Loss of Ezh2 Promotes the Development of Mutant-RUNX1 Induced MDS

Abstract 402

Polycomb group proteins are transcriptional repressors that epigenetically regulate transcription via histone modifications. There are two major polycomb-complexes, the Polycomb Repressive Complexes 1 and 2 (PRC1, PRC2). PRC2 contains SUZ12, EED, and EZH1/EZH2, and catalyzes the trimethylation of histone H3 at lysine 27 (H3K27me3), silencing target-genes. We have shown that the self-renewal of Ezh2-deficient HSCs is not compromised and H3K27me3 marks are not completely depleted in the absence of Ezh2, possibly as a result of Ezh1 complementation. EZH2 is generally thought to act as an oncogene in lymphoma and solid tumors by silencing tumor suppressor genes. Recently however, loss-of-function mutations of EZH2 have been found in myeloid malignancies such as AML, MDS and MPN, suggesting that EZH2 also functions as a tumor suppressor, although it remains unclear how EZH2 prevents the transformation of myeloid malignancies. RUNX1 is a critical transcription factor in the regulation of the self-renewal and differentiation of HSCs. RUNX1 mutations are frequently found in MDS, AML following MDS (MDS/AML) and de novo AML patients. One of the most frequent mutations, RUNX1S291fs, lacks the transactivation domain in C-terminus, but retains the RUNT DNA biding domain, resulting in a dominant negative phenotype. RUNX1S291fs-transduced bone marrow cells have been shown to generate MDS/AML in vivo. Given that RUNX1 and EZH2 mutations coexist in MDS and AML patients as reported recently, we generated a novel mouse model of MDS utilizing RUNX1S291fs retrovirus and Ezh2 conditional knockout mice in order to understand how EZH2 loss contributes to the pathogenesis of MDS upon genetic mutation of RUNX1.

We first harvested CD34-Lin-Sca1+c-Kit+(LSK) HSCs from tamoxifen-inducible Cre-ERT;Ezh2^wild/wild (EW) and Cre-ERT;Ezh2^flox/flox (EF) mice (CD45.2) and transduced these cells with RUNX1S291fs retrovirus or an empty vector, which contains IRES-GFP. Then, we transplanted RUNX1S291fs-transduced Cre-ERT;Ezh2^wild/wild (S291EW) or Cre-ERT;Ezh2^flox/flox (S291EF) HSCs into lethally irradiated recipient mice (CD45.1) together with life saving dose 1×10⁵ CD45.1 bone marrow cells. At 6 weeks post transplantation, we deleted Ezh2 via administration of tamoxifen, and observed disease progression until 12 months post transplantation.

The empty vector transduced control mice with or without Ezh2 (EW and EF) did not develop myeloid malignancies. Two out of 16 S291EW mice died due to MDS progression, while 12 out of 16 and 1 out of 17 S291EF mice developed MDS and MDS/AML, respectively. S291EF mice showed significantly shorter median survival than S291EW mice (314 days versus undefined, p=0.037). In the peripheral blood, we observed significantly lower CD45.2+GFP+ chimerism in S291EF mice; however S291EF mice eventually showed macrocytic anemia and variable white blood cell counts accompanied with dysplastic features of MDS. Despite low CD45.2+GFP+ chimerism in peripheral blood, S291EF mice showed a higher chimerism of CD45.2+GFP+ cells in the bone marrow and had a significantly increased number of LSK and CD34-LSK cells compared to EW, EF, and S291EW mice, indicating that Ezh2 loss promoted HSCs/progenitors expansion, but impaired myeloid differentiation in the presence of RUNX1S291fs. We also saw enhanced apoptosis of CD71+Ter119+ erythroblasts in S291EF MDS mice, which may account for the anemia we observed. Since S291EF MDS bone marrow cells were transplantable in secondary experiments, we performed limiting-dilution assays to evaluate the frequency of MDS initiating cells and found that the frequency of MDS initiating cells was much higher in S291EF pre-MDS Lin-Mac1-Kit+ cells compared to S291EW pre-MDS Lin-Mac1-Kit+ cells. To understand this molecular mechanism, we performed gene expression analysis during MDS progression. S291EF MDS LSKs showed 1979 and 1875 dysregulated (>5-fold) genes, compared to EW LSK and S291EF pre-MDS LSK, respectively. We are now working to understand how these dysregulated genes are involved in the development of RUNX1S291fs-induced MDS after deletion of Ezh2. In summary, we have successfully recapitulated the clinical feature of MDS in mice reconstituted with Ezh2 null HSCs expressing a RUNX1 mutant, and demonstrated that Ezh2 functions as a tumor suppressor in this context.