Lethal myelofibrosis induced by Bmi1-deficient hematopoietic cells unveils a tumor suppressor function of the polycomb group genes

Abstract

Polycomb-group (PcG) proteins form the multiprotein polycomb repressive complexes (PRC) 1 and 2, and function as transcriptional repressors through histone modifications. They maintain the proliferative capacity of hematopoietic stem and progenitor cells by repressing the transcription of tumor suppressor genes, namely Ink4a and Arf, and thus have been characterized as oncogenes. However, the identification of inactivating mutations in the PcG gene, EZH2, unveiled a tumor suppressor function in myeloid malignancies, including primary myelofibrosis (PMF). Here, we show that loss of another PcG gene, Bmi1, causes pathological hematopoiesis similar to PMF. In a mouse model, loss of Bmi1 in Ink4a-Arf(-/-) hematopoietic cells induced abnormal megakaryocytopoiesis accompanied by marked extramedullary hematopoiesis, which eventually resulted in lethal myelofibrosis. Absence of Bmi1 caused derepression of a cohort of genes, including Hmga2, which is an oncogene overexpressed in PMF. Chromatin immunoprecipitation assays revealed that Bmi1 directly represses the transcription of Hmga2. Overexpression of Hmga2 in hematopoietic stem cells induced a myeloproliferative state with enhanced megakaryocytopoiesis in mice, implicating Hmga2 in the development of pathological hematopoiesis in the absence of Bmi1. Our findings provide the first genetic evidence of a tumor suppressor function of Bmi1 and uncover the role of PcG proteins in restricting growth by silencing oncogenes.

Figure 1.

Loss of Bmi1 augments the repopulating capacity of BM cells in the absence of Ink4a/Arf. (A) To perform a competitive repopulating assay, 10⁶ pooled test BM cells from 4-wk-old mice (CD45.2⁺) of the indicated genotype were mixed with 10⁶ competitor BM cells (CD45.1⁺) and injected into lethally irradiated recipient mice (CD45.1⁺). The percent chimerism of donor cells in the recipient PB is presented as the mean ± SD (n = 5). *, P < 0.05; **, P < 0.01; ***, P < 0.001. (B) Relative numbers of donor-derived BM and splenic LSK cells, and BM CMPs, GMPs, and MEPs. Lethally irradiated wild-type recipient mice were infused with 2 × 10⁶ BM cells of the indicated genotype and analyzed at 4 mo after transplantation. Data were normalized relative to wild type and are shown as the mean ± SD (BM LSK cells, n = 12; splenic LSK cells, n = 6; CMPs, GMPs, and MEPs, n = 5; *, P < 0.05; **; P < 0.01; ***, P < 0.001). (C) Survival curve of the wild-type recipient mice repopulated by BM cells from indicated mutant mice. The data from four independent experiments were combined (wild type, n = 12; Ink4a-Arf^−/−, n = 11; Bmi1^−/−Ink4a-Arf^−/−, n = 14). The significance of the difference in survival curves was calculated by log-rank test. *, P = 0.0007. (D) PB analysis of white blood cells (WBC), platelets (PLT), and hemoglobin (HGB) of the wild-type recipient mice in C. Data are shown as the mean ± SD. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 2.

Enhanced megakaryocytopoiesis and massive myelofibrosis induced by Bmi1^−/−Ink4a-Arf^−/− hematopoietic cells. (A) Hematoxylin and eosin (H&E) staining of BM, spleen, and liver sections. BM, spleen, and liver of representative wild-type recipient mice repopulated by 2 × 10⁶ BM cells of the indicated genotype were analyzed at 174 d after transplantation. Bars: 125 µm (BM); 500 µm (spleen and liver). (B) Silver staining of BM and spleen sections in A. Bars, 50 µm. (C) H&E staining of BM sections of representative recipient mice repopulated with the indicated mutant BM cells analyzed at 138 d after transplantation. Bars, 50 µm. (D) CD41 (green) and DAPI (blue) staining of spleen sections of representative recipient mice repopulated with indicated mutant BM cells analyzed at 138 d after transplantation. Bars, 125 µm.

Figure 3.

Derepression of Hmga2 in Bmi1^−/−Ink4a-Arf^−/− hematopoietic cells. (A) List of top 10 genes up-regulated in LSK cells and CMPs in the absence of Bmi1. IL-7Rα LSK cells purified from BM of 4-wk-old Ink4a-Arf^−/− (DKO) and Bmi1^−/−Ink4a-Arf^−/− (TKO) mice and CMPs from recipients’ BM at 4 mo after infusion of DKO and TKO BM cells were subjected to microarray analyses and their profiles were compared. Highlighted genes were further characterized in this study. (B) Quantitative RT-PCR analysis of Hmga2 expression. mRNA levels in each progenitor fraction from the recipients’ BM repopulated by BM cells of the indicated genotype at 4 mo after transplantation were normalized to Hprt1 expression. Expression levels relative to those in the wild-type Flt3⁻ LSK cells are shown as the mean ± SD for triplicate analyses. The fold change in expression levels between Bmi1^−/−Ink4a-Arf^−/− cells and Ink4a-Arf^−/− cells is also indicated. *, P < 0.05; **, P < 0.01. (C) Derepression of Hmga2 in Bmi1-deficient hematopoietic cells. Lineage⁻c-Kit⁺ cells were purified from BM of 4-wk-old wild-type and Bmi1^−/− mice. mRNA levels of Hmga2, Ink4a, and Arf were normalized to Hprt1 expression. Expression levels relative to those in the Bmi1^−/− cells are shown as the mean ± SD for triplicate analyses. N.D. indicates not detected. **, P < 0.01. (D) ChIP analysis at the Hmga2 promoter in wild-type Lineage⁻c-Kit⁺ cells. The Hmga2 locus indicating its genomic structure (based on the Ensemble data, transcript ID ENSMUST00000072777) is depicted in the top panel. Exons are demarcated by black boxes. The regions 1–6 amplified from the precipitated DNA by site-specific quantitative PCR are indicated. The binding of Bmi1 and the levels of H2Aub were determined by ChIP using control mouse IgG (mIgG), anti-Bmi1, and anti-H2Aub antibodies and site-specific real-time PCR. The relative amount of immunoprecipitated DNA is presented as a percentage of input DNA (bottom). The data are shown as the mean ± SD for four independent experiments. The β-actin promoter (Actb) served as a negative control. *, P < 0.05. (E) ChIP analysis at the Hmga2 promoter in wild-type or Bmi1^−/−Ink4a-Arf^−/− Lineage⁻ cells. The binding of Bmi1 and the levels of H2Aub were determined by ChIP using anti-Bmi1 or anti-H2Aub antibodies and site-specific real-time PCR as in D. The data are shown as the mean ± SD for triplicate PCRs from two independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

Hmga2 promotes expansion of progenitor cells and enhances megakaryocytopoiesis in vitro. (A) Growth of CD34⁻LSK HSCs transduced with the Hmga2 retrovirus. 100 CD34⁻LSK HSCs were transduced with either the empty vector or Hmga2 retrovirus and cultured in the presence of SCF, TPO, IL-3, FP6, and EPO for 14 d. Numbers of cells are shown as the mean ± SD for triplicate cultures. **, P < 0.01. (B) Number of LSK cells in culture. Flow cytometric analysis of the culture in A was performed at day 10 of culture. Absolute numbers of LSK cells are indicated as the mean ± SD (n = 3). **, P < 0.01. (C) Morphology of CD34⁻LSK HSCs at day 12 of culture. CD34⁻LSK HSCs transduced with the indicated retroviruses in A were recovered, cytospun onto slide glasses, and subjected to May-Grünwald Giemsa staining. The morphology of the cells was observed under a light microscope. Bars, 100 µm. (D) The frequency of multinucleated megakaryocytes (MgK) and mononuclear micromegakaryocytes (microMgK) in culture. The cytospun cells on slide glasses in C were examined under a light microscope and absolute numbers in culture are indicated as the mean ± SD. Counts of 500 cells were independently performed three times. **, P < 0.01. (E) Effect of Hmga2 on CFC numbers in culture. 100 CD34⁻LSK HSCs were transduced with the indicated retrovirus and cultured in the presence of SCF and TPO. At day 10 of culture, colony assays were performed to evaluate the number of CFCs in culture. Absolute numbers of low proliferative potential (LPP; diameter <1 mm) and HPP (diameter ≥1 mm) CFCs are shown as the mean ± SD for triplicate cultures. *, P < 0.05; **, P < 0.01.

Figure 5.

Forced expression of Hmga2 causes myeloproliferative hematopoiesis with enhanced megakaryocytopoiesis. (A) Donor chimerism and lineage contribution in PB (n = 5). CD34⁻LSK HSCs (CD45.2) transduced with either control or Hmga2 retrovirus were transplanted into lethally irradiated CD45.1 mice together with 2 × 10⁵ BM competitor cells from 8-wk-old CD45.1 mice. The chimerism of CD45.2⁺GFP⁺-transduced cells in PB of recipient mice was examined at 6 mo after transplantation (left). Lineage contribution of CD45.2⁺GFP⁺ donor cells to myeloid (Gr-1⁺ and/or Mac-1⁺), B (B220⁺), or T (CD4⁺ or CD8⁺) cells (right). Data are presented as the mean ± SD (n = 6). (B) Absolute numbers of BM mononuclear cells (from bilateral femurs and tibiae) and spleen weights at 6 mo after transplantation. Data are presented as the mean ± SD (n = 6). *, P < 0.05. (C) Frequency of LSK cells and absolute numbers of myeloid progenitor cells in BM. The frequency of LSK cells in PB, BM, and spleen (left) and absolute numbers of myeloid progenitor cells in BM (right) of recipient mice examined 6 mo after transplantation. The relative number of LSK cells is presented in the middle panel. Data are presented as the mean ± SD (n = 6). *, P < 0.05; **, P < 0.01; ***, P < 0.001. (D) PB cell counts of recipients repopulated with CD34⁻LSK HSCs transduced with the indicated retrovirus. Cell counts at 6 mo after transplantation are presented for white blood cells (WBC) and red blood cells (RBC). The time course of the increase in platelet (PLT) counts is also depicted. Data are presented as the mean ± SD (n = 6). *, P < 0.05.