Leukemic transformation by the MLL-AF6 fusion oncogene requires the H3K79 methyltransferase Dot1l

Genomic rearrangements of the human 11q23 chromosomal band, involving the mixed lineage leukemia (MLL) gene, are observed in a diverse range of myeloid as well as lymphoblastic leukemia subtypes, with a strong association with pediatric and therapy-related leukemias.^1-3  In these leukemias, the MLL gene is fused to one of more than 60 different partner genes, resulting in the formation of dominantly acting MLL fusion-oncoproteins.^3-5  The partners of MLL are nuclear-, cytoplasmic-, or membrane-associated proteins involved in diverse functional processes ranging from chromatin modification and transcriptional elongation to cellular adhesion, endocytosis, cytoskeleton organization, and signal transduction (reviewed in Meyer et al⁴ ). A number of MLL fusion partners, especially nuclear proteins such as AF4, AF9, ENL, ELL, and AF10, fusions of which together account for the vast majority of MLL patients, are components of large, multi-subunit, protein complexes that control gene expression. Several such complexes have been identified, including the family of elongation-assisting proteins, the super-elongation complex,⁶  the related AF4/ENL/p-Tefb (AEP) complex,⁷  and the chromatin-modifying Dot1l complex.⁸  In MLL-rearranged leukemias, constitutive recruitment of one or more of these complexes by chimeric MLL fusion proteins is believed to facilitate sustained expression of MLL-target genes, resulting in leukemic transformation. These complexes represent possible targets for pharmacologic inhibition, with several studies demonstrating the importance of different components of these complexes to MLL fusion-mediated transformation (reviewed in Muntean et al³). One such promising candidate for therapeutic intervention is the histone methyltransferase Dot1l, the central component of the Dot1l complex. Dot1l is the only known enzyme catalyzing the methylation of histone 3 at lysine 79 (H3K79). Studies using various human MLL-rearranged leukemia cells demonstrate high levels of H3K79 methylation on MLL-fusion target genes, suggesting that DOT1L may play an important role in leukemia driven by a variety of MLL-fusion proteins.^9-15  Moreover, a specific small-molecule inhibitor of DOT1L has been shown to have selective activity against MLL-rearranged human leukemia cell lines,¹⁶  raising hopes for therapeutic inhibition of Dot1l as a novel strategy for patients with MLL-rearranged leukemias.

Leukemias with t(6;11)(q27;q23), which encodes the MLL-AF6 fusion protein, constitute the largest subgroup of MLL-rearranged leukemias in which MLL is fused with a predominantly cytoplasmic protein.⁵  Retrospective studies have shown that presence of the t(6;11)(q27;q23) translocation predicts a particularly poor prognosis,^17,18  and thus new therapeutic approaches are clearly needed. Given the recent studies demonstrating an important role for Dot1l in leukemias driven by various MLL-fusion proteins and the development of small-molecule Dot1l inhibitors, we wondered if the MLL-AF6 fusion protein, where the MLL fusion partner is normally cytoplasmic and thus unlikely to be associated with transcriptional complexes, might also require Dot1l to maintain the oncogenic gene expression program. MLL-AF6 leukemias have been modeled in mice, where it was recently shown that the MLL-AF6 fusion can transform hematopoietic progenitors in vitro and in vivo, a process dependent on the dimerizing activity of the Ras-interacting domain of AF6.¹⁹  We sought to analyze a potential role for H3K79 methylation by conducting a genome-wide analysis of H3K79 dimethylation in MLL-AF6 leukemia cells. We also assessed whether MLL-AF6 fusion-mediated transformation was dependent on aberrant H3K79 methylation by genetic or pharmacologic inhibition of the Dot1l histone methyltransferase.

The MSCV-neo MLL-AF6 plasmid consisting of amino acids 35 to 347 comprising the AF6 N-terminal conserved region cloned in the MSCV-neo 5′ MLL construct has been described before¹⁹  and was a kind gift from Ruud Delwel (Erasumus, Rotterdam). The Mi-Tomato plasmid and the CRE-Mi-Tomato plasmids have been described before.¹⁵  Sorted Lin^-Sca-1⁺cKit⁺ (LSK) cells from mouse bone marrow cells were transduced with the MLL-AF6 retrovirus and expanded for 2 weeks in methylcellulose M3234 (Stem Cell Technologies) supplemented with cytokines (6 ng/mL interleukin [IL]-3, 10 ng/mL IL6, and 20 ng/mL stem cell factor) and 1 mg/mL of G418. After 2 weeks of selection, MLL-AF6–transformed cells were either injected into syngenic recipients to generate leukemias or transduced with either Cre-Mi-Tomato or the empty Mi-Tomato control vector. At 48 hours after transduction with Mi-Tomato or Cre-Mi-Tomato, tdTomato-positive cells were sorted and used for colony-forming assays. For leukemia maintenance experiments, bone marrow cells harvested from primary leukemic mice were transduced with Mi-Tomato or Cre-Mi-Tomato and 72 hours later, 200 000 sorted tdTomato-positive cells were injected into sublethally (650 Rad) irradiated BL6 × 129 recipient mice. All mice used in this study were housed in the Animal Research Facility at Children’s Hospital Boston. Animal experiments and protocols were approved by the Internal Animal Care and Use Committee.

Dot1l conditional knockout mice in which the active site of Dot1l (exon5) is flanked by LoxP sites have been previously described ¹² . Bone marrow cells from 7- to 10-week-old mice in Dot1l^+/+ or Dot1l^f/f backgrounds were used for transformation assays and subsequent biochemical experiments.

Colony-forming cell assays were performed by plating 1000 tdTomato-positive cells/mL of cytokine-supplemented methylcellulose M3234 (6 ng/mL IL3, 10 ng/mL IL6, and 20 ng/mL stem cell factor). On day 6 to 7 after plating, colonies were scored using the Nikon Eclipse TS100 inverted microscope (Nikon, Tokyo, Japan). Because almost all colonies were either compact or hypercellular (blast-like) or small and diffuse (consistent with differentiation), colonies were classified into these 2 categories.

Total RNA was isolated using Trizol (Invitrogen, Carlsbad, CA). RNA from 3 independently derived murine MLL-AF6 leukemic bone marrow cells and from the human MLL-AF6–positive cell line ML2 was used for gene expression profiling using Affymetrix 430 2.A and U133 HG-U133_Plus_2 microarrays, respectively (accession no. GSE43069). Data were analyzed using Genepattern.²⁰  Complementary DNA was generated using the tetro cDNA synthesis kit (Bioline, Taunton, MA) using oligo-dT primers. Real-time polymerase chain reaction (PCR) was performed using prevalidated Taqman probes (Applied Biosystems) on the ABI 7700 Sequence Detection System (Applied Biosystems). Primer and probe information will be provided upon request. Average C_t values were normalized to the housekeeping gene GAPDH.

Histones were extracted by overnight acid extraction using the following protocol outlined on the Abcam website (www.abcam.com). Western blotting was done with standard procedures using a 10% Bis-Tris Gel (Invitrogen) and transferred onto polyvinylidene difluoride membranes. Chromatin IP-sequencing (ChIP-seq) was performed as previously described.^9,12 

More details regarding reagents and methods for cell proliferation assays, cytospins, western blotting, ChIP-seq and analysis, cell cycle and apoptosis, etc. can be found in the supplemental Methods section.

The presence of abnormally high levels of H3K79 dimethylation (H3K79me2) at MLL-fusion target genes has been shown to be a characteristic of cells bearing MLL fusions with predominantly nuclear proteins such as AF9 and AF4. In 2011, Yokoyama et al⁷  demonstrated the co-occurrence of the AEP complex, H3K79me2, as well as the MLL-AF6 fusion protein at the chromatin of select MLL-target genes in the MLL-AF6–positive cell line ML2. We decided to probe genome-wide H3K79me2 in MLL-AF6–transformed leukemia cells in order to assess whether H3K79 methylation might be involved in driving MLL-target gene expression more broadly in MLL-AF6 leukemias. We established MLL-AF6–driven leukemias using a bone marrow transplantation model in which lineage LSK cells were transformed with the MLL-AF6 fusion gene. MLL-AF6 expression was confirmed by western blot following overexpression in 293-T cells (supplemental Figure 1). All mice that developed leukemia were found to have acute myelogenous leukemia (AML), with >90% of cells expressing the Gr-1 and Mac-1 myeloid markers in the bone marrow and spleen (supplemental Figure 2). We performed gene expression profiling of 3 independently derived MLL-AF6 leukemias and conducted a genome-wide analysis of H3K79me2 by ChIP-seq using H3K79me2-specific antibodies on the same leukemic bone marrow cells. We observed high levels of H3K79me2 at well-characterized MLL-target genes in all the MLL-AF6 leukemias studied (Figure 1A). Expectedly, genes showing high expression levels in the MLL-AF6 leukemias as assessed by microarray also exhibited high levels of H3K79me2 (red line) in contrast to nonexpressed genes that had little H3K79 dimethylation (blue line). To analyze whether MLL-target loci possessed higher relative levels of H3K79me2 than other highly expressed genes, we compared the average distribution of H3K79me2 on a set of previously defined MLL-core target genes¹²  with 3 randomly chosen sets of size- and expression-matched genes as control (gray lines). We observed a consistently higher deposition of H3K79me2 associated with MLL-fusion core target genes (cyan line) compared with controls (gray lines) (Figure 1B).

To assess whether H3K79 methylation is required for MLL-AF6–mediated transformation, we transduced LSK cells from mice harboring homozygous floxed Dot1l alleles (Dot1l^fl/fl) or Dot1l wild-type controls (Dot1l^+/+) with the retrovirus encoding MLL-AF6. Subsequently, the MLL-AF6–transformed cells were selected with neomycin and transduced with an MSCV-based retrovirus that expresses Cre recombinase and tdTomato (Cre-Mi-Tomato). tdTomato-positive cells were sorted and plated in methylcellulose-based media supplemented with cytokines and replated every week for up to 3 weeks (schematic in Figure 1C). Controls were MLL-AF6–transformed Dot1l^+/+ LSKs transduced with Cre-Mi-Tomato or MLL-AF6–transformed Dot1l^fl/fl LSK cells transduced with the MSCV-based retrovirus without the Cre-recombinase (Mi-Tomato). Immunoblot analysis demonstrated that H3K79me2 levels were significantly reduced in MLL-AF6–transformed Dot1l^fl/fl bone marrow cells after Cre expression compared with controls (Figure 1D).

To characterize the effects of Dot1l loss on MLL-AF6 transformation, we first performed colony-forming assays. We found that Dot1l excision significantly diminished the clonogenic capacity of MLL-AF6–transformed cells in the first week (Figure 2A). In contrast to the dense and compact “blast-like” colonies observed in control vector-transduced cells, colonies generated upon Dot1l deletion were mostly small and diffuse, and Wright Giemsa-stained cytospins of these colonies showed features characteristic of monocytic differentiation (Figure 2B). Genotyping PCR of the Cre-transduced colonies at week 1, 2, and 3 of replating showed a progressive emergence of the unexcised Dot1l allele, demonstrating eventual outgrowth of cells that had escaped deletion of Dot1l (supplemental Figure 3). These results indicate a strong selective pressure against Dot1l-deleted cells. As we and others have previously published, we noted the lack of such strong selection in HoxA9-Meis1–transformed LSK cells, and individual HoxA9-Meis1–transformed Dot1l-excised colonies could be picked and propagated for several weeks (data not shown). These results demonstrate a selective requirement for Dot1l in MLL-AF6–transformed cells, but not all transformed cells. Because aberrant H3K79 methylation at target chromatin is believed to result in inappropriate expression of MLL-fusion target genes, we sought to assess the effects of Dot1l deletion on H3K79me2 at select MLL-fusion target genes. Consistent with the ChIP-seq data from MLL-AF6 leukemias, we found high enrichment of H3K79me2 on the promoter proximal regions of the Hoxa cluster genes compared with input chromatin (Figure 2C dark bars). Cre-mediated excision of Dot1l from these MLL-AF6–transformed cells decreased H3K79me2 levels at the promoters of MLL-target genes and β-actin (Figure 2C). Reduction in the levels of H3K79me2 resulted in a specific reduction in the transcript levels of MLL-target genes Hoxa9, Hoxa10, and Meis1, but not the housekeeping gene β-actin (Figure 2D). This result demonstrates that similar to other MLL-fusion–driven leukemias, continued expression of MLL-fusion target genes in MLL-AF6 leukemias requires activity of the Dot1l methyltransferase.

We then investigated the requirement of Dot1l for the maintenance of established murine MLL-AF6 leukemias. We observed that mice injected with leukemia cells harboring wild-type Dot1l alleles succumbed to acute myeloid leukemia in an average of 175 d after injection. The leukemic mice showed highly elevated white blood cell counts and anemia accompanied by multi-organ blast infiltration and splenomegaly (Figure 2E; supplemental Figure 4; data not shown). In stark contrast, mice injected with Dot1l-deleted leukemic blasts failed to show any signs of leukemic engraftment and retained completely normal white blood cell and red blood cell counts for up to 210 days of observation (Figure 2E; supplemental Figure 4).

Because genetic inactivation of Dot1l severely impaired MLL-AF6 transformation, we sought to assess whether a specific small-molecule inhibitor of Dot1l would similarly affect MLL-AF6–transformed cells. The specific, small-molecule Dot1l inhibitor EPZ004777 has been recently described to selectively impair the proliferation of several MLL-rearranged human cell lines, although its impact on MLL-AF6–transformed cells has not yet been tested. We incubated MLL-AF6–transformed LSK cells with increasing concentrations (0.1-10 µM) of EPZ004777 or dimethylsulfoxide vehicle control and assessed cell number over several days. MLL-AF9–transformed LSKs and HoxA9/Meis1–transformed LSKs served as positive and negative controls, respectively. A dose-dependent reduction of H3K79me2 was seen in all the transformed cell populations by western blotting after incubation with EPZ004777 (Figure 3A). We used 10 µM of the EPZ004777 inhibitor for the proliferation assays, because this is a dose that consistently affects the proliferation of MLL-AF9–transformed cells (used as a positive control) but is still much lower than the >50-µM dose that is needed for off-target effects. As shown in Figure 3B, the expansion of MLL-AF6 as well as MLL-AF9 cells was dramatically impaired by EPZ004777, whereas the growth of HoxA9/Meis1–transformed cells was not significantly altered. Consistent with previous findings with human MLL-AF9 and MLL-AF4 leukemia cell lines, the effects on MLL-AF6–transformed primary murine hematopoietic progenitor cells became apparent only after 7 days. Nevertheless, when exposed to EPZ004777 for a longer time, MLL-AF6–transformed cells showed a dramatic progressive decrease in cell number. In contrast, expansion of the Hoxa9/Meis1–transformed cells was not significantly affected in the presence of the EPZ004777 inhibitor, excluding a drop in viable cell numbers at day 10, which was seen in some but not all experiments (Figure 3B). We then used quantitative reverse-transcription-PCR to quantify expression levels of Hoxa9, Hoxa10, and Meis1 in cells exposed to EPZ004777. A decrease in Hoxa9, Hoxa10, and Meis1 transcript levels was observed in 3 days, with a significant drop at day 7 and 10 after drug exposure (Figure 3C).

To assess the effect of EPZ004777 on the MLL-AF6–transformed cells in more detail, we analyzed changes in cell cycle and apoptosis upon exposure of the MLL-AF6–transformed cells to EPZ004777 by flow cytometry for DNA content and Annexin V staining. Dot1l inhibition reduced the number of actively proliferating MLL-AF6 cells, with a progressive increase in the percentage of cells in the sub-G1 fraction and a concomitant decrease in the S-phase fraction after 10 days of incubation with 10 µM EPZ004777 (Figure 3D). These changes in the cell cycle were accompanied by an increase in the percentage of Annexin V-positive cells, consistent with apoptotic cell death after EPZ004777 treatment (compared with dimethylsulfoxide control) (Figure 3E). These findings demonstrate that MLL-AF6–driven leukemias are selectively sensitive to pharmacologic inhibition of Dot1l.

To determine whether human leukemia cells bearing the MLL-AF6 fusion gene also display high H3K79me2 on MLL-fusion target genes, we performed microarray analysis coupled with a genome-wide ChIP-seq for H3K79me2 on the human MLL-AF6–positive cell line ML2. ChIP-seq for H3K79me2 in the MLL-germline cell line HL60 and the MLL-AF9–positive cell line MOLM-13 served as negative and positive controls, respectively. ML2 cells showed very high levels of H3K79 dimethylation on the HOXA gene cluster, similar to the MLL-AF9–positive MOLM-13 cell line, whereas HL60 cells displayed low to undetectable levels of enrichment of this modification at the same locus (Figure 4A). H3K79me2 levels at other MLL-target genes such as MEF2C were similarly high in ML2 and MOLM-13 compared with HL60 (Figure 4A). Of note, unlike MOLM-13 cells, the ML2 cell line did not display high H3K79 methylation on the MEIS1 locus, consistent with a previous report.⁷  On a more global scale, MLL-fusion core targets showed significantly high levels of mean H3K79me2 signal in the ML2 cell line (Figure 4B). A recent study identified a subset of direct binding targets of the MLL-AF6 fusion protein using ChIP-Chip in the human t(6;11)(q27;q23)-positive ML2 cell line.²¹  We observed that mean H3K79me2 levels at these 24 MLL-AF6 target genes were markedly higher than even the MLL-fusion core target genes (Figure 4B). These findings demonstrate that similar to murine leukemias, the human MLL-AF6–positive cell line ML2 harbors elevated H3K79me2 on MLL-fusion target genes. We then sought to assess whether the expression of MLL-target genes in human MLL-AF6 patients also correlates with high H3K79 dimethylation. For this, we performed an integrative analysis of gene expression from published MLL-AF6–positive human AML patient samples¹⁸  with our ChIP-seq data from the ML2 cell line. Expectedly, when analyzed on a global scale, the levels of H3K79 dimethylation in ML2 positively correlated with the level of transcript expression in these patients (supplemental Figure 5, left panel). Also, mean levels of H3K79 dimethylation on both MLL-core targets as well as MLL-AF6 targets were significantly higher compared with 3 randomly selected size- and expression-matched sets of highly expressed genes (supplemental Figure 5, right panel), suggesting a correlation of H3K79me2 and MLL-target gene expression also in human MLL-AF6–positive leukemia.

To determine the effects of small-molecule inhibition of DOT1L in human leukemia cells, we assessed whether the DOT1L inhibitor EPZ0004777 affected the proliferation, differentiation, or survival of the ML2 cell line with HL60 and MOLM13 as negative and positive controls, respectively. The ML2 cell line showed sensitivity to EPZ004777, with a significant reduction in cell numbers starting from d 10 after drug exposure (Figure 5A). Despite the fact that H3K79me2 was significantly diminished upon exposure to EPZ004777 in all 3 tested cell lines (Figure 5B), only MOLM-13 and ML2 showed significant sensitivity to EPZ004777, further supporting the selectivity of this small molecule for MLL-rearranged cell lines. Exposure of ML2 cells to 10 µM EPZ004777 resulted in a progressive reduction in cells entering the S-phase starting from 3 days after drug exposure with an accumulation in the subG1 phase (Figure 5C left). EPZ004777 treatment also led to a modest but statistically insignificant increase in the percentage of Annexin V-positive cells, indicating that in contrast to cell cycle changes, the effects of EPZ00477 on apoptosis of ML2 cells was not as profound (Figure 5C right). Moreover, the expression of MLL-fusion target genes such as HOXA9 and HOXA10 was significantly down-regulated starting from d 7 after exposure of the ML2 cells to EPZ004777 (Figure 5D). Taken together, the results from the murine and human MLL-AF6–positive leukemia cells indicate that MLL-AF6–driven leukemias may show sensitivity to pharmacologic DOT1L inhibition.

MLL-AF6 fusions comprise a significant proportion of MLL-rearranged AMLs.⁵  Still, barring notable exceptions, there have been few studies assessing mechanisms of oncogenic transformation in MLL-AF6–driven leukemia. In 2 recent studies on pediatric leukemia, MLL-AF6 fusions were shown to be predictive of a significantly poor prognosis compared with most other common MLL-fusions.^17,18  The identification of molecular mechanisms of transformation in the MLL-AF6 leukemias, therefore, is of significant clinical relevance. In this study, we demonstrate that both mouse and human MLL-AF6 leukemia cells harbor elevated H3K79 methylation on MLL-target genes, indicating aberrant DOT1L activity in these leukemias. We further show that loss of Dot1l function via conditional gene inactivation or inhibition of Dot1l using specific, small-molecule inhibitors reduces H3K79 on MLL-target genes and dramatically impairs the transforming ability of the MLL-AF6 fusion gene. The human MLL-AF6–positive cell line ML2 also reflects our findings in the murine models, suggesting that DOT1L plays a critical role in human MLL-AF6 leukemia.

Several MLL fusion proteins recruit transcriptional elongation complexes such as the AEP complex and the DOT1L complex to chromatin near MLL-fusion target genes, resulting in their constitutive and aberrant activation in MLL-rearranged leukemia. A number of MLL fusion partners such as AF9, ENL, and AF10 are normally components of these multi-protein complexes and thus when fused to MLL may directly recruit DOT1L, resulting in the pathological activation of downstream MLL-fusion target genes.^12,22,23  This explanation suggests that MLL-fusion proteins consisting of a partner gene that is not normally part of a chromatin-associated transcriptional complex (such as MLL-AF6) might not be critically dependent on Dot1l and H3K79 methylation. The AF6 protein in its nonrearranged form is localized primarily at adherens junctions and binds to F-actin.^24,25  Even though a shorter spliced form of AF6 has been reported to shuttle between the nucleus and the cytoplasm²⁶  and therefore could potentially be involved in DOT1L and/or AEP recruitment, AF6 has not been co-purified in any of the described transcriptional elongation complexes that include DOT1L (reviewed in Deshpande²⁷ ). A recent study showed that in the MLL-AF6–positive human cell line ML-2, the AEP complex co-localizes, along with MLL-AF6, on MLL-fusion target genes even though the MLL-AF6 fusion cannot directly recruit this complex.⁷  The ML2 cell line also showed H3K79 methylation on select MLL-fusion targets, but in the absence of genome-wide data it was not possible to determine if this level was above what would be expected for other highly expressed genes.⁷  Thus, although both the AEP and DOT1L complexes co-localize along with the MLL-AF6 fusion on MLL-target loci through a hitherto unknown mechanism, it has remained unclear if MLL-AF6 would depend on DOT1L and H3K79 for continued proliferation. Our data using genetic as well as pharmacological Dot1l inhibition strategies on mouse and human MLL-AF6 leukemia cells clearly demonstrate a critical role for DOT1L in MLL-AF6–mediated leukemia.

The requirement of DOT1L for transformation by cytoplasmic MLL-fusion partners such as MLL-AF6 suggests that direct biochemical interaction with an MLL-fusion partner may not be the only mechanism by which DOT1L contributes to MLL-leukemogenesis. Indeed, it cannot be ruled out that different mechanisms of DOT1L recruitment operate in nuclear vs cytoplasmic fusion partners of MLL. Alternatively, it is also possible that the dysregulation of DOT1L stability, enzymatic activity, or interactions between different transcriptional complexes may play an important role in MLL-fusion–mediated transformation, perhaps in addition to DOT1L recruitment. In this regard, it is conceivable that recruitment of the polymerase-associated factor complex by the N-terminal part of MLL is an alternative, indirect way of deregulating H3K79 methylation as previously suggested,²⁸  thereby highlighting the role of the N-terminal MLL part of the fusion as opposed to the fusion partner in establishment of the aberrant H3K79 “lesion” at MLL-target genes. Even as the exact mechanisms of DOT1L recruitment by MLL fusion proteins as well as the role of transcriptional complexes in MLL-leukemogenesis warrant further investigation, there is mounting evidence indicating that DOT1L is an attractive target for therapy in the MLL leukemias. The FDA approval of small molecules targeting epigenetic enzymes such as the DNA methyltransferase inhibitors 5-azacytidine and decitabine and the HDAC inhibitors vorinostat and romidepsin have raised hopes for a new wave of targeted epigenetic therapies against human malignancies. Several efforts are being made to discover small-molecule inhibitors of histone methyltransferases such as DOT1L,^16,29  and clinical trials with some of these inhibitors have begun. We and others have previously demonstrated that transformation mediated by other oncogenes such as Hoxa9 and Meis1, E2A-PBX, as well as E2A-HLF is insensitive to Dot1l inhibition.^11-13  Also, a number of human cell lines that do not harbor MLL fusions are refractory to DOT1L inactivation using short hairpin RNAs or small-molecule inhibition.^12,16  These results strongly suggest that aberrant H3K79 methylation is a molecular feature characteristic to certain, but not all, hematological malignancies. Against this backdrop, our observation that MLL-AF6 leukemias, which make up a substantial proportion of MLL-rearranged leukemia, are dependent on aberrant DOT1L activity is therefore of high clinical relevance.

In summary, the studies discussed above indicate that several MLL fusions, independent of the subcellular localization of the C-terminal partner, may directly or indirectly recruit common transcriptional effector complexes for constitutive activation of the MLL transcriptional program. This potentially shared mechanism of transformation could prove to be an attractive therapeutic target in the MLL leukemias. The findings of this study showing that MLL-AF6–mediated transformation is dependent upon Dot1l and can be inhibited using small-molecule inhibitors could help inform future clinical trials with clinically effective DOT1Ll inhibitors.

The MLL-AF6 plasmid used in this study was a kind gift from the laboratory of Ruud Delwel (Erasmus MC, Rotterdam).The authors thank Ronald Mathieu and Mahnaz Pakhtinat for their excellent technical assistance with flow cytometry.

This work was supported by grants (to S.A.A.) from the National Institutes of Health (National Cancer Institute grants CA66996 and CA140575; and National Institute of Diabetes and Digestive and Kidney Diseases grant DK049216), and grants from the Harvard Stem Cell Institute, and the Leukemia and Lymphoma Society. A.J.D. is supported by the NCI Howard Temin Pathway to Independence Award. M.F. was supported by a summer fellowship from the Giovanni Armenise-Harvard foundation.

Contribution: A.J.D., L.C., K.M.B., and S.A.A. designed the study; A.J.D., L.C., M.F., D.B., J.C., and S.D. performed the mouse experiments; A.U.S. analyzed the genomics data and performed the statistical and bioinformatics analyses; A.J.D., L.C., and S.A.A. wrote the manuscript; E.J.O., S.R.D., V.M.R., and R.M.P. participated in analyzing and interpreting the DOT1L inhibitor data; and all authors reviewed and approved the final version of the manuscript.

Conflict-of-interest disclosure: E.J.O., S.R.D., V.M.R., and R.M.P. are employees of Epizyme Inc. and S.A.A. is a consultant for Epizyme Inc. The remaining authors declare no competing financial interests.

Correspondence: Scott Armstrong, 1275 York Ave, Box 20, New York, NY 10065; e-mail: armstros@mskcc.org.