(517c) Development of a Kynureninase Clinical Candidate, a First-in-Class Enzymatic Checkpoint Inhibitor

Tumors inhibit immune responses largely through two mechanisms: (1) Through protein-protein interactions (e.g. PD1:PD-L1) and (2) By cancer-induced accumulation of immunosuppressive metabolites in the tumor microenvironment. The most potent such metabolite is the tryptophan-oxidation product kynurenine (Kyn) that induces tolerance through many mechanisms (e.g. promotes Treg differentiation and inhibits CD8+ T cell function) and promotes tumor survival and motility. Kyn is synthesized in vivo by three different enzymes: IDO1, IDO2, and TDO. Pharmacological blockade of the Kyn pathway is highly sought after, and three small molecule inhibitors of IDO1 are in clinical trials with TDO inhibitors expected to enter the clinic soon. However, because of the redundancy of the Kyn synthesis pathway, inhibition of IDO1 or TDO alone cannot eliminate Kyn accumulation, a fact underscored by the modest effects reported for inhibitors to date. Additionally, complete blockade of Kyn synthesis would disrupt intracellular Kyn homeostasis leading to toxicities (through loss of NAD synthesis etc.). Fortuitously, excess Kyn must be physically exported out of tumor cells and into immune cells in order to have an immunosuppressive effect, giving the opportunity to prevent Kyn immune suppression by eliminating it from the tumor microenvironment and serum. Therefore, we resolved to eliminate the extracellular Kyn pool though a completely different mechanism - the systemic administration of a Kyn-degrading enzyme, kynureninase.

As a proof of concept, we utilized a highly active bacterial kynureninase to demonstrate that enzymatic elimination of tumor-produced Kyn is an effective anti-cancer therapy. Administration of the bacterial kynureninase to B16-OVA melanoma syngrafts in C57BL/6J mice reduced serum Kyn level, resulted in significant tumor growth retardation, and extended survival in a manner indistinguishable from that observed with the immune checkpoint inhibitor antibodies, α-PD-1 or α-CTLA-4. Despite its effectiveness, the bacterial kynureninase is not a realistic therapy for clinical use because it is highly immunogenic and has low serum stability (T_1/2 = ~1 hour). There exists a human kynureninase enzyme, but it is ~700X weaker than the bacterial enzyme so it elicits no in vivo anti-tumor effect. Therefore, we have undertaken a large-scale protein engineering effort to increase the catalytic active and stability of the non-immunogenic, human kynureninase enzyme to develop a clinical candidate. We first developed parallel genetic selection systems in E. coli and S. cerevisiae that enable high throughput screening of >10⁸ variants per library. We then performed numerous rounds of directed evolution, employing error-prone library construction, comprehensive codon mutagenesis, targeted mutagenesis of disparate residues, phylogenetic guided mutations, and DNA shuffling. Through a step-wise process of improving enzyme activity, introducing key substrate docking residues important for substrate binding, and then further directed evolution, we successfully engineered human variants having >600X improved activity over wildtype that show in vivo efficacy. Further efforts to reduce mutational burden, improve stability, and optimize expression have enabled development of kynureninase variants with clinical potential.