Structure-guided protein evolution to develop improved natural ligand-based chimeric antigen receptor t cell therapy against CD70 in acute myeloid leukemia

The National Cancer Institute awarded The Regents of the University of California, San Francisco $172,889 on January 12, 2026, under the Cancer Research Manpower program (CFDA 93.398) to develop and optimize chimeric antigen receptor (CAR) T-cell therapies for acute myeloid leukemia (AML).

The funded research applies structure-guided protein evolution to improve natural ligand-based CAR-T therapies targeting CD70 in AML. The project uses computational deep learning algorithms and high-throughput screening to identify coding sequence alterations in CD27-based CAR-T variants that enhance efficacy without increasing off-tumor toxicity. This approach addresses a critical gap: existing CAR-T therapies for AML that target CD33, CD123, or CLEC12A cause severe toxicities because these antigens are also expressed on healthy hematopoietic stem and progenitor cells. CD70 is highly expressed on AML tumor cells and leukemic stem cells but absent from healthy hematopoietic progenitors, offering a superior safety profile. Preclinical models demonstrate that CD27-based natural ligand CAR-Ts achieve improved tumor cytotoxicity, CAR-T proliferation, and survival compared to single-chain variable fragment approaches.

Performance occurs at the University of California, San Francisco in San Francisco, California, with a period of performance ending March 31, 2027. The Cancer Research Manpower program supports biomedical training opportunities for individuals pursuing careers in basic, clinical, and prevention research relevant to the National Cancer Program.

Abstract

This project seeks to develop and optimize chimeric antigen receptor (car) t-cell therapies for acute myeloid leukemia (AML). AML is the most prevalent acute leukemia in adults, yet has a dismal 5-year survival rate UN- DER 30% with current standard-of-care therapies. Despite their well-documented efficacy in the treatment of other hematologic malignancies, car-t therapies have shown limited efficacy in treating AML, highlighting the need for additional innovative research in this area. The most well-studied AML cars in development use single-chain variable fragments (scfvs) to target CD33, CD123, or CLEC12A, resulting in severe toxicities in patients due to their concurrent expression on hematopoietic stem and progenitor cells (hspcs). Instead, this study proposes to target CD70, a marker highly expressed on AML tumor cells and leukemic stem cells but not on hspcs, using a "natural ligand" car-t approach, based on CD70's unique native receptor, CD27. This approach has shown improved tumor cytotoxicity, car-t proliferation, and survival in preclinical models compared to scfv-based anti- CD70 car-TS. Given their superior safety profile, CD27-based car-TS merit further optimization for efficacy. Furthermore, there have been no comprehensive attempts to produce variants of natural ligand cars with AL- tered binding affinity and, thus, potentially improved car-t efficacy. Therefore, i hypothesize that by employing structure-based deep learning algorithms and high-throughput screening techniques, i can identify coding SE- quence alterations in CD27-based cars that enhance their efficacy without increasing off-tumor toxicity. In specific aim 1, i will evaluate the in vitro and in vivo efficacy of a set of CD27-variant cars with computation- ally predicted alterations in affinity to CD70. The phenotypic differences of these CD27-variant cars, particularly those influencing tumor cytotoxicity, will also be analyzed to elucidate the principles guiding natural ligand car- t function. In specific aim 2, i will perform an in vivo competition assay of car-t proliferation to screen a large library of CD27-variant cars. The leading candidates will undergo further validation for tumor cytotoxicity, SUR- vival benefits, and car-t phenotypes, along with verification that the low off-tumor toxicity of the CD27-based car is maintained. Successful execution of these aims is expected to yield superior CD27-based car designs that would be beneficial in various AML scenarios and potentially other CD70-expressing malignancies, with the long-term potential for translation to the clinic. Additionally, this natural ligand car development pipeline could be extended to additional cancer antigens, signifying a broad therapeutic potential. I will lead this project under the sponsorship of DR. Arun wiita, a physician scientist and expert in hematological malignancies and cancer antigen identification, and the CO-sponsorship of DR. Justin eyquem, an expert in car- t development and car screening technologies. Completion of my research plan, alongside my comprehensive training plan and robust mentorship from my research environment, will strongly equip me to become an inde- pendent physician scientist focused on cancer immunology and immunotherapy research.