Structurally Optimized, IL-2-Armored CLL1 CAR-NK Cells Are Highly Potent Effectors Against AML without Hpsc Toxicity

Introduction: Acute myeloid leukemia (AML) is an aggressive hematological malignancy with a poor prognosis, especially for patients with relapsed or refractory disease. Natural killer (NK) cells are emerging as promising platform for immunotherapy of AML because in contrast to T-cell based treatment strategies, they appear to have more manageable safety profiles and fewer graft restrictions, hereby providing the opportunity for off-the-shelf products. Chimeric antigen receptor (CAR) based immunotherapy of AML has not been as successful as for other types of cancer, mainly due to the lack of a truly tumor cell-specific target antigen. CLL-1 (encoded by CLEC12A) is a promising target antigen in AML because it is strongly expressed on malignant cells but absent on normal hematopoietic stem and progenitor cells (HSPC). We hypothesized that NK cells genetically modified to express a design-optimized CLL-specific CAR and provide autocrine cytokine stimulation could demonstrate enhanced cytotoxicity and in-vivo persistence, potentially leading to a durable anti-leukemic efficacy.

Methods: A CLL1-specific scFv was used to generate four CARs with varying spacer structures, designated short, intermediate, long and long-flexible in rising order of size and flexibility. Using multiparameter flow cytometry and multiplex cytokine analysis, we compared the viability, growth, phenotype and secretome of CLL1-CAR-NK during production with a focus on differences between the four constructs. Tonic signaling was investigated with a combination of Jurkat^{NFkB/NFAT/AP-1} triple-reporter cells, confocal imaging and phospho-Western Blots. To test the antileukemic potency of the CAR-NK, we used a panel of eight commonly used AML cell lines as well as six primary AML samples. Additionally, we overexpressed full-length CLL1 on the CLL1^neg K562 erythroleukemia cell line, which was sorted to yield variable CLL1^high/med/low K562 populations for specificity and stable immune conjugate testing. Using these models, we performed in-vitro cytotoxicity, secretion and serial challenge assays with a FACS-based readout. CFU assays after CLL1 CAR-NK co-culture were used to investigate potential myelosuppression. Finally, we deployed CLL1 CAR-NK with or without endogenous interleukin support in-vivo, using an aggressive U937 NSG xenograft model.

Results: The four constructs showed comparable transduction efficiency, but the MFI differed significantly, with the shortest construct showing the strongest expression. We observed that the short construct caused tonic signaling, leading to activation-induced cell-death through the upregulation of CD95. In the triple reporter model, the construct was associated with tonic signaling along the NFkB pathway. Tonic signaling could be abrogated by pharmacological inhibition, leading to a cessation of NFkB signaling, thereby improving viability and expansion. Stable conjugate formation and short-term cytotoxicity were similar across the four constructs and significantly better than non-transduced controls. In contrast, long-term in-vitro re-challenge assays demonstrated superiority of the long-flexible construct, even though it showed the lowest expression by MFI. Long-flexible CLL1 CAR-NK outperformed NT-NK controls in both cytotoxicity and proliferation against a panel of AML cell lines and primary samples. CLL1 CAR-NK were non-toxic to allogeneic HPSC as measured in a standard CFU assay. In-vivo, structurally optimized CLL1 CAR-NK supported with an endogenous IL-15 or IL-2 secretion cassette were safe, effective and prolonged the median OS from 26 to 37 days as compared to non-transduced controls (p<0.001). Interestingly, IL-2 outperformed IL-15 and achieved a 72% reduction in tumor burden over the first 21 days after treatment as well as a moderate, but not significant improvement in overall survival.

Conclusion: Our data emphasizes the pivotal role of CAR design for the efficacy of cellular immunotherapy. Varying spacer designs mediate comparable short-term cytotoxicity, but tonic signaling hampers both their growth and their long-term persistence. In-vivo, CLL1 CAR-NK are safe and effective. They benefit significantly from endogenous interleukin stimulation, with IL-2 outperforming IL-15. We demonstrate that structurally optimized allogeneic CLL1 CAR-NK are a promising new strategy for the off-the-shelf treatment of AML.