Design and optimization of TCR-T therapeutics by an advanced T cell receptor-antigen binding prediction platform

Abstract Nightstar Biotechnologies has developed an Artificial Intelligence (AI) platform for therapeutic T-cell receptor (TCR) design to aid in the design of novel therapeutic TCR-T cells for cholanigiocarcinoma (CCA) treatment. Emerging biotherapies, such as TCR-T therapy, create a new avenue for T-cell-based biotherapeutics. While CAR-T cells are limited to targeting TAAs expressed on the cell surface, TCR-T cells allow for a larger repertoire of TAA targets including intracellular oncogenes. By exploiting the intrinsic features of TCRs, T-cells can be engineered with much higher sensitivity to improve both cytotoxicity and therapeutic T-cell persistence. Traditional experimental design methods for TCRs generally rely on TAAs that are highly expressed; however, design of TCRs allows targeting of TAAs with low expression levels, thereby promoting higher specificity and therapeutic efficacy. This design tactic is currently beyond the reach of the traditional approaches. The ability to flexibly increase or decrease the binding affinity of TCRs against multiple antigens is critical to (a) achieve detection and targeting of minimally expressed tumor antigens, while (b) avoiding off-target, off-tumor toxicities from recognition of non-specific antigen recognition. NightStar Biotechnologies has developed an AI foundation model, pMTnet, for predicting the interactions between TCRs and T-cell antigens (pMHCs) for designing optimal therapeutic TCRs and aid in the design of successful therapeutic TCR-T cells. The pMTnet platform can address these difficulties and dramatically narrow down the scope of required experimental validations to ultimately yield in better, faster, and cheaper design of TCR-T therapies to benefit cancer patients and those diagnosed with autoimmune disorders. As proof-of-concept for AI-assisted design and optimization of TCR-T cells, NightStar Biotechnologies intends to use their pMTnet model to optimize a TPX2-TCR-T cell as a new immunobiologic therapy for CCA. CCA is a rare, heterogenous cancer of the hepatobiliary tract that affects around 8,000 people a year with a five-year survival rate of only 9%. Approximately 75% of CCA patients are diagnosed at late stages with locally advanced or metastatic disease, resulting in very few patients that qualify for the only effective therapy: surgical resection. Expression of targeting protein for Xklp2 (TPX2) is associated with advancing CCA disease and is linked to aberrant cell proliferation, promotion of DNA damage repair, as well as migration and invasion in CCA. In this proposal, Nighstar Biosciences will (1) Identify high affinity TCR sequences against TPX2, (2) evaluate the functional ability of the TPX2-TCR-T cells against CCA cell lines in vitro, and then, (3) validate the anti-tumor efficacy of the lead TPX2-TCR-T candidate using mouse models engrafted with CCA patient-derived xenografts. Upon completion of this proposal, Nightstar Biotechnologies will have shown proof- of-concept for their pMTnet TCR discovery platform to develop therapeutic TCRs, as well as characterized the feasibility of a TPX2-targeted TCR-T therapy for CCA, providing a hopeful strategy for patients diagnosed with this disease.