High-Level EVI1 Perturbs Hematopoietic Function By Skewing Cells Towards Myeloid Phenotype Via Induction of PU.1/ Spi1 Â Transcription

Inv(3q26) and t(3:3)(q21;q26) are specific to poor prognosis acute myeloid leukemia (AML) and myelodysplastic syndrome, and result in marked overexpression of EVI1 isoforms. Despite extensive study, the mechanism by which this induces myeloid malignancy is unclear. We describe a mouse model that mimics the transcriptional effects of 3q26 rearrangement, and study of this has revealed a novel mechanism of leukemogenesis, specifically myeloid skewing with concomitant suppression of erythropoiesis and lymphopoiesis. In an effort to characterize the effects of high-level EVI1 expression on hematopoietic cells in vivo, we created a tetracycline (Tet)-inducible allele of Evi1 (termed Evi1^TO) in the mouse by inserting seven repeats of the Tet operon within the first exon of Evi1. The advantage of this design is that it allows for the induction of all spliced isoforms of the locus, of which at least three exist, those encoding 135, 123, and 103 kDa. To assess whether Evi1 can be induced in vivo, we tested induction in mice harboring Evi1^TO/TO / Rosa26^rtTA/rtTA alleles: treatment of these mice with DOX results in dramatic up-regulation of the major Evi1 transcript in hematopoietic cells to levels commonly seen in leukemic cells. This accurately models the pattern of Mecom gene expression most commonly seen in human AML with rearrangements at 3q26, in which setting the EVI1 isoforms are upregulated, but typically not the longer, MDS1-EVI1/PRDM3 isoform; and this particular expression pattern is associated with the poorest prognosis. To determine the direct effect of EVI1 overexpression on hematopoietic stem/progenitor cells (HSPCs) and to minimize secondary and non-cell autonomous effects, we enumerated HSPCs in Evi1^TO/TO / Rosa26^rtTA/rtTA mice following three days of induction administered by DOX chow. This revealed a doubling of LSK cells, due primarily to increases in HSCs and MPP3 cells, the latter representing myeloid-poised progenitors. In long-term competitive transplantation experiments, EVI1 overexpression resulted in marked suppression of both erythropoiesis and lymphopoiesis, due to both increases in apoptosis and decreases in proliferation. In the same setting, there was a marked expansion of myeloid cells, such that EVI1-overexpressing cells constituted the majority of myeloid cells in the marrow, despite the presence of normal competitor bone marrow. Our data indicate that when Evi1 is induced in vivo in an admixture with normal hematopoietic cells, there is expansion of the Evi1 -expressing cells indicating they have a proliferative and/or survival advantage, with specific predilection for myeloid differentiation; we found that EVI1 overexpression led to significant increases in both bone marrow and circulating granulocytes and their progenitors, and marked decreases in erythrocytes and T and B lymphocytes; monocytes and platelets were largely unaffected. To understand the mechanism, we employed EML cells, which are immortalized progenitor cells. EVI1 overexpression caused marked upregulation of Sca-1, a marker of myeloid skewing, with concomitant inhibition of erythropoiesis; these effects are dependent on EVI1 binding to DNA. This myeloid skewing correlated with a two-fold increase in PU.1 expression. ChIP-seq experiments revealed several binding sites for EVI1 upstream of PU.1, and knock-out of one of these sites in vivo led to blunting of PU.1 induction by EVI1, and abrogated EVI1-mediated upregulation of Sca-1; in vitro knockdown studies showed similar results. These data are consistent with a mechanism whereby EVI1 overexpression upregulates PU.1, and together these factors suppress erythropoiesis and promote myelopoiesis. However, by itself EVI1 overexpression does not result in leukemia. To see if leukemia would result if additional mutations were induced, we treated competitively transplanted mice with the mutagen, ethylnitrosourea (ENU). This induced myeloid leukemias in 100% of the mice, and all were derived from EVI1-overexpressing cells. These data suggest that EVI1 overexpression requires additional cooperating genetic events for induction of leukemia.