Bmi-1 Is Dispensable for the Development of Acute Myeloid Leukemia Mediated by MLL-AF9

Abstract 63

The Polycomb group 1 gene Bmi1 has been shown to be required for several normal adult stem cell types, including hematopoietic stem cells. High expression of Bmi1 is correlated with adverse prognosis in human AML and MDS. Bmi1 null murine fetal liver cells transformed with HoxA9 and Meis1 give rise to a primary leukemia of expanding blasts that fails to expand in secondary recipient mice thus demonstrating a defect in leukemia self-renewal (Nature. 2003; 423:255). More recently, Bmi1 has also been shown to bind PML-RARA, and to be required for in vitro replating of PML-RARA immortalized murine colony forming cells (cfc).

Our laboratory recently demonstrated that MLL-AF9, a recurrent translocation commonly found in human acute leukemia, more efficiently transforms murine granulocyte macrophage progenitors (GMP) compared to the combination of HoxA9 and Meis1. We decided to test if MLL-AF9 is able to bypass the requirement for Bmi1 previously reported in other murine leukemia models. Lineage marker negative (Lin-) bone marrow cells from WT and Bmi1 −/− mice were transduced with an MSCV-based ecotropic retroviral vector expressing MLL-AF9 and linked via an internal ribosomal entry site (IRES) the selectable marker, GFP. Cells were harvested and injected into 5 sublethally (200 cGy) irradiated NOD/SCID mice per group. All mice developed AML based on GFP expression, spleen size, forward/side scatter profile and surface marker staining (Mac1 positive/Gr1 positive, B220 negative, CD3 negative). Median latency in the WT group was 54 days, whereas mice receiving a graft consisting of Bmi1 −/− cells transduced with MLL-AF9 developed leukemia with a median latency of 89 days. The experiment was repeated with very similar results with median latencies of 50 days (WT group) and 70 days (Bmi1 −/−). To confirm sustained self-renewal, we then proceeded to secondary transplantation experiments. Secondary leukemia developed with similar latency in recipients of primary WT (31.5d) and Bmi1^−/− (32d) leukemic bone marrow cells. The experiment was repeated with bone marrow from mice with primary leukemias from the second cohort and again similar latencies were observed (42.5 days in the WT group vs 29.5 days in the Bmi1 −/− group.). Finally we proceeded to tertiary transplantation experiments. All recipients of secondary AML cells developed AML with median latencies of 17.5 d (WT) and 21 d (KO).

To compare our results with a system closer to the original studies of leukemogenesis in a Bmi1 −/− background, we chose to transduce lin- bone marrow from WT and Bmi1 −/− with retroviral vectors for HoxA9-IRES-GFP and Meis1-pgk-puro. Both, WT and HoxA9/Meis1 transduced cells replated in in vitro methylcellulose cultures for at least 4 rounds. In the first of 2 in vivo experiments, 5/5 recipients of wild type cells developed acute myeloid leukemia with a median latency of 88 days. In contrast, none of the five recipients of HoxA9/Meis1a transduced Bmi1−/− cells developed leukemia, and these mice were sacrificed 6 months after transplantation. In a second independent experiment, 5/5 recipients of WT HoxA9/Meis1 transduced cells developed AML with a median latency of 62 days. One recipient of Bmi1 −/− cells transduced with HoxA9/Meis1 developed leukemia. One mouse died from a spontaneous lymphoma. The remaining 3 mice in this cohort did not develop leukemia. These findings are consistent with the previously published data and suggest that MLL-AF9 is more efficient in generating AML in Bmi1−/− bone marrow compared to the combination of HoxA9 and Meis1.

To further characterize the Bmi1 −/− leukemias, we analyzed expression of the primary Bmi1 target p16 by Western blot. In primary, secondary and tertiary AMLs from Bmi1 −/− mice, derepression of p16 could be demonstrated. We analyzed the cell cycle distribution of cultured tertiary leukemia cells. WT cells showed a S phase proportion of 44–45 percent (n=2), whereas the Bmi1 −/− cells demonstrated a S-phase proportion of 30–38% (n=2). These results demonstrate modest effects of p16 derepression on cell cycle without catastrophic failure to divide. This is mirrored by the in vivo development of secondary and tertiary leukemias in the presence of elevated p16 levels. Preliminary data suggest a similar but less pronounced derepression of p19. Future experiments will focus on the mechanisms that allow MLL-AF9 transduced cells to grow in the presence of derepressed p16 and p19.