Macrocycles are a therapeutic modality of interest for drugging challenging disease targets due to their high binding affinity and membrane permeability, enabling the discovery of orally delivered and efficacious therapeutics. Recent methods have enabled the purification-free synthesis of chemically diverse macrocycles and the generation of de novo libraries for high-throughput screening and identification of functional macrocycle inhibitors. However, bottlenecks existed in these approaches which limited peptide synthesis to only thousands of compounds. The goal of my thesis project was to further increase the throughput and diversity of macrocycle library synthesis and to exploit these methods to discover antagonists of important disease targets.

In my first project, I upgraded the commercial peptide synthesizer used by our lab to enable the simultaneous synthesis of 4 times as many peptides while using more than 3 times as many amino acids building blocks. While the machine is constructed to use 96-well plates, I found a 384-well filter plate which could be installed in the same machine with minor modifications. Furthermore, instead of storing amino acids in 50 mL tubes as designed, a new amino acid holder was constructed to enable storage of amino acids in 96-well plates. This now enables the use of 100s of amino acids to synthesize 1,536 peptides at once, offering a significant improvement in both library size and diversity towards the discovery of hits against challenging disease targets.

In my second project, I have applied these new tools of high-throughput synthesis to prepare a chemical macrocycle library of unprecedented diversity. A design was conceived in which 4-mer peptides would be synthesized from 4 separate pools of 100 amino acid building blocks. These pools were filled from a total of 242 and then the 1,536 resulting disulfide-cyclized peptides were received in high purity. These peptides were then coupled at nanomole-scale with 20 carboxylic acids in solution from a total set of 347. In total, a library of 30,720 macrocycles was created from 589 chemical building blocks. This library was screened against the MDM2/p53 protein-protein interaction from which a sub-micromolar inhibitor was identified.

In my third project, I again applied these tools to prepare a macrocycle library which was targeted for binding to the IL-23 receptor at the binding site of IL-23. This library was synthesized in a similar fashion in which 1,536 disulfide-cyclized peptides were made from 241 amino acids and then diversified into 30,720 products by coupling to a pool of 320 carboxylic acids. One of the disulfide-cyclized peptides was found to yield activity in a highly sequence-specific fashion. The scaffold was further optimized by the synthesis and screening of several sub-libraries in which each of the five building blocks was substituted with many additional analogues, yielding a 100-fold improvement in activity. This binding was confirmed by fluorescence polarization, however it was found to depend on the presence of a recombinant Fc tag. When the Fc tag was removed, a 100-fold loss in activity was observed with only a moderate binding affinity to the IL-23 receptor itself (IC50 of 1.6 µM).