Epigenetic Regulation of Protein Translation in KMT2A-Rearranged AML

bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Epigenetic regulation of protein translation in KMT2A-rearranged AML.

Alexandra Lenard1, Hongbo Michael Xie1, Simone S. Riedel1, Zuo-Fei Yuan1, Nan Zhu2, Tobias

Neff3, and Kathrin M. Bernt1,4,5

1Division of Pediatric Oncology, Department of Pediatrics, Center for Childhood Cancer

Research, Children's Hospital of Philadelphia, Philadelphia, PA.

2Current address: Stem Cell Biology and Hematopoiesis Program, Blood Research Institute,

Blood Center of Wisconsin, Milwaukee, WI 53226, USA.

3Division of Pediatric Hematology/Oncology/BMT, University of Colorado School of Medicine and

Children's Hospital Colorado Aurora, CO, 80045; current address: Glaxo Smith Kline, 1250 S

Collegeville Rd, Collegeville, PA 19426.

4 Department of Pediatrics, Perelman School of Medicine at the University of Pennsylvania and

Abramson Cancer Center, Philadelphia, PA

5Corresponding author: Kathrin M. Bernt, MD, Division of Oncology and Center for Childhood

Cancer Research, Children’s Hospital of Philadelphia, 3501 Civic Center Boulevard, CTRB

3064, Philadelphia, Pennsylvania 19104, Phone: 215.370-3171, email: [email protected]

ABSTRACT

Inhibition of the histone methyl-transferase DOT1L (KMT4) has shown encouraging activity in

preclinical models of KMT2A (MLL)-rearranged leukemia. The DOT1L inhibitor pinometostat

(EPZ5676) was well tolerated in early phase clinical trials and showed modest clinical activity,

including occasional complete responses (CRs) as single agent. These studies support the

development of combinatorial therapies for KMT2A-rearranged leukemias. Here, we investigated

two novel combinations: dual inhibition of the histone methyltransferases DOT1L and EZH2, and

the combination of a DOT1L inhibitor with the protein synthesis inhibitor homoharringtonine

(HHR).

EZH2 is the catalytic histone methyltransferase in the polycomb repressive complex 2 (PRC2),

and inhibition of EZH2 has reported preclinical activity in KMT2A-rearranged leukemia. We found

that the H3K79 and H3K27 methyl marks are not dependent on each other, and that DOT1L and

EZH2 inhibition affect largely distinct gene expression programs. In particular, the KMT2A/DOT1L

target HOXA9, which is commonly de-repressed as a consequence of PRC2 loss or inhibition in

other contexts, was not re-activated upon dual DOT1L/EZH2 knockout or inhibition. Despite

encouraging data in murine KMT2A-MLLT3 transformed cells suggesting synergy between

DOT1L and EZH2 inhibition, we found both synergistic and antagonistic effects on a panel of

human KMT2A rearranged cell lines. Combinatorial inhibition of DOT1L and EZH2 is thus not a

promising strategy. We identified opposing effects on ribosomal gene transcription and protein

translation by DOT1L and EZH2 as a mechanism that is partially responsible for observed

antagonistic effects. The effects of DOT1L inhibition on ribosomal gene expression prompted us

to evaluate the combination of EPZ5676 with a protein translation inhibitor. EPZ5676 was

synergistic with the protein translation inhibitor homoharringtonine (HHR), supporting further

preclinical/clinical development of this combination.

INTRODUCTION

Rearrangements of KMT2A (MLL1) occur in 10% of AML and ALL, and 70% of infant ALL. With

few exceptions, KMT2A rearrangements are associated with a poor prognosis, and KMT2A-

rearranged (KMT2A-r) leukemias have been the target of substantial drug development efforts

and clinical research without much impact yet on survival [1]. KMT2A fusions are strongly

transforming [2], likely due to the profound epigenetic changes they induce. Key histone lysine

methyltransferases (KMTs) involved in KMT2A-fusion mediated leukemogenesis are the H3K79

KMT4 (hereafter referred to as DOT1L) [3-6] and the H3K27 KMT6 (hereafter referred to as EZH2)

[7-11]. Both have been proposed as therapeutic targets, and pharmacologic inhibitors for both are

currently in clinical trials (although EZH2 inhibitors are not being studied in AML).

Direct genomic targets of the KMT2A fusion display aberrantly high H3K79 methylation levels

compared with other highly expressed loci [3], and knockdown, knockout, or pharmacologic

inhibition of DOT1L results in the transcriptional downregulation of fusion target gene expression

in KMT2A fusion leukemia models, cells lines and patient samples [3-6]. These findings formed

the basis for two phase I/II clinical trials with the DOT1L inhibitor pinometostat (EPZ5676), as well

as ongoing combination trials with hypomethylating agents and chemotherapy. Pinometostat was

well tolerated and induced single agent responses including two CRs, one of them durable for

many months. However, a substantial number of patients failed to respond, and in others,

resistance rapidly developed [12]. Several mechanisms of resistance have been reported in

model systems since the initiation of these trials. These include classic drug efflux pump

mechanisms (ABCB1A/MDR-1/PGP) [13], epigenetic resistance through downregulation of

SIRT1/SUV39 [14], and independence of H3K79 methylation through unknown mechanisms [13].

Nevertheless, the clinical experience provided proof of concept that DOT1L inhibition had activity

in KMT2A-r disease, with largely non-overlapping toxicities with other agents. A critical next bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

question is whether synergistic combinations with other agents can be identified that ultimately

improve outcomes for a larger number of patients.

In this study, we investigated the interplay between DOT1L and EZH2 in KMT2A-r AML and ALL

cells. Inhibitors of both, DOT1L and EZH2 have been reported to have single agent efficacy in

models of KMT2A-r leukemia, with different mechanisms of action [1, 3-11]. This raises the

question whether dual inhibition could be additive or synergistic. At the same time, potential

interplay on several levels could also result in antagonistic effects of dual DOT1L/EZH2 inhibition:

DOT1L inhibition results in downregulation of direct KMT2A targets such as the later HOXA

cluster genes. During embryonic and hematopoietic development as well as in several subtypes

of leukemia, the HOXA cluster is silenced and acquires H3K27 tri-methylation by the polycomb

repressive complex 2 (PRC2) [15]. Loss of PRC2 function has been linked to increased HOXA

cluster expression in leukemia [16, 17]. Inhibition of EZH2, the main KMT of the PRC2 complex,

might thus interfere with the silencing of direct KMT2A fusion targets in a manner similar to SIRT-

1/SUV39 [14]. Furthermore, AF10, a core member of the DOT1L complex and rate limiting co-

factor for the di- and tri-methylation of H3K79 [18], was reported to bind unmodified, but not H3K27

methylated H3K27 [19]. Inhibition of H3K27 methylation might thus increase the ability of AF10 to

bind to KMT2A fusion targets and increase H3K79 methylation, resulting in greater difficulty to

achieve profound inhibition of H3K79 methylation. In this study, we set out to answer the following

questions: Does EZH2 inhibition modulate H3K79 methylation? Is EZH2/PRC2 required for the

silencing of KMT2A fusion direct target genes upon DOT1L inhibition? Is dual DOT1L/EZH2

inhibition synergistic or antagonistic in KMT2A-fusion driven leukemia?

In brief, we found that the combination of DOT1L /EZH2 inhibition induced pleiotropic and context

dependent effects, and did not consistently synergize with each other. However, these studies

identified previously underappreciated effects of these inhibitors on protein translation and

suggest that protein translation inhibition acts synergistic with DOT1L inhibition. bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

METHODS:

For primer sequences, antibodies and detailed experimental procedures please refer to the

supplemental materials.

Cell lines: Human leukemia cell lines were obtained from ATCC or DSMZ and maintained in

culture as detailed in the supplemental materials. Cell lines were re-authenticated every 6

months in culture.

Drug Assays: Compounds were dissolved in DMSO, and all dilution series were prepared

keeping the DMSO exposure euqal accross all conditions and at <0.1% (<1:1000). Human

leukemia cell lines were exposed to EPZ4777, EPZ5676 (DOT1L inhibitor), GSK 126 (EZH2

Inhibitor) or homoharringtonine (HHT) at the indicated concentrations, and cells were replated at

equal densities in fresh compound containing media every 3-4 days. Cell growth and viability

was assessed either by serial replating and trypan blue exclusion, or XTT assay.

Dot1l and Ezh2 knockout mice, breeding: Animals were maintained at the Animal Research

Facility at the University of Colorado and Boston Children’s Hospital. Animal experiments were

approved by the Internal Animal Care and Use Committee. Dot1l [3] and Ezh2 ([7]) conditional

knockout mice were previously described and were maintained on a C57BL/6 background.

Generation of transformed murine cells: Ecotropic retroviral vectors containing murine

KMT2A-MLLT3-IRES-GFP, Cre-IRES-pTomato (Cre) and MSCV-IRES-pTomato (MIT) were

generated by cotransfection of 293 cells. Lin-Sca-1+c-Kit+ (LSC) cells were transduced with

KMT2A-MLLT3-GFP. After 2-7 days, GFP+ cells were sorted and transduced with Cre or control.

2-3 days after transduction, cells were sorted and plated in colony assays or liquid culture.

Biochemical Assays (cell growth, colony growth, apoptosis, cell cycle, western blotting):

For colony assays, sorted transduced cells were plated in methylcellulose M3234 containing IL3,

IL6 and SCF at 1000 cells per plate in duplicate, and colonies were scored after 7 days of culture. bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

For liquid culture of murine cells, cells were maintained in media with cytokine support, and

replated every 2-3 days. Dot1l and Ezh2 deletion was verified by PCR at each replating beyond

day 7. Cell growth and viability were followed by serial cell counts using Trypan blue exclusion.

Apoptosis and cell cycle analysis were performed using the Annexin-staining and the Click-IT

EdU kit. Protein translation was analyzed using the Click-IT OP-Puro kit. Western blotting for

histone modifications was performed on purified histones isolated by acid extraction using the

indicated antibodies and controls.

Histone Mass Spectrometry: Histones were isolated, chemically derivatized and analyzed by

mass spectrometry as previously described [20].

qPCR analysis of HOXA9 and CDKN2A: RNA was isolated from sorted murine transformed

progenitor cells or compound treated human cell lines using RNeasy mini columns (Qiagen).

Please refer to the supplement for primer sequences. Fold-change of is shown compared

vehicle treated control (human cell lines) or wild type control (murine knockout studies).

Chromatin immunoprecipitation (ChIP): Chromatin immunoprecipitation for H3K79me2 and

H3K27me3 in murine KMT2A-MLLT3 leukemias was performed using rabbit polyclonal antibodies

from abcam (ab3594 Cambridge, MA) similarly as described [21].

RNA amplification and RNA-Seq: RNA was isolated from MV4;11 cells exposed to 7 days of

EPZ5676 or GSK126 using RNeasy mini columns (Qiagen). RNA for RNA-Seq was submitted to

the UC-Denver genomics core for library preparation and sequencing.

Data analysis and statistical methods : Histone PTMs were analyzed using the EpiProfile 2.0

computational algorithm [22]. Drug interactions were evaluated for synergy or antagonism using

CompuSyn (http://www.combosyn.com/) [23, 24]. RNA-Seq raw Fastq files were aligned using

STAR [25] against Human GRCh37 reference and quantified by applying Kallisto [version 0.45.0,

PMID: 27043002]. Output from Kallisto was then directly imported into DESeq2 [26] in order to

detect differentially expressed genes (DEG). DEGs were deemed as genes with False Discovery

Rate (FDR) less than 0.05 level. bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Gene expression data was deposited at the NCBI Gene Expression Omnibus and is accessible

under GSE134369 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE134369).

RESULTS:

No interplay between H3K27 and H3K79 methylation.

In order to assess whether inhibition of EZH2 affects H3K79 methylation (or DOT1L inhibition

impacted H3K27 methylation), we exposed the KMT2A rearranged cell line MV4;11 to small

molecule inhibitors of DOT1L or EZH2 (Figure 1A). Western blotting over a wide range of inhibitor

doses revealed no changes of H3K79 methylation upon inhibition of EZH2, or H3K27 methylation

upon inhibition of DOT1L. Results were confirmed by mass spectrometry (Figure 1B) at a fixed

dose in an extended cell line panel. Despite a higher availability of H3K27 unmodified histones

that could serve as binding sites for AF10 [19], H3K79 methylation is not increased upon inhibition

of EZH2.

Synergy between DOT1L and EZH2 deletion in murine KMT2A-MLLT3 (MLL-AF9)

transformed bone marrow cells.

We next asked if dual knockout of Dot1l and Ezh2 would act synergistic or antagonistic in a

defined genetic mouse model driven by MLL-AF9 (KMT2A-MLLT3). MLL-AF9 was retrovirally

introduced into lin- cKit+ Sca-1+ (LSK) cells from mice carrying conditional alleles for Dot1l and/or

Ezh2. Deletion of either Dot1l or Ezh2 alone resulted in smaller and more differentiated colonies

(which we previously showed to be devoid of replating capacity, therefore colonies were not

replated). Dual inactivation of Dot1l and Ezh2 resulted in a near complete failure to yield any

colony formation (Figure 2A), and particularly rapid outgrowth of cells that had failed to undergo

complete deletion of the Ezh2 allele (Figure 2B). Transcriptionally, inactivation of Ezh2 was

accompanied by de-repression of Cdkn2a, which was independent of Dot1l and H3K79

methylation. In turn, HoxA9 expression was decreased upon loss of Dot1l, which was not rescued bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

by concomitant loss of Ezh2 (Figure 2C). Consistent with these results, ChIP-qPCR of the later

HoxA cluster showed no substantial increase in H3K27 trimethylation 5 days after Dot1l deletion,

while HoxA9 expression was already profoundly decreased. We therefore conclude that PRC2 is

not the primary mechanism by which the HoxA cluster is silenced upon loss or inhibition of DOT1L.

DOT1L and EZH2 inhibition results in both synergistic and antagonistic effects in human

cell lines.

Our data in the murine conditional knockout system suggest potential synergy between

inactivation of DOT1L and EZH2 at high doses, mediated by de-repression of CDKN2A. We

therefore asked whether dual inhibition of DOT1L and EZH2 would act synergistic in a panel of

cell lines with intact or deleted CDKN2A locus (Figure 3A and S1, mutational information from

ATCC, DSMZ and CCLE). Molm14, Monomac6, MV4;11 and THP1cells were exposed to a range

of difference concentrations of DOT1L and EZH2 inhibitor alone or in combination. We observed

strong synergy in Molm14 and Monomac6 cells (Figure 3B), and moderate to strong antagonism

in MV4;11 and THP1 cells (Figure 3C). Thus, CDKN2A status did not correlate with synergy

versus antagonism in the human cell lines. As in the murine system, the expression of KMT2A-

fusion target genes was not affected by inhibition of EZH2. The substantial antagonistic effects

between DOT1L and EZH2 inhibition in half the cell lines tested suggest that this is not a

combination that should be pursued clinically.

DOT1L and EZH2 inhibitors have opposing effects on the transcription of ribosomal genes.

The antagonism of DOT1L and EZH2 in MV4;11 cells was particularly striking. We therefore

performed RNA-Seq to investigate potential mechanisms of antagonism. Samples with dual

inhibition clustered separate from single inhibition, but were closer to DOT1L than EZH2 exposed

samples (Figure 4A). Principal component analysis revealed that, for the most part, DOT1L and

EZH2 act on distinct, non-overlapping gene sets (Figure 4B). H3K79me2 (catalyzed by DOT1L)

is associated with actively transcribed genes, while H3K27me3 (catalyzed by EZH2) is associated bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

with silencing. In order to interrogate which pathway(s) might be responsible for the observed

antagonistic effects, we therefore focused on genes that were downregulated upon DOT1L

inhibition, de-repressed upon EZH2 inhibition, and normalized upon dual inhibition. The gene set

in cluster 1 and 5 follows this pattern (Figure 4C+D, red and light blue cluster/line). KEGG pathway

analysis showed that this gene set is highly enriched for ribosomal genes (Figure 4E). In fact, a

role for PRC2 in repressing Polymerase II transcribed RNA transcription (rRNAs, tRNAs) has

been previously documented [27].

DOT1L inhibition reduced protein translation.

We next asked whether DOT1L inhibition reduced protein translation in MV4;11 cells. Protein

translation was measured by incorporation of OP-Puro and found to be affected by DOT1L

inhibition (Figure 5A). The rate of protein translation is different in different phases of the cell

cycle, and we and others showed decreased cycling as one of the most prominent effects of

DOT1L inhibition or deletion. We therefore chose a very early time point, and measured cell cycle

distribution at the same time as protein translation. In fact, decreases in protein translation were

one of the earliest effects measured when exposing MV4;11 cells to DOT1L inhibition (Figure 5A),

occurring before any substantial effects on cell cycle (Figure 5B), and at a dose much lower than

doses that affect HOXA cluster expression (compare to Figure 3D). EZH2 inhibition not only

rescued ribosomal gene expression (Figure 4), but also partially rescued protein synthesis (Figure

5C). The antagonism between DOT1L and EZH2 inhibitors in MV4;11 cells may thus in part be

driven by their opposing effects on ribosomal gene expression and, as a consequence, protein

translation.

DOT1L inhibition acts synergistically with the protein translation inhibitor

homoharringtonine (HHR).

The observation that DOT1L inhibition affects ribosomal gene expression and protein translation

raised the possibility that EPZ5676 exposure would sensitize cells to the effect of a protein bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

translation inhibitor. We first asked whether Inhibitory effects on protein translation were observed

across all KMT2A-rearranged cell lines used in this study, and found this to be the case (Figure

5D),. Non-KMT2A-rearranged control cell lines were not affected (Figure 5D). The protein

translation inhibitor homoharringtonine (omacetaxine) was FDA approved for chronic myeloid

leukemia (CML) in 2012 [28, 29]. It also has published preclinical activity in several AML models

and small clinical trials [30-32]. We exposed two KM2A-rearranged AML cell lines (MV4;11,

Molm14) to 4-7 days of EPZ5676, followed by exposure to EPZ5676 and HHR. We found the dual

inhibition to be synergistic, suggesting that decreased ribosomal gene expression upon DOT1L

inhibition sensitized leukemia cells to the effect of a protein translation inhibitor.

DISCUSSION:

In this study, we set out to interrogate the interplay between DOT1L and PRC2 inhibition in

KMT2A-rearranged leukemias. We discovered that:

H3K27 and H3K79 methylation do not affect each other. Despite a potential rationale that

modulation of H3K27 methylation might affect H3K79 methylation via binding of AF10 [19], we

find no such effect over a wide range of doses of the EZH2 inhibitor GSK126. Furthermore, our

results are consistent with the previously reported lack of effects of DOT1L loss or inhibition on

H3K27me3 [3, 4].

The polycomb repressive complex 2 is not responsible for silencing the HOXA cluster in

KMT2A-fusion driven leukemia. Despite a clear role of PRC2 mediated silencing of the HOXA

cluster in other contexts, the HOXA cluster does not acquire H3K27 methylation in a time frame

consistent with a primary silencing mechanism. Furthermore, dual inhibition/inactivation of DOT1L

and EZH2 does not rescue HOXA cluster expression. These results were consistent across the

murine retroviral model as well as human cell lines. Given the well documented regulation of the bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

HOXA by PRC2 during development and in the context of other subtypes of leukemia, our data

is highly relevant in that is suggests a more complex and context dependent regulation of the

HOXA cluster than previously appreciated.

The polycomb repressive complex 2 is not responsible for silencing the CDKN2A in human

KMT2A-rearranged cell lines. CDKN2A is a major canonical target of PRC2. Although CDKN2A

independent effects of loss of PRC2 on leukemogenesis have been reported [33], de-repression

of CDKN2A was described as a major mechanism for the anti-leukemic effect of PRC2

inactivation [7, 9]. In the murine model, persistent downregulation of the KMT2A-fusion target

HOXA9 and simultaneous de-repression of CDKN2A provide a good potential mechanism for the

observed synergy between inactivation of DOT1L and EZH2. However, synergistic versus

antagonistic effects in a panel of human cell lines did not correlate with CDKN2A status, and in

cells with an intact CDKN2A locus no de-repression was observed upon inhibition of EZH2.

Inhibition of DOT1L affects ribosomal gene expression, protein translation, and sensitized

KMT2A-rearranged cell lines to homoharringtonine. More in depth analysis of the antagonistic

effects between DOT1L and EZH2 inhibition in MV4;11 cells revealed that EZH2 inhibition

mediates resistance to DOT1L inhibition through partial rescue of ribosomal gene expression and

protein translation. With respect to EZH2, our data is consistent with the previously reported effect

of PRC2 on Polymerase III transcribed non-translated RNA gene transcription [27]. Furthermore,

we found that DOT1L inhibition resulted in decreased protein synthesis in KMT2A-rearranged

AML cell lines. This resulted in sensitization to the effect of the protein translation inhibitor HHR.

A wealth of preclinical data as well as several clinical trials [30-32] support HHR as an active

agent in AML. Our data suggests that the combination of DOT1L inhibition with HHR merits further

investigation, with the potential for rapid clinical translation given that HHR is FDA.

ACKNOWLEDGEMENTS bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

We thank Vikram Paralkar at the University of Pennsylvania for helpful critical discussions.

This work was supported by start-up funds from University of Colorado Denver,

Hematology/Oncology Section and the Children’s Hospital Colorado Research Institute (KMB and

TN), start-up funds by the Division of Pediatric Oncology and the Abramson Cancer Research

Center at the Children’s Hospital of Philadelphia (KMB), as well as NHLBI K08HL102264 (KMB).

AUTHOR CONTRIBUTIONS

KMB and TN designed the study, AL, SSR, NZ and KMB performed the experiments and

analyzed data, ZFY performed and analyzed mass spectrometry, HMX performed bioinformatic

analysis, AL and KMB wrote the manuscript with input from all co-authors.

REFERENCES

1. Winters, A.C. and K.M. Bernt, MLL-Rearranged Leukemias-An Update on Science and Clinical Approaches. Front Pediatr, 2017. 5: p. 4. 2. Andersson, A.K., et al., The landscape of somatic mutations in infant MLL-rearranged acute lymphoblastic leukemias. Nat Genet, 2015. 47(4): p. 330-7. 3. Bernt, K.M., et al., MLL-rearranged leukemia is dependent on aberrant H3K79 methylation by DOT1L. Cancer Cell, 2011. 20(1): p. 66-78. 4. Daigle, S.R., et al., Selective killing of mixed lineage leukemia cells by a potent small-molecule DOT1L inhibitor. Cancer Cell, 2011. 20(1): p. 53-65. 5. Jo, S.Y., et al., Requirement for Dot1l in murine postnatal hematopoiesis and leukemogenesis by MLL translocation. Blood, 2011. 117(18): p. 4759-68. 6. Nguyen, A.T., et al., DOT1L, the H3K79 methyltransferase, is required for MLL-AF9-mediated leukemogenesis. Blood, 2011. 117(25): p. 6912-22. 7. Neff, T., et al., Polycomb repressive complex 2 is required for MLL-AF9 leukemia. Proc Natl Acad Sci U S A, 2012. 109(13): p. 5028-33. 8. Shi, J., et al., The Polycomb complex PRC2 supports aberrant self-renewal in a mouse model of MLL-AF9;Nras(G12D) acute myeloid leukemia. Oncogene, 2013. 32(7): p. 930-8. 9. Tanaka, S., et al., Ezh2 augments leukemogenicity by reinforcing differentiation blockage in acute myeloid leukemia. Blood, 2012. 120(5): p. 1107-17. 10. Xu, B., et al., Selective inhibition of EZH2 and EZH1 enzymatic activity by a small molecule suppresses MLL-rearranged leukemia. Blood, 2015. 125(2): p. 346-57. 11. Kim, W., et al., Targeted disruption of the EZH2-EED complex inhibits EZH2-dependent cancer. Nat Chem Biol, 2013. 9(10): p. 643-50. 12. Stein, E.M., et al., The DOT1L inhibitor pinometostat reduces H3K79 methylation and has modest clinical activity in adult acute leukemia. Blood, 2018. 131(24): p. 2661-2669. 13. Campbell, C.T., et al., Mechanisms of Pinometostat (EPZ-5676) Treatment-Emergent Resistance in MLL-Rearranged Leukemia. Mol Cancer Ther, 2017. 16(8): p. 1669-1679. bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

14. Chen, C.W., et al., DOT1L inhibits SIRT1-mediated epigenetic silencing to maintain leukemic gene expression in MLL-rearranged leukemia. Nat Med, 2015. 21(4): p. 335-43. 15. Brand, M., et al., Polycomb/Trithorax Antagonism: Cellular Memory in Stem Cell Fate and Function. Cell Stem Cell, 2019. 24(4): p. 518-533. 16. Abdel-Wahab, O., et al., ASXL1 mutations promote myeloid transformation through loss of PRC2- mediated gene repression. Cancer Cell, 2012. 22(2): p. 180-93. 17. Danis, E., et al., Ezh2 Controls an Early Hematopoietic Program and Growth and Survival Signaling in Early T Cell Precursor Acute Lymphoblastic Leukemia. Cell Rep, 2016. 14(8): p. 1953-65. 18. Deshpande, A.J., et al., AF10 regulates progressive H3K79 methylation and HOX gene expression in diverse AML subtypes. Cancer Cell, 2014. 26(6): p. 896-908. 19. Chen, S., et al., The PZP Domain of AF10 Senses Unmodified H3K27 to Regulate DOT1L-Mediated Methylation of H3K79. Mol Cell, 2015. 60(2): p. 319-27. 20. Sidoli, S., et al., Complete Workflow for Analysis of Histone Post-translational Modifications Using Bottom-up Mass Spectrometry: From Histone Extraction to Data Analysis. J Vis Exp, 2016(111). 21. Krivtsov, A.V., et al., H3K79 methylation profiles define murine and human MLL-AF4 leukemias. Cancer Cell, 2008. 14(5): p. 355-68. 22. Yuan, Z.F., et al., EpiProfile 2.0: A Computational Platform for Processing Epi-Proteomics Mass Spectrometry Data. J Proteome Res, 2018. 17(7): p. 2533-2541. 23. Chou, T.C., Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacol Rev, 2006. 58(3): p. 621-81. 24. Chou, T.C., Drug combination studies and their synergy quantification using the Chou-Talalay method. Cancer Res, 2010. 70(2): p. 440-6. 25. Dobin, A., et al., STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 2013. 29(1): p. 15-21. 26. Love, M.I., W. Huber, and S. Anders, Moderated estimation of fold change and dispersion for RNA- seq data with DESeq2. Genome Biol, 2014. 15(12): p. 550. 27. Liu, C., et al., PRC2 regulates RNA polymerase III transcribed non-translated RNA gene transcription through EZH2 and SUZ12 interaction with TFIIIC complex. Nucleic Acids Res, 2015. 43(13): p. 6270-84. 28. Li, Y.F., et al., Prolonged chronic phase in chronic myelogenous leukemia after homoharringtonine therapy. Chin Med J (Engl), 2009. 122(12): p. 1413-7. 29. Quintas-Cardama, A., H. Kantarjian, and J. Cortes, Homoharringtonine, omacetaxine mepesuccinate, and chronic myeloid leukemia circa 2009. Cancer, 2009. 115(23): p. 5382-93. 30. Gu, L.F., et al., Low dose of homoharringtonine and cytarabine combined with granulocyte colony- stimulating factor priming on the outcome of relapsed or refractory acute myeloid leukemia. J Cancer Res Clin Oncol, 2011. 137(6): p. 997-1003. 31. Xie, M., et al., HAG (Homoharringtonine, Cytarabine, G-CSF) Regimen for the Treatment of Acute Myeloid Leukemia and Myelodysplastic Syndrome: A Meta-Analysis with 2,314 Participants. PLoS One, 2016. 11(10): p. e0164238. 32. Kantarjian, H., et al., Effectiveness of homoharringtonine (omacetaxine mepesuccinate) for treatment of acute myeloid leukemia: a meta-analysis of Chinese studies. Clin Lymphoma Myeloma Leuk, 2015. 15(1): p. 13-21. 33. Danis, E., et al., Inactivation of Eed impedes MLL-AF9-mediated leukemogenesis through Cdkn2a- dependent and Cdkn2a-independent mechanisms in a murine model. Exp Hematol, 2015. 43(11): p. 930-935 e6.

Figure 1

A H3K79me2

H3K27me3

Total H3 0.3 0.6 1.2 2.5 0.3 0.6 1.2 2.5 0.3 0.6 1.2 2.5 0.3 0.6 1.2 2.5 0.3 0.6 1.2 2.5 0.3 0.6 1.2 2.5

mM iEZH2 0.08 0.16 0.08 0.16 0.08 0.16 0.08 0.16 0.08 0.16 0.08 0.16 DMSO DMSO DMSO DMSO DMSO DMSO mM iDOT1L DMSO 0.08 0.16 0.3 0.6 1.2

B H3K27um Molm14 MV4;11 RS4;11 SEMK2 H3K79um Molm14 MV4;11 RS4;11 SEMK2 H3K27me3 Molm14 MV4;11 RS4;11 SEMK2 H3K79me2 Molm14 MV4;11 RS4;11 SEMK2 % histones % % histones %

DMSO iDOT1L iEZH2 DMSO iDOT1L iEZH2

Figure 1: A: MV4;11 cells were exposed to the indicated concentrations of EPZ4777 and GSK126 for 4 days, and H3K79me2 / H3K27me3 were detected by Western Blotting. B: the indicated cell lines were exposed to 10 mM EPZ5676 (iDOT1L) or 3 mM GSK126 (iEZH2) for 4 days, and H3K79me2 and H3K27me3 were determined by mass-spectrometry. Figure 2 Figure C A

N=2 N=2 (technical), error bars: SD. over the indicated for loci H3K79me2 and H3K27me2 5 days after N=3 independent experiments, error bar: SEM, *p<0.05, **p<0.01. monitored at the indicated days **p<0.01 plated in methylcellulose. N=3 plated duplicate, in error bar: SEM, by transduction with or homozygous conditional heterozygous or wildtype conditional backgrounds with either homozygous wildtype A: Figure2: Ezh2 Dot1l Fold expression 10 certified bypeerreview)istheauthor/funder,whohasgrantedbioRxivalicensetodisplaypreprintinperpetuity.Itmadeavailableunder colonies/plate bioRxiv preprint 8 0 2 4 6 MLL Dot1l Ezh2 100 150 200 250 300 50 0 wt wt Cdkn2A wt wt - AF9 driven leukemias were established driven AF9 were established on leukemias +/+ +/ +/+ +/+ +/+ +/+ +/++/+ +/+ . B: B: ko wt * * - cells from were A maintained in liquid culture, and Ezh2 alleles were doi: wt wt ko - / - https://doi.org/10.1101/790287 ko ko +/+ - / - Cre - + / / - - , and , and leukemia cells the with indicated genotypes were - - / Ezh2 Dot1l -

Fold expression / - 0.5 1.5 Ezh2 0 1 large diff large diff small blast wt wt C: C: wt wt HoxA9 alleles. Deletion alleles. Deletion of conditional alleles was induced ** qPCR for for qPCR HoxA9 and Cdkn2A in from cells A. ko wt a B ** CC-BY-NC-ND 4.0Internationallicense ; wt wt this versionpostedOctober2,2019. ko Ezh2 Dot1l D12 D10 D7 ko ko

D +/+ Dot1l

relative enrichment (%) +/+

10 15 20 25 +/f 0 5

HoxA7 HoxA9 HoxA10 Meis1 HoxB1 Actin HoxB1Meis1 HoxA10 HoxA9 HoxA7 f/f homozygous, homozygous, +/+ f/f f/f Day 5 Day control5 Day H3K79me2 +/f f/f *p<0.05, *p<0.05, Dot1l D: D: Cre The copyrightholderforthispreprint(whichwasnot

+/+ . ChIP +/- +/+ deletion. deletion. -/- - qPCR qPCR +/+ - / +/- -

Day 5 Day control5 Day H3K27me3 -/- Cre Ezh2 flox wt del flox wt del flox wt del bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Figure 3

A B Molm14 Monomac6 KMT2A-(MLL) FLT3 CDKN2A Fusion status status * * * * Molm14 MLL-AF9 FLT3-ITD deleted 100 100 * Monomac6 MLL-AF9 FLT3-V592A wild type 80 * 80 * * * MV4;11 MLL-AF4 FLT3-ITD wild type 60 60 THP1 MLL-AF9 wild type deleted 40 40 20 20 0

0 DMSO D E C D E C DMSO D E C D E C Viable Viable cells [% control] - - - - – - – – 0.16 0.3 0.16 3 0.08 0.3 0.001 0.01

C THP1 MV4;11 * Drug treatment [mM] Drug treatment [mM] * * * * 100 100 * Molm14 Monomac6 * 80 * 80 * Isobologram Isobologram 60 * 60 * 1 1 40 40

20 20 0.5 0.5

0 0 iEZH2 iEZH2 DMSO D E C D E C DMSO D E C D E C Viable Viable cells [% control] ------1 3 0.16 0.3 1 3 0.16 0.08 0 0 0 0.5 1 0 0.5 1 Drug treatment [mM] Drug treatment [mM] iDOT1L iDOT1L

D MV4;11, 0.3 mM MV4;11, 1 mM Molm14, 1 mM HOXA9 MEIS1 MEF2C HOXA9 MEIS1 MEF2C HOXA9 MEIS1 MEF2C ns ns ns 1.5 2.0 ns 1.5 ns 1 1.5 1 ns * ns * 1 * 0.5 * * 0.5 * 0.5 * * * * Fold Fold expression * Fold expression * Fold expression * * 0 0 0 iDot1l - + - + - + - + - + - + iDot1l - + - + - + - + - + - + iDot1l - + - + - + - + - + - + iEzh2 - - + + - - + + - - + + iEzh2 - - + + - - + + - - + + iEzh2 - - + + - - + + - - + +

Figure 3: A: KMT2A (MLL), FLT3 and CDKN2A status of Molm14, Monomac6, MV4;11 and THP1 cell lines (source: ATCC, DSMZ and CCLE). B+C: Cells were treated with a range of doses of EPZ5676 (“D”, DOT1L inhibitor) and GSK126 (“E”, EZH2-inhibitor) alone and in combination (“C”). Cells were replated at equal concentrations every 3-4 days. Cell numbers and viability was assess on day 14 by cell counting and trypan blue exclusion. B top panel: Molm14 and THP1 viable cells at the indicated dose levels. N = 3 independent experiments, error bar: SEM, *p<0.05. Bottom panel: CI-isobologram over the entire dose range (CompuSyn). C: THP1 and MV4;11 viable cells at the indicated dose levels. N = 3 independent experiments, error bar: SEM, *p<0.05 D: qPCR analysis for HOXA9 (black), MEIS1 (grey) and MEF2C (white) after exposure to of EPZ5676 (“D”, DOT1L inhibitor) and GSK126 (“E”, EZH2-inhibitor) alone and in combination at the indicated dose levels (0.3 or 1 mM) in MV4;11 and MOLM14 cells. N = 3 independent experiments, error bar: SEM, *p<0.05 compared to DMSO, ns = no significant difference between indicated samples. bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Figure 4

A Selected DEGs C Expression clusters

0.8 iEzh2 D-low D-high C-low C-high 0.4 0.0 iEzh2 1 low low - - high high - - C D C D 2 3 B Principal component analysis 4 of individual array variation 5 6 10 7

C-high D 1.5 C-low 0 D-high D-low 1.0 Ezh2 PC2, PC2, 21.65% NC 0.5

-10 0.0

-0.5

-1.0

-20 -10 0 10 20 30 Normalized group average PC1, 36.58% -1.5 NC Ezh2 D-low D-high C-low C-high

E Cluster 1 KEGG pathways: Protein processing in ER (p=0.000043, FDR =0.047) (only pathway with significant FDR) Cluster 5 KEGG pathways: Ribosome, eukaryotes (p=1.6e-13, FDR = 6.1E-11) Ribosome (p=3.4e-13, FDR = 9.1E-11) PI3K-AKT signaling (p=0.00016, FDR=0.0085)

Figure 4: RNA-Seq of MV4;11 exposed to 0.08 mM (D-low) or 0.16 mM (D-high) DOT1L inhibitor EPZ5676 or EZH2 inhibitor GSK126 (iEzh2) alone or in combination (C-low and C- high). A: unsupervised clustering of drug exposed cells. B: Principal component analysis. C+D: Clusters of gene expression patterns in drug exposed cells. Cluster 1 and 5 include genes who’s expression increases upon EZH2 inhibition, decreases upon DOT1L inhibition, and is rescued by the combination. E: KEGG pathway analysis of cluster 1 and cluster 5 gene sets. bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Figure 5

A OPP incorporation day 3 B Cell cycle analysis day 3

4x105 p<0.005 G0/G1 S G2 DMSO control

iDOT1L 3x105

MFI 2x105 Cell count count Cell 1x105

0x105 DMSO iDOT1L DMSO iDOT1L DMSO iDOT1L DMSO iDOT1L OPP incorporation

C OPP incorporation day 7 4x105 * ns ns 3x105

MFI 2x105 DMSO control iDOT1L. 1x105

iDOT1L. + iEZH2 5 Cell count count Cell 0x10 DMSO iDOT1L iEZH2 + iDOT1L iEzh2

OPP incorporation

D KMT2A-rearranged non-KMT2A-r

5 4x105 4x10 DMSO control 5 3x105 3x10 iDOT1L

5 MFI 2x105 2x10

5 1x105 1x10

5 0x105 0x10 MV4;11 Molm14 Monomac6 THP1 Kasumi HL60

Figure 5: A: Effect of DOT1L inhibition (0.08 mM EPZ5676) on protein translation measured as OPP incorporation on day 3 after exposure in MV4;11 cells. B: Cell cycle analysis of MV4;11 cells on day 3 of exposure to 0.08 mM EPZ5676 C: Partial rescue of the effect of DOT1L inhibition on protein translation by EZH2 inhibitor. D: Effect of DOT1L inhibition on protein translation measured as OPP incorporation in KMT2A- rearranged cell lines: MV4;11 (0.08 mM), Molm14 (1 mM), THP1 (1 mM), Monomac6 (0.1 mM) (left panel, p<0.02, 2-way Anova) and non-KMT2A-rearranged cell lines: Kasumi (5 mM), HL60 (5 mM) (right panel, p=not significant, 2-way Anova ). bioRxiv preprint doi: https://doi.org/10.1101/790287; this version posted October 2, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Figure 6

A MV4;11 Isobologram * * 100 * 80 * 1 * * 60 0.5

40 HHR

Viable Viable cells [% control] 20

0 DMSO D H C D H C

- - - - 0 0 0.5 1 0.006 0.16 0.006 0.08 iDOT1L

Drug treatment [mM]

B Molm14 Isobologram * * 100 * 80 * 1 * 60 40 0.5 HHR

Viable Viable cells [% control] 20

0 DMSO D H C D H C

- - - - 0 0.003 0.6 0.006

0.3 0 0.5 1 iDOT1L

Drug treatment [mM]

Figure 6: A: MV4;11 and. B: Molm14 cells were pretreated with a range of doses of EPZ5676 (“D”, DOT1L inhibitor). Cells were then exposed to Homoharringtonine (“H”) alone and in combination (“C”), and viability was assessed at 72 hours. Left panel: synergistic effects at the indicated dose levels. N=3 independent experiments, error bar: SEM, *p<0.05. Right panel: CI-isobologram over the entire dose range (CompuSyn).