Normal Myeloid Cells Are Required for Sustained CAR T Cell Activity Against Myeloid Tumor in a Humanized Mouse Model

Chimeric antigen receptor (CAR) T cells are effective against B cell malignancies and multiple myeloma, but their efficacy has been limited to date for acute myeloid leukemia (AML). We sought to investigate whether there were fundamental differences in targeting B cell antigens as compared to myeloid antigens with CAR T cells, that may shed light on the mechanism of CAR T cell resistance in patients with AML.

For these studies, we utilized human CAR T cells targeting CD19 (CART19) and CD33 (CART33), canonical B cell and myeloid cell antigens, respectively. To ensure that the potency of the two CAR constructs were equivalent, we generated dual CD19 and CD33 expressing cell lines, by adding CD33 to Ramos, a CD19+ B lymphoblastic cell line, and adding CD19 to THP-1, a CD33+ myeloid cell line. We confirmed that CART19 and CART33 were equally potent against CD33+Ramos and CD19+THP-1 cells.

To investigate the influence of normal hematopoietic cells on CAR T cell behavior, we incubated CD19+THP-1 cells with CART19 and CART33 in the presence of peripheral blood (PB) or bone marrow (BM) mononuclear cells. We found that both PB and BM enhanced tumor clearance to a similar degree for each CAR construct. Additionally, IL-6 was detected in the supernatant of PB or BM co-cultured with CART19 and CART33, and these levels were markedly increased in the presence of tumor cells. Notably, THP-1 cells by themselves produced high levels of IL-6 upon exposure to CAR T cells, likely reflecting the myeloid origin of this cell line, while Ramos cultured with these same CAR T cells did not produce IL-6. We assessed other myeloid cell lines (U937, KG-1, Kasumi-3, Molm13, HL-60, and K562) and also noted IL-6 production when co-cultured with CART33, although the levels were significantly lower than that produced by THP-1. Of note, IL-6 levels were slightly but consistently higher with CART19 than with CART33 in these in vitro assays, which we attribute to the loss of normal myeloid cells from CART33-mediated killing.

To study the effects of normal hematopoiesis on human CAR T cell activity in vivo, we injected NSGS mice with human cord blood CD34+ hematopoietic stem cells (HSCs) to generate a human hematopoietic system in these mice, followed by administration of untransduced (UTD) control T cells, CART19 or CART33. To prevent the confounding effect of allogeneic killing, CAR T cells were generated from T cells of the same cord blood product as the CD34+ cells. We confirmed the expected loss of human CD19+ B cells and CD33+ myeloid cells in the peripheral blood after CART19 and CART33 treatment, respectively. Surprisingly, we found that only CART33 treatment led to elevated plasma human IL-6 levels in this model. We then injected CD19+THP-1 cells to the mice after HSC engraftment, to assess the anti-tumor activity of the CAR T cells and to increase the potential for toxicity. Consistent with our in vitro data, mice with a human hematopoietic system cleared tumor more efficiently than mice without prior HSC engraftment after treatment with CART19 or CART33. However, while we observed mild weight loss and IL-6 elevation in mice after CART19 treatment, this effect was much more pronounced in mice that received CART33.

We hypothesized that the presence of antigen on normal myeloid cells both increased the toxicity and decreased the efficacy of CART33, due to a massive release of inflammatory cytokines from myeloid cells in the immediate aftermath of CART33 treatment, followed by loss of the augmentation of CAR T cell activity mediated by myeloid cells in the long term. To test this hypothesis, we engrafted mice with either control HSCs or CD33 KO HSCs, followed by injection of THP-1 and CART33. Only mice with CD33 KO HSCs maintained myeloid cells after CART33, as expected. CD33 KO HSC-engrafted mice exhibited less toxicity after CART33 treatment than mice with control HSCs, in that they did not lose weight or demonstrate elevated IL-6 levels. Furthermore, absence of CD33 on myeloid cells led to enhanced CAR T cell expansion and persistence, that resulted in better long-term tumor control.

In summary, our data suggests that targeting myeloid antigens with CAR T cells may be intrinsically self-defeating due to loss of myeloid cells that are required for sustained CAR T cell activity. These studies illuminate the challenges when extending CAR T cell therapy to myeloid malignancies, and highlight the importance of normal myeloid cells in augmenting T cell-based immunotherapies.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal