Aggressive leukemia driven by MLL-AF9

ABSTRACT

We recently showed that cellular origin impacts the aggressiveness and the phenotype of acute myeloid leukemia (AML). Direct induction of the MLL-AF9 fusion in various hematopoietic compartments in vivo using a doxycycline (DOX) regulated mouse model (iMLL-AF9) led to an invasive chemoresistant AML expressing several genes known to be involved in epithelial to mesenchymal transition (EMT) in solid cancers. Many of these genes play important roles in migration and invasion and are significantly associated with poor overall survival in AML patients.

Chromosomal translocations involving the mixed lineage leukemia (MLL) gene are associated with aggressive pediatric and adult acute leukemia with poor prognosis. The clinical heterogeneity of MLL-fusion driven acute myeloid leukemia (AML) has been attributed to several parameters including the patient’s age and health status, the existence of secondary mutations, and, more recently, the cell of origin.¹

In vivo induction of the MLL-AF9 fusion using a novel inducible mouse model (iMLL-AF9) led to a fully reversible doxycycline (DOX)-dependent AML phenotype.² Transplantation of naïve flow-sorted bone marrow-derived hematopoietic stem and progenitor subpopulations into irradiated wild-type recipients always resulted in AML. In contrast to a similar DOX-inducible mouse model for MLL-ENL, we succeeded in initiating AML after transplantation of long term-hematopoietic stem cells (LT-HSC), probably due to expression of critical fusion levels.³ Notably, in 10–20% of the recipients transfer of LT-HSC resulted in a particularly aggressive AML (named “LT-HSC-early” AML) characterized by short latency, excessive organ infiltration, increased leukemia-initiating potential, and resistance to cytarabine chemotherapy. Ex vivo immortalization of iMLL-AF9 LT-HSC led to enhanced clonogenic capacity and the formation of colonies with invasive morphology. Transcriptome analysis linked these origin-related differences to distinct gene expression signatures. As in previous studies, we found that the aggressive LT-HSC-derived AML blasts expressed higher levels of the transcription factor ecotropic viral integration site 1 (Evi1) and the transcription factor ETS-related (Erg) genes.^4,5 In addition, they expressed many genes associated with invasion and/or epithelial to mesenchymal transition (EMT) in solid cancers that were not previously discussed in acute leukemia, such as zinc finger E-box-binding homeobox 1 (Zeb1) or transcription factor 4 (Tcf4).² The relevance of Zeb1 for phenotype maintenance was confirmed by knockdown experiments that impaired the invasive capacity of LT-HSC-early AML cells in vitro and leukemia induction in vivo.

Coupling expression levels of Erg and Evi1 allowed further classification of mouse AMLs based on cellular hierarchy and aggressiveness (Fig. 1). Extensive bioinformatic analyses of mouse and human transcriptomes revealed a significant overlay and led to the identification of genes linked to migration, invasion, and EMT that were associated with poor overall survival of AML patients.

Figure 1.

Cell-of-origin dependent heterogeneity of acute myeloid leukemia (AML) in mice revealed similarities to the human disease. Activation of MLL-AF9 in transplanted naïve long term-hematopoietic stem cells (LTHSC) or granulocyte-macrophage progenitor (GMP) cells using a doxycycline regulated mouse model (iMLL-AF9) results in AML. About 10-20% of LT-HSC–derived AMLs developed an invasive chemoresistant AML characterized by high expression of the transcription factors ecotropic viral integration site 1 (Evi1) and ETS-related gene (Erg) and of genes related to epithelial to mesenchymal transition (EMT). Akin to the clustering of iMLL-AF9 mouse AML based on Evi1/Erg expression, EVI1/ERG expression characterized human AMLs with poor overall survival.

Transfer of 500–1000 LT-HSCs was required to induce this particularly aggressive AML. It is currently unclear why only 10–20% of the irradiated recipients developed this phenotype although we hypothesize that this may be due to activation of iMLL-AF9 in a rare LT-HSC subpopulation with different engraftment potential. For example, single-cell transplantation experiments revealed a small fraction of myeloid restricted HSCs with particular long-term repopulating activity that could be candidates for iMLL-AF9 LT-HSC-early AML.⁶ Given that the absence of more cell type-specific surface markers prevents further dissection of the LT-HSC sub-population conferring the aggressive AML, sorting for c-kit/CD34/FcγRII/III cell surface markers allowed us to enrich for leukemia initiating cells with advanced migratory properties.

EMT is a process in which epithelial cells undergo loss of cellular polarity and cell-to-cell contacts and gain mesenchymal characteristics such as increased migratory and invasive capacity and increased self-renewal.⁷ Physiologically, EMT plays a pivotal role in regulating embryogenesis, wound healing, and fibrosis. Pathologically, EMT is crucial for cancer progression and metastasis by allowing cancer cells to acquire stemness, migration and invasive capabilities, evade apoptosis, and initiate metastasis in distant organs. Given that hematopoietic cells do not need to exit an epithelial state to acquire mesenchymal characteristics we hypothesize that the EMT-like gene signature in the aggressive LT-HSC-early iMLL-AF9 AML may be reminiscent of a circuit facilitating a physiological process in HSC that was inherited by the leukemic blasts upon transformation. Several of the expressed genes, including Zeb1, matrix metallopeptidase 9 (Mmp9), and vimentin, were previously implicated in the above-mentioned processes, including wound healing.⁷ A recent study underlined that the contribution of HSCs in steady state hematopoiesis significantly differs from that under stress conditions.⁸ Notably, researchers identified HSC subtypes with fundamentally distinct molecular signatures and reconstitution potentials, especially following transplantation in irradiated recipients. It remains to be investigated whether iMLL-AF9 activation in a rare HSC subpopulation is linked to development of the particularly aggressive phenotype upon severe stress stimuli.

In human cancers, several investigators suggested that tumor cells inherit significant portions of the transcriptome of disease-initiating cells from which they arose.⁹ ChIP-assays on LT-HSC-early AML and granulocyte-macrophage progenitor (GMP)-derived AML cells showed potent binding of MLL-AF9/MLL1 at the homeobox protein Hox-a9 (Hoxa9) locus, a known MLL-AF9 target in both AMLs, whereas binding to Zeb1 and Evi1 loci was only found in LT-HSC-early AML cells. In line with these data, histone H3 lysine 4 trimethylation (H3K4me3) marking transcriptional activity correlated with iMLL-AF9/MLL1 promoter binding only in LT-HSC-early-AML, whereas histone H3 lysine 27 tri-methylation (H3K27me3) marking transcriptional repression was higher in GMP AML. Therefore, the cellular state at the point of transformation may influence the disease phenotype by allowing access of the fusion protein to different promoters depending on cellular origin. In line with our findings, a recent study using retroviral MLL-AF9 induction in hematopoietic stem and progenitor cells showed that open chromatin profiling may reliably define the tumor cell of origin from the primary bulk tumor cells and reveal open chromatin patterns with prognostic relevance.¹⁰

Collectively, we substantiated the results of previous studies suggesting that cellular origin is a critical determinant of the biology of iMLL-AF9 leukemia and further demonstrated that fusion activation in LT-HSCs induced a highly invasive disease characterized by the expression of EMT-like genes. The anticipation from studies exploring origin-related gene signatures is to provide a source of potential novel biomarkers and targets for personalized therapeutic strategies for this disease that currently lacks effective treatments.

Funding Statement

This work was supported by: Swiss National Science Foundation (SNF-31003A_130661 and 31003A_149714/1), Swiss Cancer League (OCS-2357-02-2009, OCS-02778-02-2011, KFS-3019-08-2012), Wilhelm Sander Foundation (Munich), Swiss Bridge Foundation, the Gertrude Von Meissner Foundation Basel, Novartis Research Foundation, SystemsX.ch (Cell plasticity) and EMBO YIP program.

Disclosure of potential conflicts of interest

No potential conflicts of interests were disclosed.