EVI1 is critical for the pathogenesis of a subset of MLL-AF9-rearranged AMLs

Abstract

The proto-oncogene EVI1 (ecotropic viral integration site-1), located on chromosome band 3q26, is aberrantly expressed in human acute myeloid leukemia (AML) with 3q26 rearrangements. In the current study, we showed, in a large AML cohort carrying 11q23 translocations, that ∼ 43% of all mixed lineage leukemia (MLL)-rearranged leukemias are EVI1(pos). High EVI1 expression occurs in AMLs expressing the MLL-AF6, -AF9, -AF10, -ENL, or -ELL fusion genes. In addition, we present evidence that EVI1(pos) MLL-rearranged AMLs differ molecularly, morphologically, and immunophenotypically from EVI1(neg) MLL-rearranged leukemias. In mouse bone marrow cells transduced with MLL-AF9, we show that MLL-AF9 fusion protein maintains Evi1 expression on transformation of Evi1(pos) HSCs. MLL-AF9 does not activate Evi1 expression in MLL-AF9-transformed granulocyte macrophage progenitors (GMPs) that were initially Evi1(neg). Moreover, shRNA-mediated knockdown of Evi1 in an Evi1(pos) MLL-AF9 mouse model inhibits leukemia growth both in vitro and in vivo, suggesting that Evi1 provides a growth-promoting signal. Using the Evi1(pos) MLL-AF9 mouse leukemia model, we demonstrate increased sensitivity to chemotherapeutic agents on reduction of Evi1 expression. We conclude that EVI1 is a critical player in tumor growth in a subset of MLL-rearranged AMLs.

Figure 1

Supervised gene expression profiling of MLL-rearranged AMLs uncovers 2 different subgroups. Pearson correlation clustering of 35 MLL-rearranged leukemias defined 2 subgroups, an EVI1^pos group and an EVI1^neg group (supplemental Figure 1). Supervised analysis revealed 22 probe sets that were differentially expressed between the 2 MLL-rearranged subgroups. Red color corresponds to high correlation, whereas blue color corresponds to low correlation of mRNA expression of genes in patient samples. Note that arrows (←) point to probe sets for EVI1 mRNA expression, which are part of the distinctive signature separating the 2 clusters.

Figure 2

EVI1 is frequently expressed in MLL-rearranged leukemias. (A) Relative EVI1 expression of patients with different 11q23 aberrations, corresponding to MLL-AF4, MLL-AF6, MLL-AF9, MLL-AF10, MLL-ENL, and other MLL-fusions. A cutoff value of 0.1 relative to the calibrator ovarian carcinoma cell line SKOV3 was chosen to classify EVI1^pos from EVI1^neg cases. Per MLL translocation, the percentage (%) of EVI1^pos patients is indicated. (B) EVI1 protein expression is shown in selected patient samples with MLL-rearranged leukemias. *For the MLL-AF6 sample, we observed EVI1 protein expression at longer exposure time. A 3q26/EVI1-rearranged AML and an EVI1-negative case served as, respectively, positive and negative control for EVI1 protein expression. Actin staining shows equal protein loading. MW marker indicates molecular weight marker; marking bands at, respectively 170, 130, 95 kDa on EVI1 Western and 55, 43 kDa on actin Western blot.

Figure 3

Aberrant expression of EVI1 mRNA in sorted CD34/CD38 fractions of MLL-AF6–rearranged AMLs. BM cells from (B) 2 MLL-AF6 rearranged AMLs and (A) normal BM were FACS sorted for the distinct CD34/CD38 populations, with subsequent isolation of mRNA. Relative expression was normalized against the reference gene PBGD. Each measurement was carried out in triplicate and SD is shown per measurement.

Figure 4

High Evi1 expression in MLL-AF9–transformed mononucleated normal mouse BM. (A) Transduction of normal mBM with MLL-AF9 leads to colony formation and sustained replating capacity compared with empty vector control (EV). Normal mBM transduced with E2A-PBX served as a positive control for transformation. Experiments were performed in triplicate and presented as average number of CFUs with SD. (B) Evi1 and Meis1 relative mRNA expression was determined from colonies at day 7, 14, or 21. An average of 3 measurements is depicted with SD. (C) Detection of Evi1 protein by Western blot analysis in lysates of pooled colonies from day 7 or 14. Blots were reprobed with an α-actin Ab to show equal protein loading.

Figure 5

Enrichment of CMPs in Evi1^pos clonal MLL-AF9–transformed BM cell cultures. (A) Example of FACS-sorting strategy to select for CMP, GMP, and MEP derived from cultured Evi1^pos and Evi1^neg MLL-AF9 clones. (B) Percentages of CMP, GMP, and MEP relative to the total number of progenitor cells are calculated for each clone. (C) Evi1 expression in LSK, CMP, GMP, and MEP subpopulations of normal mononucleated mBM cells. The MLL-AF9 cell line 4166 served as positive control for Evi1 expression. (D) Evi1 mRNA expression in lineage-positive subfractions of MLL-AF9 clones. The expression of Evi1 was calculated relative to the lineage-negative fraction of MLL-AF9 clone #1. Evi1 relative expression was normalized using Hprt as a reference gene. For the last 2 panels, the average of 3 experiments with its SD is shown.

Figure 6

Enrichment of H3K79me2 on the Evi1 promoter of MLL-AF9–transformed cells. ChIP using (A) H3K79me2 and (B) H3K27me3 Abs showing enrichment of the H3K79me2 mark on the Evi1 promoter of Evi1^pos vs Evi1^neg clones. Conversely, the H3K27me3 mark is enriched on the Evi1 promoter of Evi1^neg clones compared with Evi1^pos MLL-AF9 clones. ChIP using H3K4me3, H3K79me2, and H3K27me3 Abs performed on 3 EVI1^pos and 3 EVI1^neg MLL-AF9 rearranged leukemias showed higher H3K79me2 marks present on the EVI1 promoter of EVI1^pos MLL-AF9–rearranged human AMLs compared with EVI1neg cases. Q-PCRs were performed to monitor relative enrichment of each histone mark on the promoters of either (C) EVI1, (D) HoxA9 promoter, or (E) β-actin. Experiments show average relative enrichment with SD of either (A-B) 3 biologic replicates or (C-E) 2 independent ChIP pulldowns with triplicate Q-PCR measurements performed on the same cross-linked patient material.

Figure 7

Evi1 knockdown (KD) in MLL-AF9 cell line 4166 leads to reduced growth in vitro and in vivo. Transduction of 4166 with lentiviral vectors containing shRNA against Evi1 leads to reduced (A) Evi1 mRNA and (D top) protein expression. (B) Reduced Evi1 levels were accompanied with a significant reduction of colony formation. Fold change of 4 experiments (mean + SD) is shown for panels A and B. (C) Evi1 shRNA-mediated inhibition of cell growth was attributed to (bottom panel) an induction of apoptosis, (top panel) with no changes in cell-cycle profile. (D) Western blotting shows an increase of (pro)apoptotic markers after knockdown of Evi1. HSP90 and actin were used to show equal cell viability and loading of protein lysates. SHC indicates short hairpin control. (E) Survival curve of lethally irradiated mice transplanted with 1 × 10⁵ 4166 cells, transduced with either Evi1 shRNA or control virus supplemented with 5 × 10⁵ normal mBM cells. As a control group, untreated 4166 cells were transplanted to monitor offset of normal 4166-mediated MLL-AF9 tumorigenesis. All 3 groups contained 9 mice each. By posttransplantation day 170, the majority of the mice in the untreated (8 of 9) or control virus (8 of 9) transduced group had succumbed to leukemia, while all mice receiving Evi1 shRNA-transduced cells were still alive. Leukemia development was significantly delayed in animals receiving Evi1 shRNA-transduced 4166 cells (log-rank P < .001) with 44% of the mice being long-term survivors.