inhibiting the oncogenic translation program is an effective … › content › scitransmed › 9...

SC I ENCE TRANS LAT IONAL MED I C I N E | R E S EARCH ART I C L E

CANCER

1Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA 02215, USA.2Department of Hematology, Lille Hospital, 59000 Lille, France. 3INSERM UMR-S1172, University of Lille 2, 59000 Lille, France. 4Boston University Center for Mo-lecular Discovery, Boston, MA 02215, USA. 5Department of Hematology, UniversityHospital of Poitiers, 86021 Poitiers, France. 6Whitehead Institute for BiomedicalResearch, Cambridge, MA 02142, USA.*These authors contributed equally to this work.†Corresponding author. Email: [email protected] (I.M.G.); [email protected] (S.M.)

Manier et al., Sci. Transl. Med. 9, eaal2668 (2017) 10 May 2017

2017 © The Authors,

some rights reserved;

exclusive licensee

American Association

for the Advancement

of Science.

Dow

nloaded fr

Inhibiting the oncogenic translation program is aneffective therapeutic strategy in multiple myelomaSalomon Manier,1,2,3*† Daisy Huynh,1* Yu J. Shen,1 Jia Zhou,1 Timur Yusufzai,1 Karma Z. Salem,1

Richard Y. Ebright,1 Jiantao Shi,1 Jihye Park,1 Siobhan V. Glavey,1 William G. Devine,4

Chia-Jen Liu,1 Xavier Leleu,5 Bruno Quesnel,3 Catherine Roche-Lestienne,3 John K. Snyder,4

Lauren E. Brown,4 Nathanael Gray,1 James Bradner,1 Luke Whitesell,6

John A. Porco Jr.,4 Irene M. Ghobrial1†

Multiple myeloma (MM) is a frequently incurable hematological cancer in which overactivity of MYC plays acentral role, notably through up-regulation of ribosome biogenesis and translation. To better understand theoncogenic program driven by MYC and investigate its potential as a therapeutic target, we screened a chemicallydiverse small-molecule library for anti-MM activity. The most potent hits identified were rocaglate scaffold inhibitorsof translation initiation. Expression profiling of MM cells revealed reversion of the oncogenic MYC-driven tran-scriptional program by CMLD010509, the most promising rocaglate. Proteome-wide reversion correlated withselective depletion of short-lived proteins that are key to MM growth and survival, most notably MYC, MDM2,CCND1, MAF, and MCL-1. The efficacy of CMLD010509 in mouse models of MM confirmed the therapeuticrelevance of these findings in vivo and supports the feasibility of targeting the oncogenic MYC-driven translationprogram in MM with rocaglates.

hom
by guest on July 23, 2020
ttp://stm.sciencem

ag.org/

INTRODUCTIONMultiple myeloma (MM) is a hematological malignancy characterizedby a clonal proliferation of plasma cells in the bone marrow micro-environment (1). According to the National Cancer Institute (NCI),the prevalence of MM was estimated at 89,650 people in the UnitedStates in 2012, with an annual incidence of 6.3 new cases per 100,000individuals and a 5-year overall survival (OS) of 46.6% (2). Despite no-table therapeutic advances, MM remains a frequently incurable hema-tological malignancy. Therefore, there is an urgent unmet need for thedevelopment of therapeutic optionswith newmechanisms of action inMM.

The transcription factorMYC plays a central role in the progressionof the disease. About two-thirds of patients who are newly diagnosedwithMMharborMYC activation, which correlates with adverse clinicaloutcome (3). In about 20 to 40% of the patients, this is explained bytranslocations that involve theMYC locus (4, 5). Another 10% of pa-tients harbor a gain of 8q24 comprising MYC (6). MYC is centrallypositioned in cell growth and cancer regulatory networks. MYC hetero-dimerizes withMAX, to bind the E-box element CACGTG. Functionalcategories of MYC‐induced genes include cell growth (ribosomebiogenesis and protein synthesis), cell cycle control, energy production(glycolysis, glutaminolysis, and mitochondrial biogenesis), anabolicmetabolism (synthesis of amino acids, nucleotides, and lipids), andDNA replication (7).

Alterations in ribosome synthesis and translation activity are adefining feature of most cancer cells, satisfying the increased anabolicdemands associated with malignant transformation and tumor growth.MYC directly enhances transcription genes encoding ribosomal pro-

teins, ribosomal RNA, and proteins of the translation initiation complex.The ribosome and translation initiation complexes have historicallybeen perceived as passive complexes, supporting protein synthesis with-out having an active role in directing functional expression of thegenome. However, recent studies have revealed that ribosome biogenesisand translation activity are highly regulated to drive specific translationprograms, independent of the geneticmakeup of the cancer (8–10), andrepresent a crucial node for hyperactivation of oncogenic downstreamsignaling in cancer cells (11). Accordingly, ribosome biogenesis andtranslation initiation potentially provide very attractive targets forcancer therapy.

Here, we sought to identify chemical compounds that disrupt ribo-some biogenesis and translation activation in MM. We also describeanalogs of the rocaglate natural product class as very potent compounds,which inhibit the oncogenic translation program supporting MM.

RESULTSMYC and translation activation in MMTo examine potential correlation between MYC expression andtranslation activation in MM, we analyzed the Cancer Cell Line En-cyclopedia (CCLE), a large database of gene expression profiling formore than 1000 human cancer cell lines. We first generated a Z scorefor each cell line by combining two KEGG (Kyoto Encyclopedia ofGenes and Genomes) canonical pathway gene sets: ribosomal biogenesisand translation. We found a significant correlation between MYCexpression and translation activation across the ~1000 CCLE cell lines(R = 0.478, P < 0.001) (Fig. 1A). In particular, many MM cell lines hadincreasedMYC expression and an enrichment of ribosomal biogenesisand translation (high Z score), which was also observed in other hem-atological malignancies (fig. S1). We used MM patient tumor cell–derived gene expression profiling (GSE6477 and GSE16558) to furtherconfirm the enrichment of translation (Fig. 1B) and ribosomal biogenesis(Fig. 1C) in the context of highMYC expression. These results establish astrong correlationbetweenMYC expression and translation activity inMM.

1 of 13

http://stm.sciencemag.org/


by guest on July 23, 2020http://stm

.sciencemag.org/

Dow

nloaded from

Identification of rocaglate derivativesTo identify a chemical biological probe with which to perturb lymphoidmalignancies in the context of MYC overexpression, we screened adiverse small-molecule library (2812 compounds) provided by theBoston University Center for Molecular Discovery (BU-CMD). TheBU-CMD screening collection is composed largely of compounds


synthesized using diversity-oriented syn-thesis techniques, as well as syntheticanalogs of bioactive natural productscaffolds, assembled with the goal of ac-cessing greater structural complexityand a broader coverage of chemical spacethan is possible with conventionalcombinatorial chemistry–based libraries(12). Two lymphoid cell lines with highexpression of MYC were used: NCI-H929(MM) and NAMALWA (Burkitt lympho-ma). We identified 45 compounds, whichpotently inhibited proliferation of at leastone of the cell lines (Fig. 2A and fig. S2).We validated these 45 hits in five MM celllines and one Burkitt lymphoma cell line(Fig. 2B). On the basis of their structure-activity relationships (SARs), this secondaryvalidation highlighted three compounds(CMLD010331, CMLD010332, andCMLD009433), all totally synthetic ana-logs of the rocaglate natural product class(fig. S2B). These three compounds werethe most active inhibiting proliferationof MM cells to a level similar to that ofbortezomib (Fig. 2B).

Rocaglate (flavagline) natural pro-ducts (and synthetic analogs) are rec-ognized to inhibit protein synthesis (13, 14).To better define the activity of rocaglatederivatives against MM cells, we per-formed a follow-up SAR study involving40 additional structurally related com-pounds (Fig. 3A and fig. S2C). Theseincluded synthetic samples of bothnatural and nonnatural rocaglates. Ofthe set, the compound CMLD010509[SDS-1-021 (13, 15)] was identified asthe most potent compound, with amedian inhibitory concentration (IC50)below 10 nM for most MM cell linestested (Fig. 3, B and C). The compoundwas less effective against lung and breastcancer cell lines, with an IC50 of ~30 nM(Fig. 3D).

Gene expression altered byCMLD010509 inhibiting translationin MMTo further define the mode of action forCMLD010509 againstMM, we performedRNA sequencing (RNA-seq) on drug-versus dimethyl sulfoxide (DMSO)–treated

cells in five differentMMcell lines.We identified 845 and 475 genes thatwere significantly up-regulated (table S1A) and down-regulated (tableS1B), respectively, with a fold change higher than 2 and a P value of<0.05 (Fig. 4A). Several zinc finger transcription factors were amongthe top up-regulated genes, whereas ribosomal proteins and heat shockproteins were among the most significantly down-regulated (Fig. 4A).

A

B

P = 0.0008FDR = 0.011

TRANSLATION

0

0.5

MYC high MYC low

KEGG_RIBOSOME

ES

P = 0.01FDR = 0.009

TRANSLATION

0

0.6

MYC high MYC low

ES

CKEGG_RIBOSOME

P = 0.005FDR = 0.015

0

0.7

MYC high MYC low

P = 0.02FDR = 0.03

0

0.8

MYC high MYC low

6

8

10

12

14

16

−10 −5 0 5 10

MY

C e

xpre

ssio

n

Ribosomal biogenesis and translation Enrichment scoreZ

CCLE

Cancer linesMM lines

R = 0.478P < 0.0001

Fig. 1. Correlation of MYC expression with ribosomal biogenesis and translation activity. (A) Correlation anal-ysis of MYC expression on the y axis and the Z-score enrichment across over 1000 cell lines from the CCLE database.A Z score was generated for each cell line by combining two KEGG canonical pathway gene sets: ribosomalbiogenesis and translation. Red dots indicate MM cell lines, and black circles indicate all CCLE cell lines except MM. Asignificant correlation between MYC expression and the translation activation was observed; R = 0.478, P < 0.0001. GeneSet Enrichment Analysis (GSEA) of MM patient tumor cell-derived gene expression profiling (GSE6477 and GSE16558)shown in (B) and (C), confirming enrichment of translation and ribosomal biogenesis in context of high expression ofMYC. ES, enrichment score; FDR, false discovery rate.

2 of 13




.sciencemag.org/

Dow

nloaded from

Furthermore, we defined four clusters by arranging the significantlyaltered gene sets in a network enrichment map (16). The transcriptionand posttranslational clusters were enriched in CMLD010509-treatedcells, whereas oxidative phosphorylation and translation clusters wereenriched in control cells, indicating down-regulation in CMLD010509-treated cells (Fig. 4B and table S1C). Themost significantly up-regulatedgenes were highly enriched for transcriptional activation, and the mostsignificantly down-regulated genes were highly enriched for ribosomecomponents and translation (Fig. 4C and table S1, D and E). This tran-scriptome profile has been associated with low translational flux andconsequent inactivation of heat shock factor 1 (HSF1) (17). Here, inhibi-tionof translation reverses constitutiveHSF1 activation in cancers, therebyleading to extensive remodeling of the transcriptome. We confirmed that


CMLD010509 inhibits the heat shock response in MM cells, as indi-cated by decreased HSP70 and HSP90 mRNA and down-regulationof the HSF1 program in CMLD010509-treated cells by pathway analysis(fig. S3A). To assess whether HSF1 is constitutively activated in MM,we used a previously published HSF1 activation signature (18) to inter-rogate publicly available data sets. We observed that this HSF1 activa-tion signaturewas increased inMM cells compared to normal plasmacells (19, 20) and correlatedwith poor clinical outcome inMM(fig. S3, Band C) (20, 21).

To further characterize the link between CMLD010509 andtranslation inhibition, we used our CMLD010509 expression signa-ture to query the Library of Integrated Network–Based Cellular Sig-natures (LINCS) National Institutes of Health (NIH) program

0

20

40

60

80

100

120

140

0 1000 2000 3000Number of compounds

NAMALWA

10331103329433

A

0

20

40

60

80

100

120

0 10 20 30 40 50Number of compounds

U266

9433

1033110332

Bortezomib

0

20

40

60

80

100

120


Bortezomib 9433 1033110332

0

20

40

60

80

100

120


Bortezomib9433

1033110332

0

20

40

60

80

100

120

0 10 20 30 40 50

Rel

ativ

e su

rviv

al

Number of compounds

Bortezomib 943310331

10332

0

20

40

60

80

100

120

0 10 20 30 40 50

Rel

ativ

e su

rviv

al

Number of compounds

NCI-H929

Bortezomib

94331033110332

0

20

40

60

80

100

120


NAMALWA

Bortezomib

94331033110332

B

0

20

40

60

80

100

120

140

0 1000 2000 3000

Rel

ativ

e s

urv

ival

Number of compounds

NCI-H929

1033110332

9433

Fig. 2. Chemical library screen. (A) Small-molecule library generated by the Boston University Center (~3000 synthetic compounds) against NCI-H929 (MM) andNAMALWA (Burkitt lymphoma). We identified 45 compounds with a potent inhibition of proliferation for at least one of the cell lines. (B) Validation of 45 hits insix cell lines harboring diverse MM driver genomic features of MM, showing that the most effective compounds to inhibit proliferation were in the rocaglate class,namely, CMLD010331, CMLD010332, and CMLD09433 (red dots). These compounds had similar potency to bortezomib.

3 of 13




.sciencemag.org/

Dow

nloaded from

(www.lincscloud.org). The LINCS database is a large catalog of per-turbational gene expression profiles across 77 cell lines. It includes over20,000 chemical compounds and more than 22,000 genetic perturba-tions (knockdown and overexpression of genes). The most significantpositive correlations with the CMLD010509 signature were representedby known translation inhibitors such as emetine, omacetaxine, andcephaeline, as well as knockdown of ribosome subunits and translationinitiation factors (Fig. 4D and table S1F). These results are consistentwith the potent activity of CMLD010509 as a translation inhibitor.

MYC knockdown was one of the perturbations with the most pos-itive correlation with our compound signature. The GSEA of our com-pounds’ effects using the RNA-seq data revealed that severalMYC genesets were among the most significantly enriched compared to the


DMSO control cells, indicating a down-regulation of the entire MYCpathway in CMLD010509-treated cells (Fig. 4E). Several MYC targetgenes were down-regulated by the drug; however, the amounts ofMYC transcript were not decreased in CMLD010509-treated cells,indicating posttranscriptional regulation of MYC activity (Fig. 4F).Together, these results reinforce the role ofMYC in translation acti-vation in MM and the mechanistic action of CMLD010509 as atranslation inhibitor in MM cells.

Disruption of the oncogenic translation program in MMby CMLD010509Tobetter define the consequences ofCMLD010509-induced translationinhibition, we performed an unbiased, proteome-wide experiment in

A

10 µµM 100 nM

10515105361051210514105101053510582105061050910517103311051810501

756410483

75659433

10531105071051610332105421058310527104261048410584105341058510533105251052810523105221058710499105191058610520104931076210498

Bortezomib 1007550250

% Survival

BNCI-H929

125

100

75

50

25

00 0.001 0.003 0.01 0.03 0.1 0.3 1 3 10

Dose (µM)

Cel

l via

bili

ty (

%)

C

CMLD010509

D10509 nM:

100 60 20 0

MM1R

KMS-18

MM1S

NCI-H929

OPM-2

U266

RPMI8226

NCIH-1650

KMM-1

NCIH-23

NCIH-1473

CAL51

ZR-751

MDAMB-231

NCIH-1750

Fig. 3. High potency of rocaglate inhibitors in MM. (A) Heat map of the effects of 40 rocaglate derivatives on relative survival of MM cells (NCI-H929, KMS-18, MM1R,MM1S, OPM-2, RPMI8226, and U266) along with lymphoma cell lines (NAMALWA and MEC1) shows strong activity with low doses of the rocaglate compounds in thesecell lines. (B) IC50 of these 40 compounds in NCI-H929 showing that CMLD010509 was the most potent compound, with an IC50 below 10 nM. (C) CMLD010509 is asynthetic rocaglate derivative containing the cyclopenta[b]tetrahydrobenzofuran core structure. (D) IC50 of CMLD010509 showing an IC50 below 10 nM for most MM celllines tested (indicated in red), whereas it was relatively resistant in lung and breast cancer cell lines (in black) with an IC50 of ~30 nM.

4 of 13

http://www.lincscloud.org




.sciencemag.org/

Dow

nloaded from

a quantitative and highly parallel format (22). We compared the im-mediate impact of CMLD010509 treatment (100 nM) to that of vehiclecontrol in NCI-H929 cells. Exposure for 2 hours was selected to identify


primary, immediate consequences of compound action and to miti-gate expected, confounding effects of suppressed CMLD010509 targetproteins. Each treatment condition was prepared in three biological

A

0

1

2

3

4

5

6

7

8

–4 –2 0 2 4

−Lo

g10

(P

val

ue)

Log2 (FC)

0 20 40 60

REGULATION_OF_TRANSCRIPTION

REGULATION_OF_RNA_METABOLIC…

REGULATION_OF_NUCLEOBASENU…

REGULATION_OF_TRANSCRIPTION…

TRANSCRIPTION

TRANSCRIPTION_DNA_DEPENDENT

RNA_BIOSYNTHETIC_PROCESS

NUCLEOBASENUCLEOSIDENUCLEO…

RNA_METABOLIC_PROCESS

BIOPOLYMER_METABOLIC_PROCESS

−Log10 (q value)

Up-regulated genes against MSigDB C5

0 20 40 60

MACROMOLECULE_BIOSYNTHETIC_PROCESS

BIOSYNTHETIC_PROCESS

CELLULAR_BIOSYNTHETIC_PROCESS

TRANSLATION

STRUCTURAL_MOLECULE_ACTIVITY

RNA_BINDING

PROTEIN_METABOLIC_PROCESS

CELLULAR_MACROMOLECULE_METABOLIC_P…

CELLULAR_PROTEIN_METABOLIC_PROCESS

STRUCTURAL_CONSTITUENT_OF_RIBOSOME

−Log10 (q value)

Down-regulated genes against MSigDB C5

C

Transcription Posttranslationalmodification

TranslationOxidative

phosphorylation

Down Up10509-treated H929

vs DMSO

Enriched in

B

D

ZNF619

ZNF434

ZNF148FRYL

PGAM1

HSPA8

RPL27A

HSP90AA1

–150

–100

–50

0

50

100

150

1 10 100 1000 10,000

Mea

n c

on

nec

tivi

ty s

core

Number of perturbagen

Translation inhibitors

Ribosome subunits KDa

MYC KDa

E

F

0

1

2

3

DM

SO

-no

rmal

ized

MY

C

tran

scri

pt

MYCWT 10509

MESSEN_MYC

ES

0

−0.8

10509 DMSO

P < 0.001FDR = 0.024−0.7

DANG_MYC

ES

0

10509 DMSO

P < 0.001FDR = 0.022

HALLMARK_MYC

ES

0

10509 DMSO

P < 0.001FDR = 0.019

SCHUMACHER_MYC

ES

0

−0.7

10509 DMSO

P = 0.014FDR = 0.032−0.7

Fig. 4. Transcription activation and translation inhibition induced by the rocaglate derivative CMLD010509. (A) Volcano plot of RNA-seq of drug- versus DMSO-treated cells in five different MM cell lines showing 845 and 475 genes significantly up- and down-regulated, respectively, with a fold change (FC) higher than 2. (B) Networkenrichment map identifying two clusters enriched in CMLD010509-treated cells (transcription and posttranslational modification clusters) and two clusters enriched incontrol cells (oxidative phosphorylation and translation clusters). (C) Bar graphs showing the most significantly up- and down-regulated genes in CMLD010509-treatedcells determined by querying the Molecular Signatures Database (MSigDB) C5 gene sets. (D) Connectivity score using LINCS NIH program with the gene signature ofCMLD010509 against 10,000 “pertubagen” signatures (corresponding to short hairpin RNA, open reading frame, and compounds). (E) GSEA analysis of CMLD010509using RNA-seq data. Several MYC gene sets were among the most significantly enriched compared to DMSO control cells. (F) Bar graphs of MYC transcript expression inMM cell lines and NAMALWA in CMLD010509-treated cells, normalized to DMSO control. WT, wild type; KD, knockdown.

5 of 13



replicates and individually labeled using isobaric tagging. We identi-fied 7312 proteins, of which 54 were significantly depleted by morethan twofold after CMLD010509 exposure (P < 0.05; fold change, >2)(Fig. 5A and table S2A). Several key oncoproteins in MM were amongthe proteins most depleted by the compound, including MYC, MDM2,CCND1, MCL-1, and MAF. MYC is overexpressed in about 40% ofMM by either translocations or gain of 8q24, among other potentialepigenetic regulations (4, 5). CCND1 is overexpressed in about 20%


of MM in the case of t(11;14) or focal gain in 11q13 (4, 23).MDM2 isan ubiquitin ligase (E3) that acts as a negative regulator of p53 (24).MCL-1 belongs to the Bcl-2 family and is one of the driver oncogenesin gain-of-1q cancers (25).MAF is frequently overexpressed in MMin relation to t(14;16) (26).

To further study the mechanism of action of CMLD010509, wecompared the protein fold changes to the transcript fold changes inCMLD010509- versus vehicle-treated cells (Fig. 5B). The large majority


.sciencemag.org/

Dow

nloaded from

A

MDM2

MCL-1

0.0625

0.125

0.25

0.5

1

2

0.00

0000

1

0.00

0001

0.00

001

0.00

01

0.00

1

0.010.

11

Fo

ld c

han

ge

P value

MYC

CCND1

MAF

7312 quantified proteins B

0.0625

0.125

0.25

0.5

1

2

4

0.12

5

0.25 0.

5 1 2 4 8

Pro

tein

s fo

ld c

han

ge

mRNA fold change

Transcription induction

Translation induction

Translation inhibition

Transcription inhibition

MYCMDM2CCND1

MCL-1 MAF

MDM2

Tubulin

MCL-1MYC CCND1 MAFC

DMSO 10509 DMSO 10509DMSO 10509 DMSO 10509 DMSO 10509

0

1

2

3

DMSO 10509

Rel

ativ

e tr

ansc

rip

t le

vel

MYC

0

1

2

3

DMSO 10509

MDM2D

0

1

2

3

4

DMSO 10509

MCL-1

0

1

2

DMSO 10509

CCND1

0

1

2

DMSO 10509

MAF

Fig. 5. Regulation of the oncogenic translation program of MM by CMLD010509. (A) Expression proteomics showing the impact of CMLD010509 treatment (100 nM)relative to vehicle control in NCI-H929 cells after a 2-hour incubation (three biological replicates were used and individually tagged using isobaric tagging). We identified7312 proteins, of which 54 were significantly depleted by more than twofold through CMLD010509 exposure (P < 0.05; fold change, >2). The most depleted proteinsincluded MYC, MDM2, CCND1, MCL-1, and MAF (red dots). (B) Comparison of the protein fold changes to the transcript fold changes in CMLD010509- versus vehicle-treatedcells. A large majority of depleted proteins had a transcript fold change lower than twofold, suggesting a specific translational mechanism of action for CMLD010509.(C) Immunoblots and (D) qRT-PCR analysis of MYC, MDM2, CCND1, MCL-1, and MAF in compound-treated (100 nM, 3 hours) or vehicle-treated NCI-H929. All five geneswere depleted at the protein level, whereas their transcripts were unchanged or slightly overexpressed.

6 of 13



Dow

n

of depleted proteins had a change in transcript amount of less thantwofold, indicative of a selective translation inhibitory mode of actionfor CMLD010509. To validate these findings, we measured MYC,MDM2, CCND1, MCL-1, andMAF by immunoblotting and quantita-tive real-time polymerase chain reaction (qRT-PCR) after compoundtreatment as above (100 nM, 2 hours). All five genes were depleted atthe protein level, whereas the amounts of each transcript were un-changed or slightly increased (Fig. 5, C and D), consistent with post-transcriptional effects of CMLD010509. Similar results were observed insix additional MM cell lines (fig. S4, A to G). To assess whether thetargeted proteins were directly depleted by inhibiting their synthesis,we pulse-labeled cells with 35S-cysteine/methionine and immuno-precipitated the proteins of interest. In the presence of CMLD010509,we observed selective inhibition of new translation forMYC, CCND1, andMCL-1 as opposed to housekeeping proteins such as glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (fig. S4H).

To further establish pathways affected by CMLD010509, we queriedthe 54 proteins with more than twofold depletion against the KEGGdatabase. Cancer pathways were highly enriched among these depleted


proteins, representing a distinct cluster by network enrichment analysis(Fig. 6A and table S2B). Moreover, when interrogating the wholeMSigDB C2 canonical pathway database (representing 1330 gene sets),we found that the most highly enriched pathways consisted of MMpathways: IRF4 andMYC targets. This result indicates that a significantnumber of CMLD010509-depleted proteins are present in the IRF4 andMYC pathways (P < 0.001), as defined by the GSEA transcriptionalpathways. IRF4 is a key transcription factor essential for plasma celldifferentiation, upon which MM cells are dependent on (Fig. 6B andtable S2C) (27), suggesting that CMLD010509 inhibits translation ofkey oncoproteins in MM. To determine whether this translationalprogram is specific for MM, we used the signature of 54 proteins to cal-culate Z scores for over 1000 cell lines across the CCLE database. TheCMLD010509 signature was enriched inMM cell lines at the transcrip-tional level compared to other cancer cell lines (Fig. 6C). Furthermore,the signature was also significantly enriched in plasma cells of patientswith MM compared to that of healthy donors (P < 0.05), in twoindependent gene expression profiling experiments including patients atdiagnosis (GSE16558) and at relapse (GSE6477) (Fig. 6D). Together, these


.sciencemag.org/

loaded from

MAPK pathway

JAK/STAT pathway

Colorectal cancer

Melanoma

Thyroid cancer

Pathways in cancer

AML

Lung cancer

Lung small cells cancer

Pancreatic cancer

Endometrial cancerBladder cancer

CML

Prostate cancer

Glioma

Down-regulated proteins against KEGG pathways

FDR < 0.25

A

0 2 4 6 8

SHAFFER_IRF4_TARGETS_IN_MYELOMA

HELLER_HDAC_TARGETS_SILENCED_BY

SHAFFER_IRF4_TARGETS_IN_PLASMA_CE

SHAFFER_IRF4_MULTIPLE_MYELOMA_PRO

BOYLAN_MULTIPLE_MYELOMA_PCA1_UP

BAKER_HEMATOPOIESIS_STAT3_TARGETS

KIM_WT1_TARGETS_12HR_DN

NAGASHIMA_EGF_SIGNALING_UP

BERENJENO_TRANSFORMED_BY_RHOA_UP

DANG_BOUND_BY_MYC

−Log10 (P value)

Down-regulated proteins against MSigDB C2

B

C DGSE16558GSE6477

–2

–4

0

2

Enr

ichm

ent Z

sco

re

–2.5

0

2.5

P = 0.004 P = 0.03

Fig. 6. Pathway analysis of CMLD010509. (A) Network enrichment analysis of 54 depleted proteins with more than twofold change analyzed in the KEGG database,showing that cancer pathways were highly enriched among these depleted proteins. AML, acute myeloid leukemia; CML, chronic myeloid leukemia; MAPK, mitogen-activatedprotein kinase; JAK/STAT, Janus kinase/signal transducer and activator of transcription. (B) The 54 depleted proteins were queried against MSigDB C2 canonical pathwaydatabase, showing an enrichment of MM specific oncoproteins including IRF4 pathway. (C) Z scores of the 54 proteins evaluated on 1000 cell lines of the CCLE database. (D) Zscores of the 54 proteins evaluated on two independent gene expression profiles including patients at diagnosis (GSE16558) and at relapse (GSE6477).

7 of 13




.sciencemag.org/

Dow

nloaded from

results demonstrate that CMLD010509 in-hibits the translation of key oncoproteins,supporting the proliferation and survivalof MM.

Induction of apoptosis in MM cellsby CMLD010509We next explored the antiproliferativeconsequences of inhibiting the oncogenictranslation program in MM cells withCMLD010509. The drug was associatedwith a strong induction of an apoptoticresponse in NCI-H929 and MM1S cells,as measured by caspase-3 and caspase-7activation (Fig. 7A) and cleavage of bothpoly[adenosine 5′-diphosphate–ribose]polymerase (PARP) and caspase-3 (Fig.7B). Kinetic studies revealed a rapidapoptotic response to CMLD010509 at3 hours, associated with depletion ofMYC and MCL-1 even at low concentra-tions (10 nM) (Fig. 7C). We next assessedthe in vitro activity of CMLD010509against human peripheral blood mono-nuclear cells (PBMCs) and found no cyto-toxicity, suggesting the potential for auseful therapeutic window for this agentagainst MM (fig. S5).

In vivo suppression of MMby CMLD010509On the basis of the mechanism of actionof CMLD010509 in MM, we assessedwhether the drug would be effective andtolerable in animal models. We firstestablished a xenograft model by intra-venous injection of MM1S cells expressingluciferase (luc) into severe combined im-munodeficient (SCID)mice. After engraft-ment, themice were randomized to vehicleand compound treatment groups, withequal amounts of bioluminescence imaging(BLI) intensity in eachgroup.CMLD010509or vehicle control was administered via in-traperitoneal injection at 0.7 mg/kg twice aweek (Fig. 8A). CMLD010509 caused amarked reduction in tumor burden despitethe very rapid growth of this aggressiveMM cell line, as monitored by BLI andprolonged survival (median OS, 35 versus47 days; P < 0.001) (Fig. 8, B and C). No-tably, 4 weeks of CMLD010509 treatmentwere well tolerated with preservation of
body weight and normal complete blood counts (fig. S6, A to C). Toconfirm the in vivo efficacy of CMDL010509, we used a second xeno-graft model consisting of SCID mice inoculated intravenously withKMS-11 cells. The mice were treated with intraperitoneal injectionsof either CMLD010509 at 0.7 mg/kg or vehicle control twice a week,starting at week 4, and were sacrificed at week 8. Immunohistochemistry

studies were conducted on bone marrow samples to assess the presenceof CD138+ tumor cells as well as amounts of MYC and Ki-67 protein. Adecrease inCD138+plasmacells, depletionofMYC, and inhibitionofpro-liferation (Ki-67 staining) were observed in mice treated withCMLD010509 compared to those treated with vehicle control (Fig. 8D),further validating the activity of CMLD010509 on MM cell lines in vivo.

A

B

C

0

1

2

3

4

5

6

7

8

0 10 100

1000

Ap

op

tosi

s fo

ld c

han

ge

(Cas

pas

e-G

lo 3

/7)

NCI-H929–10509 (nM)

0 10 100

1000

3 hours 24 hours

0123456789

0 10 100

1000

MM1S–10509 (nM)

0 10 100

1000

3 hours 24 hours

NCI-H929–10509 (nM) MM1S–10509 (nM)

Full Caspase 3

Full PARP

GAPDH

Cleaved Caspase 3

Cleaved PARP

0 10 100

1000

0 10 100

1000

0 10 100

1000

0 10 100

1000

3 hours 24 hours 3 hours 24 hours

GAPDH

MCL-1

MYC

3 hours 24 hours 3 hours 24 hours

0 10 100

1000

0 10 100

1000

0 10 100

1000

0 10 100

1000

NCI-H929–10509 (nM) MM1S–10509 (nM)

Fig. 7. Induction of apoptosis by CMLD010509. (A) Apoptosis assay measured by caspase-3 and caspase-7activation (Caspase-Glo 3/7) at 3 and 24 hours in NCI-H929 and MM1S cells treated with three concentrationsof CMLD010509 versus vehicle control. (B) Immunoblots for cleavage of both PARP and caspase-3 in response to threeconcentrations of CMLD010509 in NCI-H929 and MM1S cells at 3 and 24 hours. (C) Immunoblots for MYC and MCL-1 inresponse to three concentrations of CMLD010509 in NCI-H929 and MM1S cells at 3 and 24 hours.

8 of 13




.sciencemag.org/

Dow

nloaded from

Finally, to complete the assessment of CMLD010509’s effects onMYC-driven MM cells in vivo, we used an immunocompetent mousemodel, in which transplantable mouse MM cells were injected intoC57BL/6 mice. The tumor cells are driven by plasma cell–specific


MYC overexpression (Vk*Myc cells) (28).Micewere treatedwith eitherCMLD010509 at 0.7 mg/kg or vehicle control intraperitoneally twicea week, beginning 1 week after cell tumor inoculation. Treatment waswell tolerated, as indicated by normal transthyretin and albumin serum

0 15 32Study day:Survival follow-upBLI BLI BLI

DMSO or 105090.7 mg/kg i.p.

Cell injection

Randomization

A

P < 0.001

10509DMSO control

15 d

ays

25 d

ays

32 d

ays

B

C

D E

F

10509DMSO control

P = 0.0019

10509DMSO control

CD138

MYC

Ki-67

Per

cen

t su

rviv

al

Per

cen

t su

rviv

alB

iolu

min

esce

nce

co

un

t

0

50

100

100806040200Time (days)

Time (days)

DMSO

10509

DMSO10509

DMSO10509

00

50

100

10 20 30 40 50 60

Time (days)15 20 25 30 35

0

2 × 107

4 × 107

6 × 107

8 × 107

P < 0.001

P < 0.001

Fig. 8. In vivo suppression of MM by CMLD010509. (A) Diagram of in vivo studies performed. SCID mice were injected with MM1S GFP (green fluorescent protein)/Luc+. After engraftment, the mice were randomized to two groups based on BLI, and CMLD010509 or vehicle control was administered by intraperitoneal (i.p.) injec-
tions at 0.7 mg/kg two times a week. Tumor growth was assessed by weekly bioluminescence (BLI) counts, (B) showing a significant difference in tumor growthbetween the DMSO-treated group and the CMLD010509 group; P < 0.001. (C) Kaplan-Meier curve showing a significant survival difference between vehicle- anddrug-treated mice, with a median OS of 35 versus 47 days, P < 0.001. (D) Immunohistochemistry for CD138, Ki-67, and MYC in bone marrow samples from theKMS-11–injected mice. Scale bars, 50 mm. (E) Electrophoresis of serum proteins from transplantable Vk*Myc mice indicating the presence of a g-globulin (M-spike)in the DMSO control group (arrow) at 4 weeks. C57BL/6 syngeneic mice were intravenously injected with 3.5 × 106 Vk*Myc cells. At day 7 after tumor cells injection,mice were randomized into two groups and received DMSO (n = 10) or CMLD010509 drug (n = 10, 0.7 mg/kg per mouse) through intraperitoneal injections twice aweek. (F) Kaplan-Meier curve showing a significant survival difference between vehicle- and drug-treated mice, with a median OS of 39 days versus not reached, P = 0.0019.Statistical analyses were performed using log-rank test.
9 of 13



concentrations at 6 weeks (fig. S6D). Serum electrophoresis proteins(SPEPs) revealed a robust decrease ofM-spike in CMLD010509-treatedmice as shown at week 4 (Fig. 8E), which was associated with a signif-icantly prolonged survival—median survival of 39 days in the controlgroup versus not reached in the CMLD010509-treated group at thisend of the experiment (90 days) (P = 0.0019) (Fig. 8F). Together, thesedata indicate that CMLD010509 is highly active and well tolerated invivo in several mouse models of MM.


.sciencemag.org/

Dow

nloaded from

DISCUSSIONThe oncogenic translation program of tumors is supported by aberrantactivation of ribosome biogenesis and dysregulation of the translationinitiation machinery (8–10). Here, we report the correlation betweenMYC activation and the translation of an oncogenic program in MM,which can be inhibited by the rocaglate derivative CMLD010509. Weidentified the rocaglate derivative CMLD010509 (SDS-1-021) as a high-ly specific inhibitor of the oncogenic translation program supportingMM—including key oncoproteins such as MYC, MDM2, CCND1,MAF, and MCL-1.

Although ribosome biogenesis and translation initiation complexfunction have historically been perceived as passive processes, recentstudies have revealed that they are actively regulated to support specifictranslation programs (8, 9). Here, using a proteome-wide approach, wereport that targeting translation initiation in MM selectively depletesmultiple oncoproteins while leaving other housekeeping proteinslargely unperturbed. The selective inhibition of translation initiationfor oncogenic mRNA in the context of cancer can be explained byseveral mechanisms. First, some messengers are translated in a cap-dependent manner, whereas others are translated through the internalribosome entry site process. Cap-dependent translation relies on theability of the eukaryotic translation initiation factor 4F (eIF4F) complexto bind to the 5′ 7-methylguanosine cap present on mature mRNAs(29). Moreover, the presence of complex secondary structures withinthe 5′ untranslated region (5′UTR) influences mRNA translation.Transcripts with complex 5′UTR secondary structures are subjectto greater dependence on the initiation machinery and thereforemight be more sensitive to its inhibition. Subsequent studies haveshown that the limiting factor is often eIF4E, mainly through its abilityto recruit the eIF4A helicase (30–33). Complex 5′UTR structures havebeen identified in several prooncogenicmRNAs such as those encodingMYC, MDM2, and cyclins (10).

Rocaglate natural products, including silvestrol and rocaglamide A,comprise a class of protein synthesis inhibitors (14, 17) that interactwiththe eIF4A—a protein subunit of the translation initiation complex(14, 34). These compounds increase the affinity between eIF4A andmRNA and specifically clamp eIF4A onto polypurine sequences inmRNA 5′UTR, thus blocking ribosome subunit scanning and reducingprotein expression from transcripts bearing the rocaglamide A–eIF4Atarget sequence (35). The compound CMLD010509 is a potent andselective translation inhibitor through an eIF4E phosphorylation–independent mechanism (15). Here, we demonstrate that CMLD010509inhibits translation of a key suite of oncoproteins in MM that constitutean oncogenic translation program supporting MM oncogenesis. More-over, this compound is highly effective in vivo in several mouse models,suggesting an effective therapeutic approach toward targeting MYCin MM.

The present work was aimed to assess the translational potential ofrocaglates inMM.Wedid not focus on the specific relationship between


MYC and translation activation inMM.Moreover, the specific effect ofCMLD010509 on the translational oncogenic program inMMcould befurther explored to fully determine its mechanism of action.

In summary, we present an indirect but highly effective mechanismfor targeting an oncogenic translation program frequently driven byMYC in MM. Even with the most recent advances in the managementof plasma cell dyscrasias, MM remains a frequently incurable disease.Therefore, there is an urgent need for additional treatment options withnew modes of activity. As shown here, rocaglates inhibit a specifictranslation oncogenic program related to high expression ofMYC, witha very potent activity in vitro and in vivo. Thus, targeting dysregulatedtranslation initiation with rocaglates rather than targeting the elonga-tion machinery with other translation inhibitors might be less toxic tonormal tissues. Several characteristics of CMLD010509 support its fur-ther preclinical development. First, the compound can bemade by totalsynthesis from simple starting materials, which avoids reliance onlimited bioresources (36). Second, its mode of action suggests a simplestrategy for therapeutic drug monitoring to minimize toxicity in pa-tients based on routinely available measurements of short-lived serumproteins such as prealbumin (transthyretin) or factor VII. Finally,successful clinical development might be improved by use of the onco-genic signature we have proposed as a potential companion diagnosticto identify patients most likely to benefit from rocaglate treatment. Thetargeting of protein degradation inMMwith proteasome inhibitors hasbecome a mainstay in the management of this disease. On the basis ofthe preclinical data presented here, we suggest that targeting proteinhomeostasis in the opposite way, through selective inhibition of pro-tein synthesis, is worthy of further development and may provide anequally effective approach in the management of MM and possiblyother hematological malignancies.

MATERIALS AND METHODSStudy designThe objectives of this study were as follows: (i) to identify compoundsable to interfere withMYC activation in MM, (ii) to assess the mech-anistic role of rocaglate compounds in MM, and (iii) to determinethe preclinical efficacy of CMLD010509 as an effective treatmentagainst MM. CellTiter-Glo was used to evaluate the efficacy of asmall-molecule screen against two cell lines with high expression ofMYC, to validate the best hits against six cell lines, and to generateIC50 values of rocaglate compounds. Functional evaluation ofCMLD010509 was performed by RNA-seq of five MM cell linesand tandem mass tag (TMT) mass spectrometry using NCI-H929exposed to CMLD010509 or DMSO in both cases. Further, qRT-PCRand Western blotting were performed to validate the data. We usedGSEA and the LINCS cloud databases for pathway analyses. The effectof CMLD010509 in vitro was assessed on MM cell lines by Caspase-Glo 3/7 andWestern blotting for caspase-3 and PARP. To determinethe preclinical efficacy of CMLD010509, we used three differentmouse models. The first one was a xenograft model where MM1Sluc/GFP cells were injected intravenously into SCIDmice. After 2weeksof engraftment, tumor formation was monitored by BLI. To allocateanimals to experimental groups, we measured BLI using Living Image4.0 software and randomized mice to obtain equal amounts of BLIintensity in each group. Mice were treated with either CMLD010509or DMSO and monitored weekly for BLI, blood count, and bodyweight and for survival. The second xenograftmodel consisted of inject-ing KMS-11 cells into SCID mice and treating them with either

10 of 13




.sciencemag.org/

Dow

nloaded from

CMLD010509 or DMSO.Mice were all sacrificed at week 6 and studiedfor immunohistochemistry of the bonemarrowwithCD138,MYC, andKi-67 antibodies. Finally, we developed an immunocompetent mousemodel by injecting Vk*Myc cells into C57BL/6 mice. After 1 week ofengraftment, mice were randomized into two groups and treated witheither CMLD010509 or DMSO. Mice were monitored for SPEP andsurvival. The experiment was terminated after 3 months of follow-up.

Cell lines and primary cellsNCI-H929, NAMALWA, MM1S, MM1R, U266, and RPMI8226were purchased from American Type Culture Collection, OPM-2 wasfrom DSMZ, and KMS-11 and KMM-1 were from Japanese Collectionof Research Bioresources Cell Bank. MM1S luc/GFP cells were giftsfrom A. Kung (Dana-Farber Cancer Institute), and CAL51, NCI-1650,NCI-1750, NCI-23, DU4475, and ZR-751 cells were gifts from the JänneLab (Dana-Farber Cancer Institute). Vk*Myc cells used in this studywere isolated from late-stage Vk*Myc mice (gift from L. BergsagelandM. Chesi) and expanded in vivo. All cells were maintained under5% CO2 in a medium according to their specifications. PMBCs wereobtained from the peripheral blood of healthy volunteers by Ficollgradient centrifugation.

Small-molecule screenMM cell lines were treated with 2812 compounds provided by the BU-CMD (www.bu.edu/cmd). Amicroplate dispenser, theMatrixWellMate(Thermo Fisher Scientific), was used to dispense 2000 adherent cellsor 5000 suspension cells per well into 384-well microplates. Compoundswere added using Biomek FX pin tool (Beckman Coulter). After treat-ment, cytotoxicity wasmeasured byCellTiter-Glo (Promega) accordingto the manufacturer’s protocol, and luminescence signals were readusing EnSpire (PerkinElmer), a plate reader.

Cell viability assayRelative cell growth and survival weremeasured in a 96-well microplateformat by using the luminescent detection of CellTiter-Glo or Caspase-Glo as the end point. For IC50, 5000 adherent cells and 30,000 suspen-sion cells were plated and incubatedwith compounds for 72hours. Eachanalysis was performed three times.

Protein and RNA isolationHarvested cellswere lysedusing lysis buffer (Cell Signaling) supplementedwith 5mMNaF and 1mMphenylmethylsulfonyl fluoride for 30minon ice, and lysates were centrifuged at 14,000g for 30min. Protein con-centration was measured using Bio-Rad Protein Assay kit. Total RNAwas isolated from cells using the RNeasy Mini Kit (Qiagen) accordingto the manufacturer’s protocol and evaluated for quantity and qualityby NanoDrop spectrophotometer.

Quantitative real-time PCRFor qRT-PCR, 100 ng of total RNA per sample was used for reversetranscription using the SuperScript III First-Strand Synthesis Kit (Invi-trogen). qRT-PCR was done on a StepOnePlus Real-Time PCR System(Applied Biosystems) using SYBRGreen IMasterMix.Gene expressionwas normalized to 18S. The primer sequences are as follows: humanMYC, TCCGTCCTCGGATTCTCTGCTCT (forward) and GCCT-CCAGCAGAAGGTGATCCA (reverse); humanMDM2, CCCAAGA-CAAAGAAGAGAGTGTGG (forward) and CTGGGCAG-GGCTTATTCCTTTTCT (reverse); human CCND1, GCTGCGA-AGTGGAAACCATC (forward) and CCTCCTTCTGCACACATTT-


GAA (reverse); human MAF, CTGGCAATGAGCAACTCCGA(forward) and AGCCGGTCATCCAGTAGAGT (reverse); humanMCL-1, TGCTTCGGAAACTGGACATCA (forward) and TAGC-CACAAAGGCACCAAAAG (reverse); and human 18S, TCAA-CTTTCGATGGTAGTCGCCGT (forward) and TCCTTGGATGTG-GTAGCCGTTTCT (reverse).

RNA sequencingNCI-H929, NAMALWA, U266, MM1S, and OPM-2 cells were treatedwith 50 nM CMLD010509 or vehicle control for 6 hours. RNA wasprepared as mentioned earlier. A starting amount of 500 ng of RNAwas used to prepare polyadenylate-enriched, single bar–coded librariesusing theNEBNext Kit. Quality control of the libraries was evaluated byBioanalyzer analysis withHigh Sensitivity chips (Agilent Technologies).Sequencing was performed on a HiSeq 2500 (Illumina) by 2× 50–basepair paired-end reads at the Biopolymers Facility of Harvard MedicalSchool. We used Bcbio_nextgen (https://github.com/chapmanb/bcbio-nextgen/) to process the RNA-seq data. Briefly, cutadapt (https://github.com/marcelm/cutadapt/) was used to trim adapters; trimmedreads were aligned to human reference genome (GRCh37) with to-phat2; read count for each gene was calculated byHTSeq. Genes withlow expression (fragments per kilobasemillion of <1 across all samples)were filtered out. GSEA was used to identify significantly enrichedpathways (16), with an FDR of <0.25 and a P value of <0.05. Genesets were downloaded from the Broad Institute’s MSigDB (www.broadinstitute.org/gsea/index.jsp).

ImmunoblottingCell lysates (50 mg) were subjected to SDS–polyacrylamide gel electro-phoresis (PAGE), transferred to a polyvinylidene fluoride membrane,blocked with 5% nonfat dry milk, and incubated with primary anti-bodies overnight at 4°C. Primary antibodies used included anti-MYC,anti–MCL-1, anti-CCDN1 (all Cell Signaling), anti-MDM2, anti-MAF(all Santa Cruz Biotechnology), and anti-tubulin (Sigma-Aldrich).

TMT spectrometryNCI-H929 cells were treated in triplicate with CMLD010509 (100 nM)or vehicle control for 2 hours. Proteins were isolated and further pro-cessed for quantitative proteomic analysis by TMT (Thermo FisherScientific) as per the manufacturer’s protocol, with mass spectrometry.

35S labelingNCI-H929 and MM1S cells were preincubated in methionine-freeRPMI medium supplemented with dialyzed fetal bovine serum (FBS)and penicillin-streptomycin (Thermo Fisher Scientific) for 10 min.35S-methionine was added to the medium at a final concentrationof 100 mCi/ml (~9 mM methionine), and the cells were labeled for20 min at 37°C. An equal volume of complete RPMI medium(containing an excess of unlabeled methionine; ~50 mM final) wasadded, and the cells were incubated for an additional 10 min at 37°C.The cells were then collected by centrifugation and washed with cold1× phosphate-buffered saline. Whole-cell extracts were prepared byresuspending the cells in 1× Cell Lysis Buffer (Cell Signaling) supple-mented with Halt protease and phosphatase inhibitors (Thermo FisherScientific) followed by a 30-min incubation on ice. The extracts werecentrifuged for 15 min at 18,000g to remove the insoluble material.For the immunoprecipitation (IP) experiments, the whole-cell extractswere incubated with primary antibodies [anti-MYC (Abcam #ab32),anti-cyclin D1 (Abcam #ab134175), anti–MCL-1 (Cell Signaling

11 of 13

www.bu.edu/cmd

https://github.com/chapmanb/bcbio-nextgen/

https://github.com/chapmanb/bcbio-nextgen/

https://github.com/marcelm/cutadapt/

https://github.com/marcelm/cutadapt/

http://www.broadinstitute.org/gsea/index.jsp

http://www.broadinstitute.org/gsea/index.jsp




.sciencemag.org/

Dow

nloaded from

#94296), anti-GAPDH (Abcam #ab181602), or anti–control immuno-globulin G (Cell Signaling #3900)] overnight at 4°C with rotation.The following morning, 40 ml of DynaBeads Protein G (Invitrogen)was added, and the samples were rotated for an additional hour. Thebeads were separated using a magnetic rack and washed three timeswith Cell Lysis Buffer. The immunoprecipitated material was eluted in25 ml of 1× SDS sample buffer. The whole-cell extracts and the IPsamples were resolved by SDS-PAGE and transferred to a nitrocellulosemembrane (GE Healthcare). The membrane was stained with PonceauS to visualize total proteins in samples. The membrane was then air-dried and exposed to a phosphorimager screen to visualize the 35S-labeled proteins.

In vivo studiesSCID mice (n = 10 per group) used for xenograft experiments wereinjected intravenously with 5 × 106 MM1S GFP-Luc+ and monitoredby blood cell count, body weight, BLI, and survival. After engraftment,the mice were randomized into two groups on the basis of BLI, andCMLD010509 or vehicle control was administered by intraperitonealinjection at 0.7 mg/kg twice a week. Similar studies were performedin SCID mice (n = 10 per group) by injecting 7 × 106 KMS-11, butbecause these cells are not luc+, the mice were sacrificed at week 6,and the bonemarrow of femurs was subjected to immunohistochemistryas described below. For the immunocompetent mousemodel, C57BL/6mice were purchased from the Jackson Laboratory. The transplantableVk*MYC cell line, which was a gift from M. Chesi and P. L. Bergsagel(MayoClinic, Scottsdale, AZ), wasmaintained and expanded in vitrousingRPMI1640 containing10%FBS (Sigma-Aldrich), 2mML-glutamine,penicillin (100U/ml), and streptomycin (100mg/ml) (GIBCO).Vk*Myccells (3.5 × 106) were injected intravenously into C57BL/6mice, and theperipheral blood from the mice was collected for SPEPmeasurement atserial time points by submandibular bleeding. SPEP was performedusing the QuickGel Protein Kit (Helena Laboratories) according tothe manufacturer’s protocol. After 1 week of Vk*Myc cell injection,themice were randomized to therapy or vehicle control (intraperitonealinjection at 0.7 mg/kg twice a week, n = 10 mice per group).

Study approvalAll mice were treated, monitored, and sacrificed in accordance with anapproved protocol of the Dana-Farber Cancer Institute Animal Careand Use Committee.

ImmunohistochemistryMurine tissues were fixed overnight in 10% buffered formalin and storedin 70% ethanol before proceeding to paraffin-embedding, sectioning,and staining with hematoxylin and eosin. Immunohistochemistry wasperformed using anti-CD138, anti–Ki-67, and anti–c-Myc, as previouslyreported (37).

Statistical analysisUnpaired Student’s t test was used to compare two independent groups.One-way analysis of variance (ANOVA) was used when three or moreindependent groups were compared. For survival data, Kaplan-Meiercurves were plotted and compared using a log-rank test. All tests weretwo-sided. A P value of <0.05 was considered statistically significant.Pearson coefficient was used to assess the correlation between twogroups. Analysis was performed with GraphPad Prism Software(GraphPad Software Inc.). Mann-Whitney U test was used to quantifythe enrichment of a given signature in a ranked list of genes, which was


represented as a Z score. GSEA was performed as described previouslyusing GSEA v2.0.10 (www.broadinstitute.org/gsea/) (16). MSigDBcollections (http://software.broadinstitute.org/gsea/msigdb/index.jsp)and specific gene sets (KEGG Translation and Ribosome and severalMYC gene sets) were tested for their enrichment in data sets—MMpa-tients with high expression of MYC versus low expression or MM celllines treated with CMLD010509 versus DMSO. Investigate_GeneSets(http://software.broadinstitute.org/gsea/msigdb/annotate.jsp) was usedto investigate pathways enriched in a list of genes against MSigDBcollections. The analysis was performed by applying two-tailed Fisher’stest as implemented in the Investigate_GeneSets module at MSigDB.For network enrichment mapping, gene sets with significant enrich-ment in CMLD010509- or DMSO-treated cells by GSEA were selectedon the basis of P < 0.05 and FDR < 0.25 and visualized with Cytoscapev3.3.0 (www.cytoscape.org/cy3.html). This software organizes the sig-nificant gene sets into a network, where nodes correspond to gene sets,and the edges reflect significant overlap between the nodes according toFisher’s test. The size of the nodes is proportional to the number ofgenes in the gene set, as previously described (38).

SUPPLEMENTARY MATERIALSwww.sciencetranslationalmedicine.org/cgi/content/full/9/389/eaal2668/DC1Fig. S1. Correlation between MYC expression and translation activation in differenthematologic malignancies.Fig. S2. Three potent rocaglate derivatives identified by an initial drug screen of a small-molecule compound library.Fig. S3. Association of HSF1 activation with poor outcomes in MM.Fig. S4. RNA-seq and TMT proteomic data validation in several MM cell lines.Fig. S5. Assessment of CMLD010509 toxicity on PBMCs.Fig. S6. Evaluation of CMLD010509 toxicity in vivo.Table S1. RNA-seq of MM cells treated with CMLD010509 (provided as an Excel file).Table S2. TMT proteomic analysis of NCI-H929 treated with CMLD010509 (provided as an Excel file).

REFERENCES AND NOTES1. A. Palumbo, K. Anderson, Multiple myeloma. N. Engl. J. Med. 364, 1046–1060 (2011).2. N. Howlader, A.-M. Noone, M. Yu, K. A. Cronin, Use of imputed population-based cancer

registry data as a method of accounting for missing information: Application to estrogenreceptor status for breast cancer. Am. J. Epidemiol. 176, 347–356 (2012).

3. W.-J. Chng, G. F. Huang, T. H. Chung, S. B. Ng, N. Gonzalez-Paz, T. Troska-Price,G. Mulligan, M. Chesi, P. L. Bergsagel, R. Fonseca, Clinical and biological implications ofMYC activation: A common difference between MGUS and newly diagnosed multiplemyeloma. Leukemia 25, 1026–1035 (2011).

4. B. A. Walker, C. P. Wardell, A. Murison, E. M. Boyle, D. B. Begum, N. M. Dahir, P. Z. Proszek,L. Melchor, C. Pawlyn, M. F. Kaiser, D. C. Johnson, Y.-W. Qiang, J. R. Jones, D. A. Cairns,W. M. Gregory, R. G. Owen, G. Cook, M. T. Drayson, G. H. Jackson, F. E. Davies, G. J. Morgan,APOBEC family mutational signatures are associated with poor prognosis translocationsin multiple myeloma. Nat. Commun. 6, 6997 (2015).

5. M. Affer, M. Chesi, W. D. Chen, J. J. Keats, Y. N. Demchenko, K. Tamizhmani, V. M. Garbitt,D. L. Riggs, L. A. Brents, A. V. Roschke, S. Van Wier, R. Fonseca, P. L. Bergsagel, W. M. Kuehl,Promiscuous MYC locus rearrangements hijack enhancers but mostly super-enhancers todysregulate MYC expression in multiple myeloma. Leukemia 28, 1725–1735 (2014).

6. D. R. Carrasco, G. Tonon, Y. Huang, Y. Zhang, R. Sinha, B. Feng, J. P. Stewart, F. Zhan,D. Khatry, M. Protopopova, A. Protopopov, K. Sukhdeo, I. Hanamura, O. Stephens,B. Barlogie, K. C. Anderson, L. Chin, J. D. Shaughnessy Jr., C. Brennan, R. A. Depinho, High-resolution genomic profiles define distinct clinico-pathogenetic subgroups of multiplemyeloma patients. Cancer Cell 9, 313–325 (2006).

7. T. R. Kress, A. Sabò, B. Amati, MYC: Connecting selective transcriptional control to globalRNA production. Nat. Rev. Cancer 15, 593–607 (2015).

8. M. L. Truitt, D. Ruggero, New frontiers in translational control of the cancer genome. Nat.Rev. Cancer 16, 288–304 (2016).

9. M. L. Truitt, C. S. Conn, Z. Shi, X. Pang, T. Tokuyasu, A. M. Coady, Y. Seo, M. Barna,D. Ruggero, Differential requirements for eIF4E dose in normal development and cancer.Cell 162, 59–71 (2015).

10. M. Bhat, N. Robichaud, L. Hulea, N. Sonenberg, J. Pelletier, I. Topisirovic, Targeting thetranslation machinery in cancer. Nat. Rev. Drug Discov. 14, 261–278 (2015).

12 of 13

http://www.broadinstitute.org/gsea/

http://software.broadinstitute.org/gsea/msigdb/index.jsp

http://software.broadinstitute.org/gsea/msigdb/annotate.jsp

http://www.cytoscape.org/cy3.html

http://www.sciencetranslationalmedicine.org/cgi/content/full/9/389/eaal2668/DC1




.sciencemag.org/

Dow

nloaded from

11. C. Vogel, E. M. Marcotte, Insights into the regulation of protein abundance fromproteomic and transcriptomic analyses. Nat. Rev. Genet. 13, 227–232 (2012).

12. L. E. Brown, K. Chih-Chien Cheng, W.-G. Wei, P. Yuan, P. Dai, R. Trilles, F. Ni, J. Yuan,R. MacArthur, R. Guha, R. L. Johnson, X.-z. Su, M. M. Dominguez, J. K. Snyder, A. B. Beeler,S. E. Schaus, J. Inglese, J. A. Porco Jr., Discovery of new antimalarial chemotypes throughchemical methodology and library development. Proc. Natl. Acad. Sci. U.S.A. 108,6775–6780 (2011).

13. C. M. Rodrigo, R. Cencic, S. P. Roche, J. Pelletier, J. A. Porco Jr., Synthesis of rocaglamidehydroxamates and related compounds as eukaryotic translation inhibitors: Synthetic andbiological studies. J. Med. Chem. 55, 558–562 (2012).

14. M.-E. Bordeleau, F. Robert, B. Gerard, L. Lindqvist, S. M. H. Chen, H.-G. Wendel, B. Brem,H. Greger, S. W. Lowe, J. A. Porco Jr., J. Pelletier, Therapeutic suppression of translationinitiation modulates chemosensitivity in a mouse lymphoma model. J. Clin. Invest. 118,2651–2660 (2008).

15. J. Chu, R. Cencic, W. Wang, J. A. Porco Jr., J. Pelletier, Translation inhibition by rocaglates isindependent of eIF4E phosphorylation status. Mol. Cancer Ther. 15, 136–141 (2016).

16. A. Subramanian, P. Tamayo, V. K. Mootha, S. Mukherjee, B. L. Ebert, M. A. Gillette,A. Paulovich, S. L. Pomeroy, T. R. Golub, E. S. Lander, J. P. Mesirov, Gene set enrichmentanalysis: A knowledge-based approach for interpreting genome-wide expression profiles.Proc. Natl. Acad. Sci. U.S.A. 102, 15545–15550 (2005).

17. S. Santagata, M. L. Mendillo, Y.-c. Tang, A. Subramanian, C. C. Perley, S. P. Roche, B. Wong,R. Narayan, H. Kwon, M. Koeva, A. Amon, T. R. Golub, J. A. Porco Jr., L. Whitesell,S. Lindquist, Tight coordination of protein translation and HSF1 activation supports theanabolic malignant state. Science 341, 1238303 (2013).

18. M. L. Mendillo, S. Santagata, M. Koeva, G. W. Bell, R. Hu, R. M. Tamimi, E. Fraenkel,T. A. Ince, L. Whitesell, S. Lindquist, HSF1 drives a transcriptional program distinct fromheat shock to support highly malignant human cancers. Cell 150, 549–562 (2012).

19. L. Agnelli, S. Bicciato, M. Mattioli, S. Fabris, D. Intini, D. Verdelli, L. Baldini, F. Morabito,V. Callea, L. Lombardi, A. Neri, Molecular classification of multiple myeloma: A distincttranscriptional profile characterizes patients expressing CCND1 and negative for 14q32translocations. J. Clin. Oncol. 23, 7296–7306 (2005).

20. F. Zhan, Y. Huang, S. Colla, J. P. Stewart, I. Hanamura, S. Gupta, J. Epstein, S. Yaccoby,J. Sawyer, B. Burington, E. Anaissie, K. Hollmig, M. Pineda-Roman, G. Tricot, F. van Rhee,R. Walker, M. Zangari, J. Crowley, B. Barlogie, J. D. Shaughnessy Jr., The molecularclassification of multiple myeloma. Blood 108, 2020–2028 (2006).

21. G. Mulligan, C. Mitsiades, B. Bryant, F. Zhan, W. J. Chng, S. Roels, E. Koenig, A. Fergus,Y. Huang, P. Richardson, W. L. Trepicchio, A. Broyl, P. Sonneveld, J. D. Shaughnessy Jr.,P. L. Bergsagel, D. Schenkein, D.-L. Esseltine, A. Boral, K. C. Anderson, Gene expressionprofiling and correlation with outcome in clinical trials of the proteasome inhibitorbortezomib. Blood 109, 3177–3188 (2007).

22. E. L. Huttlin, M. P. Jedrychowski, J. E. Elias, T. Goswami, R. Rad, S. A. Beausoleil, J. Villén,W. Haas, M. E. Sowa, S. P. Gygi, A tissue-specific atlas of mouse protein phosphorylationand expression. Cell 143, 1174–1189 (2010).

23. J. G. Lohr, P. Stojanov, S. L. Carter, P. Cruz-Gordillo, M. S. Lawrence, D. Auclair, C. Sougnez,B. Knoechel, J. Gould, G. Saksena, K. Cibulskis, A. McKenna, M. A. Chapman, R. Straussman,J. Levy, L. M. Perkins, J. J. Keats, S. E. Schumacher, M. Rosenberg; Multiple MyelomaResearch Consortium, G. Getz, T. R. Golub, Widespread genetic heterogeneity in multiplemyeloma: Implications for targeted therapy. Cancer Cell 25, 91–101 (2014).

24. M. Wade, Y.-C. Li, G. M. Wahl, MDM2, MDMX and p53 in oncogenesis and cancer therapy.Nat. Rev. Cancer 13, 83–96 (2013).

25. R. Beroukhim, C. H. Mermel, D. Porter, G. Wei, S. Raychaudhuri, J. Donovan, J. Barretina,J. S. Boehm, J. Dobson, M. Urashima, K. T. Mc Henry, R. M. Pinchback, A. H. Ligon, Y.-J. Cho,L. Haery, H. Greulich, M. Reich, W. Winckler, M. S. Lawrence, B. A. Weir, K. E. Tanaka,D. Y. Chiang, A. J. Bass, A. Loo, C. Hoffman, J. Prensner, T. Liefeld, Q. Gao, D. Yecies,S. Signoretti, E. Maher, F. J. Kaye, H. Sasaki, J. E. Tepper, J. A. Fletcher, J. Tabernero,J. Baselga, M.-S. Tsao, F. Demichelis, M. A. Rubin, P. A. Janne, M. J. Daly, C. Nucera,R. L. Levine, B. L. Ebert, S. Gabriel, A. K. Rustgi, C. R. Antonescu, M. Ladanyi, A. Letai,L. A. Garraway, M. Loda, D. G. Beer, L. D. True, A. Okamoto, S. L. Pomeroy, S. Singer,T. R. Golub, E. S. Lander, G. Getz, W. R. Sellers, M. Meyerson, The landscape of somaticcopy-number alteration across human cancers. Nature 463, 899–905 (2010).

26. E. M. Hurt, A. Wiestner, A. Rosenwald, A. L. Shaffer, E. Campo, T. Grogan, P. L. Bergsagel,W. M. Kuehl, L. M. Staudt, Overexpression of c-maf is a frequent oncogenic event inmultiple myeloma that promotes proliferation and pathological interactions with bonemarrow stroma. Cancer Cell 5, 191–199 (2004).

27. A. L. Shaffer, N. C. T. Emre, L. Lamy, V. N. Ngo, G. Wright, W. Xiao, J. Powell, S. Dave, X. Yu,H. Zhao, Y. Zeng, B. Chen, J. Epstein, L. M. Staudt, IRF4 addiction in multiple myeloma.Nature 454, 226–231 (2008).

28. M. Chesi, D. F. Robbiani, M. Sebag, W. J. Chng, M. Affer, R. Tiedemann, R. Valdez,S. E. Palmer, S. S. Haas, A. K. Stewart, R. Fonseca, R. Kremer, G. Cattoretti, P. L. Bergsagel,


AID-dependent activation of a MYC transgene induces multiple myeloma in a conditionalmouse model of post-germinal center malignancies. Cancer Cell 13, 167–180 (2008).

29. N. Sonenberg, A. G. Hinnebusch, Regulation of translation initiation in eukaryotes:Mechanisms and biological targets. Cell 136, 731–745 (2009).

30. M. Kozak, Influences of mRNA secondary structure on initiation by eukaryotic ribosomes.Proc. Natl. Acad. Sci. U.S.A. 83, 2850–2854 (1986).

31. J. Pelletier, N. Sonenberg, Insertion mutagenesis to increase secondary structure withinthe 5′ noncoding region of a eukaryotic mRNA reduces translational efficiency. Cell 40,515–526 (1985).

32. A. E. Koromilas, A. Lazaris-Karatzas, N. Sonenberg, mRNAs containing extensive secondarystructure in their 5′ non-coding region translate efficiently in cells overexpressinginitiation factor eIF-4E. EMBO J. 11, 4153–4158 (1992).

33. K. Feoktistova, E. Tuvshintogs, A. Do, C. S. Fraser, Human eIF4E promotes mRNArestructuring by stimulating eIF4A helicase activity. Proc. Natl. Acad. Sci. U.S.A. 110,13339–13344 (2013).

34. A. L. Wolfe, K. Singh, Y. Zhong, P. Drewe, V. K. Rajasekhar, V. R. Sanghvi, K. J. Mavrakis,M. Jiang, J. E. Roderick, J. Van der Meulen, J. H. Schatz, C. M. Rodrigo, C. Zhao, P. Rondou,E. de Stanchina, J. Teruya-Feldstein, M. A. Kelliher, F. Speleman, J. A. Porco Jr., J. Pelletier,G. Ratsch, H.-G. Wendel, RNA G-quadruplexes cause eIF4A-dependent oncogenetranslation in cancer. Nature 513, 65–70 (2014).

35. S. Iwasaki, S. N. Floor, N. T. Ingolia, Rocaglates convert DEAD-box protein eIF4A into asequence-selective translational repressor. Nature 534, 558–561 (2016).

36. N. J. Lajkiewicz, S. P. Roche, B. Gerard, J. A. Porco Jr., Enantioselectivephotocycloaddition of 3-hydroxyflavones: Total syntheses and absolute configurationassignments of (+)-ponapensin and (+)-elliptifoline. J. Am. Chem. Soc. 134,13108–13113 (2012).

37. E. A. Akbay, J. Moslehi, C. L. Christensen, S. Saha, J. H. Tchaicha, S. H. Ramkissoon,K. M. Stewart, J. Carretero, E. Kikuchi, H. Zhang, T. J. Cohoon, S. Murray, W. Liu, K. Uno,S. Fisch, K. Jones, S. Gurumurthy, C. Gliser, S. Choe, M. Keenan, J. Son, I. Stanley,J. A. Losman, R. Padera, R. T. Bronson, J. M. Asara, O. Abdel-Wahab, P. C. Amrein,A. T. Fathi, N. N. Danial, A. C. Kimmelman, A. L. Kung, K. L. Ligon, K. E. Yen, W. G. Kaelin Jr.,N. Bardeesy, K.-K. Wong, D-2-hydroxyglutarate produced by mutant IDH2 causescardiomyopathy and neurodegeneration in mice. Genes Dev. 28, 479–490 (2014).

38. A. A. Lane, B. Chapuy, C. Y. Lin, T. Tivey, H. Li, E. C. Townsend, D. van Bodegom, T. A. Day,S.-C. Wu, H. Liu, A. Yoda, G. Alexe, A. C. Schinzel, T. J. Sullivan, S. Malinge, J. E. Taylor,K. Stegmaier, J. D. Jaffe, M. Bustin, G. te Kronnie, S. Izraeli, M. H. Harris, K. E. Stevenson,D. Neuberg, L. B. Silverman, S. E. Sallan, J. E. Bradner, W. C. Hahn, J. D. Crispino, D. Pellman,D. M. Weinstock, Triplication of a 21q22 region contributes to B cell transformationthrough HMGN1 overexpression and loss of histone H3 Lys27 trimethylation. Nat. Genet.46, 618–623 (2014).

Acknowledgments: We are thankful to M. Mendillo (Northeastern University, Boston, MA) foranalysis of the HSF1 activation signature in MM data sets, to A. Kung (Dana-Farber CancerInstitute, Boston, MA) for providing the MM1S luc/GFP cells, to P. Jänne (Dana-Farber CancerInstitute, Boston, MA) for providing lung and breast cell lines, and to L. Bergsagel and M. Chesi(Mayo Clinic, Scottsdale, AZ) for providing Vk*Myc cells. Funding: This work was supported,in part, by NCI R01 CA181683-01A1. Work at Boston University is supported by R35GM118173and R24GM111625. We thank S. Stone for the initial synthesis of CMLD010509. R.Y.E. wassupported by the National Institute of General Medical Sciences T32GM007753. Authorcontributions: S.M., N.G., J.B., L.W., and I.M.G. designed the study. S.M., D.H., Y.J.S., K.Z.S., R.Y.E.,S.V.G., T.Y., and J.Z. performed the experiments. S.M., D.H., J.P., J.S., L.W., and I.M.G. analyzedthe data. W.G.D., L.E.B., J.K.S., and J.A.P. generated reagents. I.M.G. provided funding. Allauthors wrote, reviewed, and edited the manuscript. Competing interests: The authorsdeclare that they have no competing interests. Data and materials availability: RNA-seq datahave been deposited to the Gene Expression Omnibus (www.ncbi.nlm.nih.gov/geo/) underaccession no. GSE94827. All compounds tested are from J.A.P. under a material transferagreement with Boston University.

Submitted 23 October 2016Accepted 16 March 2017Published 10 May 201710.1126/scitranslmed.aal2668

Citation: S. Manier, D. Huynh, Y. J. Shen, J. Zhou, T. Yusufzai, K. Z. Salem, R. Y. Ebright, J. Shi,J. Park, S. V. Glavey, W. G. Devine, C.-J. Liu, X. Leleu, B. Quesnel, C. Roche-Lestienne, J. K. Snyder,L. E. Brown, N. Gray, J. Bradner, L. Whitesell, J. A. Porco Jr., I. M. Ghobrial, Inhibiting the oncogenictranslation program is an effective therapeutic strategy in multiple myeloma. Sci. Transl. Med. 9,eaal2668 (2017).

13 of 13

http://www.ncbi.nlm.nih.gov/geo/


multiple myelomaInhibiting the oncogenic translation program is an effective therapeutic strategy in

Jr.. and Irene M. GhobrialRoche-Lestienne, John K. Snyder, Lauren E. Brown, Nathanael Gray, James Bradner, Luke Whitesell, John A. Porco,Jihye Park, Siobhan V. Glavey, William G. Devine, Chia-Jen Liu, Xavier Leleu, Bruno Quesnel, Catherine Salomon Manier, Daisy Huynh, Yu J. Shen, Jia Zhou, Timur Yusufzai, Karma Z. Salem, Richard Y. Ebright, Jiantao Shi,

DOI: 10.1126/scitranslmed.aal2668, eaal2668.9Sci Transl Med

rocaglates as potential therapeutic candidates for multiple myeloma.decreased tumor cell proliferation. The treatment was safe and effective in multiple mouse models, suggesting showed that it reversed the effects of MYC, blocked excessive translation, induced tumor cell apoptosis, androcaglate derivatives are active in multiple myeloma. The authors focused on one lead rocaglate derivative and

. determined that small-moleculeet alproliferating cancer cells. By performing a high-throughput screen, Manier key role by stimulating ribosome production and up-regulating protein translation to satisfy the needs of rapidly

Multiple myeloma is a difficult-to-treat hematologic malignancy, where the MYC oncoprotein often plays aRocaglates rocking multiple myeloma

ARTICLE TOOLS http://stm.sciencemag.org/content/9/389/eaal2668

MATERIALSSUPPLEMENTARY http://stm.sciencemag.org/content/suppl/2017/05/08/9.389.eaal2668.DC1

CONTENTRELATED

http://stm.sciencemag.org/content/scitransmed/11/494/eaau9087.fullhttp://stm.sciencemag.org/content/scitransmed/11/484/eaar5012.fullhttp://stm.sciencemag.org/content/scitransmed/10/453/eaan0941.fullhttp://stm.sciencemag.org/content/scitransmed/3/78/78cm11.fullhttp://stm.sciencemag.org/content/scitransmed/7/288/288ra78.fullhttp://stm.sciencemag.org/content/scitransmed/8/363/363ra147.full

REFERENCES

http://stm.sciencemag.org/content/9/389/eaal2668#BIBLThis article cites 38 articles, 10 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

registered trademark of AAAS. is aScience Translational MedicineScience, 1200 New York Avenue NW, Washington, DC 20005. The title

(ISSN 1946-6242) is published by the American Association for the Advancement ofScience Translational Medicine

Copyright © 2017, American Association for the Advancement of Science


.sciencemag.org/

Dow

nloaded from

http://stm.sciencemag.org/content/9/389/eaal2668

http://stm.sciencemag.org/content/suppl/2017/05/08/9.389.eaal2668.DC1

http://stm.sciencemag.org/content/scitransmed/8/363/363ra147.full

http://stm.sciencemag.org/content/scitransmed/7/288/288ra78.full

http://stm.sciencemag.org/content/scitransmed/3/78/78cm11.full

http://stm.sciencemag.org/content/scitransmed/10/453/eaan0941.full

http://stm.sciencemag.org/content/scitransmed/11/484/eaar5012.full

http://stm.sciencemag.org/content/scitransmed/11/494/eaau9087.full

http://stm.sciencemag.org/content/9/389/eaal2668#BIBL

http://www.sciencemag.org/help/reprints-and-permissions

http://www.sciencemag.org/about/terms-service


inhibiting the oncogenic translation program is an effective … › content › scitransmed › 9...

Documents