Preclinical Efﬁcacy and Molecular Mechanism of Targeting ... · Preclinical Efﬁcacy and Molecular Mechanism of Targeting CDK7-Dependent Transcriptional Addiction in Ovarian Cancer

Small Molecule Therapeutics

Preclinical Efficacy and Molecular Mechanism ofTargeting CDK7-Dependent TranscriptionalAddiction in Ovarian CancerZhenfeng Zhang1,2, Huixin Peng1, Xiaojie Wang3, Xia Yin4,5, Pengfei Ma1, Ying Jing1,Mei-Chun Cai2, Jin Liu2, Meiying Zhang4,5, Shengzhe Zhang6, Kaixuan Shi6,Wei-Qiang Gao1,6,Wen Di4,5, and Guanglei Zhuang1,5

Abstract

Ovarian cancer remains a significant cause of gynecologiccancer mortality, and novel therapeutic strategies are urgentlyneeded in clinic as new treatment options. We previouslyshowed that BET bromodomain inhibitors displayed promis-ing efficacy for the treatment of epithelial ovarian cancer bydownregulating pivot transcription factors. However, thepotential antitumor activities and molecular mechanisms ofother epigenetic or transcriptional therapies have not beensystematically determined. Here, by performing an unbiasedhigh-throughput drug screen to identify candidate compoundswith antineoplastic effects, we identified THZ1, a recently

developed covalent CDK7 inhibitor, as a new transcription-targeting compound that exerted broad cytotoxicity againstovarian tumors. Mechanistically, CDK7 represented a previous-ly unappreciated actionable vulnerability in ovarian cancer,and CDK7 inhibition led to a pronounced dysregulation ofgene transcription, with a preferential repression of E2F-regu-lated genes and transcripts associated with super-enhancers.Our findings revealed the molecular underpinnings of THZ1potency and established pharmaceutically targeting transcrip-tional addiction as a promising therapeutic strategy in aggres-sive ovarian cancer. Mol Cancer Ther; 16(9); 1739–50. �2017 AACR.

IntroductionEpithelial ovarian cancer (EOC) represents themost life-threat-

ening gynecologicmalignancy, causingmore than140,000deathsannually worldwide (1, 2). For the past decades, the standardtreatment has been surgery and platinum-based chemotherapy.The survival improvement achieved with conventional cytotoxicagents has reached a plateau, posing formidable challenges forcurrent translational researches to invent novel targeted thera-peutic regimens (3, 4). However, ovarian cancer is characterizedby low prevalence of actionable genetic alterations (5). As a result,

targeted approaches engaging tumor-specific mutations are oftennot an attractive option. Therefore, further elucidation of diseasepathogenesis is imperative to identify unique molecular vulner-abilities in ovarian cancer for developing state-of-the-art therapies(6, 7).

Recent pan-cancer analysis of tumor genomic traits has revealeda striking cancer genome hyperbola. Almost all high-grade serousovarian carcinoma samples (HGS-OvCa), displaying large-scalecopy number aberrations, are positioned in the C class anddistinct from tumors of M class, which are dominated by somaticmutations (8). On the basis of these new insights into mechan-isms of oncogenesis, a promising therapeutic framework emergesby exploring recurrent chromosomal gains and losses in ovariancancer. Along this line, we and others have recently demonstratedthat the bromodomain and extra terminal (BET) protein BRD4 isfrequently amplified in HGS-OvCa and that pharmacologic inhi-bition of BRD4 using BET bromodomain inhibitors JQ1 or I-BET151 shows antiproliferative efficacy in preclinical ovariantumormodels. As in other various diseases, the anticancer activityof BET inhibitors has been attributed to transcriptional suppres-sion of key proto-oncogenes, such as MYC or FOXM1 (9, 10).Despite these tremendous progressions on uncovering gene tran-scription regulation as an emerging therapeutic vulnerabilityagainst ovarian malignancy, the potential antitumor activitiesand molecular mechanisms of other epigenetic or transcriptionaltherapies remain largely elusive.

Here we used a small-molecule screen and discovered THZ1, acovalent inhibitor of cyclin-dependent kinase 7 (CDK7), as a drugcandidate for the treatment of ovarian tumors. Mechanistically,THZ1 led to a profound downregulation of gene transcripts,preferentially those containing E2F-binding motifs and associat-ed with super-enhancers (SE). These findings provided a

1State Key Laboratory of Oncogenes and Related Genes, Renji-Med X ClinicalStem Cell Research Center, Ren Ji Hospital, School of Medicine, Shanghai JiaoTong University, Shanghai, China. 2State Key Laboratory of Oncogenes andRelated Genes, Shanghai Cancer Institute, Ren Ji Hospital, School of Medicine,Shanghai Jiao Tong University, Shanghai, China. 3Department of Obstetrics andGynecology, Tongren Hospital, Shanghai Jiao Tong University School of Med-icine, Shanghai, China. 4Department of Obstetrics and Gynecology, Ren JiHospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China.5Shanghai Key Laboratory of Gynecologic Oncology, Ren Ji Hospital, School ofMedicine, Shanghai Jiao TongUniversity, Shanghai, China. 6School of BiomedicalEngineering and Med-X Research Institute, Shanghai Jiao Tong University,Shanghai, China.

Note: Supplementary data for this article are available at Molecular CancerTherapeutics Online (http://mct.aacrjournals.org/).

Z. Zhang, H. Peng, X. Wang, and X. Yin contributed equally to this article.

Corresponding Author: Guanglei Zhuang, School of Medicine, Shanghai JiaoTong University, Shanghai 200240, China. Phone: 86-21-34208660; Fax: 86-21-58394262; E-mail: [email protected]

doi: 10.1158/1535-7163.MCT-17-0078

�2017 American Association for Cancer Research.

MolecularCancerTherapeutics

www.aacrjournals.org 1739

on June 25, 2020. © 2017 American Association for Cancer Research. mct.aacrjournals.org Downloaded from

Published OnlineFirst June 1, 2017; DOI: 10.1158/1535-7163.MCT-17-0078

http://crossmark.crossref.org/dialog/?doi=10.1158/1535-7163.MCT-17-0078&domain=pdf&date_stamp=2017-8-14

http://mct.aacrjournals.org/

conceptual rationale for exploiting transcriptional addiction as anew therapeutic opportunity in ovarian cancer.

Materials and MethodsCell culture and reagents

Tumor cell lineswere originally obtained fromATCCor JCRB in2014, where cell characterization was performed using polymor-phic short tandem repeat (STR) profiling. No subsequent authen-tication was done by the authors. Cells were cultured inRPMI1640 (Invitrogen) supplemented with 10% FBS (Invitro-gen). For gene knockout experiments using CRISPR-Cas9 system,cells were infected with lentivirus encoding sgRNAs (Supplemen-tary Table S9, for details of the sgRNA sequences). JQ1 waspurchased from Millipore (11). THZ1 was purchased from Med-Chem Express. Small-molecule inhibitor library was purchasedfrom Selleck Chemicals. All inhibitors were reconstituted inDMSO (Sigma-Aldrich) at a stock concentration of 10 mmol/L.

High-throughput inhibitor screen and THZ1 IC50measurementHigh-throughput small-molecule screen was performed as

previously reported (12). Cells were seeded at optimal densityand treated with the indicated inhibitors at the same concentra-tion (500nmol/L). Freshmediumanddrugswere changed every 3days. After 6 days of drug exposure, cellswere imaged and viabilitywas calculated using ArrayScan Infinity (Thermo Scientific). Todetermine the IC50 of THZ1 in ovarian cancer cell lines, sevenconcentrations of compounds were applied at a stepwise three-fold dilution series. Cell viability was evaluated using CellTiter-Glo reagent according to the manufacturer's instructions (Pro-mega). Estimates of IC50 were derived from the seven dose–response curves plotted by GraphPad Prism 6 (GraphPad Soft-ware, Inc.).

Cell cycle and apoptosis analysisCell-cycle analysis was performed 24 hours after THZ1 treat-

ment. Cells were fixed in cold ethanol and incubated with pro-pidium iodide (PI)/RNase Staining Solution (Cell SignalingTechnology) for 15 minutes. Cell apoptosis was assayed usingDead Cell Apoptosis Kit with Annexin V-FITC and PI followingmanufacturer's instructions (Life Technologies). Flow cytometricanalysis was performed on a FACS AriaII cytometer (BD Bios-ciences). Flow cytometry data were analyzed by using FlowJosoftware.

Western blot analysisCells were lysed in RIPA buffer (Tris pH 7.4 50 mmol/L, NaCl

150 mmol/L, NP-40 1%, SDS 0.1%, EDTA 2 mmol/L) containingproteinase inhibitors (Roche) and phosphatase inhibitors(Roche), and subjected to SDS-PAGE and Western blot analysis.Antibodies against the following proteins were used: CDK4,CDK6, CDK7, CDK9, cleaved PARP, H3, actin, tubulin (CellSignaling Technology); RNAPII S2, RNAPII S5, RNAPII S7(Millipore).

RNA sequencing and analysisCells were treated with DMSO or THZ1 (100 nmol/L) for 6

hours, and cell numbers were measured by manually counting ofTrypan blue-stained cells before lysed in buffer. ERCC Spike-InRNA Mix (Life Technologies) was added to the cell lysates inproportion to cell number as previously described (13). Total

RNA (three biological replicates per condition) was extractedusing RNeasy Plus Mini Kit (Qiagen) according to the manufac-turer's protocol. RNA quality was assessed using an Agilent 2100Bioanalyzer. Sequencing libraries were generated using NEBNextUltra RNA Library Prep Kit for Illumina (NEB). The index-codedlibraries were clustered on a cBot Cluster Generation SystemusingTruSeqPECluster Kit v3-cBot-HS (Illumina) and sequenced on anIllumina Hiseq 2500 platform to generate 30 million 125 bppaired-end reads (Novogene). Quality control was performed onraw data in fastq format and clean reads were obtained byremoving adapters, ploy-N sequences and low-quality reads.Index of the reference genome was built using Bowtie v2.2.3(14) and clean reads were aligned to the reference genome usingTopHat v2.0.12 (15). We used HTSeq v0.6.1 (16) to countthe reads mapped to each gene and further normalized readcounts to the control ERCC reads. Differential expression analysiswas performed using theDESeq R package (1.18.0). P-values wereadjusted using the Benjamini–Hochberg procedure for control-ling the false discovery rate. Genes with an adjusted P-value<0.05were considered differentially expressed. The sequencing datahave been deposited in NCBI BioProject database (http://www.ncbi.nlm.nih.gov/bioproject/) under the accession numberSRP106825.

Chromatin precipitation, sequencing, and analysisChromatin precipitation was performed as previously

described. Briefly, cells were cross-linked with 1% formaldehydeand quenched with 2.5 M glycine. Cell pellets were lysed andsonicated using Sonics Vibra-Cell 505 ultrasonicator. Sonicatedlysates were cleared and incubated overnight withmagnetic beadsbound with H3K27ac antibody (Abcam) to enrich for associatedDNA fragments. Precipitated complexes were washed and cross-links were reversed overnight. RNA and protein were digestedusing RNase A and Proteinase K, respectively. DNA was purifiedwithQiaquick PCR Purification Kit. DNA libraries were generatedusing the Illumina TruSeq DNA Sample Preparation v2 Kit andsequenced on the Illumina HiSeq 2000 to generate 20 million 50bp reads (Novogene). Clean reads were aligned to the referencegenome using BWA v0.7.12 (17). Wiggle files for gene tracks werecreated and peaks were called using MACS v2.1.0 (18). Super-enhancers were identified using ROSE package as previouslydescribed (19).

Tumor xenograft modelsFor THZ1 efficacy studies, tumor cells (1 � 106) were mixed

with Matrigel (BD Biosciences) and subcutaneously implantedin the dorsal flank of BALB/c Nude mice. When tumor sizesreached approximately 150 mm3, mice were randomized intotwo groups of 10 mice each. One group of mice was treatedwith vehicle control (10%DMSO in D5W), and the other groupwas treated with THZ1 10 mg/kg twice daily. To test the effectsof CDK7, control or CDK7 knockout cells were subcutaneouslyimplanted into BALB/c Nude mice. Tumor volumes (10 ani-mals per group) were measured with digital caliper and calcu-lated as length � width2 � 0.5. All animal protocols wereapproved by the Institutional Animal Care and Use Committeeof Ren Ji Hospital.

Histology and IHCHistology and IHC were performed using formalin-fixed, par-

affin-embedded 5-mm-thick tumor sections. Slides were

Zhang et al.

Mol Cancer Ther; 16(9) September 2017 Molecular Cancer Therapeutics1740



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

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


deparaffinized in xylene, passed through graded alcohols, andantigen retrieved with 10 mmol/L citrate buffer (pH 6.0) in asteam pressure cooker. Preprocessed tissues were either stainedwith hematoxylin and eosin (H&E), or treated with PeroxidaseBlock (Dako) to quench endogenous peroxidase activity, blockedusing Protein Block (Dako), and subsequently incubated with Ki-67 or cleaved caspase-3 antibodies (Cell Signaling Technology).Slides were then washed in 50 mmol/L Tris-Cl (pH 7.4) andincubated with horseradish peroxidase-conjugated secondaryantibody. Immunoperoxidase staining was developed using a3,3'diaminobenzidine (DAB) chromogen (Sigma). Slides werethen counterstained with hematoxylin, dehydrated in gradedalcohol and xylene, and coverslipped using mounting solution.

Statistical analysisGene set enrichment analysis was performed using the GSEA

software (20). Gene ontology and pathway analyses were per-

formed with DAVID Bioinformatics Resources (21). In all experi-ments, comparisons between two groups were based on two-sided Student t test and one-way ANOVA was used to test fordifferences among more groups. P values of <0.05 were consid-ered statistically significant.

ResultsIdentification of THZ1 as a potent inhibitorin ovarian cancer

We reasoned that an antitumor inhibitor susceptibility pro-file of ovarian tumor cells might unveil additional therapeuticvulnerabilities of ovarian cancer. We therefore subjected mul-tiple ovarian cancer cell lines, including COV 413B, OVCA420,and SKOV3, to a high-throughput small-molecule screenwith a library of 181 FDA-approved drugs or other clinicallyrelevant compounds (Supplementary Table S1). Using an

A

B

D

C EDMSO JQ1 THZ1

JQ1 (µmol/L)0 0.1 0.3 1 3 0 0.03 0.1 0.3 1

THZ1 (µmol/L)

CO

V 4

13B

OV

CA

420

SK

OV

3

THZ1

THZ1

THZ1

COV 413B0 0.05 0.1 0.5 1

RNAPII S2

CDK7

Tubulin

RNAPII S5

RNAPII S7

RNAPII S2

CDK7

Tubulin

RNAPII S5

RNAPII S7

RNAPII S2

CDK7

Tubulin

RNAPII S5

RNAPII S7

SKOV30 0.05 0.1 0.5 1

OVCA4200 0.05 0.1 0.5 1

THZ1Dinaciclib

FlavopiridolRomidepsinPanobinostatPF−04929113

AT13387Torin 2

INK 128BEZ235BI 6727BI 2536

MLN2238BortezomibDasatinibYM155

PemetrexedAbitrexate

Gemcitabine RaltitrexedPaclitaxelDocetaxel Topotecan

DoxorubicinEpirubicin

CDK

INHIBITOR PubChemID TARGET COV 413B OVCA420 SKOV3

CDKCDK

HDACHDACHSPHSP

mTORmTORmTORPLKPLK

ProteasomeProteasome

SrcIAP

ChemoChemoChemoChemoChemoChemoChemoChemoChemo

7360282746926350545921953520626918837

441955711195571651358113

4537595311977753104615081136442125183872

3874473062316

469310003944931269416075010475836314148124607003170341867

High

IC50 < 0.5 µmol/L

96-well plate

Imaging using ArrayScan Infinity

Ovarian cancer cells

Inhibitor library

inhibitor activity

Low

CO

V 4

13B

OV

CA

420

SK

OV

3

Figure 1.

Identification of THZ1 as a potent inhibitor in ovarian cancer. A, Schematic overview of high-throughput small-molecule screen. B, Top-ranked compoundssuppressing ovarian cancer cell growth and their corresponding targets were presented in an inhibitory activity heatmap format. C,Representative images of cancercells exposed to indicated inhibitors and counter stained by DAPI. D, Crystal violet staining of tumor cells treated with various concentrations of JQ1 or THZ1for 10 days. E, Western blot analysis of RNAPII CTD phosphorylation in ovarian cancer cells that were treated with THZ1.

CDK7-Dependent Transcriptional Addiction in Ovarian Cancer

www.aacrjournals.org Mol Cancer Ther; 16(9) September 2017 1741




imaging-based viability assay, cells were screened against eachinhibitor for drug candidates with a IC50 of less than 0.5 mmol/L(Fig. 1A). A small subset of the tested compounds exhibitedpronounced toxicity toward each cell line (SupplementaryFig. S1A) and we identified a list of top-ranked 25 compoundswith IC50 < 0.5 mmol/L across all the three cell lines (Fig. 1B).Nine of these compounds were chemotherapeutic agents, inkeeping with the generally exceptional chemosensitivity ofovarian cancer. Other major putative categories of these effec-tive inhibitors included transcription, proteostasis, and epige-netic regulation (Supplementary Fig. S1B).

BRD4 regulates gene transcription through linking histoneacetylation and core transcriptional components, and transcrip-tional perturbation of oncogenic drivers has been implicated inthe wide range of sensitivity to BRD4 inhibition (22). As tran-scriptional CDKs also play essential roles in transcription initi-ation and elongation, we focused our further investigations onthe promising therapeutic merit of CDK inhibitors, among whichTHZ1, a novel covalent CDK7 inhibitor (23), was particularlyinteresting. THZ1 has recently been demonstrated to displaytherapeutic activity against an emerging list of cancers, such asT-cell acute lymphoblastic leukemia (23), MYCN-amplified neu-roblastoma (24), small-cell lung cancer (25), triple-negativebreast cancer (TNBC; ref. 26), and esophageal squamous cellcarcinoma (27). Likewise, all three ovarian cancer cell lines werehighly sensitive to THZ1 in our drug screen (Fig. 1C) and wevalidated these findings using crystal violet staining (Fig. 1D).CDK7 regulates transcriptional initiation by directly phosphory-lating serine 5 (S5) and serine 7 (S7) at the carboxyl-terminaldomain (CTD) of RNA polymerase II (RNAPII). In addition,CDK7 exerts its control on transcriptional elongation via activat-ing other CDKs to phosphorylate serine 2 (S2) of RNAPII CTD(28–30). As expected, THZ1 treatment resulted in a dose-depen-dent inhibition of RNAPII CTD phosphorylation at S2, S5, and S7(Fig. 1E).

Antineoplastic effects of THZ1 across variousovarian cancer cell lines

To further validate the potency of THZ1 as a drug candidate,we extended our findings in COV 413B, OVCA420, and SKOV3cells to a large set of distinct ovarian cancer cell lines. Acomprehensive panel of 18 ovarian cancer cell lines was assem-bled to represent the major histologic subtypes of EOC (sevenserous, five nonserous, and six unspecified). We conducted apharmacologic screen to determine their response to THZ1.These studies confirmed that the majority of these cell linesshowed a general hypersensitivity to THZ1-induced cytotoxicity(Fig. 2A and B; Supplementary Table S2). It is noteworthy thatalthough many cell models also responded well to JQ1 (11),THZ1 was usually more potent to suppress tumor cell prolif-eration (Supplementary Fig. S2A), and there was a synergisticeffect of growth inhibition with JQ1 and THZ1 in combination(Supplementary Fig. S2B). Furthermore, we found that multiplecell lines including A2780, COV 362, and ES-2, which wereintrinsically resistant to JQ1, were efficaciously inhibited byTHZ1 treatment (Fig. 2C and D).

To further characterize the antineoplastic effects of THZ1,cell-cycle progression and apoptotic index were determinedusing flow cytometry in three representative cell lines(A2780, COV 413B, and OVCA420). THZ1-induced cell-cyclearrest, particularly a pronounced S phase reduction, was evi-

dent in all three cases (Supplementary Fig. S3A). We alsoobserved a concomitant increase of cell apoptosis/necrosisin a dose-responsive and time-dependent manner as assessedby cleaved PARP Western blotting (Fig. 2E) and Annexin V-PIstaining (Supplementary Fig. S3B). Finally, we explored the invivo therapeutic potential of THZ1 in two independent xeno-transplantation models (A2780 and HEY). Once tumorsreached an optimal volume of �150 mm3, the animals wererandomized into two cohorts (n ¼ 10/group) and treated withvehicle or THZ1. Notably, mice receiving THZ1 showed astatistically significant reduction in tumor burden (Fig. 2F).Histologic analysis of THZ1-treated tumors revealed wide-spread necrotic tissues, decreased cell proliferation indicatorKi-67, and activation of the apoptotic marker cleaved caspase-3,whereas vehicle-treated tumors were exclusively composedof viable epithelial cells (Supplementary Fig. S4A). It wasnotable that although THZ1 exhibited some inhibitory effectsin noncancerous cells, including HEK293T, BEAS-2B, and pri-mary normal ovarian epithelial cells, the IC50 values in thesecells were relatively higher than those in EOC cell lines (Sup-plementary Fig. S4B). In addition, THZ1 was not as efficaciousin lung cancer cells as compared with ovarian cancer cells(Supplementary Fig. S4C) and there was no noticeable miceweight loss during continuous THZ1 administrations (Supple-mentary Fig. S4D), consistent with previous reports demon-strating the limited toxicity of THZ1 and supporting a feasibletherapeutic window to use CDK7 inhibitors in clinic. Weconcluded that THZ1 impeded both cell-cycle and cell survivalto attenuate tumor growth in different histologic subtypes ofovarian cancer.

Widespread dependence of ovarian cancer cells on CDK7THZ1 selectively inhibits CDK7 with additional cross-reac-

tivity against CDK12/13 at higher concentrations (23). BecauseCDK7 kinase activity has been implicated in the regulation ofboth gene transcription and cell cycle (31), the capacity of THZ1to target ovarian cancer presumably relates to its covalentinhibition of CDK7. However, a formal investigation on thepotentially critical role of CDK7 in ovarian cancer is necessary toestablish CDK7 as a valid molecular target and to complementthe pharmacologic studies of THZ1 in this indication. To thisend, we first examined the expression levels of CDK7 protein inour ovarian cancer cell panel. CDK7, as well as RNAPII CTDphosphorylation at S2, S5, and S7, were ubiquitously expressedin essentially all cell lines (Fig. 3A). Next, we used CRISPR–Cas9system to genetically deplete CDK7 in 11 ovarian cancer celllines. Three independent single-guide RNA sequences targetingCDK7 (sgCDK7) led to a substantial reduction of CDK7 protein(Fig. 3B and Supplementary Fig. S5A). Strikingly, in all tested cellmodels, CDK7 downregulation profoundly suppressed cell via-bility in vitro (Fig. 3C; Supplementary Fig. S5B). Accordingly,ablation of CDK7 in HEY or ES-2 cells significantly impairedtumor xenograft formation in vivo (Fig. 3D). These data creden-tialed CDK7 as a functionally relevant target of THZ1 in ovariancancer.

CDK7belongs to a large family of�20 serine/threonine kinasesthat can be roughly divided into cell-cycle CDKs and transcrip-tional CDKs (31). In addition to phosphorylating RNAPII CTD,CDK7 is a pivot component of CDK-activating kinase (CAK),which is involved in regulating all CDKs (32). We determined therequirement of other CDKs for ovarian tumorigenicity by

Zhang et al.





knocking out each CDK gene using CRISPR-Cas9 technique.Interestingly, several cell-cycle CDKs (CDK1, CDK2, CDK4, andCDK6) and transcriptional CDKs (CDK7, CDK9, and CDK12)were indispensable for tumor cell growth (Supplementary Fig.S6A and S6B). Therefore, unlike TNBC (26), ovarian cancerexhibited a universal dependence on multiple CDKs, reinforcingCDK7 as a master CDK that provided a unique target for con-trolling ovarian cancer.

To assess the clinical relevance of CDK7 in ovarian cancerpatients, we conducted a meta-analysis using the curatedOvar-

ianData database (33). Some CDKs, including CDK7, were asso-ciated with poor prognosis in ovarian cancer (Supplementary Fig.S7; overall HR for CDK7 ¼ 1.07, 95% confidence interval, 1.01–1.15). We also performed immunohistochemistry on a cohort offifty EOC specimens, and found that more than 40% of thesesamples showed positive staining of CDK7 (Supplementary TableS3). When the patients were stratified into two groups based onCDK7 expression, the CDK7 positive group had a significantmedian overall survival disadvantage compared with the CDK7negative group (Fig. 3E). Together, these analyses established a

A

C D

E F

B

Pol II S2

Cleaved PARP

CDK7

Tubulin

Pol II S5

Pol II

Pol II S7

4 8 12 16 200

300

600

900

1,200

1,500

1,800Vehicle

HEY XenograftA2780 Xenograft

Time (days) Time (days)

THZ1

**

**

*

**

* **

n = 10 n = 10

3 6 9 12 15 18 21 24 270

200

400

600

800

1,000

1,200VehicleTHZ1

Tum

or v

olum

e (m

m3 )

A2780 COV 413B OVCA420

0 0.05 0.1 0.5 1 0 0.05 0.1 0.5 1 0 0.05 0.1 0.5 1 THZ1 (µmol/L)

THZ1 Concentration (µmol/L)

0.010.001 0.1 1

Inhibitor concentration (µmol/L)0.010.001 0.1 1 10

0

20

40

60

80

120

100

Cel

l via

bilit

y (%

)

0

20

40

60

80

120

100

Cel

l via

bilit

y (%

)

COV 362A2780TOV-112DCOV 413BDOV 13OV90SW982OVCA433PA1OVCAR5HEYIGROV1SKOV3OVCA420OVTOKOOVCAR8COV 318ES-2

SW

982

DMSO THZ1

OVC

A433

PA1

OV

CA

R5

HE

YIG

RO

V1

SK

OV

3

DMSO THZ1

OVC

A420

OVT

OKO

OV

CA

R8

CO

V 3

18ES

-2

CO

V 36

2

DMSO THZ1

A278

0TO

V-11

2DC

OV

413B

DO

V 13

OV9

0

A27

80C

OV

362

ES

-2

JQ1 (µmol/L)

0 0.1 0.3 1 103 0 0.030.01 0.1 0.3 1

THZ1 (µmol/L)

A2780+JQ1A2780+THZ1COV 362+JQ1COV 362+THZ1ES-2+JQ1ES-2+THZ1

Figure 2.

Antineoplastic effects of THZ1 across various ovarian cancer cell lines. A, Cell viability in the indicated panel of ovarian cancer cell lines treated with variousconcentrations of THZ1. B, Crystal violet staining of ovarian cancer cells treated with THZ1 (100 nmol/L) versus DMSO for 10 days. C, A2780, COV 362, and ES-2 cellswere treatedwith JQ1 or THZ1 for 10 days. The remaining cells were stained with crystal violet.D, Cell viability of A2780, COV 362, and ES-2 cells treated with variousconcentrations of JQ1 or THZ1. E, Western blot analysis of cleaved PARP and RNAPII CTD phosphorylation in A2780, COV 413B, and OVCA420 cells treatedwith THZ1. F, Tumor growth of A2780 and HEY xenografts (10 mice/group) that were treated with THZ1 or vehicle control. Each line represents an individual tumorgrowth curve (blue: vehicle; red: THZ1), and the thick blue/red lines indicate mean tumor volume of the treatment group. Tumor volumes of the THZ1 groupwere statistically significantly lower than those of the vehicle control (�P < 0.05, unpaired Student t test).






A

C

D E

B

Pol II S2

CDK7CDK7

OVCA420

H3

CDK7

H3

SW98

2

HEY

A278

0

DO

V 13

OVT

OKO

CO

V 31

8

SKO

V3

OVC

A420

MC

AS

TOV-

112D

PA1

IGR

OV1

OVC

AR5

OVC

A433

OVC

AR8

CO

V 41

3B

ES-2

OV9

0

sgEG

FP

sgC

DK7

-1

sgC

DK7

-2

sgC

DK7

-3

Pol II S5

Pol II S7

CDK4

CDK6

CDK9

Actin

PositiveNegative

A2780

sgEGFP sgCDK7-1 sgCDK7-2 sgCDK7-3 sgEGFP sgCDK7-1 sgCDK7-2 sgCDK7-3

COV 413B

HEY

ES-2

SKOV3

OVCA420

sgEGFP

sgCDK7-1

sgCDK7-2

sgEGFP

sgCDK7-1

sgCDK7-2

ES-2

HEY

0 10 20 30 40 50 60

0.0

0.2

0.4

0.6

0.8

1.0

Follow-up (months)

Epithelial OvCa

Ove

rall

surv

ival

(pro

b.)

P = 0.036

n = 28

n = 22

CDK7 NegativeCDK7 Positive

sgEGFP

sgCDK7-1

sgCDK7-2

sgEGFP

sgCDK7-1

sgCDK7-2

0.00

0.05

0.10

0.15

0.20

0.25

0.30 *

Tum

or w

eigh

t (g)

Tum

or w

eigh

t (g)

0.00

0.05

0.10

0.15

0.20

0.25 *

A2780

sgEG

FP

sgC

DK7

-1

sgC

DK7

-2

sgC

DK7

-3

COV 413B

sgEG

FP

sgC

DK7

-1

sgC

DK7

-2

sgC

DK7

-3

ES-2

sgEG

FP

sgC

DK7

-1

sgC

DK7

-2

sgC

DK7

-3

HEY

sgEG

FP

sgC

DK7

-1

sgC

DK7

-2

sgC

DK7

-3

SKOV3

sgEG

FP

sgC

DK7

-1

sgC

DK7

-2

sgC

DK7

-3

0.0

0.5

1.0

* * *

* * *

* * *

* * *

* * *

* * *

Rel

ativ

e pr

olife

ratio

n

0.0

0.5

1.0

Rel

ativ

e pr

olife

ratio

n

0.0

0.5

1.0

Rel

ativ

e pr

olife

ratio

n

0.0

0.5

1.0

Rel

ativ

e pr

olife

ratio

n

0.0

0.5

1.0

Rel

ativ

e pr

olife

ratio

n

0.0

0.5

1.0

Rel

ativ

e pr

olife

ratio

n

sgEGFP

sgCDK7-1

sgCDK7-2

sgCDK7-3

Figure 3.

Widespread dependence of ovarian cancer cells on CDK7.A,Western blot analysis of RNAPII CTD phosphorylation andCDKs in a panel of ovarian cancer cell lines.B,CDK7 was knocked out in the indicated cells using CRISPR-Cas9 system. Western blot analysis demonstrated CDK7 downregulation. C, Impaired cell viabilityfollowing depletion ofCDK7 in ovarian cancer cells. The left, middle, and right panels showed the bright-field images, the crystal violet staining, and the quantificationof cell viability (�P<0.05, ANOVA followedbyTukeypost-test), respectively.D, Images andweights of tumor xenografts derived fromHEYandES-2 cells editedwithsgCDK7 or sgEGFP control (� , P < 0.05, ANOVA followed by Tukey post-test). E, IHC analysis of CDK7 expression in a cohort of ovarian cancer specimens.Representative images of CDK7-negative or CDK7-positive samples were shown in the top panel. The bottom panel showed Kaplan–Meier plot of overall survival inCDK7-negative versus CDK7-positive ovarian cancer patients.

Zhang et al.





therapeutic rationale for targeting CDK7 in a subset of ovariancancer patients.

THZ1-inhibited gene transcription in ovarian cancerConsidering the prominent role of CDK7 in regulating RNAPII-

mediated transcription, we next investigated the impact of THZ1treatment on genome-wide gene expression in ovarian cancerusing RNA sequencing. Treatment with 100 nmol/L THZ1 for 6hours resulted in a dramatic decrease of global messenger RNAlevels, with 41% to 87%of active transcripts showing greater than1.5-fold reduction across the three tested cell lines (Fig. 4A). Weperformed gene set enrichment analysis and found that the mostdifferentially expressed genes were associated with pathways thatinvolve in generic transcription regulation and cell-cycle progres-sion (Supplementary Fig. S8A). Moreover, it was of great interestto observe that gene sets containing E2F-binding motifs weresignificantly altered by THZ1 inhibitor in all the three cell linemodels (Supplementary Fig. S8B).

We focused on the 2,257 differentially expressed genes thatwere affected by THZ1 in all the three cell lines (SupplementaryTable S4). Gene ontology analysis of these core transcripts inhib-ited by THZ1 revealed a significant enrichment of genes involvedin cell cycle, DNA repair process and transcription regulation (Fig.4B; Supplementary Table S5). Notably, many vital transcriptionfactors with established roles in oncogenesis were uniformlydownregulated upon THZ1 treatment, including numerous zincfinger molecules, ETVs, and most interestingly, several E2F genes.In addition, enrichment analysis of common transcription factorbinding motifs showed that genes containing binding sites oftranscription factors such as E2F, NRF2, and ELK1 were prefer-

entially enriched in this gene set (Supplementary Table S5). Takentogether, these data demonstrated that targeting CDK7 usingTHZ1 led to a widespread transcriptional shutdown in ovariancancer cells, with E2F transcription factors being crucial candidatemediators of the inhibitory effects.

Preferential inhibition of super-enhancer-driven geneexpression by THZ1

Previous work has established that the expression of genesassociated with super-enhancers is disproportionally vulnerableto transcriptional inhibitors, which underlies the extreme sensi-tivity toward TZH1 in several types of cancer (24, 25). However,the identification and functional annotation of super-enhancersin ovarian cancer are elusive. We therefore sought to investigatethe possible contribution of super-enhancers to the remarkablesusceptibility to THZ1 in our ovarian cancer models. We per-formed chromatin immunoprecipitation followed by next-gen-eration sequencing (ChIP-seq) using an antibody recognizinghistone H3K27 acetylation (H3K27ac), to estimate the positionsand sizes of active enhancer regions in COV 413B,OVCA420, andSKOV3 cells. Stitched enhancer domainswere rankedbyH3K27acsignal intensity and in each cell line (Supplementary Tables S6–S8), a subset of enhancers contained an appreciably higherH3K27ac load than regular enhancers and was identified assuper-enhancers (Fig. 5A). GSEA revealed that genes with top-ranked enhancers involved in regulation of embryonic develop-ment, gene transcription, and cancer pathways (SupplementaryFig. S9A). In addition, these genes were significantly enrichedfor transcripts featured by E2F-binding motifs (SupplementaryFig. S9B). There was a considerable overlap of SE-associated genes

A BTHZ1 (nmol/L)

All

11,0

14 a

ctiv

e tra

nscr

ipts

(FP

KM

>1)

COV 413B

Gene expressionLog2 Fold Change

0

0-4 4

100

All

10,9

49 a

ctiv

e tra

nscr

ipts

(FP

KM

>1)


OVCA420

0-4 4

0 100

All

9,75

3 ac

tive

trans

crip

ts (F

PK

M>1

)


SKOV3

0-4 4

0 100

0 5 10 15 20 25 30 35 40 45

Spliceosome

Homologous recombination

Cell cycle

RNA transport

Fanconi anemia pathway

Double-stranded DNA binding

Poly(A) RNA binding

Protein binding

Centrosome

Mitochondrion

Nucleolus

Nucleus

Nucleoplasm

Transcription, DNA-templated

Double-strand break repair via synthesis-dependent strand annealing

Organelle organization

tRNA processing

Mitotic nuclear division

DNA replication

Double-strand break repair via homologous recombination

Gene expression

Double-strand break repair

Mitotic cell cycle

DNA repair

Cell division

Biological process

Cellular component

Molecular function

KEGG Pathway

-Log10 FDR

Figure 4.

THZ1-inhibited gene transcription in ovarian cancer. A, Heatmaps of global gene expression values in COV 413B, OVCA420, and SKOV3 cells that were treated withTHZ1 (100 nmol/L for 6 hours) versus DMSO control. B, Gene ontology categories and KEGG pathways overrepresented in the core set of differentiallyexpressed transcripts that were inhibited by THZ1 across all three cell lines.






in three cell lines (Supplementary Fig. S9C), although each celltype displayed a distinct super-enhancer profile.

We found that many super-enhancers were adjacent toknown oncogenic molecules and lineage-specific transcriptionfactors previously implicated in ovarian cancer progression,such as PAX8, FGF18, and SRC (refs. 34–36; Fig. 5B; Supple-mentary Fig. S9D). Furthermore, we identified a number of newSE-associated genes whose functions had not been described inthe context of ovarian cancer (Fig. 5A). For example, Zinc FingerMIZ-Type Containing 1 (ZMIZ1), encoding a PIAS-like proteinrelated to NOTCH and androgen receptor signaling (37, 38),was associated with super-enhancers in all the three cell lines.We also identified a common super-enhancer at the ITGB6gene, a less studied integrin subunit with proposed functional

role in breast and colon cancer (refs. 39, 40; Fig. 5B). Todetermine whether THZ1 conferred preferential inhibition ofSE-driven gene expression in ovarian cancer, GSEA was con-ducted and showed a significant enrichment of genes associatedwith super-enhancers in THZ1-targeted transcripts (Fig. 5C).Similar results were achieved in COV 413B, OVCA420, andSKOV3 cell lines, emphasizing the special sensitivity of SE-driven gene expression to THZ1.

Ovarian cancer genes defined by E2F-regulated and super-enhancer-associated transcripts

To further corroborate the biological significance of E2F-medi-ated transcription in the context of super-enhancers, we function-ally inferred a cluster of genes that were defined by being both

A

B

C

COV 413B

LAMA5 (4)ZMIZ1 (6)ITGB6 (9)TRIO (39)SOX4 (125)EGFR (226)PAX8 (287)KLF13 (301)PAX2 (348)FGF18 (349)MYC (439)HES1 (483)

CDK5 (750)E2F3 (862)

KLF9 (528)

0.0

0 5,000 10,000

Ranked enhancers Ranked enhancers Ranked enhancers15,000 20,000 0 5,000 10,000 15,000 0 5,000 10,000 15,000 20,000

1.0

2.0

3.0

0.0

0.5

1.0

1.5

0.0

0.5

1.0

1.5

OVCA420 SKOV3

H3K

27A

c S

igna

l (×1

05 )

979

supe

r-en

hanc

ers

752

supe

r-en

hanc

ers

237

supe

r-en

hanc

ers

COV 413B OVCA420 SKOV3

THZ1 versus DMSOTHZ1 versus DMSO THZ1 versus DMSO

Enr

ichm

ent s

core

-0.10-0.15-0.20

-0.050.000.050.10

Enr

ichm

ent s

core

-0.20-0.15-0.10

-0.25-0.30

-0.050.000.05

Enr

ichm

ent s

core

-0.10-0.15-0.20-0.25

-0.050.000.05

FDR=0 FDR=0 FDR=0

COV 413B

OVCA420

SKOV3

−1

0

1

2

3

−1

0

1

2

3

−1

0

1

2

3

0 50 100 150 200 250 kb

ZMIZ1

rpm

/bin

0 50 100 150 200 kb

ITGB6

COV 413B

OVCA420

SKOV3

0

2

4

0

2

4

0

2

4

rpm

/bin

7550250 kb

PAX8

COV413B

OVCA420

SKOV3

−1

0

1

2

3

−1

0

1

2

3

−1

0

1

2

3

rpm

/bin

SRC (14)LAMA5 (34)ITGB6 (54)BCL2L1 (65)FGFR3 (89)ID2 (225)ZMIZ1 (233)IGF1R(297)FOSL2 (331)SMAD3 (440)TGFA (466)

CCND1 (497)

JUN (659)HES1 (569)

TRIO (488)

HOXB5 (2)ITGB6 (16)

BCL2L1 (50)CENPA (37)

EHF (79)SLC2A1 (60)

JUNB (87)IGFBP4 (103)FZD2 (112)ZMIZ1 (154)HNF1B (156)SREBF1 (166)

PDGFC (209)PAX8 (242)

LIF (199)

Figure 5.

Preferential inhibition of super-enhancer-driven gene expression by THZ1. A, CHIP-seq identified enhancer regions were rank-ordered on the basis of input-normalized H3K27ac signal (length � density) in COV 413B, OVCA420, and SKOV3 cell lines. B, CHIP-seq profiles for H3K27ac occupancy of representative super-enhancer-associated genes in COV 413B, OVCA420, and SKOV3 cells. The x-axis showed gene tracks, and the y-axis showed the signal of H3K27ac bindingwithin 50 bp bins in units of reads per million bin (rpm/bin). C, GSEA plots indicated downregulation of super-enhancer-associated gene sets in THZ1 versusDMSO-treated ovarian cancer cells.

Zhang et al.





putative E2F-regulated and SE-associated transcripts. These genesincludedMYC, FOSL1, FOSL2, andELK3, someofwhichhadbeenpreviously implicated in ovarian cancer and other malignancies(9, 41). CHIP-seq profiles for H3K27ac occupancy of these genesconfirmed that theywere indeed adjacent to super-enhancers (Fig.6A). As expected, the gene expression levels of MYC, FOSL1,FOSL2, and ELK3 were uniformly sensitive to THZ1 treatmentin COV 413B, OVCA420, and SKOV3 cells, based on RNAsequencing quantification (Fig. 6B). We determined the biolog-ical function of these potential ovarian cancer genes by employingCRISPR-Cas9 system to genetically knock out each of them inCOV 413B, OVCA420, or SKOV3 cell lines. Two independentsingle-guide RNA sequences were tested for each gene and bothresulted in a statistically significant reduction of cell viability,albeit to different extent (Fig. 6C andD). Therefore, E2F-regulatedand super-enhancer-associated transcripts included candidateessential cancer genes towhich ovarian tumorsmight be addicted,

and as a result, targeting CDK7-dependent transcription enabledpreferential and collective suppression ofmultiple oncogenes thatwere critical for ovarian tumorigenesis.

DiscussionThe lack of actionable recurrent genetic alterations has enor-

mously impeded the development of effective targeted thera-peutics in EOC, a highly aggressive disease representing anoutstanding medical challenge. In this study, through unbiasedsmall-molecule screen approach, we found that different his-totypes of ovarian cancer were exceptionally sensitive to CDK7inhibition. Further transcriptomic and epigenetic analyses sug-gested that mechanism of action underlying the efficacy ofCDK7 inhibitors might relate to the preferential repression ofE2F-regulated gene sets and transcripts associated with super-enhancers. Our functional and mechanistic investigations

A

B D

C

ELK3FOSL2FOSL1MYC

MYC FOSL1 FOSL2 ELK3 MYC FOSL1 FOSL2 ELK3 MYC FOSL1 FOSL2 ELK3

sgEGFP

CO

V 4

13B

OV

CA

420

SK

OV

3

sgMYC-1 sgMYC-2 sgFOSL1-1 sgFOSL1-2 sgFOSL2-1 sgFOSL2-2 sgELK3-1 sgELK3-2

0.0

0

50

100

0

50

100

0

50

100

0.5

1.0

0.0

0.5

1.0

0.0 sgEG

FP

sgMY

C-1

sgMY

C-2

sgFOS

L1-1

sgFOS

L1-2

sgFOS

L2-1

sgFOS

L2-2

sgELK

3-1

sgELK

3-2

0.5

1.0

COV 413B

OVCA420

SKOV3

−1012

−1012

−1012

0 25 50 75

COV 413B

OVCA420

SKOV3

−10123

−10123

−10123

0 10 20 30

COV 413B

OVCA420

SKOV3

0

2

4

0

2

4

0

2

4

0 10 20 30

COV 413B

OVCA420

SKOV3

−101234

−101234

−101234

0 5 10 15 kb kb kb kb

rpm

/bin

COV413B

**

*

*

*

**

*

**

*

**

*

* ** * * *

* *

* *

* ** *

* *

* * * ** *

Rel

ativ

e pr

olife

ratio

n

Rel

ativ

e ge

ne e

xpre

ssio

n (%

)

OVCA420

Rel

ativ

e pr

olife

ratio

n

SKOV3

Rel

ativ

e pr

olife

ratio

n

COV413B OVCA420 SKOV3DMSOTHZ1

Figure 6.

Ovarian cancer genes defined by E2F-regulated and super-enhancer-associated transcripts. A, CHIP-seq profiles for H3K27ac occupancy of representative super-enhancer-associated genes in COV 413B, OVCA420, and SKOV3 cells. The x-axis showed gene tracks, and the y-axis showed the signal of H3K27ac bindingwithin 50 bp bins in units of reads per million bin (rpm/bin). B, Relative gene expression ofMYC, FOSL1, FOSL2, and ELK3 in THZ1-treated COV 413B, OVCA420, andSKOV3 cells compared with DMSO control (�P <0.05, Student t test). C,Crystal violet staining showed impaired cell viability following depletion of selected genes inovarian cancer cells. D, Quantification of cell viability upon knockout of indicated genes (� , P < 0.05, Student t test).






unambiguously supported that targeting CDK7-dependenttranscriptional program represented a potential therapeuticstrategy in patients with EOC.

These data unraveled the transcriptional addiction of ovariancancer, which might have important therapeutic implications inthe treatment of this disease. Our previous study and other recentreports have identified BRD4 as a candidate oncogenic driver ofepithelial ovarian carcinoma, whose inhibition by BET bromo-domain inhibitors such as JQ1 resulted in disrupted BET-depen-dent transcriptional machinery and profound cell-cycle arrest ofovarian cancer (9, 10, 42). Here, we further discovered that aselective CDK7 inhibitor, THZ1, exerted potent antineoplasticactivity against both JQ1-responsive and JQ1-refractory ovariancancer cells. We further identified a widespread dependence onCDK7 and consequently an extraordinary therapeutic effect ofCDK7 inhibition across different histologic subtypes of EOC.BRD4 is a crucial regulator of transcriptional elongation thatrecruits the positive transcription elongation factor b (P-TEFb)complex to chromatin, whereas Cdk7 is a pivot subunit of RNAPIIinitiation factor TFIIH which phosphorylates CTD of RNAPIIduring transcription initiation (43, 44). It was noteworthy thatcertain EOC cell lines exhibited discrepancy in their sensitivity toJQ1 and THZ1, indicative of differential dependency on differenttranscriptional regulators. Nevertheless, because both JQ1 andTHZ1 induced profound changes in the transcription landscape,this research unequivocally illustrated that inhibition of tran-scriptional addiction was an effective approach to target ovariancancer that has been shown to lack obvious actionable drivergenetic mutations (5).

The unique property of transcriptional addiction and conspic-uous sensitivity to THZ1 presumably is mechanistically depen-dent on certain factors participating in and/or downstream ofgene transcription that drives ovarian cancer. Our resultshighlighted E2F as a possible functional basis for the therapeuticeffects of CDK7 inhibition. Specifically, gene expression profilingof multiple THZ1-treated EOC cell lines consistently revealeddownregulation of functionally defined E2F target genes. Thesefindingswere in agreementwith decreased abundance of gene setswith E2F-binding sites in the presence of BET bromodomaininhibitors, as reported by several recent studies including oursin ovarian cancer (10, 45). Emerging data indicate that E2Ftranscription factors, which play important roles in cell-cyclecontrol and other tumorigenic processes, directly promote path-ogenesis of ovarian cancer (46, 47). In addition to E2Fs, cell-cycle-related CDKs, for example, CDK1, CDK2, CDK4, and CDK6, werealso found indispensable for ovarian tumor cell growth, implyingthat the intimate interaction between transcriptional machineryand cell-cycle regulation was crucial for EOC progression. Weprovided further insights into the preferential inhibition of E2Ftarget genes with THZ1 by discovering that these genes tended tobe adjacent to large enhancers or even super-enhancers in ovariancancer. Therefore, at least twomodelsmay be proposed to explainthe broad downregulation of THZ1-sensitive genes enriched forE2F-binding motifs. On one hand, E2Fs could function as coac-tivators of SE-mediated gene transcription, andon the other hand,E2F expression is likely altered byCDK7 inhibitors, as exemplifiedby some E2F genes uniformly downregulated across ovariancancer cell lines upon THZ1 treatment. Of note, there are morethan nine members in the mammalian E2F family (48) and thesedistinct gene products may have specific or redundant activitiesfollowing CDK7 inhibition. Moreover, decreased E2F target gene

expression should not be solely responsible for the pleiotropicefficacy of THZ1. We corroborated this point by identifying, forthe first time, a set of SE-marked genes encoding both knownovarian oncoproteins and novel candidate tumor-promotingmolecules. It would be interesting to further investigate thepotential contribution of these previously unrecognized factorsto ovarian malignancy in future work.

Super-enhancers by definition are large clustered enhancerregions, which are frequently associated with genes essential forcell lineage identity and tumor pathogenesis (19, 49). Pertur-bation of transcriptional components typically has a dispro-portionally profound effect on SE-associated genes, whichunderlies the extreme sensitivity toward TZH1 in other cancertypes (24, 25). Nevertheless, a catalogue of super-enhancersand their biological significance in ovarian cancer had not beenthoroughly characterized. By performing CHIP-seq profiling forH3K27ac occupancy, we showed the super-enhancer landscapeof EOC for the first time, to our knowledge, and found thatmany super-enhancers were adjacent to known or novel centralregulators of ovarian cancer progression and were enriched atTHZ1-targeted transcripts. In accordance with previous reports(19, 49), SE-associated genes were considerably different indistinct cell lines, suggestive of oncogenic dependency ondisparate SE-regulated molecules. However, we enumerated agroup of ovarian cancer-driving genes that were uniformlydefined by being both putative E2F-regulated and SE-associatedtranscripts in all three cell models. We illustrated the function-ally important role of these genes including MYC, FOSL1,FOSL2, and ELK3, among which only MYC and FOSL1 hadbeen previously implicated in ovarian tumorigenicity (9, 50).Notably, genetic depletion of each candidate ovarian cancergene generally led to a modest decrease in cell viability, arguingthat downregulation of individual genes by THZ1 was unlikelysolely responsible for its antiproliferative effects. Therefore,preferential and collective suppression of multiple oncogenesconceivably contributed to rendering ovarian cancer exception-ally susceptible to THZ1 treatment.

In conclusion, we have provided extensive preclinical evidencethat transcriptional inhibitors THZ1 serves as a promising ther-apeutic option for the control of diverse ovarian cancers. Althoughtumor cells may display tremendous heterogeneity and plasticityof transcriptional machinery, ovarian cancer generally remainsexquisitely addicted to vibrant transcription ofmyriad oncogenes,especially those associated with super-enhancers. Our study pin-points CDK7 as a new molecular target to modulate pathologicgene expression in EOC and defines a feasible treatment strategyagainst ovarian cancer through the pharmaceutical disruption oftranscriptional program.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: X. Wang, W.-Q. Gao, W. Di, G. ZhuangDevelopment of methodology: Z. Zhang, H. PengAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): Z. Zhang, H. Peng, X. Yin, M. Zhang, W. DiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): X. Wang, X. Yin, M.-C. Cai, W. DiWriting, review, and/or revision of the manuscript: X. Yin, W.-Q. Gao, W. Di,G. Zhuang

Zhang et al.





Administrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): Z. Zhang, H. Peng, P. Ma, Y. Jing, J. Liu,M. Zhang, S. Zhang, K. ShiStudy supervision: W. Di, G. Zhuang

AcknowledgmentsThe authors thank all lab members for helpful discussions and advice

regarding this work.

Grant SupportThis work was supported by the National Natural Science Foundation of

China (81472537 and 81672714, to G. Zhuang; 81502597, to Y. Jing;81472426, to W. Di; 81130038 and 81372189, to W.Q. Gao), the Grants fromthe State Key Laboratory of Oncogenes and Related Genes (no. 91-15-12, to G.Zhuang; SB201510, to Z. Zhang), the grants fromShanghai Jiao TongUniversity

School of Medicine (14XJ10020, to G. Zhuang; DLY201505, to W. Di;YG2016MS51, to X. Yin), Shanghai Natural Science Foundation(16ZR1434600, to Z. Zhang), the Shanghai Institutions of Higher Learning(Eastern Scholarship to G. Zhuang), Shanghai Rising-Star Program(16QA1403600, to G. Zhuang), Shanghai Municipal Commission of Healthand Family Planning (2013ZYJB0202 and 15GWZK0701, to W. Di), the grantfrom Science and Technology Commission of Shanghai Municipality(16140904401, to X. Yin), the Shanghai Health Bureau Key Discipline andSpecialty Foundation and the KC Wong foundation (to W.Q. Gao).

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received January 24, 2017; revised April 10, 2017; accepted May 16, 2017;published OnlineFirst June 1, 2017.

References1. Hennessy BT, Coleman RL, Markman M. Ovarian cancer. Lancet

2009;374:1371–82.2. Banerjee S, Kaye SB. New strategies in the treatment of ovarian cancer:

current clinical perspectives and future potential. Clin Cancer Res2013;19:961–8.

3. Vaughan S, Coward JI, Bast RC Jr, Berchuck A, Berek JS, Brenton JD, et al.Rethinking ovarian cancer: recommendations for improving outcomes.Nat Rev Cancer 2011;11:719–25.

4. Bowtell DD, Bohm S, Ahmed AA, Aspuria PJ, Bast RC Jr, Beral V, et al.Rethinking ovarian cancer II: reducing mortality from high-grade serousovarian cancer. Nat Rev Cancer 2015;15:668–79.

5. Cancer Genome Atlas Research Network. Integrated genomic analyses ofovarian carcinoma. Nature 2011;474:609–15.

6. Yap TA, Carden CP, Kaye SB. Beyond chemotherapy: targeted therapies inovarian cancer. Nat Rev Cancer 2009;9:167–81.

7. Liu J, Matulonis UA. New strategies in ovarian cancer: translating themolecular complexity of ovarian cancer into treatment advances. ClinCancer Res 2014;20:5150–6.

8. Ciriello G, Miller ML, Aksoy BA, Senbabaoglu Y, Schultz N, Sander C.Emerging landscape of oncogenic signatures across human cancers. NatGenet 2013;45:1127–33.

9. Baratta MG, Schinzel AC, Zwang Y, Bandopadhayay P, Bowman-Colin C,Kutt J, et al. An in-tumor genetic screen reveals that the BET bromodomainprotein, BRD4, is a potential therapeutic target in ovarian carcinoma. ProcNatl Acad Sci U S A 2015;112:232–7.

10. Zhang Z, Ma P, Jing Y, Yan Y, Cai MC, Zhang M, et al. BET bromodomaininhibition as a therapeutic strategy in ovarian cancer by downregulatingFoxM1. Theranostics 2016;6:219–30.

11. Filippakopoulos P, Qi J, Picaud S, Shen Y, Smith WB, Fedorov O, et al.Selective inhibition of BET bromodomains. Nature 2010;468:1067–73.

12. Jing Y, Zhang Z, Ma P, An S, Shen Y, Zhu L, et al. Concomitant BET andMAPK blockade for effective treatment of ovarian cancer. Oncotarget2016;7:2545–54.

13. Loven J,OrlandoDA, Sigova AA, Lin CY, Rahl PB, Burge CB, et al. Revisitingglobal gene expression analysis. Cell 2012;151:476–82.

14. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. NatMethods 2012;9:357–9.

15. Trapnell C, Roberts A, Goff L, Pertea G, KimD, Kelley DR, et al. Differentialgene and transcript expression analysis of RNA-seq experiments withTopHat and Cufflinks. Nat Protoc 2012;7:562–78.

16. Anders S, Pyl PT, Huber W. HTSeq—a Python framework to work withhigh-throughput sequencing data. Bioinformatics 2015;31:166–9.

17. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 2009;25:1754–60.

18. Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, et al.Model-based analysis of ChIP-Seq (MACS). Genome Biol 2008;9:R137.

19. Loven J, Hoke HA, Lin CY, Lau A, Orlando DA, Vakoc CR, et al. Selectiveinhibition of tumor oncogenes by disruption of super-enhancers. Cell2013;153:320–34.

20. SubramanianA, TamayoP,Mootha VK,Mukherjee S, Ebert BL,GilletteMA,et al. Gene set enrichment analysis: a knowledge-based approach for

interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A2005;102:15545–50.

21. HuangdaW, ShermanBT, Lempicki RA. Systematic and integrative analysisof large gene lists using DAVID bioinformatics resources. Nat Protoc2009;4:44–57.

22. Shi J, Vakoc CR. The mechanisms behind the therapeutic activity of BETbromodomain inhibition. Mol Cell 2014;54:728–36.

23. Kwiatkowski N, Zhang T, Rahl PB, Abraham BJ, Reddy J, Ficarro SB, et al.Targeting transcription regulation in cancer with a covalent CDK7 inhib-itor. Nature 2014;511:616–20.

24. Chipumuro E, Marco E, Christensen CL, Kwiatkowski N, Zhang T, Hathe-way CM, et al. CDK7 inhibition suppresses super-enhancer-linked onco-genic transcription in MYCN-driven cancer. Cell 2014;159:1126–39.

25. Christensen CL, Kwiatkowski N, Abraham BJ, Carretero J, Al-Shahrour F,Zhang T, et al. Targeting transcriptional addictions in small cell lung cancerwith a covalent CDK7 inhibitor. Cancer Cell 2014;26:909–22.

26. Wang Y, Zhang T, Kwiatkowski N, Abraham BJ, Lee TI, Xie S, et al. CDK7-dependent transcriptional addiction in triple-negative breast cancer. Cell2015;163:174–86.

27. Jiang YY, Lin DC, Mayakonda A, Hazawa M, Ding LW, Chien WW, et al.Targeting super-enhancer-associated oncogenes in oesophageal squamouscell carcinoma. Gut 2016.

28. Akhtar MS, Heidemann M, Tietjen JR, Zhang DW, Chapman RD, Eick D,et al. TFIIH kinase places bivalent marks on the carboxy-terminal domainof RNA polymerase II. Mol Cell 2009;34:387–93.

29. Glover-Cutter K, Larochelle S, Erickson B, Zhang C, Shokat K, Fisher RP,et al. TFIIH-associated Cdk7 kinase functions in phosphorylation of C-terminal domain Ser7 residues, promoter-proximal pausing, and termi-nation by RNA polymerase II. Mol Cell Biol 2009;29:5455–64.

30. Larochelle S, Amat R, Glover-Cutter K, Sanso M, Zhang C, Allen JJ, et al.Cyclin-dependent kinase control of the initiation-to-elongation switch ofRNA polymerase II. Nat Struct Mol Biol 2012;19:1108–15.

31. AsgharU,Witkiewicz AK, TurnerNC, Knudsen ES. The history and future oftargeting cyclin-dependent kinases in cancer therapy. Nat Rev Drug Discov2015;14:130–46.

32. Schachter MM, Fisher RP. The CDK-activating kinase Cdk7: taking yes foran answer. Cell Cycle 2013;12:3239–40.

33. Ganzfried BF, Riester M, Haibe-Kains B, Risch T, Tyekucheva S, Jazic I, et al.curatedOvarianData: clinically annotated data for the ovarian cancertranscriptome. Database 2013;2013:bat013.

34. Cheung HW, Cowley GS, Weir BA, Boehm JS, Rusin S, Scott JA, et al.Systematic investigation of genetic vulnerabilities across cancer cell linesreveals lineage-specific dependencies in ovarian cancer. Proc Natl Acad SciU S A 2011;108:12372–7.

35. Wei W, Mok SC, Oliva E, Kim SH, Mohapatra G, Birrer MJ. FGF18 as aprognostic and therapeutic biomarker in ovarian cancer. J Clin Invest2013;123:4435–48.

36. Wiener JR, Nakano K, Kruzelock RP, Bucana CD, Bast RC Jr, Gallick GE.Decreased Src tyrosine kinase activity inhibits malignant human ovariancancer tumor growth in a nude mouse model. Clin Cancer Res1999;5:2164–70.






37. Pinnell N, Yan R, Cho HJ, Keeley T, Murai MJ, Liu Y, et al. The PIAS-likecoactivator zmiz1 is a direct and selective cofactor of notch1 in T celldevelopment and leukemia. Immunity 2015;43:870–83.

38. SharmaM, Li X,Wang Y, ZarnegarM,HuangCY, Palvimo JJ, et al. hZimp10is an androgen receptor co-activator and forms a complex with SUMO-1 atreplication foci. EMBO J 2003;22:6101–14.

39. Moore KM, Thomas GJ, Duffy SW, Warwick J, Gabe R, Chou P, et al.Therapeutic targeting of integrin alphavbeta6 inbreast cancer. JNatl CancerInst 2014;106:1–14.

40. Bates RC, Bellovin DI, Brown C, Maynard E, Wu B, Kawakatsu H, et al.Transcriptional activation of integrin beta6 during the epithelial-mesen-chymal transition defines a novel prognostic indicator of aggressive coloncarcinoma. J Clin Invest 2005;115:339–47.

41. LockwoodWW, Zejnullahu K, Bradner JE, Varmus H. Sensitivity of humanlung adenocarcinoma cell lines to targeted inhibition of BET epigeneticsignaling proteins. Proc Natl Acad Sci U S A 2012;109:19408–13.

42. Yokoyama Y, Zhu H, Lee JH, Kossenkov AV, Wu SY, Wickramasinghe JM,et al. BET inhibitors suppress ALDH activity by targeting ALDH1A1 super-enhancer in ovarian cancer. Cancer Res 2016;76:6320–30.

43. Shu S, Polyak K. BET bromodomain proteins as cancer therapeutic targets.Cold Spring Harb Symp Quant Biol 2016;81:123–9.

44. Heidemann M, Hintermair C, Voss K, Eick D. Dynamic phosphorylationpatterns of RNA polymerase II CTD during transcription. Biochim BiophysActa 2013;1829:55–62.

45. Chapuy B, McKeown MR, Lin CY, Monti S, Roemer MG, Qi J,et al. Discovery and characterization of super-enhancer-associateddependencies in diffuse large B cell lymphoma. Cancer Cell 2013;24:777–90.

46. De Meyer T, Bijsmans IT, Van de Vijver KK, Bekaert S, Oosting J, VanCriekinge W, et al. E2Fs mediate a fundamental cell-cycle deregulation inhigh-grade serous ovarian carcinomas. J Pathol 2009;217:14–20.

47. Reimer D, Sadr S,Wiedemair A, Stadlmann S, ConcinN, Hofstetter G, et al.Clinical relevance of E2F family members in ovarian cancer–an evaluationin a training set of 77 patients. Clin Cancer Res 2007;13:144–51.

48. Chen HZ, Tsai SY, Leone G. Emerging roles of E2Fs in cancer: an exit fromcell cycle control. Nat Rev Cancer 2009;9:785–97.

49. Hnisz D, Abraham BJ, Lee TI, Lau A, Saint-Andre V, Sigova AA, et al.Super-enhancers in the control of cell identity and disease. Cell 2013;155:934–47.

50. Kurimchak AM, Shelton C, Duncan KE, Johnson KJ, Brown J, O'Brien S,et al. Resistance to BET Bromodomain Inhibitors Is Mediated by KinomeReprogramming in Ovarian Cancer. Cell Rep 2016;16:1273–86.


Zhang et al.




2017;16:1739-1750. Published OnlineFirst June 1, 2017.Mol Cancer Ther Zhenfeng Zhang, Huixin Peng, Xiaojie Wang, et al. CDK7-Dependent Transcriptional Addiction in Ovarian CancerPreclinical Efficacy and Molecular Mechanism of Targeting

Updated version

10.1158/1535-7163.MCT-17-0078doi:

Access the most recent version of this article at:

Material

Supplementary

http://mct.aacrjournals.org/content/suppl/2017/06/01/1535-7163.MCT-17-0078.DC1

Access the most recent supplemental material at:

Cited articles

http://mct.aacrjournals.org/content/16/9/1739.full#ref-list-1

This article cites 48 articles, 12 of which you can access for free at:

Citing articles

http://mct.aacrjournals.org/content/16/9/1739.full#related-urls

This article has been cited by 3 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://mct.aacrjournals.org/content/16/9/1739To request permission to re-use all or part of this article, use this link



http://mct.aacrjournals.org/lookup/doi/10.1158/1535-7163.MCT-17-0078

http://mct.aacrjournals.org/content/suppl/2017/06/01/1535-7163.MCT-17-0078.DC1

http://mct.aacrjournals.org/content/16/9/1739.full#ref-list-1

http://mct.aacrjournals.org/content/16/9/1739.full#related-urls

http://mct.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mct.aacrjournals.org/content/16/9/1739


Documents

Preclinical Efﬁcacy and Molecular Mechanism of Targeting ... · Preclinical Efﬁcacy and Molecular Mechanism of Targeting CDK7-Dependent Transcriptional Addiction in Ovarian Cancer