CD33-TIM3 Dual CAR T Cells: Enhancing Specificity While Maintaining Efficacy Against AML

Background: The myeloid associated antigen CD33 is overexpressed in over 88% of Acute myeloid leukemia (AML) patients, making it a suitable target antigen for CAR T cell therapy across different genetic subtypes. Despite the development of clinical trial data, there remains a concern about on-target off-leukemia toxicity. To enhance specificity, we generated dual targeting CAR T cells that target CD33 and/or T-cell immunoglobulin and mucin-domain containing-3 (TIM3). The latter is known to be expressed on LSCs, but not on healthy hematopoietic precursor or stem cells. Notably, TIM3 is recognized for its inhibitory immunomodulatory role further enhancing the suitability of this target antigen. The objective of this research was to explore various CD33 and TIM3 dual-targeting CAR T cells approaches that enhance AML specificity and maintain high anti-AML activity.

Methods: The hybridoma technology was used to generate antibodies against TIM3 immunized mice. Antibodies were screened for TIM3 specificity via ELISA and FACS. The anti-TIM3 single-chain variable fragment (scFv) DNA sequence was sequenced from hybridoma cells. Furthermore, the binding ability to TIM3 antigen (PDB ID: 5F71) was evaluated using AlphaFold. The CD33 scFv was derived from Gemtuzumab ozogamicin (clone: hP67.6). ScFv sequences and co-stimulation domains (CD28 or 4-1BB) were cloned into a pMP71 vector. Retrovirus for transduction was produced using the 293Vec-GALV and RD114 retroviral production system. In vitro co-culture assays of CAR T cells and target cells were performed to study the efficacy of CAR T cells. The cytotoxicity against wild-type and TIM3 transduced AML cell lines (THP-1 and OCI-AML3) was assessed by multiparameter flow cytometry (MPFC). The secretion of effector cytokines (IFN-γ, TNF, IL-2) was analyzed via CBA assays. In addition, avidity between CAR T cells and target cells was determined by Z-Movie analyzer. The different CAR constructs were screened for on-target off-tumor toxicity in colony forming unit assays (CFU) using isolated CD34⁺ hematopoietic stem and progenitor cells (HSPCs) from healthy doners after 14 days. In order to compare CAR constructs regarding long-term efficacy in antigen restimulation assays, CAR T cells were co-cultured with irradiated TIM3 transduced OCI-AML3 every 4 days at an E: T ratio of 1:1 for 24 days. Moreover, CAR T-cell proliferation, checkpoint marker expression and T-cell subset differentiation were analyzed via MPFC.

Results: All dual CAR T cells (compound, split, tandem, pooled) were generated with high and robust transduction efficacy (Figure 1A). TIM3-dependent fratricide was not observed during CAR T cell manufacturing. Compared to single antigen targeting CAR T cells, dual CAR T cells showed enhanced cytotoxicity against AML cell lines and primary AML cells (Figure 1B). Moreover, we also observed a strong increase in the secretion of proinflammatory cytokines (IFN-γ and IL-2) and higher avidity of dual CAR T cells. Notably, split CAR T cells demonstrated greater specificity in cocultures directed against mono vs. dual-target antigen expression cell lines. In accordance with these observations, split CAR T cells did not exert on-target off-leukemia toxicity against healthy HSC in CFU assays. In the antigen restimulation assay, we observed that compound CAR T cells exhibited diminished expanding capacity and heightened expression of exhaustion markers in comparison to the other CAR T constructs.

Conclusion: We developed multiple CD33 and TIM3 dual CAR T cells using both "AND" and "OR" gating strategies in this study. Our findings revealed that dual CAR T cells provided higher avidity and cytotoxicity compared to single targeting CAR T cells against dual antigen expressing target cells in vitro. Importantly, only the split CAR T constructs demonstrated specific killing of CD33⁺TIM3⁺ cell lines and primary AML cells while sparing HSPCs. Prospectively, compound, tandem, pooled CD33 TIM3 specific CAR T cells might be helpful in a bridge to transplant setting, whereas split CAR T cells might provide a higher safety profile, thereby allowing a transplant independent approach. To advance our concepts, we are currently conducting in vivo experiments in an NSG mouse model.