• Open Access

Ecotropic viral integration site 1, stem cell self-renewal and leukemogenesis



It has become evident that acute myeloid leukemia (AML) is organized as a cellular hierarchy initiated and maintained by a subset of self-renewing leukemia stem cells. Recent gene expression profile analysis of human leukemia stem cells and hematopoietic stem cell (HSC) populations identified a key transcriptional program shared by leukemia stem cells and HSC, which is associated with adverse outcomes in AML patients. One molecule that has been established as a pivotal regulator in fine-tuning of stem cell properties as well as a potent oncogenic determinant is ecotropic viral integration site 1 (EVI1). EVI1 is a transcription factor that has stem cell-specific expression pattern and is essential for the regulation of HSC self-renewal. This gene is notorious for its involvement in AML, as its activation confers extremely poor prognosis in patients with AML. Molecular analysis has identified a variety of gene products that are involved in HSC regulation as downstream targets or interacting proteins of EVI1. Thus, exploration of the molecular pathogenesis underlying EVI1-related leukemogenesis provides insight into how shared stemness transcriptional programs contribute to leukemia progression and therapeutic resistance in AML. Here, we review the current knowledge regarding the role of EVI1 in HSC self-renewal and leukemogenesis and highlight the relationship between stem cell self-renewal properties and adverse outcome in myeloid malignancies. (Cancer Sci, doi: 10.1111/j.1349-7006.2012.02303.x, 2012)

Despite advances in therapeutic modalities and supportive care, long-term outcomes of patients with acute myeloid leukemia (AML) remain dismal.[1] Clinical investigation has focused on the identification of prognostic subgroups to stratify patients into risk-adapted treatment groups. Such clinical studies have discovered that the molecular genetic aberrations, including mutations in the FLT3, NPM1 and IDH1/2 genes, or increased expression of the EVI1, BAALC and MN1 genes, are predictive of clinical outcomes in AML, especially cytogenetically normal AML (CN-AML).[2] AML is organized as a cellular hierarchy initiated and maintained by a subset of self-renewing leukemia stem cells (LSC).[3] Gene expression profile analysis of human LSC and hematopoietic stem cell (HSC) populations has identified a core transcriptional program shared by LSC and HSC, revealing the molecular machinery underlying “stemness” properties.[4, 5] These stem cell-specific gene signatures are highly significant predictors of patient survival. Using murine genetic models, Kataoka et al.[6] establish a specific relationship between ecotropic viral integration site 1 (Evi1) and HSC activity; therefore, EVI1 can be regarded as one of the key molecules in the stem cell machinery that are associated with poor prognosis in AML. This review will summarize the role of EVI1 in HSC self-renewal and leukemogenesis, with specific focus on the relationship between stem cell self-renewal properties and adverse outcome in myeloid malignancies.

Ecotropic Viral Integration Site 1 as a Pivotal Regulator of Hematopoietic Stem Cell Self-Renewal

Accumulating evidence suggests an essential role of Evi1 in HSC function. Evi1-deficient embryos exhibit a marked reduction in hematopoietic stem/progenitor cells (HSPC) in the para-aortic splanchnopleura (P-Sp) region, while losing their long-tem repopulating capacity in vivo.[7] In addition, HSC in Evi1-deficient fetal livers are severely reduced in number with defective multilineage reconstitution ability.[8] Conditional deletion of Evi1 in adult mice causes a profound loss of HSC self-renewal activity, but does not affect blood cell lineage commitment.[8] These findings suggest that Evi1 is indispensable for HSC self-renewal in the fetal and adult hematopoietic system. In a detailed analysis of Evi1 expression pattern using Evi1-green fluorescent protein reporter mice, Kataoka et al.[6] demonstrate that Evi1 expression marks HSC possessing long-term repopulating activity, suggesting a specific relationship between Evi1 expression and HSC function throughout ontogeny. Consistent with this stem cell-specific expression pattern of Evi1, Evi1 heterogeneity causes a marked reduction of long-term HSC (LT-HSC) with a specific defect of self-renewal capacity, indicating that Evi1 has a dominating effect on the regulation of LT-HSC activity.

Because, in the above described Evi1 knockout models, the targeted disruption results in transcriptional loss of both Evi1 and Mds1-Evi1, a fusion variant generated through intergenic splicing with Mds1,[8, 9] relative contributions of these gene products to HSC homeostasis remain unexplored. A knockout study by Zhang et al.[10] implies that Mds1-Evi1 deletion alone results in a significant reduction in HSC numbers with impairment of long-term reconstitution activity due to loss of quiescence. In contrast, it is reported that retroviral transfer of Evi1, but not Mds1-Evi1, is required to rescue LT-HSC defects caused by heterozygosity of both Evi1 and Mds1-Evi1, suggesting a distinct role of Evi1 in HSC self-renewal.[6] Therefore, although there has been controversy regarding functional differences between Evi1 and Mds1-Evi1, these proteins may cooperate in regulating HSC function and execute their roles through different mechanisms.

Ecotropic Viral Integration Site 1 Activation Confers Poor Prognosis in Myeloid Neoplasms

The Evi1 locus was initially identified as a common target of retroviral integration site in murine myeloid leukemias.[11] Aberrant expression of EVI1 has been implicated in the development of various human myeloid malignancies, including AML, chronic myeloid leukemia (CML) and myelodysplastic syndrome (MDS).[12] High EVI1 expression occurs in approximately 5–10% of patients with de novo AML.[13-15] As EVI1 gene maps to human chromosome 3q26, its rearrangements, such as inv(3)(q21q26.2) or t(3;3)(q21;q26.2) (inv[3]/t[3;3]), cause aberrant EVI1 expression in myeloid neoplasms. In the recent World Health Organization classification, AML with inv(3)/t(3;3) is incorporated as a new entity (Table 1), which is associated with elevated white blood cell and platelet counts, and marked hyperplasia with dysplastic megakaryopoiesis; this confers adverse prognostic value.[16, 17] Deregulated EVI1 expression is also observed in a subset of AML without 3q26 rearrangements, and is frequently associated with monosomy 7 and 11q23 translocations involving mixed lineage leukemia (MLL).[13-15] Consistent with these clinical observations, in vivo studies with MLL-AF9 knock-in mice also show Evi1 overexpression after leukemic transformation.[18] As the underlying mechanism of EVI1 activation in MLL-rearranged leukemias, analysis of the Evi1 promoter region by Arai et al.[19] reveals that MLL fusion proteins directly activate Evi1 transcription.

Table 1. Acute myeloid leukemia (AML) with inv(3)(q21q26.2) or t(3;3)(q21;q26.2); RPN1-EVI1
Epidemiology1–2% of all AML
Clinical featuresHigh white blood cell count, anemia and normal or occasionally elevated platelet count
May present de novo or have a prior myelodysplastic syndrome
Morphology(Peripheral blood)
Hypogranular neutrophils with a pseudo-Pelger–Huët anomaly
Giant and hypogranular platelets with bare megakaryocyte nuclei
(Bone marrow)
Multilineage dysplasia
Increased atypical megakaryocytes (small monolobed or bilobed forms)
Variable fibrosis
ImmunophenotypeCD13(+), CD33(+), HLA-DR(+), CD34(+) and CD38(+)
CD7(±), CD41(±) and CD61(±)
GeneticsAssociated with monosomy 7, 5q deletions and complex karyotypes
Exhibit dissociate EVI1 versus MDS1/EVI1 overexpression
Associated with N-RAS mutations
Postulated normal counterpartHematopoietic stem cell with multilineage potential
PrognosisAggressive disease with short survival

Elevated EVI1 expression predicts poor treatment response and adverse outcome in patients with AML, irrespective of the presence of typical 3q26 rearrangements.[13-15] This adverse prognostic impact is more pronounced in intermediate cytogenetic risk AML, and especially CN-AML.[15] In addition, EVI1 activation has been reported in CML in blast crisis, and less commonly in chronic phase (CP). Detection of EVI expression in CML-CP patients who have failed imatinib is a strong predictor for disease progression after therapy with second-generation tyrosine kinase inhibitors.[20] Five splice variants of EVI1 have been reported, that is, EVI1-1A, -1B, -1C, -1D and -3L, as well as the MDS1-EVI1 intergenic splice variant, differing mainly in their 5′ untranslated region,[21] and these different EVI1 splice forms should be taken into account in EVI1 screening. In particular, recent studies have reported a disproportionate pattern of elevated EVI1 and non-elevated MDS1/EVI1 expression in AML patients with inv(3)/t(3;3), which is consistent with the observation that MDS1/EVI1 expression is abrogated, resulting from a breakpoint between MDS1 and EVI1.[13, 14, 16] In addition, cryptic 3q26 aberrations may be identified by FISH in patients without any cytogenetically detectable 3q26 abnormalities who have disproportionate EVI1 and MDS1/EVI1 expression levels.[13, 16] This dissociated EVI1 and MDS1/EVI1 expression is associated with an extremely poor prognosis, which is distinguishable from the general AML patients with high EVI1 expression,[13] suggesting that the EVI1 gene product is a potent oncogenic determinant in AML.

Ecotropic Viral Integration Site 1 Expression Enhances Stem Cell Self-Renewal Capacity, Leading to Myeloid Leukemogenesis

How EVI1 activation contributes to leukemogenesis and the development of therapeutic resistance remains speculative. In a murine bone marrow transplant model, constitutive expression of Evi1 in bone marrow cells induces myelodysplasia or leukemia after long latency.[22, 23] Conversely, disruption of Evi1 in various transformed leukemic cells leads to a significant loss of their proliferative activity both in vitro and in vivo.[8] These findings implicate EVI1 as a major contributing factor in leukemogenesis. In contrast to the physiological states, aberrant expression of EVI1 has been reported to affect hematopoietic differentiation toward various lineages, such as interfering with granulocytic and erythroid differentiation, or promoting megakaryocytic commitment.[12, 24] However, as there have been conflicting reports on this point, no definitive conclusion regarding the role of EVI1 in differentiation into individual hematopoietic lineages has so far been reached.[24] Conversely, experimental and clinical studies have provided consistent evidence that EVI1 activation confers enhanced self-renewal capacity to HSC, potentially leading to leukemia development. In agreement with knockout studies, ectopic expression of EVI1 in bone marrow progenitors enhances their replating capacity in methylcellulose medium.[25] Furthermore, Evi1 overexpression blocks differentiation and induces HSPC expansion in liquid culture.[6] More information on the role of EVI1 in HSC self-renewal and leukemogenesis has been gained in retroviral insertion studies. MDS1-EVI1 genomic locus is highly overrepresented as a retrovirus insertion site in vitro and in vivo. In nonhuman primates and mice receiving HSPC transduced with retrovirus vectors, recurrent integration into the MDS1-EVI1 locus has been observed, which is related to enhanced self-renewal potential of HSPC, leading to in vivo clonal dominance of nonmalignant cells.[26, 27] Notably, the insertional activation of Evi1 can trigger immortalization of bone marrow progenitors in vitro[28] or result in leukemic transformation in vivo.[29] Retroviral insertional mutagenesis in mice has identified Evi1 as a cooperative gene for other oncogenic events, such as Hoxa9/Meis1 coactivation,[30] and Runx1[31] or Nras mutation[32] in myeloid leukemogenesis. In humans, a clinical trial on gene therapy for X-linked chronic granulomatous disease reports that activating insertions in MDS1-EVI1 influences regulation of long-term hematopoiesis by expanding gene-corrected myelopoiesis.[33] A follow-up investigation revealed that, in some patients, the specific clone with retroviral insertion in EVI1 dominated, eventually resulting in clonal progression toward MDS/AML with monosomy 7 after the development of genomic instability.[34] Taken together, the studies discussed above underscore the significance of EVI1 in the self-renewal and competitive fitness of HSC, whereas EVI1-mediated leukemogenesis is likely to require additional genetic events, considering its relatively long latency.

Consistent with the specific relationship between EVI1 and HSC activity in normal hematopoiesis, a gene expression analysis of human AML cases revealed that an AML subset with EVI1 activation has overall expression patterns closest to normal CD34+ HSPC, indicating a stem cell phenotype for EVI1-expressing leukemias.[35] Eppert et al. and Gentles et al.[4, 5] report that the stemness transcriptional profile shared by LSC and HSC confers poor prognosis in human AML. The above observations suggest that EVI1 is a critical determinant of stem cell properties whose activation has adverse prognostic significance in AML. Although it has been postulated that LSC can originate from normal HSC or more committed progenitors,[3] the cellular origin may influence aggressiveness and therapeutic resistance in AML. In murine leukemia models, MLL fusion proteins can more efficiently transform HSC than committed myeloid progenitors.[36] In addition, Evi1 upregulation by MLL fusion proteins occurs exclusively in HSC, but not in myeloid progenitors, suggesting that MLL oncoprotein-mediated Evi1 activation requires the properties of normal HSC.[19] Indeed, in human cases, approximately half of patients with 11q23 translocations show aberrant EVI1 expression.[15] As AML with high EVI1 expression exhibit HSC-like gene expression patterns among the cases with MLL-rearranged leukemias,[19] EVI1 upregulation may occur only in AML in which MLL-fusion proteins target HSC. Because EVI1-positive AML are associated with adverse outcomes in these patients, it is tempting to speculate that certain oncoproteins can augment stem cell self-renewal capacity, such as EVI1, when targeting HSC, which leads to leukemia progression and therapeutic resistance (Fig. 1). Therefore, exploration of the molecular pathogenesis underlying EVI1-related leukemogenesis provides a clue that should lead to the understanding of how shared stemness transcriptional programs contribute to therapeutic resistance in human AML.

Figure 1.

Proposed model depicting how the initial genetic events influence aggressiveness and therapeutic resistance in acute myeloid leukemia (AML). Some leukemic oncoproteins (such as BCR-ABL and AML1-ETO) can initiate AML in hematopoietic stem cell (HSC), but not committed progenitors, whereas a set of potent fusion proteins (such as mixed lineage leukemia [MLL] fusion proteins) can transform not only HSC, but also myeloid progenitors. The former oncoproteins may require the properties of normal HSC and maintain their self-renewal capacity in leukemic transformation. The latter can confer self-renewal potential to committed progenitors, while they further activate inherent self-renewal machineries when targeting HSC, which leads to aggressiveness and therapeutic resistance.

Molecular Machinery Governed by Ecotropic Viral Integration Site 1 in Hematopoietic Stem Cells

In line with the observations that implicate EVI1 as a core molecule of stem cell regulation, a gene expression analysis demonstrates that Evi1-binding sites are enriched in the upstream region of genes that are preferentially expressed in LT-HSC.[37] Furthermore, a variety of molecules that are involved in regulating HSC self-renewal and leukemogenesis have been identified as direct downstream targets of EVI1, including GATA2,[7, 38] PBX1[39] and PTEN.[23] GATA2 is essential for the maintenance and proliferation of HSC,[40] and its mutations have been reported in familial syndromes characterized by predisposition to MDS and AML.[41] In HSC of Evi1-deficient embryos, Gata2 expression is profoundly reduced, and restoration of Gata2 expression in Evi1-deficient HSC can prevent the failure of in vitro maintenance and proliferation of HSC.[7, 38] An analysis of the Gata2 promoter region reveals that Evi1 directly binds to Gata2 promoter as an enhancer.[7] PBX1 is a TALE class homeodomain transcription factor that regulates HSC self-renewal by maintaining their quiescence,[42] and PBX TALE proteins are required for many HOX-dependent developmental and oncogenic programs, such as MLL transformation.[43, 44] Promoter region analysis reveals that Evi1 directly upregulates Pbx1 transcription, which is important for Evi1-induced leukemic transformation.[39] The PTEN tumor suppressor is commonly deleted or otherwise inactivated in diverse cancers, including hematopoietic malignancies. Whereas Pten deletion generates transplantable leukemia-initiating cells, it leads to HSC depletion through a cell-autonomous mechanism.[45] As opposed to GATA2 and PBX1, Evi1 directly represses Pten transcription, which leads to activation of AKT/mTOR signaling.[23] In mouse bone marrow, Pten expression is significantly elevated in Evi1-deficient HSPC, whereas PTEN expression is inversely correlated with EVI1 expression in human leukemia. Taken together, these observations suggest that EVI1 is a central transcription factor in HSC that can function as an activator or repressor of gene transcription, depending on the molecular context.

Along with its DNA-binding activity, EVI1 can recruit diverse proteins, including critical hematopoietic transcription factors and chromatin remodeling proteins (discussed in more detail below), generating higher-order complexes for transcriptional regulation. EVI1 physically interacts with RUNX1, a versatile regulator involved in HSC establishment and leukemic transformation, and this interaction represses RUNX1-dependent transactivation, while blocking granulocytic differentiation.[46] Furthermore, EVI1 directly binds to PU.1, an Ets family transcription factor that is essential for lymphoid and myeloid development, and impairs myelopoiesis through suppressing PU.1-dependent activation of myeloid genes,[47] whereas EVI1 interaction with GATA1, a master regulator of erythropoiesis, blocks efficient DNA binding of GATA1 and inhibits proper erythroid differentiation.[48] Generally, the balance between self-renewal and differentiation is regulated by the competition between transcription factor complexes. Cross-antagonism of transcription factors by direct binding to each other, and subsequent competition for specific target genes, could lead to a rapid shift resulting in blocked differentiation and maintained self-renewal.[49] The above findings implicate EVI1 as a pivotal player in this competitive transcription-factor model for HSC self-renewal.

In addition to intrinsic factors, the interactions between HSC and their microenvironment are implicated in HSC maintenance.[49] Among them, decreased expression of Tie2/angiopoietin signaling molecules, a key extrinsic pathway that regulates HSC quiescence in the bone marrow niche, is observed in Evi1-deficent P-Sp cells.[7] Furthermore, Tie2/angiopoietin signaling contributes to the maintenance of quiescence in AML cells with EVI1 activation.[50] Yamazaki et al.[51] report that active transforming growth factor-β (TGF-β) produced by nonmyelinating Schwann cells in the bone marrow microenvironment is critical for maintaining HSC hibernation. EVI1 has been shown to interfere with TGF-β signaling through direct interaction with Smad3.[52] Hence, through these interactions, EVI1 may have the potential to coordinate and modulate the cellular and molecular response to extrinsic signals in the bone marrow niche.

Epigenetic Perturbations and Underlying Mechanistic Links in Ecotropic Viral Integration Site 1-Related Leukemias

Genome-wide DNA methylation profiling has revealed that promoter DNA methylation distributes into specific and distinct patterns in human AML, which reflects biologically and clinically relevant differences.[53] AML blasts with EVI1 activation have been shown to display a distinct DNA hypermethylation signature of CpG-rich promoters that differs entirely from those of normal HSPC and any other generically well-defined AML.[54] Furthermore, EVI1 consensus DNA binding sequences are overrepresented in hypermethylated promoters in EVI1-positive AML. The relationship between EVI1 and DNA methylation is supported by a mechanistic link through physical interaction between EVI1 and DNA methyltransferases DNMT3A and DNMT3B, which mediate de novo DNA methylation,[54] suggesting a role for EVI1 in directing de novo DNA hypermethylation in human AML (Fig. 2).

Figure 2.

Mechanistic link of epigenetic machinery and ecotropic viral integration site 1 (EVI1). EVI1 physically interacts with multiple components of various epigenetic machineries. (A) EVI1 interacts with histone deacetylases (HDAC) and C-terminal binding protein (CtBP), leading to transcriptional repression. (B) EVI1 interacts with histone acetyltransferases (HAT), such as CBP and P/CAF, which mediates transcriptional activation via EVI1 acetylation. (C) EVI1 binds to H3K9-specific histone methyltransferases (HMT), including SUV39H1, G9a and polycomb group proteins, which results in transcriptional silencing through histone methylation. (D) EVI1 interacts with BRG1, a component of the SWI/SNF ATP-dependent nucleosome remodeler, leading to transcriptional activation. (E) EVI1 binds to DNA methyltransferases (DNMT), which results in transcriptional repression through DNA methylation. (F) EVI1 interacts with Mbd3 protein and this interaction inhibits HDAC activity of Mi-2/NuRD complex. Ac, acetylation; Me, methylation.

EVI1 is also involved in chromatin remodeling through direct interactions with multiple components of histone modification complexes (Fig. 2). In particular, Evi1 binds to several polycomb group proteins, which mediate H3K27 methylation and function as transcriptional repressors, and this interaction epigenetically represses Pten transcription.[23] In addition, EVI1 physically interacts and colocalizes with H3K9-specific histone methyltransferases (HMT), SUV39H1 and G9a.[55] Remarkably, specific knockdown of these HMT significantly reduces the colony-forming activity of Evi1-induced immortalized murine progenitor cells.[23, 55] In addition, EVI1 also directly binds to C-terminal binding protein (CtBP), and this binding results in recruitment of histone deacetylases (HDAC), leading to transcriptional silencing.[56] EVI1 also interacts with several HDAC proteins directly and through sites different from those required for CtBP binding.[57] In contrast to these corepressors, EVI1 physically interacts with coactivators CBP and P/CAF, both of which have histone acetyltransferase (HAT) activity, resulting in subsequent transcriptional activation via EVI1 acetylation.[57, 58] Furthermore, EVI1 can directly bind to nucleosome remodelers, such as Mbd3/Mi-2/NuRD repressor complex,[59] or BRG1 SWI/SNF chromatin remodeling complex.[60] These findings support the hypothesis that EVI1 integrates functions in histone modification complexes and DNA methylation to mediate transcriptional repression or activation (Fig. 2). The mechanistic contribution of EVI1 to a more aggressive disease phenotype remains elusive, but may be associated with its interaction with these epigenetic regulators.

Novel Molecular Targets in Acute Myeloid Leukemia with Ecotropic Viral Integration Site 1 Activation

As a retrospective study reveals that AML patients with high EVI1 expression might benefit from allogeneic hematopoietic stem cell transplantation in first complete remission,[15] this treatment strategy might represent a viable treatment option. However, development of molecularly targeted therapies is of particular importance for this patient cohort with extremely poor prognosis. The above findings on the molecular mechanisms underlying EVI1-mediated leukemogenesis provide several potential molecular targets. In fact, inhibition of the Akt/mTOR pathway antagonizes the leukemogenic properties of Evi1-expressing leukemic cells,[23] suggesting that PI3K/AKT/mTOR signaling inhibitors may have a therapeutic effect on AML with EVI1 activation. Furthermore, targeting epigenetic components, such as DNMT or chromatin remodeling complexes, might be of therapeutic benefit, because epigenetic changes are potentially reversible processes. Indeed, an HDAC inhibitor, trichostatin A, has been shown to alleviate EVI1-mediated repression of TGF-β signaling.[56] Specific inhibitors targeting epigenetic regulators have been developed in recent years, and these inhibitors have potential to expand our therapeutic options for EVI1-expressing AML in the near future.

A clinical study by Raza et al.[61] provides another potential target for therapeutic intervention, and shows that an arsenic trioxide (ATO) and thalidomide combination produces multilineage hematologic responses in MDS patients with high EVI1 expression. These findings are supported by biochemical and in vitro studies that demonstrate that ATO degrades EVI1 via the ubiquitin–proteasome pathway, and is effective against the murine myeloid cell line forced to express EVI1.[61, 62] One potential strategy to develop therapies that effectively target LSC is to identify cell surface markers that can distinguish LSC from normal HSC. It has been shown that CD52, which is mainly expressed on lymphocytes, is highly expressed in most cases of EVI1-expressing AML, and anti-CD52 monoclonal antibody (alemtuzumab) exhibits in vitro and in vivo antitumor effect against AML with high EVI1 expression.[63] Thus, these biological processes and related molecules are promising therapeutic targets for AML with high EVI1 expression (Fig. 3).

Figure 3.

Potential therapeutic targets of acute myeloid leukemia with ecotropic viral integration site 1 (EVI1) activation. Studies regarding the molecular mechanism underlying EVI1-related leukemias identified several molecular targets, including the PI3K/AKT/mTOR pathway, surface molecule CD52, EVI1 degradation pathway, and epigenetic regulators, including histone acetyltransferases (HAT), histone deacetylases (HDAC), histone methyltransferases (HMT) and DNA methyltransferases (DNMT).

Concluding Remarks

It has become evident that a transcriptional program shared by LSC and HSC is associated with adverse outcomes in AML. The above observations have established EVI1 as a key regulator in fine-tuning of stem cell properties as well as a potent oncogenic determinant that confers poor prognosis in myeloid neoplasms. Thus, investigating EVI1-mediated leukemia provides a novel insight into how molecular machinery governing stem cell function contributes to leukemia progression and therapeutic resistance in AML, which leads to the improvement of current therapeutic modalities.

Disclosure Statement

The authors have no conflict of interest to declare.