Leukemia Targeting Via Cytotoxic T Cells Expressing a CD33 Specific Chimeric Receptor.

CD33 is a surface protein expressed on the malignant blast cells of most acute myeloid leukemia (AML) and is a potential tumor-associated target antigen. Immunotoxin-based therapy against CD33⁺ myeloid leukemia has shown promising results with no long-term side effects on CD33⁺ progenitor cells, but is associated with liver toxicity. We describe a cell therapy approach targeting CD33 for AML, which should be more effective than the immunotoxin and have lower non-specific toxicity. The genetic modification of human T cells to express tumor specific chimeric immune receptors (TCR) can produce large numbers of tumor-specific T cells. Unfortunately, these modified T cells have only a limited response to tumor cells in vivo. By contrast, EBV-specific cytotoxic T cells (EBV-CTLs) expand and have long-term activity in vivo. We exploited this phenomenon and redirected EBV-CTLs against CD33. EBV-CTLs were generated from six EBV-seropositive healthy donors and transduced using a retroviral vector encoding the chimeric CD33 specific TCR. The cRCD33-EBV-CTL maintained their pre-transduction immunophenotype and specifically lyzed EBV-LCL and CD33⁺ targets (specific lysis of 30% and 35% at E:T ratio 25:1). They also produced Th1 and Th2-type cytokines on exposure to CD33⁺ targets. Addition of the CD28 intracellular domain did not significantly increase cytotoxicity to CD33⁺ targets, but resulted in a 5–7 and 2–3 fold increase in IFN-γand IL-5 production. We confirmed the specificity of the TCR interactions with CD33 by inhibiting them with MHC class I, class II and CD33 blocking monoclonal antibodies (MoAb). Preincubation of CD33⁺ target cells with the CD33-blocking MoAb resulted in 40% inhibition of lysis and 60% inhibition of cytokine release by cRCD33-EBV-CTLs, whereas preincubation of CD33⁺ targets with class I and II blocking MoAb had no significant effect. Incubation of autologous LCLs with MHC Class I and Class II MoAb showed MHC class I was the major restriction element for the native antigen receptor of both nontransduced and modified EBV-CTLS. Of note, the cRCD33-EBV-CTLs proliferated only in response to autologous LCLs and not to CD33⁺ targets. Failure to proliferate was not overcome by incorporating the CD28 endodomain into the CAR. By contrast, cRCD33-EBV-CTLs can be expanded and maintain their specific activities toward CD33⁺ targets by co-culture with LCLs, and these effects are sustained for over 30 days in culture. Despite their activity against CD33⁺ blast cells, long-term culture-initiating cell assay demonstrated that neither cRCD33-EBV-CTLs nor cRCD33-CD28-EBV-CTLs had measurable toxicity against CD33⁺ progenitor cells isolated from bone marrow samples. To assess activity in vivo, we injected NOD-SCID mice bearing a human CD33⁺ AML tumor with EBV-CTLs x4 weekly, starting 5 days after tumor inoculation. Whereas the administration of PBS or untransduced EBV-CTLs had no effect on tumor progression, cRCD33-EBV-CTLs showed a trend to reduced tumor growth, the statistical significance of which is being confirmed on a larger series of animals. Incorporation of the CD28 endodomain had no additive effect on anti-tumor activity. Importantly, no adverse effects wee seen on normal CD33⁺ progenitor cells in vivo. We conclude that EBV-CTL expressing the CD33 chimeric receptor may be appropriate as a therapeutic modality for the treatment of myeloid leukemia.