Targeting CD123 In Leukemic Stem Cells Using Dual Affinity Re-Targeting Molecules (DARTs^®)

T-cell directed killing of tumor cells using bispecific reagents is a promising approach for the treatment of hematologic malignancies. In contrast to B-cell malignancies, this approach has been limited in AML by the lack of tumor-specific antigens/targets. CD123 (IL3RA) is highly and differentially expressed in AML blasts compared to normal hematopoietic stem and progenitor cells and is a potential target for immunotherapy. We investigated the ability of a DART constructed from an antibody to CD123 (7G3) and a MacroGenics’ proprietary CD3 antibody to redirect T cells against CD123⁺ AML blasts. DARTs consist of 2 independent polypeptides, each comprising the V^H of one antibody in tandem with the V^L of the other antibody. The resultant heterodimer is stabilized by a disulfide bond at the carboxyl terminal domains of the 2 V^H regions. This construct binds to both the N-terminal extracellular domain of human CD123 and to the extracellular domain of CD3 in the human T-cell receptor complex. Our in vitro studies demonstrate that the CD3xCD123 DART induces specific aggregation of TCR/CD3⁺ Jurkat or human T cells and human CD123-transduced K562 (K562^CD123/GFP) cells compared to a control DART lacking specificity for one of the antigens (16±3.2% vs. 1.6±0.2%, p=0.0074) or when compared to CD3xCD123 DART incubated with control GFP-transduced K562 (K562^GFP) cells. Incubation of human T cells (1:1 ratio) with K562^CD123/GFP and CD3xCD123 DART vs. control DARTS (10 ng/ml) for 5 days in vitro resulted in profound T-cell activation (CD25 expression, 88.8±2.7% vs.1.2±0.2%; p=0.0009), T-cell proliferation (VPD-450 proliferation assay; 98.2 ± 0.4% vs. 2.27± 0.4%, p=0.0001), expansion of the central memory T cell compartment (T_CM, 62.6±1.5% vs. 5±1.3%, p<0.0001), and killing of K562^CD123/GFP targets as measured by 7-AAD FACS (97±0.9% apoptosis relative to the control; p<0.0001) and chromium release assays (28.5% vs.3.1%; p=0.0002). As expected, aggregation, T-cell activation and target cell killing was negligible when T cells and CD3xCD123 DART were incubated with control K562^GFP cells. Similar results were seen when A20 targets (BALB-C/H-2^d B lymphoma cell line) overexpressing CD123 were used (7-AAD⁺ after 18 hours, 91.1±2.01% vs. 28.1±0.76%, p=0.0012). In spite of very low E:T ratios (0.009:1-0.071:1), when primary frozen/thawed AML peripheral blood specimens (n=6; CD123⁺ blasts ranging from 34.7 to 87.2%) were used, there was a profound and universal CD3xCD123 DART-specific activation and expansion of the few human T cells present in these AML samples (ranging between 0.9 to 6.6% of cells). After 6 days of in vitro incubation (37°C w/o exogenous IL-2) of each AML sample (1x10⁶/ml) with CD3xCD123 DART (0.1 ng/ml) or control DARTs, there was a CD3xCD123 DART-specific increase in both T cell numbers (median: 8 fold, range:1.7-15 fold vs. 0.6-1 fold for control DARTs), and in the percentage of T cells expressing CD25 (median: 69.4 fold, range: 21.4-152.8 fold vs. 0.2-6.5 fold for control DARTs). This was accompanied by a dose-dependent reduction in blasts by 35±25% at 0.1 ng/ml DART and 99.6±0.05% at 10 ng/ml DART (p=0.0063). In addition, AML colony-forming units (L-CFU) were also inhibited by 94 ±0.6% while CFU-GEMM, CFU-GM and BFU-E from cord blood and (G-CSF)-mobilized peripheral blood from normal donors were not affected by either CD3xCD123 DART or control DARTs (0.1-10 ng/ml), suggesting limited impact on normal human hematopoietic progenitors in vitro. Bioluminescence imaging of irradiated NSG mice (n= 5/group; 300cGy) performed on days 3, 12, 19, and 28 after the infusion of 1.5 x 10⁶ Click Beetle Red (CBR) luciferase⁺- transduced K562^CD123/GFP cells on day 0 and both CD3xCD123 DART (0.5 mg/kg IV) and human T cells (3 x 10⁶) on day 3 revealed no expansion of tumor cells in sharp contrast to NSG mice receiving either control DARTs and/or no human T cells (1415- fold expansion; p<0.0001). Of interest is that at 6 weeks post-infusion of primary AML xenografts into NSG mice (E:T=0.009:1), there was near- complete elimination (>97%) of AML blasts from the peripheral blood even in the absence of exogenously added human T cells. Clearing of primary human AML blasts from the spleen and bone marrow was also significant (40-77.8%) but less than that seen in the blood. These results provide the basis for the CD3xCD123 DART as a novel reagent for the treatment of patients with CD123⁺ AML.