDP aSN using the RaPID system. The second chosen model IDP was the low complexity domain (LCD) of hnRNPA1. The LCD can mediate phase separation, a phenomenon important for the formation of membraneless organelles. Mutations in the LCD are associated with fibrillization and amyotrophic lateral sclerosis. Here, we set out to discover de novo cyclic peptide binders for the LCD of hnRNPA1 using the RaPID system. However, the peptides bound to streptavidin rather than the LCD. The LCD potentially poses a significant challenge for the RaPID system, which discards weak binders using stringent selection methods. The RaPID method may have reached its limit. 

Overall, we have increased the classes of proteins that can be targeted by RaPID-type peptides. These identified peptides represent starting points for medicinal chemistry optimization to create future research tools and therapeutics. Moreover, we discovered some of the current limits of the basic RaPID system including some RNA binding domains and IDPs.