Directed Evolution of Stabilized Peptides with Bacterial Display

Abstract

Interactions between proteins govern cellular and the body’s states, including aberrant interactions found in diseases such as in cancers and infections. Small molecule drugs are not ideal in targeting these interactions as their size generally prevents efficient blocking of contacts over large surface areas. Antibodies and related biologics have seen clinical success in the past few decades and can block large surfaces but are typically limited to extracellular targets. Intermediate-size peptides have the potential to bridge this gap, with the ability to target large surface areas inside the cell. Peptide stapling, by chemically linking two or more amino acid residues, can confer affinity improvements, resistance to degradation, and better biological transport properties. As such, stapled peptides show promise as next-generation therapeutics. Unfortunately, existing methods to screen sequence and stapling locations suffer from numerous disadvantages including limited search space, lack of real-time monitoring of selections, and difficulty in incorporating the non-canonical amino acids used for amino acid stapling.

In this dissertation, I describe my research on stapled peptide discovery with bacterial incorporation of non-canonical amino acids. To screen stapled peptides of the type desired, we incorporated azidohomoalanine (AHA) into surface displayed peptides, enabling an in situ ‘click’ chemistry reaction to bridge two turns of an alpha helical (i, i+7) amino acid library for directed evolution. Using the p53-MDM2 interaction as a model target, we developed peptides that block MDM2 degradation of the tumor suppressor protein p53, an interaction that is dysregulated in a sizeable fraction of cancers. We generated and displayed a stapled peptide library on the bacterial cell surface with fixed residues for stabilization and binding requirements, while randomizing the remaining amino acids. After multiple rounds of selection, clones were sequenced and characterized. The dissociation constants of the peptide-MDM2 interaction were measured on both the bacterial cell surface by flow cytometry and in solution by bio-layer interferometry. The highest affinity variant, named SPD-M6-V1 with sequence VCDFXCYWNDLXGY (dissociation constant = 1.8 nM; X = azidohomoalanine) was selected for structural characterization by NMR spectroscopy, revealing a bicyclic disulfide and double click-constrained peptide. Sequencing showed that peptides with two cysteines were highly enriched, further suggesting that the MDM2-binding conformation was enforced with a disulfide bond. In addition, SPD-M6-V1 was the most protease-resistant peptide from the library that we tested.

Next, we stapled the displayed peptide library with chemically distinct linkers and screened each library separately. We performed deep sequencing to better understand the relationship between amino acid sequence and linker identity in contributing to high affinity MDM2 binding. We found that both linker-specific and linker-agnostic (i.e. MDM2-specific) mutations were enhanced. Finally, we developed a dual-channel, sequential labeling selection strategy to discriminate between high-display, low-affinity peptides and low-display, high-affinity peptides, two categories that would ordinarily overlap in a typical one-color screen in the absence of an independent display marker. In summary, this thesis develops the chemical tools to screen libraries of stabilized peptides on the bacterial cell surface and applies these techniques to select stabilized alpha helices that disrupt the p53-MDM2 interaction.