Flavaglines Target Primitive Leukemia Cells and Enhance Anti-Leukemia Drug Activity

ORIGINAL ARTICLE Flavaglines target primitive leukemia cells and enhance anti-leukemia drug activity

KP Callahan1, M Minhajuddin2, C Corbett1, ED Lagadinou1, RM Rossi1, V Grose1, MM Balys1,LPan3, S Jacob4, A Frontier4, MR Grever5, DM Lucas5, AD Kinghorn3, JL Liesveld1, MW Becker1 and CT Jordan2

Identification of agents that target human leukemia stem cells is an important consideration for the development of new therapies. The present study demonstrates that rocaglamide and silvestrol, closely related natural products from the flavagline class of compounds, are able to preferentially kill functionally defined leukemia stem cells, while sparing normal stem and progenitor cells. In addition to efficacy as single agents, flavaglines sensitize leukemia cells to several anticancer compounds, including front-line chemotherapeutic drugs used to treat leukemia patients. Mechanistic studies indicate that flavaglines strongly inhibit protein synthesis, leading to the reduction of short-lived antiapoptotic proteins. Notably though, treatment with flavaglines, alone or in combination with other drugs, yields a much stronger cytotoxic activity toward leukemia cells than the translational inhibitor temsirolimus. These results indicate that the underlying cell death mechanism of flavaglines is more complex than simply inhibiting general protein translation. Global gene expression profiling and cell biological assays identified Myc inhibition and the disruption of mitochondrial integrity to be features of flavaglines, which we propose contribute to their efficacy in targeting leukemia cells. Taken together, these findings indicate that rocaglamide and silvestrol are distinct from clinically available translational inhibitors and represent promising candidates for the treatment of leukemia.

Leukemia (2014) 28, 1960–1968; doi:10.1038/leu.2014.93

INTRODUCTION cell lymphoma, at doses that caused no discernable toxicity. In the past decade, a relatively new focus in drug development In these studies, the activity of silvestrol was due at least in has been the concept of a cancer stem cell, where the diversity part to loss of the antiapoptotic protein Mcl-1, with within tumors arises at least in part as a consequence of subsequent mitochondrial depolarization and caspase-dependent 7,10 developmental programs that mirror normal tissue systems.1 apoptosis. In addition to leukemia, silvestrol has shown activity Perhaps, the best characterized cancer stem cells are those found toward lung, breast and prostate cancer cells, and thus the utility in the blood cancer acute myelogenous leukemia (AML). Several of these compounds may extend beyond hematologic 11,12 studies have identified malignant stem cells, termed leukemia malignancies. Studies have shown that silvestrol promotes stem cells (LSCs), that are central to the initiation, growth and an aberrant interaction between capped mRNA and eIF4A, relapse of AML.2,3 The high rates of relapse in AML patients thereby interfering with the assembly of the eIF4F translation suggest that current chemotherapeutic strategies do not complex and blocking translation initiation.13,14 Consistent with adequately target LSCs. Thus, there is great interest in the these observations, recent work has identified eIF4A as one of the development/identification of compounds that target LSCs, but primary targets of rocaglamide and silvestrol.15 Hence, the activity still maintain an optimal therapeutic index. of these compounds appears to be related to their ability to inhibit Nearly half of the agents used in cancer therapy today are either translation. Extensive in vitro evidence now points to the natural products or derivatives of natural products.4 A group of translational machinery as a powerful therapeutic target in flavagline compounds from the plant genus Aglaia have attracted cancer, including hematologic malignancies.16,17 The translation attention owing to their insecticidal activities and inhibition of initiation complex constitutes a major node of convergence for tumor growth.5 Two members of this family, rocaglamide and numerous signaling pathways; however, few agents impact this silvestrol, have shown toxicity toward leukemia cells.6–9 The machinery directly, leaving this avenue largely unexplored in vivo. Li-Weber group has shown that rocaglamide induces apoptosis in Thus, flavaglines are a unique set of compounds that represent malignant but not normal proliferating lymphocytes, possibly the first direct inhibitor of translation initiation with clinical attributed to its ability to suppress selectively mitogen-activated potential, as evidenced by their preclinical activity on an array of protein kinase/extracellular signal-regulated kinase survival tumor types in the nanomolar range. activity in the cancer.6,8 Silvestrol has shown efficacy in vitro Here we show that rocaglamide and silvestrol preferentially kill and in mouse models of the B-cell malignancies, chronic phenotypically and functionally defined LSCs, while sparing lymphocytic leukemia, acute lymphoblastic leukemia and mantle normal stem and progenitor cells. Importantly, these compounds

1Department of Medicine, University of Rochester School of Medicine, Rochester, NY, USA; 2Division of Hematology, University of Colorado, Denver, Aurora, CO, USA; 3College of Pharmacy, The Ohio State University, Columbus, OH, USA; 4Department of Chemistry, University of Rochester, Rochester, NY, USA and 5Department of Internal Medicine, The Ohio State University, Columbus, OH, USA. Correspondence: Dr CT Jordan, Division of Hematology, University of Colorado, Denver, 12700 E 19th Avenue, Room 10016, Aurora, CO 80045, USA. E-mail: [email protected] Received 21 August 2013; revised 19 February 2014; accepted 21 February 2014; accepted article preview online 28 February 2014; advance online publication, 1 April 2014 Flavaglines target LSCs KP Callahan et al 1961 are significantly more toxic to leukemia cells as single agents or in shRNA-mediated p53 knockdown and Myc overexpression combination with other anticancer drugs than clinically available The coding sequences of p53 (NM_001126114.1) targeted by short hairpin translational inhibitors. This difference in cytotoxicity, however, is RNAs (shRNAs) are: sh-p53 A, 50-GTCCAGATGAAGCTCCCAGAA-30; sh-p53 E, not attributable to the respective differences after inhibition of 50-GCATCTTATCCGAGTGGAAGG-30; sh-P53 H, 50GGTCTTTGAACCCTTG CTTGC30. The detailed methods for cloning shRNA, generating lentiviral global protein synthesis; rather, it appears that they more 20 efficiently decrease levels of Myc protein and also alter particles and infecting AML cells, are as described previously. Myc mitochondrial integrity via p53 activation. overexpression construct was created as follows: the c-MYC gene was excised from MSCVh c-MYC IRES GFP (addgene plasmid 18119) with EcoRI restriction enzyme and ligated into pLVX-mcherry (Clontech, Mountain View, CA, USA). MATERIALS AND METHODS Primary AML and normal hematopoietic cells Normal and leukemic human bone marrow samples were obtained after Cell cycle analysis informed consent from volunteer donors at the University of Rochester For evaluation of the cell cycle profile, cells were fixed by the BD Fixation/ Medical Center. Total bone marrow mononuclear cells were isolated by Permeabilization Kit (554714; BD Biosciences) according to the manufac- standard Ficoll procedures (GE Healthcare, Niskayuna, NY, USA) and turer’s instructions, stained with anti-Ki67-Alexa Fluor 647 (BD Biosciences) cryopreserved in freezing medium consisting of Cryostor CS10 (BioLife for 45 min at 4 1C and resuspended overnight at 4 1C in 0.5 ml of 5 mg/ml Solutions, Bothell, WA, USA). The viability of leukemic cells after thawing 7-AAD staining solution. Cells were analyzed using BD LSRII flow cytometer was 50–90%. Normal bone marrow total mononuclear cells were further (BD Biosciences). enriched for CD34-positive cells using MACS CD34 Enrichment Kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Mitochondrial depolarization Mitochondrial membrane depolarization was assessed using tetramethylr- Cell death assays hodamine, ethyl ester, percholate (TMRE) purchased from Life Technologies. For in vitro cell death assays, normal and leukemic cells were cultured in Following treatment with drugs, cells were incubated with TMRE for 30 min serum-free media for 24 or 48 h in the presence of drug and analyzed with at 37 1C, and then washed and stained with AnnexinV/7-AAD. All analyses AnnexinV/7-aminoactinomycin D (7-AAD) staining using the LSRII flow show the percentage of viable cells with depolarized mitochondrial cytometer (BD Biosciences, San Jose, CA, USA). For ex vivo toxicity assays, membranes. cells were treated in vitro with rocaglamide (Enzo Life Sciences, Farmingdale, NY, USA) for 48 h, and then harvested and injected in Measurement of endogenous reactive oxygen species levels irradiated NSG (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mice. For AML and Endogenous reactive oxygen species levels of M9-ENL cells were detected by normal bone marrow specimens, engraftment of human cells was 6 evaluated after 6–8 weeks by flow cytometry. labeling 1 Â 10 cells for 15 min at 37 1C with the redox-sensitive probes CM-H2DCFDA (5 mM), or MitoSox (5 mM) (Life Technologies). Labeled cells were resuspended in 0.2 ml of 10 mM 40,6-diamidino-2-phenylindole Colony-forming assay (Invitrogen, Carlsbad, CA, USA) solution and analyzed using BD LSRII flow A total of 5 Â 104/ml of AML or normal cells were plated in Methocult GF cytometer (BD Biosciences) and FlowJo software (Treestar, Ashland, OR, USA). H4534 (Stemcell Technologies, Vancouver, BC, Canada) as described previously.17 Colonies were scored after 21 days of culture. In vivo treatment of xenografts with silvestrol Irradiated NSG mice were transplanted with human leukemia cells (3 Â 106 Methionine incorporation assay per mouse) via tail vein in a final volume of 0.2 ml of phosphate-buffered Methionine labeling experiments and subsequent click-it chemistry were saline with 0.5% fetal bovine serum. After 3 weeks, animals were size performed using reagents and protocols provided by Invitrogen/Life matched to treatment and control groups (10 animals per group) and Sciences (Grand Island, NY, USA). Briefly, cells were grown in media lacking received either silvestrol (0.75 mg/kg) or vehicle. Drugs were injected methionine for 1 h, incubated with Click-IT AHA (Life Technologies, Grand intraperitoneally for 3 weeks with a schedule of 5 days on and 2 days off Island, NY, USA) for 2 h, and then washed and lysed. Click-it chemistry was (15 injections total). Silvestrol was formulated at of 0.1 mg/ml in 30% performed on lysates that were subsequently run on sodium dodecyl hydroxypropyl b-cyclodextrin. Following treatment, mice were killed and sulfate-polyacrylamide gel electrophoresis gel, transferred to nitrocellulose engrafted with human cells and were evaluated by flow cytometry. and probed for the presence of labeled methionine with streptavidin horse radish peroxidase antibody. RESULTS Western blot Rocaglamide selectively ablates leukemia blast cells and LSCs Primary antibodies for Mcl-1, Bcl-XL, Myc, phospho-p53, Bax and actin were To establish the in vitro cytotoxicity of flavaglines, we initially purchased from Cell Signaling Technologies (Danvers, MA, USA). Primary treated cultures of primary human AML specimens or normal antibodies for Bcl-2 and glyceraldehyde 3-phosphate dehydrogenase were bone marrow controls with varying concentrations of rocagla- purchased from Santa-Cruz Biotechnology (Dallas, TX, USA). Western blot was performed as described previously.18 mide. AML cells showed minimal cell death as measured by AnnexinV and 7-AAD at 24 h of treatment; however, toxicity increased significantly at 48 h in both total and the more primitive Combination index calculation CD34 þ populations (Figure 1a and Supplementary Figure S1A). Combination index was calculated by Calcusyn software (Biosoft, Parallel experiments on normal cells showed very modest Cambridge, UK). The results were input into Calcusyn software that applies decreases in viability (Figure 1c and Supplementary Figure S1B). a Chou–Talalay method to calculate the combination index value for each specific dose combo. Parthenolide (Enzo Life Sciences), cytarabine (Sigma, To assess functionally the effects of rocaglamide on stem cells, St Louis, MO, USA), daunorubicin (Pfizer, New York, NY, USA), ABT-737 primary AML and normal specimens were treated with rocagla- (Selleckchem, Houston, TX, USA) and PU-H71 (Calbiochem, Billerica, MA, mide for 48 h and then transplanted into sublethally irradiated USA) were used in the assays. NSG (NOD/SCID/IL2Rg) mice. At 8 weeks after transplantation, bone marrow cells were analyzed to determine the engraftment of Whole transcriptome analysis human cells. Rocaglamide-treated cells were strongly impaired Total RNA was isolated from AML 24 h after rocaglamide treatment using the with respect to xenograft engraftment (Figure 1b and RNeasy Plus Kit (Qiagen, Valencia, CA, USA), as per the manufacturer’s Supplementary Figure S1C), whereas normal controls show no recommendations. The TruSeq RNA Sample Preparation Kit v2 (Illumina, San significant reduction in engraftment potential (Figure 1d and Diego, CA, USA) was used for next-generation sequencing library construc- Supplementary Figure S1D). In agreement with xenograft experi- tion, as per the manufacturer’s protocols and as described previously.19 ments, in vitro colony assays showed that rocaglamide treatment

& 2014 Macmillan Publishers Limited Leukemia (2014) 1960 – 1968 Flavaglines target LSCs KP Callahan et al 1962 AML CD34 cells AML#1 AML xenograft the result of rocaglamide affecting mammalian target of 100 AML#2 100 rapamycin machinery directly; rather, it is a response to the 80 AML#3 80 decrease in translation caused by rocaglamide (data not shown). AML#4 60 AML#5 60 Taken together, these results clearly indicate that rocaglamide 40 AML#6 40 affects the translational status of treated cells. % Viable 20 20 Next, we examined levels of the Bcl-2 family of antiapoptotic % Human Cells 0 0 proteins Mcl-1, Bcl-2 and Bcl-XL, some of which have been 0 10 2030 40 02040 reported to decrease in other systems upon inhibition of Roca (nM) Roca (nM) translation.7,9 We observed decreases in the levels of these proteins in primary AML cells 24 h after treatment with Normal CD34 cells Normal xenograft rocaglamide (Figure 1f). The changes in expression levels of these 100 NBM#1 10 proteins following rocaglamide treatment are not consistent 80 NBM#2 NBM#3 7.5 between samples. There are clear differences in the identity of 60 NBM#4 5 the proteins that decrease as well as in the degree of change, 40 NBM#5 suggesting that the apoptosis induced by rocaglamide cannot be % Viable NBM#6 20 2.5 attributed to a decrease of one particular protein of this family, % Human Cells 22 0 0 consistent with recent genetic experiments. 0 10 2030 40 02040 Despite the inconsistency in Bcl-2 family member reduction Roca (nM) Roca (nM) caused by rocaglamide treatment, in all AML specimens tested the level of at least one of the mitochondrial-associated protein UNT Rocaglamide Cell # AML #1 AML #2 AML #3 decreased, leading us to further evaluate mitochondrial function. 15552.5 15 2.5 in millions Roca -++ -+- Using TMRE, a fluorescent dye indicative of mitochondrial membrane integrity, we observed that rocaglamide increased Mcl-1 the number of cells with depolarized mitochondria Bcl-Xl (Supplementary Figure S1F). This observation is consistent with previous studies using the closely related compound silvestrol, Bcl-2 where LNCaP prostate cancer cells and chronic lymphocytic leukemia cells also demonstrated a loss of membrane Actin 7,23 α Actin potential. Overall, our results show that rocaglamide inhibits translation in AML cells, decreases the levels of antiapoptotic Figure 1. (a) Percentage of viable CD34 hematopoietic cells in proteins and disrupts the mitochondrial integrity. primary human AML cells (N ¼ 6) exposed to rocaglamide. Cells were stained with AnnexinV and 7-AAD, and viability was measured by a flow cytometer and values were found relative to untreated Rocaglamide is more cytotoxic to leukemia cells than other controls. (b) Percentage engraftment achieved in NSG mice translational inhibitors receiving AML cells after 48 h culture with or without rocaglamide. Each symbol represents a single animal analyzed 8 weeks after The deregulation of translation has been implicated in cancer transplantation. Mean engraftment is indicated by the horizontal initiation and progression, and consequently there is significant bars. (c) Percentage of viable CD34 þ cells from healthy donors interest in targeting this process with therapeutic agents.24 To exposed to rocaglamide. (d) Percentage engraftment achieved in assess how flavaglines compare to other agents, we examined NOG mice receiving hematopoietic cells from healthy donors after temsirolimus, a rapamycin derivative that targets the mammalian 48 h culture with or without rocaglamide. (e) Methionine metabolic target of rapamycin pathway, and has been investigated in labeling of untreated leukemia cells and leukemia cells treated with numerous treatment regimens for AML25,26 and other cancers.27,28 rocaglamide for 24 h. Cells were labeled with an epitope-tagged version of methionine to quantify the extent of inhibition of new We treated the M9-ENL cell line with temsirolimus or rocaglamide protein synthesis. (f) Whole-cell lysates were generated from for 48 h and found that rocaglamide was markedly more cytotoxic primary AML cells treated with rocaglamide for 24 h for western toward leukemia cells (Figure 2a). In fact, rocaglamide was more blot analysis. toxic to leukemia cells than the combination of temsirolimus and ribavirin, another translational inhibitor that blocks the binding of eIF4E to the mRNA cap structure (Supplementary Figure S2A).26,29 resulted in only a modest reduction in normal myeloid and We performed methionine labeling in parallel with the cytotoxicity erythroid colony-forming units (Supplementary Figure S1E). Taken assays to evaluate the effect rocaglamide and temsirolimus had together, these results indicate that rocaglamide is able to induce on the rate of translation. Strikingly, temsirolimus, which is death in bulk leukemia cells, as well as in the primitive LSCs, while minimally toxic to leukemia cells, inhibits global translation to a sparing normal stem and progenitor cells. degree that is comparable to that of rocaglamide (Figure 2b). Thus, translational inhibition alone cannot explain the relative efficacy of rocaglamide. Rocaglamide inhibits translation, resulting in reduced levels of Next, we surveyed a broad range of drugs in combination with antiapoptotic proteins rocaglamide to determine its potential in combination regimens. Flavaglines have previously been shown to inhibit the translation M9-ENL cells pretreated with rocagamide for 24 h before the of nascent proteins.13,21 To verify this activity for rocaglamide, we addition of a second drug showed a strong cytotoxicity for several used an epitope-tagged form of methionine and pulse-chase agents (Figure 2c). Temsirolimus also enhanced the activity of studies to determine and quantify inhibition of new protein compounds tested; however, the degree was much less than that synthesis. Rocaglamide markedly reduced the amount of nascent observed with rocaglamide (Figure 2c). We calculated combina- translation after 24 h treatment (Figure 1e). Next, we performed tion indices according to the Chou–Talalay method, which western blot analysis on primary AML cells treated with indicated that all combinations using rocaglamide were synergistic rocaglamide and observed reduced levels of phosphorylated (Supplementary Figures S2B–E). We also observed increased 70S6K and 4EBP1, proteins critical for the initiation/regulation of toxicity when rocaglamide was given simultaneously with the cap-dependent translation (Supplementary Figure S1G). It should second drug; however, increases were not as large as those be noted that the loss of phosphorylation does not appear to be achieved using the pretreatment strategy (Supplementary

Leukemia (2014) 1960 – 1968 & 2014 Macmillan Publishers Limited Flavaglines target LSCs KP Callahan et al 1963 rocaglamide has the potential to increase the efﬁcacy of Tem

UNT multiple agents, a degree of versatility that could be useful in 100 Roca building novel treatment regimens. 80 Tem 60 Rocaglamide modulates the activity of key transcriptional factors 40

% Viable Roca To better understand the mechanism of rocaglamide-mediated 20 cell death, we performed RNA-seq-based studies to generate a 0 comprehensive proﬁle of gene expression changes that occur as a 0hr 24hr 48hr consequence of drug treatment. Using four primary AML specimens and the M9-ENL cell line, cells treated with or without M9-ENL cells 100 40 nM rocaglamide for 24 h were assayed. A total of 794 genes 100 were differentially expressed using a P 0.05 statistical cutoff 87 88 o (Supplementary Table S1). Bioinformatic analyses of the data 80 76 69 71 demonstrated that the transcription factors p53 and Myc were 63 60 58 UNT strongly inﬂuenced by rocaglamide treatment (Supplementary 60 Roca Figure S3B). 41 The tumor suppressor p53 is mutated in 10% of AML patients 39 o % Viable 40 Tem and thus its activity may be relevant to the biology of most AML cells. We observed increased phosphorylation of p53 on serine 15 17 19 20 12 following rocaglamide treatment, as well as increased levels of the 2 proapoptotic and p53-regulated protein Bax (Figure 3a). We 0 speculate that increased Bax is due to enhanced protein stability, UNT PTL PU-H71 ABT DNR as protein synthesis is strongly inhibited by rocaglamide. To test whether p53 activation is required for rocaglamide-mediated cell Total AML cells (N=4) death, we performed shRNA-mediated knockdown of p53 mRNA 100 100 expression in M9-ENL cells and tested their sensitivity to rocaglamide. We generated three independent clones, all of 80 65 59 which had reduced levels of p53 protein (Figure 3b). All three clones showed resistance to rocaglamide treatment, indicating 54 60 No Roca that p53 activation is a component of rocaglamide-mediated cell 39 Roca death (Figure 3c). It should be noted that there is no complete

% Viable 40 rescue of cell death in our p53-deficient cells, indicating that p53 17 22 may not be the sole driver of rocaglamide-induced apoptosis. The 15 20 12 potential for a p53-independent mechanism is consistent with 5 experiments performed in chronic lymphocytic leukemia and 7,23 0 LNCaP cells treated with silvestrol. UNT PTL PU-H71 ABT Ara-C The Myc oncogene contributes to the genesis of many human 33 Figure 2. (a) Percent survival of leukemia cells treated with different cancersandappearstohavearoleinLSCself-renewal. We translation inhibitors (40 nM rocaglamide and 2.5 mg/ml temsiroli- observed a strong decrease in Myc protein levels following mus). (b) Met-label experiment comparing translation after treat- rocaglamide treatment (Figure 3a), consistent with our RNA-seq ment with indicated compounds (40 nM rocaglamide and 2.5 mg/ml analysis where 485% (33 out of 38) of Myc-regulated genes were temsirolimus). Graph represents percent survival of (c) M9-ENL inhibited (Supplementary Figure S3A). Importantly, the RNA-seq data leukemia cells and (d) CD34 primary leukemia cells (N ¼ 4) treated and subsequent quantitative polymerase chain reaction analyses with indicated drugs: PTL ¼ 2.5 mM parthenolide; PU-H71 ¼ 2 mM PU-H71; (data not shown) showed no significant decreases in Myc mRNA ABT737 ¼ 25 nM ABT-737; DNR ¼ 25 nM daunorubicin. Solid bars following rocaglamide treatment, indicating that the decreases in represent percent survival of compounds given individually. Striped the level of Myc protein are the result of post-transcriptional events. and gray bars represent the percent survival when compounds were given following 24 h pretreatment with rocaglamide or temsirolimus, Many of these downregulated genes controlled by Myc are respectively. implicated in cell cycle progression, and thus we colabeled M9 cells with anti-Ki67 and 7-AAD to evaluate cell cycle status after rocaglamide treatment. Rocaglamide decreased the percentage of Figure S2G). It should be noted that not all drugs synergized with actively dividing cells, while increasing the percentage of cells in G0 rocaglamide; in fact, the proteasome inhibitor bortezomib was (Figure3dandSupplementaryFigureS3B).Totesttheimportanceof antagonistic with rocaglamide, presumably because it slowed the Myc inhibition for rocaglamide-mediated cell death, we generated turnover of short-lived antiapoptotic proteins (Supplementary M9-ENL cells in which Myc cDNA was ectopically expressed from a Figure S2F). lentiviral promoter, which resulted in a significant increase in Myc To further explore drug synergies, we examined several protein levels (Figure 3e). Despite elevated levels of Myc, combinations in primary human AML CD34 þ cell populations rocaglamide was able to reduce Myc protein and induce apoptosis in vitro (Figure 2d). Importantly, rocaglamide enhanced the to an even greater extent than cells transduced with an empty cytotoxicity of cytosine arabinoside, a mainstay of AML therapeu- vector (Figures 3e and f). The increased sensitivity of the Myc- tic regimens. In addition, rocaglamide enhanced the activity of expressing cells to rocaglamide is consistent with the synthetic lethal several experimental agents: (1) ABT-737, a Bcl-2 inhibitor,30 relationship between Myc and the eIF4F complex.34 We also tested (2) PU-H71, a new class of heat-shock protein 90 inhibitor31 and the sensitivity of the transduced M9-ENL cells to the Myc inhibitor (3) parthenolide, a potent inducer of reactive oxygen species.32 JQ1 and observed that, like rocaglamide, the Myc-expressing cells Importantly, ABT-737 and parthenolide, like rocaglamide, show were more sensitive to JQ1. The similar toxicity profiles of JQ1 and selectivity toward the LSC population as single agents, indicating rocaglamide, as well as the ability of rocaglamide to reduce the that their combination holds promise for targeting more primitive forced expression of the Myc protein, indicate that Myc inhibition is leukemia cells.18,20 Taken together, the data indicate that a component of the rocaglamide mechanism.

& 2014 Macmillan Publishers Limited Leukemia (2014) 1960 – 1968 Flavaglines target LSCs KP Callahan et al 1964

AML #1 AML #2 AML #3 Roca - + - + - +

P-P53 Vector shP53-A sh-p53-B shP53-C (Ser 15) Total P53 Bax P21 Myc Actin

Actin

M9-ENL cells 100 30 26

80 20 Vector 13 60 11 shP53-A 10 40 % Viable

shP53-E % Cells in G2/M 20 shP53-H 0 UNT 20nM 40nM 0 0 204060 Roca (nM)

M9-ENL cells 100 PLUX PLUX- Myc 80

60 PLVX UNT Roca UNT Roca 40 PLVX-MYC MYC % Viable 20 GAPDH 0 0 10203040 Roca (nM)

Figure 3. (a)Whole-cell lysates were generated from primary AML cells treated with 40 nM rocaglamide for 24 h for western blot analysis. (b) Protein expression of M9-ENL cells infected with empty vector and shRNA targeting p53. (c) Graph represents percent survival of cells treated with indicated concentrations of rocaglamide. (d) Cell cycle analysis of M9 cells treated with indicated amounts of rocaglamide. (e) Protein expression of M9 cells infected with empty vector and Myc expressing vector. (f) Graph represents percent survival of M9 cells treated with indicated amounts of rocagalmide.

Rocaglamide has a greater effect on mitochondrial integrity and effects rocaglamide drug combinations have on mitochondria Myc levels than temsirolimus events. Using the fluorescent dye Mitosox, which detects To further examine the relevance of Myc reduction, we compared mitochondrial-specific superoxide, we observed significant protein levels between rocaglamide- and temsirolimus-treated increases in superoxide levels in cells treated with rocaglamide cells. Rocaglamide more effectively reduced Myc, consistent with a and ABT (Figure 4c), but not for cells treated with either agent role for this protein in cell death (Figure 4a). individually (Figure 4c). Interestingly, the combination had no We next investigated the possibility that the mechanism of effect on reactive oxygen species as measured by the fluorescent rocaglamide may extend beyond translation inhibition and that dye CM-H2DCFDA (Supplementary Figure S4D). CM-H2DCFDA these additional activities are not shared by temsirolimus. provides a more broad measure of intracellular redox levels than Rocaglamide affects mitochondrial integrity (Supplementary Mitosox. These results indicate that the disruption of mitochon- Figure S1F), and synergizes with the Bcl-2 inhibitor ABT-737 to drial integrity/function is critical to the mechanism of rocagla- a much greater degree than temsirolimus (Figure 2c). Thus, we mide. Having already established a role for p53 activation in the compared the extent of mitochondrial depolarization induced by rocaglamide-induced cell death (Figures 3a and c), we asked if the rocaglamide and temsirolimus as single agents, as well as in combination with ABT-737. The amount of depolarization impaired mitochondrial function could be explained by p53 induced by rocaglamide is more than that of temsirolimus activation. Using our p53-deficient M9-ENL cells, we observed a (Figure 4b and Supplementary Figure S4B). Strikingly, the significant reduction in mitochondrial depolarization (and combination of rocaglamide and ABT-737 shows a significant increased viability, data not shown) induced by the combination increase in depolarization compared with ABT-737 alone or in of rocaglamide and ABT-737, compared with an empty vector combination with temsirolimus. In fact, the degree of depolar- control (Figure 4d) This result indicates that p53 activation is ization is similar to levels induced by the well-characterized critical to the mitochondrial defects induced by rocaglamide and mitochondrial depolarizer, CCCP (Figure 4b and Supplementary thus its ultimate mechanism of action, and it represents properties Figures S4B and C). These results led us to further evaluate the distinct from other translational inhibitors such as temsirolimus.

Leukemia (2014) 1960 – 1968 & 2014 Macmillan Publishers Limited Flavaglines target LSCs KP Callahan et al 1965 AML #1 AML #2 AML #3 UNT Roca Tem UNT Roca Tem UNT Roca Tem

MYC

Actin

TMRE Staining Tem Roca ABT Roca+ABT Tem+ABT

Drug Treated Untreated 2%1.85 5%4.43 2%1.36 29%28.7 3%2.47

Mitosox Staining

Roca ABT Roca+ABT

Drug Treated Untreated

TMRE Staining

Vector Only p53-shRNA

61% 9%

Figure 4. (a) Whole-cell lysates were generated from primary AML cells treated with 40 nM rocaglamide for 24 h for western blot analysis. Blots were probed with antibodies against either c-MYC or glyceraldehyde 3-phosphate dehydrogenase (GAPDH). (b) M9-ENL cells were treated with indicated drugs, and depolarized mitochondria was determined by flow cytometry following staining with TMRE dye. (c) M9-ENL cells were incubated with indicated drugs, followed by labeling with the fluorescent dye Mitosox, which detects superoxide, and analyzed by flow cytometry. (d) M9-ENL cells infected with empty vector or shRNA targeting p53 were treated with rocaglamide and ABT-737 and stained with TMRE.

The flavagline compound silvestrol reduces leukemia burden protein synthesis inhibition and a subsequent reduction in Myc in vivo protein levels, is also similar (Supplementary Figures S5E and F). Although several labs have successfully synthesized rocagla- Thus, from a mechanistic standpoint, silvestrol appears to be mide,35–40 it is still difficult to obtain rocaglamide in quantities extremely similar to rocaglamide, inducing the same molecular necessary for in vivo studies. Thus, we chose to evaluate the in vivo events and changes in cell physiology. activity of silvestrol, a closely related member of the flavagline Next, we sought to determine the efficacy of silvestrol in family (Figure 5a), which we have successfully isolated in gram reducing tumor burden in an established human–mouse xeno- quantities from the plant source.4,11,41,42 Silvestrol has previously graft. NSG mice were sublethally irradiated and injected with demonstrated in vivo toxicity toward chronic lymphocytic primary human AML cells from three independent specimens. leukemia, acute lymphoblastic leukemia and mantle cell Beginning at 3 weeks after transplant, mice were treated with lymphoma cells.7,10 We first compared the in vitro toxicity of 0.75 mg/kg silvestrol 5 days a week for 3 weeks, after which tumor silvestrol to rocaglamide using several primary AML samples and burden, as a function of percent bone marrow engraftment, was found that both compounds produced similar dose curves, determined (Figure 5b). Silvestrol significantly reduced engraft- with silvestrol showing a slight increased potency (B2-fold) ment for all three specimens (Po0.05), with minimal deleterious (Supplementary Figure S5A). Further, as observed for rocaglamide, effects on the mice as measured by viability of the marrow and silvestrol showed minimal toxicity toward normal hematopoietic weight loss during treatment (Supplementary Figure S5G). In cells, including primitive populations (Supplementary Figures S5B addition, we were able to more specifically evaluate in vivo activity and C). Next, we compared the ability of silvestrol to synergize of silvestrol for specimen no. 1, which had a distinct CD33 þ , with the collection of anticancer drugs that we previously tested CD3 þ leukemic population. As shown in Supplementary Figure with rocaglamide. Similar to rocaglamide, silvestrol significantly S5H, analysis of the CD33 þ /CD3 þ cells following treatment enhanced the toxicity of all the drugs tested (Supplementary showed a preferential reduction in this aberrant clone in Figure S5D). In addition to analogous cytotoxicity profiles, the comparison with other cells in the graft that may represent a molecular response induced by both compounds, including mixture of normal and leukemic cells. Thus, the overall decrease of

& 2014 Macmillan Publishers Limited Leukemia (2014) 1960 – 1968 Flavaglines target LSCs KP Callahan et al 1966 rocaglamide silvestrol

AML #1 AML #2 AML #3 80 30 60 70 P<0.05 50 60 P<0.005 20 50 40 40 30 30 P<0.001 10 20 % Human Cells % Human Cells 20 % Human Cells 10 10 0 0 0 Vehicle Silvestrol Vehicle Silvestrol Vehicle Silvestrol Figure 5. (a) Chemical structure of rocaglamide and silvestrol. (b) Primary AML cells were injected into sublethally irradiated mice. Three weeks after transplant, mice were either treated with silvestrol or vehicle for 3 weeks. For AML nos. 1 and 2, silvestrol was administered at 0.75 mg/kg daily, ﬁve times a week for 3 weeks. Silvestrol was administered three times a week for 3 weeks at a dose of 1.5 mg/kg for mouse engrafted with AML no. 3. Mice were killed and the bone marrow was analyzed for the percentage of human cells using anti-human CD45 antibody.

CD45 þ cells may under-represent the degree of tumor reduction, treatment with rocaglamide, as compared with temsirolimus. An which may be upwards of 10-fold (Supplementary Figure S5I). inability to reduce Myc levels has been reported for other Taken together, the data clearly indicate in vivo activity of mammalian target of rapamycin inhibitors in AML cells and may silvestrol toward primary human AML populations. represent a limitation to this class of compounds for treating AML.17 The importance of Myc activity in the biogenesis and progression of cancer has long been appreciated; however, up DISCUSSION until recently; the ability to target Myc remained elusive.45–47 Numerous lines of investigation have demonstrated that cellular Silvestrol, JQ1 and other BET bromodomain inhibitors currently in heterogeneity for leukemia is evident at multiple levels, including development present viable therapeutic options for targeting this immunophenotype, morphology, cell cycle status, growth poten- nexus. tial and response to chemotherapy.43,44 Thus, in order for a drug In addition to their potency as single agents, we also show that to provide durable remissions, it must be capable of targeting rocaglamide and silvestrol increase the cytotoxicity of several multiple cell types, each having distinct properties. Furthermore, types of drugs toward AML cells, including daunorubicin and drugs that target LSC, while sparing normal HSC, are of particular cytosine arabinoside, two chemotherapeutic drugs currently used interest. Results presented here indicate that drugs of the to treat AML patients. The broad range of drug classes that, flavagline class (rocaglamide and silvestrol) represent promising rocaglamide and silvestrol, are able to enhance is quite striking. It candidates for the development of improved AML regimens. They is possible that these compounds are blocking translation-based are toxic to total AML cells, as well as AML stem cells, as assayed survival mechanisms induced by the second drug; however, one by engraftment of NSG mice, whereas normal hematopoietic cells, would expect all translational inhibitors to synergize with these including HSCs, are largely unaffected under the same conditions. compounds, which is not the case (Figure 2c). We propose that the Further, in vivo experiments show that silvestrol is able to reduce ability of rocaglamide to synergize is related, at least in part, to the the tumor burden of mice engrafted with primary AML cells, with effect it has on mitochondrial integrity/function (Figure 4b). The minimal deleterious effects on the mice. mitochondrial defect induced by rocaglamide treatment in In comparison with temsirolimus, rocaglamide and silvestrol are essence ‘primes’ cells for death when the second drug hits. Using significantly more toxic to leukemia cells when given as single M9-ENL cells harboring shRNA constructs targeting p53, we were agents or in combination with other drugs. Interestingly, this able link the mitochondrial defects with p53 activation (Figure 4d), increased cytotoxicity does not directly correlate with the ability of indicating a critical role for this node in the mechanism of action the compounds to inhibit translation as temsirolimus inhibits of flavaglines. translation at levels equal to or greater than rocaglamide. This The deregulation of translation is a critical feature of cancer observation led us to further investigate the molecular mechanism initiation and progression, and thus there is great interest in of cell death induced by rocaglamide. Using several assays, we targeting protein synthesis therapeutically.48,49 Data presented showed that rocaglamide leads to p53 activation, Myc reduction, here and in other labs have shown that targeting the translational cell cycle inhibition and defects in mitochondrial integrity. Each of initiation complex with flavaglines can be preferentially toxic to these activities appears to contribute to the overall activity of transformed cells.50,51 Silvestrol is the first direct inhibitor of rocaglamide. translation initiation with clinical potential. It is now under The observation that rocaglamide and temsirolimus inhibit development at the National Cancer Institute’s NExT Program, global protein synthesis to a similar extent does not eliminate the with current efforts dedicated to toxicology and scale-up using possibility that the compounds differentially affect the translation supercritical fluid extraction. Adequate raw materials for rates of a subset of proteins critical for the survival of leukemia preclinical toxicology studies have already been collected by the cells. We looked at Myc levels following drug treatment and source country, and Ohio State University has received an observed a striking difference in the level of reduction following exclusive license to develop silvestrol for the treatment of

Leukemia (2014) 1960 – 1968 & 2014 Macmillan Publishers Limited Flavaglines target LSCs KP Callahan et al 1967 cancer. We believe that the therapeutic flexibility, in vivo efficacy 18 Guzman ML, Rossi RM, Karnischky L, Li X, Peterson DR, Howard DS et al. and ability to target LSCs pose intriguing possibilities for these The sesquiterpene lactone parthenolide induces apoptosis of human acute compounds to serve as adjuvants to current therapeutic regimes. myelogenous leukemia stem and progenitor cells. Blood 2005; 105: 4163–4169. 19 Lagadinou ED, Ziros PG, Tsopra OA, Dimas K, Kokkinou D, Thanopoulou E et al. c-Jun N-terminal kinase activation failure is a new mechanism of anthracycline CONFLICT OF INTEREST resistance in acute myeloid leukemia. Leukemia 2008; 22: 1899–1908. 20 Lagadinou ED, Sach A, Callahan K, Rossi RM, Neering SJ, Minhajuddin M et al. The authors declare no conflict of interests. BCL-2 inhibition targets oxidative phosphorylation and selectively eradicates quiescent human leukemia stem cells. Cell Stem Cell 2013; 12: 329–341. 21 Bordeleau ME, Robert F, Gerard B, Lindqvist L, Chen SM, Wendel HG et al. Ther- ACKNOWLEDGEMENTS apeutic suppression of translation initiation modulates chemosensitivity in a This work was supported by the Leukemia and Lymphoma Society (CTJ, 6230-11), mouse lymphoma model. J Clin Invest 2008; 118: 2651–2660. Department of Defense (CTJ, W81XWH-07-1-0601), NY State Stem Cell Foundation 22 Lindqvist LM, Vikstrom I, Chambers JM, McArthur K, Ann Anderson M, Henley KJ et al. (CTJ, C024964), National Institutes of Health (AF, RO1 GM079364), (ADK, PO1 Translation inhibitors induce cell death by multiple mechanisms and Mcl-1 reduction CA125066) and a Wilmot-Roswell Park Collaborative Seed Grant. KPC was supported is only a minor contributor. Cell Death Dis 2012; 3:e409. by American Cancer Society (PF-10-054-01-LIB) and Wilmot Cancer Center 23 Kim S, Hwang BY, Su BN, Chai H, Mi Q, Kinghorn AD et al. Silvestrol, a potential postdoctoral fellowships. anticancer rocaglate derivative from Aglaia foveolata, induces apoptosis in LNCaP cells through the mitochondrial/apoptosome pathway without activation of executioner caspase-3 or -7. Anticancer Res 2007; 27: 2175–2183. AUTHOR CONTRIBUTIONS 24 Lindqvist L, Pelletier J. Inhibitors of translation initiation as cancer therapeutics. KPC designed research, performed experiment, and wrote the paper; CTJ Fut Med Chem 2009; 1: 1709–1722. 25 Recher C, Beyne-Rauzy O, Demur C, Chicanne G, Dos Santos C, Mas VM et al. designed research and wrote the paper; MM, CC, EDL, RMR, MBB, LP, SJ, AF, Antileukemic activity of rapamycin in acute myeloid leukemia. Blood 2005; 105: MRG, ADK, JLL, MWB performed experiments and analyzed data. 2527–2534. 26 Assouline S, Culjkovic B, Cocolakis E, Rousseau C, Beslu N, Amri A et al. Molecular targeting of the oncogene eIF4E in acute myeloid leukemia (AML): REFERENCES a proof-of-principle clinical trial with ribavirin. Blood 2009; 114: 257–260. 1 Jordan CT, Guzman ML, Noble M. Cancer stem cells. N Engl J Med 2006; 355: 27 Chan S, Scheulen ME, Johnston S, Mross K, Cardoso F, Dittrich C et al. Phase II 1253–1261. study of temsirolimus (CCI-779), a novel inhibitor of mTOR, in heavily pretreated 2 Wang JC, Dick JE. Cancer stem cells: lessons from leukemia. Trends Cell Biol 2005; patients with locally advanced or metastatic breast cancer. J Clin Oncol 2005; 23: 15: 494–501. 5314–5322. 3 Becker MW, Jordan CT. Leukemia stem cells in 2010: current understanding and 28 Staehler M, Haseke N, Khoder W, Stief CG. Profile of temsirolimus in the treatment future directions. Blood Rev 2011; 25: 75–81. of advanced renal cell carcinoma. OncoTargets Ther 2010; 3: 191–196. 4 Lucas DM, Still PC, Perez LB, Grever MR, Kinghorn AD. Potential of plant-derived 29 Kentsis A, Topisirovic I, Culjkovic B, Shao L, Borden KL. Ribavirin suppresses natural products in the treatment of leukemia and lymphoma. Current Drug eIF4E-mediated oncogenic transformation by physical mimicry of the 7-methyl Targets 2010; 11: 812–822. guanosine mRNA cap. Proc Natl Acad Sci USA 2004; 101: 18105–18110. 5 Kim S, Salim AA, Swanson SM, Kinghorn AD. Potential of cyclopenta[b]benzo- 30 Oltersdorf T, Elmore SW, Shoemaker AR, Armstrong RC, Augeri DJ, Belli BA et al. furans from Aglaia species in cancer chemotherapy. Anti-cancer Agents Med Chem An inhibitor of Bcl-2 family proteins induces regression of solid tumours. Nature 2006; 6: 319–345. 2005; 435: 677–681. 6 Zhu JY, Lavrik IN, Mahlknecht U, Giaisi M, Proksch P, Krammer PH et al. The 31 Chiosis G, Kang Y, Sun W. Discovery and development of purine-scaffold Hsp90 traditional Chinese herbal compound rocaglamide preferentially induces apop- inhibitors. Expert Opin Drug Discov 2008; 3: 99–114. tosis in leukemia cells by modulation of mitogen-activated protein kinase activ- 32 Wen J, You KR, Lee SY, Song CH, Kim DG. Oxidative stress-mediated apoptosis. ities. Int J Cancer 2007; 121: 1839–1846. The anticancer effect of the sesquiterpene lactone parthenolide. J Biol Chem 2002; 7 Lucas DM, Edwards RB, Lozanski G, West DA, Shin JD, Vargo MA et al. The novel 277: 38954–38964. plant-derived agent silvestrol has B-cell selective activity in chronic lymphocytic 33 Kim J, Woo AJ, Chu J, Snow JW, Fujiwara Y, Kim CG et al. A Myc network accounts leukemia and acute lymphoblastic leukemia in vitro and in vivo. Blood 2009; 113: for similarities between embryonic stem and cancer cell transcription programs. 4656–4666. Cell 2010; 143: 313–324. 8 Zhu JY, Giaisi M, Kohler R, Muller WW, Muhleisen A, Proksch P et al. Rocaglamide 34 Lin CJ, Nasr Z, Premsrirut PK, Porco Jr JA, Hippo Y, Lowe SW et al. Targeting sensitizes leukemic T cells to activation-induced cell death by differential synthetic lethal interactions between Myc and the eIF4F complex impedes regulation of CD95L and c-FLIP expression. Cell Death Differ 2009; 16: 1289–1299. tumorigenesis. Cell Rep 2012; 1: 325–333. 9 Bleumink M, Kohler R, Giaisi M, Proksch P, Krammer PH, Li-Weber M. Rocaglamide 35 Malona JA, Cariou K, Frontier AJ. Nazarov cyclization initiated by peracid breaks TRAIL resistance in HTLV-1-associated adult T-cell leukemia/lymphoma by oxidation: the total synthesis of ( þ / À )-rocaglamide. JAmChemSoc2009; 131: translational suppression of c-FLIP expression. Cell Death Differ 2011; 18: 362–370. 7560–7561. 10 Alinari L, Prince CJ, Edwards RB, Towns WH, Mani R, Lehman A et al. Dual targeting 36 Gerard B, Sangji S, O’Leary DJ, Porco Jr. JA. Enantioselective photocycloaddition of the cyclin/Rb/E2F and mitochondrial pathways in mantle cell lymphoma with mediated by chiral Bronsted acids: asymmetric synthesis of the rocaglamides. the translation inhibitor silvestrol. Clin Cancer Res 2012; 18: 4600–4611. J Am Chem Soc 2006; 128: 7754–7755. 11 Hwang BY, Su BN, Chai H, Mi Q, Kardono LB, Afriastini JJ et al. Silvestrol and 37 Davey AE, Schaeffer MJ, Taylor RJK. Synthesis of the Novel antileukemic tetra- episilvestrol, potential anticancer rocaglate derivatives from Aglaia silvestris. JOrg hydrocyclopenta[b]benzofuran, rocaglamide. J Chem Soc Chem Commun 1991 Chem 2004; 69: 3350–3358. 1137–1139. 12 Malina A, Cencic R, Pelletier J. Targeting translation dependence in cancer. 38 Dobler MR, Bruce I, Cederbaum F, Cooke NG, Diorazio LJ, Hall RG et al. Total Oncotarget 2011; 2: 76–88. synthesis of ( þ / À )-rocaglamide and some aryl analogues. Tetrahedron Lett 2001; 13 Cencic R, Carrier M, Galicia-Vazquez G, Bordeleau ME, Sukarieh R, Bourdeau A et al. 42: 8281–8284. Antitumor activity and mechanism of action of the cyclopenta[b]benzofuran, 39 Hongsen L, Fu B, Wang MA, Li N, Liu WJ, Xie ZQ et al. Total synthesis and silvestrol. PLoS One 2009; 4: e5223. biological activity of ( þ / À )-rocaglamide and its 2,3-di-epi analogue. Eur J Org 14 Chambers JM, Lindqvist LM, Webb A, Huang DC, Savage GP, Rizzacasa MA. Chem 2008 1753–1758. Synthesis of biotinylated episilvestrol: highly selective targeting of the translation 40 Trost BM, Greenspan PD, Yang BV, Saulnier MG. An unusual oxidative factors eIF4AI/II. Org Lett 2013; 15: 1406–1409. cyclization—a synthesis and absolute stereochemical assignment of ( À )-roca- 15 Sadlish H, Galicia-Vazquez G, Paris CG, Aust T, Bhullar B, Chang L et al. Evidence glamide. J Am Chem Soc 1990; 112: 9022–9024. for a functionally relevant rocaglamide binding Site on the eIF4A-RNA complex. 41 Saradhi UV, Gupta SV, Chiu M, Wang J, Ling Y, Liu Z et al. Characterization of ACS Chem Biol 2013; 8: 1519–1527. silvestrol pharmacokinetics in mice using liquid chromatography-tandem mass 16 Grzmil M, Hemmings BA. Translation regulation as a therapeutic target in cancer. spectrometry. AAPS J 2011; 13: 347–356. Cancer Res 2012; 72: 3891–3900. 42 Pan L, Kardono LB, Riswan S, Chai H, Carcache de Blanco EJ, Pannell CM et al. 17 Tamburini J, Green AS, Bardet V, Chapuis N, Park S, Willems L et al. Protein Isolation and characterization of minor analogues of silvestrol and other synthesis is resistant to rapamycin and constitutes a promising therapeutic target constituents from a large-scale re-collection of Aglaia foveolata. J Nat Prod 2010; in acute myeloid leukemia. Blood 2009; 114: 1618–1627. 73: 1873–1878.

& 2014 Macmillan Publishers Limited Leukemia (2014) 1960 – 1968 Flavaglines target LSCs KP Callahan et al 1968 43 Jordan CT. Unique molecular and cellular features of acute myelogenous leuke- 48 Silvera D, Formenti SC, Schneider RJ. Translational control in cancer. Nat Rev mia stem cells. Leukemia 2002; 16: 559–562. Cancer 2010; 10: 254–266. 44 Dick JE. Acute myeloid leukemia stem cells. Ann NY Acad Sci 2005; 1044: 1–5. 49 Stumpf CR, Ruggero D. The cancerous translation apparatus. Curr Opin Genet Dev 45 Delmore JE, Issa GC, Lemieux ME, Rahl PB, Shi J, Jacobs HM et al. BET bromodomain 2011; 21: 474–483. inhibition as a therapeutic strategy to target c-Myc. Cell 2011; 146:904–917. 50 Graff JR, Konicek BW, Vincent TM, Lynch RL, Monteith D, Weir SN et al. Therapeutic 46 Mertz JA, Conery AR, Bryant BM, Sandy P, Balasubramanian S, Mele DA et al. suppression of translation initiation factor eIF4E expression reduces tumor growth Targeting MYC dependence in cancer by inhibiting BET bromodomains. Proc Natl without toxicity. J Clin Invest 2007; 117: 2638–2648. Acad Sci USA 2011; 108: 16669–16674. 51 Moerke NJ, Aktas H, Chen H, Cantel S, Reibarkh MY, Fahmy A et al. Small-molecule 47 Zuber J, Shi J, Wang E, Rappaport AR, Herrmann H, Sison EA et al. RNAi screen identiﬁes inhibition of the interaction between the translation initiation factors eIF4E and Brd4 as a therapeutic target in acute myeloid leukaemia. Nature 2011; 478: 524–528. eIF4G. Cell 2007; 128: 257–267.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu)

Leukemia (2014) 1960 – 1968 & 2014 Macmillan Publishers Limited