ORIGINAL ARTICLE

Sequential treatment with aurora B inhibitors enhancescisplatin-mediated apoptosis via c-Myc

Yaxi Ma & Handi Cao & Siyue Lou & Xuejing Shao &

Wen Lv & Xiaotian Qi & Yujia Liu & Meidan Ying &

Qiaojun He & Xiaochun Yang

Received: 12 June 2014 /Revised: 13 October 2014 /Accepted: 6 November 2014# Springer-Verlag Berlin Heidelberg 2014

AbstractPlatinum compound such as cisplatin is the first-line chemo-therapy of choice in most patients with ovarian carcinoma.However, patients with inherent or acquired cisplatin resis-tance often experience relapse. Therefore, novel therapies areurgently required to treat drug-resistant ovarian carcinoma.Here, we showed that compared to the non-functional tradi-tional simultaneous treatment, sequential combination of Au-rora B inhibitors followed by cisplatin synergistically en-hanced apoptotic response in cisplatin-resistant OVCAR-8cells. This effect was accompanied by the induction of poly-ploidy in a c-Myc-dependent manner, as c-Myc knockdownreduced the efficacy of the combination by suppressing theexpression of Aurora B and impairing cellular response toAurora B inhibitor, as indicated by the decreased polyploidyand hyperphosphorylation of histone H1. In c-Myc-deficientSKOV3 cells, c-Myc overexpression restored Aurora B

expression, induced polyploidy after inhibition of Aurora B,and sensitized cells to this combination therapy. Thus, ourreport reveals for the first time that sequential treatment ofAurora B inhibitors and cisplatin is essential to inhibit ovariancarcinoma by inducing polyploidy and downregulating c-Mycand that c-Myc is identified as a predictive biomarker to selectcells responsive to chemotherapeutical combinations targetingAurora B. Collectively, these studies provide novel ap-proaches to overcoming cisplatin chemotherapy resistance inovarian cancer.

Key Message& Pretreatment of Aurora B inhibitors augment apoptotic

effects of cisplatin.& The synergy of Aurora B inhibitor with cisplatin is depen-

dent on c-Myc expression.& c-Myc-dependent induction of polyploidy sensitizes cells

to cisplatin.

Keywords Aurora B . c-Myc . Cisplatin . Chemotherapyresistance . Ovarian carcinoma . Drug in combination

Introduction

Ovarian carcinoma is the leading cause of death for gyneco-logical malignancies in the western world. Depending on thestage, the 5-year-survival rate ranges from 80 to 90 % (Inter-national Federation of Gynecology and Obstetrics (FIGO) I)to 11 % (FIGO IV) [1]. Conventional treatment for late-stageovarian carcinoma is surgical resection followed by platinum/taxane combination chemotherapy. Although this treatmentstrategy is effective as the first-line therapy of choice, resis-tance to chemotherapy ultimately results in recurrence, whichoccurs in up to 75 % of ovarian carcinoma patients [2]. Thus,

Yaxi Ma and Handi Cao contributed equally to this work.

Electronic supplementary material The online version of this article(doi:10.1007/s00109-014-1228-0) contains supplementary material,which is available to authorized users.

H. Cao : S. Lou :X. Shao :X. Qi :Y. Liu :M. Ying :Q. He :X. Yang (*)Institute of Pharmacology and Toxicology, Zhejiang Province KeyLaboratory of Anti-Cancer Drug Research, College ofPharmaceutical Sciences, Zhejiang University, Hangzhou 310058,Chinae-mail: yangxiaochun@zju.edu.cn

Y. MaDepartment of Gynecology, The Second Affiliated Hospital, Schoolof Medicine, Zhejiang University, Hangzhou 310009, China

W. LvTongde Hospital of Zhejiang Province, Hangzhou 310009, China

J Mol MedDOI 10.1007/s00109-014-1228-0

there is an urgent need to find more effective chemotherapeu-tic agents to treat highly malignant tumors.

Aurora B, a protein kinase regulated by the c-Myc onco-gene [3], is involved in a series of mitotic functions, includingsister chromatin cohesion, chromosome–microtubule interac-tions, spindle-assembly checkpoint, and cytokinesis [4, 5].Aurora B is found to be overexpressed in ovarian carcinomacells compared to normal ovarian tissues, which correlatedwith poor differentiation, positive ascites cytology, lymphnode metastases, and shorter progression-free survival andoverall survival [6]. Inhibition of Aurora B activity results inimpaired chromosome alignment, abrogation of the mitoticcheckpoint, induction of polyploidy, and cell death [7, 8].Therefore, Aurora B is thought to be a potential target forthe treatment of ovarian carcinoma [9, 10].

Several small-molecule inhibitors of Aurora have beenproposed as antitumor agents and are currently in clinical trials[7]. One of these, AZD1152 (barasertib), is an acetanilide-substituted pyrazole-aminoquinazoline phosphate pro-drugthat is rapidly converted into the more active hydroxy-quinazoline pyrazole anilide of AZD1152 (AZD1152-hQPA)in plasma and acts as a reversible ATP-competitive inhibitorof Aurora B [7, 11]. AZD1152 induces chromosomemisalign-ment, terminates cell division with the introduction of poly-ploidy, reduces cell viability, and promotes apoptosis in cancercells [11]. In immune-deficient mice, AZD1152 potentlyinhibited the growth of human breast, colon, lung, and hema-tologic tumor xenografts while efficiently inhibited pulmo-nary metastatic nodule formation in a breast cancer model [8,12]. Moreover, AZD1152 synergistically enhanced the effectsof vincristine to induce growth arrest of Burkitt lymphoma/leukemia cells [13].

Our previous study has reported the synergistic antitumoractivity of AZD1152 and cisplatin in ovarian carcinoma by theinduction of cell apoptosis [14]. However, its mechanism isstill largely unknown. The purpose of the present study is todetermine whether sequential administration of AZD1152 andcisplatin to cisplatin-insensitive ovarian carcinoma cells canreverse the chemotherapy resistance by inducing mitoticarrest-associated apoptosis and clarify the role which c-Mycplayed during this process. Our studies discovered that com-pared to the moderate level of cell apoptosis after simulta-neous treatment, sequential combination of AZD1152 andcisplatin significantly induced apoptosis in cisplatin-resistantOVCAR-8 cells, whereas the c-Myc expression level deter-mined ovarian carcinoma’s sensitivity to this combinationstrategy. Therefore, these findings reveal for the first time thatsequential treatment of Aurora B inhibitors with cisplatin inovarian carcinoma is essential for the induction of polyploidyand inhibition of proliferation and suggest the potential ofapplying c-Myc as an effective biomarker for designing ratio-nal chemotherapeutical combinations of Aurora B inhibitorsand cisplatin to target ovarian carcinoma.

Materials and methods

Reagents

AZD1152-hQPA (AZD1152) and ZM447439 were purchasedfrom Selleck Chemicals (Houston, TX, USA, Cat # S1147 andS1103). Cisplatin and carboplatin were purchased from QiluPharmaceutical Co. (JiNan, China). Taxol was purchasedfrom Xiehe Pharmaceutical Co. (Beijing, China).

Cell lines and cell culture

Human ovarian cancer cell lines A2780, OVCAR-8, SKOV3,and ES-2; osteosarcoma cell lines U2OS and KHOS; breastcancer cell line MDA-MB-468; lung cancer cell lines NCI-H1299 and H460; bladder cancer cell line 5637; and prostatecancer cell line PC3 were purchased from the Shanghai Insti-tute of Biochemistry and Cell Biology (Shanghai, China).Following receipt, cells were grown and frozen as a seed stockas they were available. All 11 cell lines were passaged for amaximum of 2 months after which new seed stocks werethawed, and all cell lines were tested for mycoplasma contam-ination at least every month. The MDA-MB-468 cells weremaintained in Leibovitz’s L-15 medium supplemented with10 % fetal bovine serum, and A2780 cells were cultured inDulbecco’s modified Eagle’s medium supplemented with10 % fetal bovine serum. ES-2 cells were maintained inMcCoy’s 5A medium supplemented with 10 % fetal bovineserum. All other cells were cultured in RPMI-1640 mediumsupplemented with 10 % fetal bovine serum.

Cell proliferation assay

Cell proliferation was assessed by sulforhodamine B (SRB)cell viability assay. Briefly, cells were plated in 96-well platesand treated with serial concentrations of Aurora B inhibitors(AZD1152 or ZM447439), with or without the combinationof platin-based chemotherapeutical agents (cisplatin orcarboplatin) or taxane-based chemotherapeutical drug (taxol)for indicated times. A complete change of medium was per-formed after a 24-h pretreatment of Aurora B inhibitors intreatment schedule C to terminate their inhibitory effect. Cellswere fixed with 10 % (wt/vol) trichloroacetic acid and stainedwith 4 mg/mL SRB for 15 min, after which the excessive dyeis removed by washing repeatedly with 1 % (vol/vol) aceticacid. The protein-bound dye was dissolved in 10 mM Tris-base solution for optical density determination at 510 nmusing a Multiskan Spectrum (Thermo Electron Co, Vantaa,Finland). The cell survival rate for each well was calculated as(A510treated cells/A510control cells)×100 %. The average IC50

values were determined by Logit method from at least threeindependent tests.

J Mol Med

DNA content analyses

Ovarian cancer cells were plated in six-well plates at a densityof 5×105/well and treated with cisplatin or Aurora B inhibi-tors (AZD1152 or ZM447439) for indicated times. Cells werethen harvested and washed with ice-cold phosphate-bufferedsaline (PBS), fixed with 75 % ethanol overnight at 4 °C. Thecell pellets were resuspended and stained in 1 mL of 0.1 %sodium citrate containing 0.01 mg propidium iodide (PI) and50 μg RNase for 30 min at room temperature protected fromlight. Flow cytometric analysis was performed onFACSCalibur cytometer (Becton Dickinson, San Jose, CA,USA). The DNA content and cell cycle status were analyzedusing ModFit software (Verity Software).

Quantitative real-time RT-PCR analysis

Total RNAwas prepared with TRIzol (Invitrogen) and cDNAwas synthesized from 2 μg of RNA with ReverTra Ace(Toyobo). Quantitative real-time RT-PCR (qRT-PCR) wasperformed with SYBR Green PCR Kits (Qiagen Inc.). Theprimers used were as follows: p21cip/kip, forward primer 5′-TGTGATGCTAGGAACATGAGCAA-3′ and reverse primer5′-GTGGAAAGCCC-AAGCCTGAA-3′; GAPDH, forwardprimer 5′-GTCATCCATGACAACTTTGG-3′ and reverseprimer 5′-GAGCTTGACAAAGTGGTCGT-3′. Values arenormalized to GAPDH and expressed as fold increases rela-tive to the reference sample.

Western blotting (WB) analyses

WB analyses were performed as described [15]. Antibodiesagainst cleaved caspase-3 and Aurora B were from Cell Sig-naling Technology (Danvers, MA, USA). Phosphorylatedhistone H1 antibody was purchased from EMD Millipore(Darmstadt, Germany). All other antibodies were purchasedfrom Santa Cruz Biotechnology (Santa Cruz, CA, USA). WBwas semiquantified by Quantity One 1-D Analysis Software(Bio-Rad, Hercules, CA). β-Actin or GAPDH was used as aninternal control for normalization of protein quantity.

Nuclear morphology analyses

Cells were washed once with PBS and then incubated with 1mLPBS containing 0.1 % Triton and 0.1 % 4',6-diamidino-2-phenylindole (DAPI). The morphology of cells’ nuclei wasobserved under a fluorescence microscope (Leica DMI 4000B,Buffalo Grove, IL, USA) at excitation wavelength of 350 nm.

c-Myc gene silencing by small interfering RNA

OVCAR-8 cells were plated in six-well plates at a density of3×104/well. The next day, the medium was replaced with

Opti-MEM I Reduced Serum Media (Invitrogen, Carlsbad,CA, USA), and cells were transfected with small interferingRNA (siRNA) against c-Myc (siRNA-c-Myc) (GenePharma,China) or control siRNA (siRNA-NC) to a final siRNA con-centration of 20 nM using oligofectamine reagent(Invitrogen). The sense sequence of the c-Myc siRNA is 5′-CGGGGCUUUAUCUAACUCG-3′.

Virion production and lentiviral transduction

Recombinant lentivirus was produced by co-transfecting293FT cells with c-Myc lentiviral construct and control plas-mid with pRD8.9 and pMD.G packaging vectors. Virionproduction, titration, and transduction were performed as pre-viously described [15].

Statistical analyses

Descriptive statistics, including means and standard devia-tions, were computed at each time point for each experimentaland control group, using Student’s unpaired two-tailed t test.Two-tailed P values of 0.05 or less were considered statisti-cally significant. Statistical data shown are means±standarddeviations that represent at least triplicate assays. Combina-tion index (CI) is well accepted for quantifying drug syner-gism based on the multiple drug effect equation of Chou-Talalay [16]. In our study, CI values were computed usingCalcusyn (Biosoft, Cambridge, UK), and values lower than0.95 indicated synergism [17].

Results

Sequential application of Aurora B inhibitors enhancesthe activity of cisplatin in ovarian cancer cells

As Aurora B is involved inmultiple processes that regulate thesensitivity of cancer cells to DNA-targeted anticancer drugs[18, 19], we asked whether different treatment schedules ofAurora B inhibitors could affect the effectiveness of combi-nation chemotherapy with cisplatin. First, we assayed thesensitivity of four different ovarian cancer cell lines againstAurora B inhibitor [AZD1152-hQPA (AZD1152 or AZD)]and cisplatin. As shown in Fig. 1a, b, in contrast to SKOV3cells, OVCAR-8 cells were resistant to cisplatin while sensi-tive to AZD1152, whereas A2780 cells showed a moderateresponse to both drugs. A high-grade serous ovarian cancerES-2 cell line [20] showed similar sensitivity to AZD1152 asthat of OVCAR-8 cells, as well as certain resistance to cis-platin. Cell cycle analysis also confirmed the varied sensitivityof OVCAR-8, ES-2, and SKOV3 cells to AZD1152 orZM447439 treatment, as indicated by the increase of poly-ploid cells with DNA content greater than 4N (Fig. 1c, d).

J Mol Med

Next, three different treatment schedules were designed totest the combination efficiency of cisplatin with Aurora B

inhibitors. In the simultaneous treatment schedule, both drugswere maintained in the cell culture medium for 72 h before

J Mol Med

assaying cell density (Fig. 1e, treatment A). In separate exper-imental schedules, to ensure the inhibition of Aurora B sig-naling, cells were pretreated with Aurora B inhibitors 24 hbefore the addition of cisplatin, and Aurora B inhibitors wereeither present (Fig. 1e, treatment B) or absent (Fig. 1e, treat-ment C) for the next 72 h. Compared to treatment A, sequen-tial Aurora B inhibitors followed by cisplatin in both treat-ments B and C synergistically enhanced the antiproliferativeeffects of cisplatin in both OVCAR-8 and ES-2 cells(Fig. 1f, g). This enhancement of inhibitory activity is partiallydue to the elevated transcription and expression of cell cycleinh ib i to r p21 (F ig . 1h) . In te res t ing ly, ano the rchemotherapeutical drug, carboplatin, but not taxol, depicteda synergistic effect in combination with AZD1152 inOVCAR-8 cells (Supplementary Fig. S1). These data suggestthat by inhibiting Aurora B signaling response and introduc-ing polyploidy in ovarian cancer cells, the sequential combi-nation of AZD1152 or ZM447439 can synergistically inhibitcell proliferation induced by cisplatin.

The sequential combination of AZD1152 with cisplatininduces apoptosis in cisplatin-resistant OVCAR-8 cells

AZD1152 is reported to inhibit the growth of human tumorxenograft by the induction of apoptosis [8]. To further verifywhether the inhibition of proliferation induced by sequentialtreatment of AZD1152 and cisplatin resulted from apoptosis,we examined the mode of cell death in cisplatin-resistantOVCAR-8 and ES-2 cells. Changes of cellular and nuclearmorphology monitored by bright-light microscopy and DAPI

staining showed the formation of large multinucleated giantcells in OVCAR-8 and ES-2 cells treated with AZD1152,which indicated endoreduplication. Sequential combinationtreatment of cisplatin with or without continuous AZD1152exposure for 72 h markedly induced characteristic apoptosis-related phenotypic responses such as cell condensation andnuclear fragmentation, as compared to that of simultaneoustreatment (Fig. 2a, b). PI staining and measurement of sub-G1fractions by cell cycle analysis further confirmed that sequentialadministration of AZD1152 followed by cisplatin significantlyincreased the percentage of apoptotic cells (Fig. 2c, d). Thus, ascells exposed to drug combinations were more prone to apo-ptosis, WB analyses of poly ADP-ribose polymerase (PARP)and caspase cleavage were performed to document the molec-ular events underlying the process. A 72-h treatment ofOVCAR-8 cells with cisplatin induced both PARP and caspase3/8 cleavage (Fig. 2e, lanes 3, 7, and 11). Combination treat-ment with AZD1152 increased the extent of PARP and caspase3/8 cleavage, with respect to either agent used alone (Figs. 2e(lanes 4, 8, and 12 vs. lanes 3, 7, and 11) and 2f). The cleavageof apoptotic markers was more evident in sequential schedulesB and C, which was consistent with the apoptosis patterndetected by microscopy and flow cytometry. Similar changesof apoptotic marker expression were also observed in ES-2 cellline (Fig. 2g). These results clearly support the concept thatcombination of AZD1152 with cisplatin induces schedule-dependent apoptosis in cisplatin-resistant ovarian cancer cells.

c-Myc expression is required for the synergistic effectof cisplatin and AZD1152 in different solid tumor cells

Since cisplatin is the first-line choice of chemotherapy in thetreatment of osteosarcoma, breast cancer, non-small-cell lungcarcinoma, bladder cancer, and prostate cancer, we next ex-amined the cellular sensitivity to cisplatin with combination ofAZD1152 in cell lines representative of these cancers bygrowth inhibition assays using the sequential treatment sched-ule B (Fig. 1e). Growth assays revealed that the degrees ofproliferation inhibition were cell-dependent (Fig. 3a). Weanalyzed the CI values reflecting synergism or antagonismin the combination of 4 μM cisplatin with 0.5 μM AZD1152using the calculation method of Chou and Talalay [16]. Apartfrom MDA-MB-468 and H460 cells representing breast can-cer and non-small-cell lung carcinoma, the CI values were lessthan 0.95 (synergism) for the other four cell lines (Fig. 3b).Thus, cisplatin in combination with AZD1152 showed syner-gistic effects in various solid tumors. To elucidate the molec-ular mechanisms underlying the different outcomes of thistreatment strategy, we focused on the expression of c-Myc, atranscription factor regulating Aurora B expression [3] andwhose overexpression is commonly found in a wide variety ofhuman malignancies [21]. WB analysis revealed that c-Myc ishighly expressed in cells in which cisplatin and AZD1152

�Fig. 1 Sequential application of Aurora B inhibitors enhances theactivity of cisplatin in ovarian cancer cells. a Cell sensitivity to cisplatinwas determined by SRB assay. The mean IC50 values of cisplatin inA2780, OVCAR-8, ES-2, and SKOV3 cells were 16.8, 37.2, 7.8, and5.9 μM, respectively. b Cell sensitivity to AZD1152 was determined bySRB assay. The mean IC50 values of AZD1152 in A2780, OVCAR-8,ES-2, and SKOV3 cells were 7.1, 0.4, 1.8, and 19.5 μM, respectively. cRepresentative cell cycle distribution of OVCAR-8, ES-2, and SKOV3cells after 24-h treatment of Aurora B inhibitors, with 4N-8N DNArepresenting polyploidy. d Quantification of cell cycle distribution inOVCAR-8, ES-2, and SKOV3 cells after 24-h treatment of AZD1152.e Three different drug combination schedules were designed to test theefficacy of combination therapy. Apart from treatment A in which twodrugs were given simultaneously, treatment schedules B and C included a1-day pretreatment of AZD1152, with (treatment B) or without (treatmentC) the presence of AZD1152 with the daily cisplatin treatment for thenext 72 h. f, g The proliferation of OVCAR-8 cells (f) or ES-2 cells (g)under different combination schedules of cisplatin with either AZD1152or ZM447439. h From OVCAR-8 cells treated with either 100 nMAZD1152 or 15 μM cisplatin alone, or both under different treatmentschedules, total RNAwas extracted and p21cip/kip mRNA expression wasanalyzed by RT-PCR using GAPDH as a housekeeping gene. Untreatedcontrols were scaled to 100 %. WB analysis depicted expression of p21protein in three different treatment schedules. Ab, antibody. *P<0.05;**P<0.01. Data are expressed as mean±SD. Data represent at least threeindependent experiments. AZD, AZD1152; ZM, ZM447439

J Mol Med

J Mol Med

displayed synergistic effect (Fig. 3c), indicating that expres-sion of c-Myc may be a useful biomarker for predicting thesensitivity of solid tumor cells to this combination strategy.Together, these results support the notion that c-Myc expres-sion is required for the synergistic effect of cisplatin andAZD1152 and suggest the possibility of applying this combi-nation strategy on solid tumors where cisplatin is the first-linechoice of chemotherapy.

c-Myc expression level determines ovarian carcinoma’ssensitivity to the sequential combination of AZD1152and cisplatin

To test whether c-Myc regulates Aurora B expression inovarian carcinoma, and clarify its function in the combinationof AZD1152 and cisplatin, we examined the expression of c-Myc and Aurora B in cisplatin-sensitive SKOV3 andcisplatin-resistant OVCAR-8 cells. We found that c-Myc pos-itively correlated with Aurora B and was highly expressed inOVCAR-8 cells responsive to the combination therapy(Fig. 4a). However, c-Myc decreased as cells undergo thesynergistic inhibition of proliferation and induction of apopto-sis (Fig. 4b, lanes 8 and 12).

In order to test whether c-Myc could influence the syner-gistic interaction between cisplatin and AZD1152, a RNAinterference approach taking advantage of a synthetic siRNAspecifically targeting the c-Myc transcript was used to inhibitits expression. While the negative control plasmid displayedno effect on c-Myc protein level, the c-Myc targeting siRNAplasmid markedly downregulated c-Myc expression (Fig. 4c).Of note, under treatment schedule B, the synergism observedin OVCAR-8 cells was impaired after c-Myc knockdown asindicated by the higher survival rate and reduced apoptosis(Fig. 4d, e). In an attempt to better define the role of c-Myc, we

carried out transient transfections in c-Myc-deficient SKOV3cells with a green fluorescent protein (GFP)-tagged lentivirusexpressing the full-length c-Myc. An increased expression ofthe c-Myc-GFP protein was achieved in SKOV3 cells, assupported by fluorescence microscopy and WB analysis(Fig. 4f). After sequential combination of AZD1152 andcisplatin in c-Myc-expressing cells and control cells (i.e., cellsexpressing GFP vector), we observed that c-Myc expressionsensitized SKOV3 cells to this combination treatment, asevident by the inhibition of proliferation and induction ofapoptosis (Fig. 4g, h). These data support the notion that c-Myc is relevant for the occurrence of a favorable combinationinteraction in cisplatin-resistant ovarian cancer cells.

Aurora B-mediated mitotic functions in ovarian carcinoma areregulated by c-Myc

AsAurora B is regulated by c-Myc, we wish to further explorethe relationship between the combination efficiency of cisplat-in and AZD1152 in ovarian carcinoma and the mitotic func-tions mediated by Aurora B. In combination-responsiveOVCAR-8 cells, knockdown of c-Myc by siRNA correspond-ingly downregulated the expression of Aurora B (Fig. 5a) andled to a decrease in polyploidy when treated with AZD1152(Fig. 5b, c). In c-Myc-deficient SKOV3 cells, the lentiviraloverexpression of c-Myc rescued the expression of Aurora B(Fig. 5d), and restored sensitivity to AZD1152, as indicated bythe marked increase of polyploid cells (Fig. 5e, f). Consistentwith the inhibition of Aurora B by AZD1152, the phosphor-ylation of histone H1, a substrate of Aurora B [22], was alsodownregulated after the treatment of AZD1152 or sequentialcombination (Fig. 5g, lanes 6, 8, and 12). In addition, c-Mycknockdown in OVCAR-8 cells retained the phosphorylationof histone H1 after treatment with AZD1152 (Fig. 5h). Theseresults demonstrate that the relevance of c-Myc to the efficacyof this combination strategy may rely on the regulation ofAurora B-mediated mitotic functions, possibly via modulatingthe phosphorylation of histone H1.

Discussion

Due to the crucial function of Aurora signaling in sustainingsurvival of ovarian carcinoma cells [6], study of howAurora Binhibition impacts on tumor cell sensitivity to platinum treat-ment is of considerable interest in understanding tumor cellbiology and finding targeted therapy. In the present study,using a cisplatin-resistant ovarian carcinoma cell line withhigh-level of Aurora B expression, we find that the sequentialcombination of cisplatin with AZD1152, a potent Aurora Bselective inhibitor, displays a synergistic effect in inhibitingcell proliferation and inducing apoptosis. The synergistic

�Fig. 2 The sequential combination of AZD1152 with cisplatin inducesapoptosis in cisplatin-resistant ovarian carcinoma cells. a Bright-fieldimages of cell morphology after treatment with cisplatin and AZD1152using three different combination schedules. Arrows indicate dying cells.b Representative fluorescent images showed nuclear fragmentation aftertreatment with cisplatin and AZD1152, clearly condensed chromatin orexhibiting fragmented nuclei in cells were taken as apoptotic (indicatedby arrow). The number of apoptotic cells per 100 cells was quantified. cFlow cytometry analysis of DNA content and DNA breaks. Peaks of lessthan 2N DNA content falling in front of G0/G1 phase which representsDNA degradation, an event reminiscent of apoptosis, were observed inOVCAR-8 and ES-2 cells treated with cisplatin and AZD1152. dQuantification of cell apoptosis in c. e WB analysis depicted expressionand cleavage of apoptosis-related proteins in OVCAR-8 cells. f Statisticalanalysis of PARP and caspase-3 cleavage levels in e. g WB analysisdepicted expression and cleavage of apoptosis-related proteins in ES-2cells. Ab, antibody. **P<0.01; ***P<0.001. AZD, AZD1152. Theconcentrations of AZD1152 and cisplatin applied to OVCAR-8 cellswere 100 nM and 15 μM, respectively, whereas 600 nM and 1 μM,respectively, for ES-2 cells. Data are expressed as mean±SD. Results arerepresentative of data generated in at least three independent experiments

J Mol Med

effect is associated with the modulation of c-Myc protein. Infact, c-Myc expression is reduced upon AZD1152 exposure,while c-Myc knockdown by siRNA in OVCAR-8 cells abol-ishes the cytotoxic effect of AZD1152, thus hampering thesynergetic efficiency. By contrast, moderate antagonism wasobserved in c-Myc-deficient SKOV3 cells, whereas overex-pression of c-Myc restores the synergistic interaction betweencisplatin and AZD1152. Therefore, the combination efficien-cy of cisplatin and the Aurora B inhibitor was most prominentin the cisplatin-resistant and c-Myc-overexpressed ovariancarcinoma cells.

Cisplatin-resistant ovarian carcinoma cells are sensitiveto the treatment with Aurora B inhibitors [9, 14, 23]. How-ever, we are the first to reveal that pretreatment with AuroraB inhibitors for 24 h followed by cisplatin either with orwithout Aurora B inhibitors for the next 72 h enhancescytotoxicity, as evident by the reduced cell survival(Fig. 1f, g) and increased apoptosis (Fig. 2). On the otherhand, the simultaneous treatment schedule does not displaysignificant synergism. Thus, the schedule of administrationinfluences the success of the combination strategy, as foundin other reports [24].

Fig. 3 c-Myc expression is required for the synergistic effect of cisplatinand AZD1152 in different solid tumor cells. a Proliferation analysis ofcisplatin and AZD1152 combination using treatment schedule B inU2OSand KHOS cells (osteosarcoma); MDA-MB-468 cells (breast cancer);NCI-H1299 and H460 cells (non-small-cell lung carcinoma); 5637 cells(bladder cancer); and PC3 cells (prostate cancer). b Combination index(CI) values obtained for combinations of 4 μM cisplatin with 0.5 μM

AZD1152 in different solid tumor cell lines using sequential treatmentschedule B. According to the CI method, CI values lower than 0.95indicate synergistic drug interactions, whereas CI values higher than1.20–1.45 and around 1 indicate antagonism and additive effect,respectively. c WB analysis of c-Myc expression level in various solidtumor cell lines. Ab, antibody

J Mol Med

One possible explanation for the schedule-dependent syn-ergism is the capability of AZD1152 to induce polyploidy thatoccurred as a result of several rounds of endoreduplication andfailed cell divisions [25]. For example, optimal schedule fortopoisomerase inhibitor SN-38 and AZD1152 in colon cancercells was when AZD1152 was given before SN-38, whereasthe reverse sequence or concurrent treatment would block theAZD1152 effect by arresting cells in the G2 phase andinhibiting cells from undergoing polyploidy [26]. Our results

show that Aurora B inhibitors induce polyploidy of ovariancarcinoma in a dose-dependent manner (Fig. 1c, d), with theend result of decreased proliferation of the cell population(Fig. 1f, g). This polyploidy can accumulate defects in themaintenance of genomic stability, increases the burden on thecells’ metabolism, and results in activation of stress pathways[27]. Since polyploid cells are genomically less stable thandiploid cells, it is likely that the platinum-DNA adductscaused by cisplatin [28] may induce more fatal DNA damage,

Fig. 4 c-Myc expression level determines ovarian carcinoma’ssensitivity to the sequential combination of AZD1152 and cisplatin. aWB analysis of c-Myc and Aurora B expression levels in A2780,OVCAR-8, and SKOV3 ovarian cancer cells. b c-Myc expression leveldetected by WB analysis and its statistical analysis in cisplatin andAZD1152-treated OVCAR-8 cells. c Knockdown efficiency of c-Mycby siRNA interference in OVCAR-8 cells, while siRNA-NC served asnegative control. d, e Proliferation (d) and flow cytometry analysis ofapoptosis (e) in OVCAR-8 cells after the knockdown of c-Myc and

sequential combination treatment of cisplatin and AZD1152 usingschedule B. f Expression levels of pCCL-c-Myc-GFP or pCCL-GFPvector in SKOV3 cells was visualized under fluorescence microscopeand detected by WB analysis. g, h Proliferation (g) and flow cytometryanalysis of apoptosis (h) in SKOV3 cells after the overexpression of c-Myc and sequential combination treatment of cisplatin and AZD1152using schedule B. Ab, antibody. **P<0.01; ***P<0.001. Data areexpressed as mean±SD. Results are representative of data generated inat least three independent experiments

J Mol Med

which is responsible for the increased cytotoxic effect. In ourstudy, the polyploid cells generated from Aurora B inhibitionpossibly account for the increased sensitivity to cisplatin(Fig. 1f, g), while the knockdown of c-Myc decreasesAZD1152-induced polyploidy (Fig. 5b) and leads to resis-tance of ovarian cancer cells to the drug combination(Fig. 4d, e). Similar patterns were observed in childhood acutelymphoblastic leukemia (ALL) [29] and melanoma cells [30],where multinucleated cells were more sensitive to DNA-damaging agents (e.g., cytarabine and cisplatin) than werenon-polyploid ALL cells or normal skin fibroblasts. There-fore, polyploidy is possibly a favorable prognostic factor inthe combination treatment with DNA-damaging agents suchas cisplatin. And our study showed that the schedule in whichAZD1152 is given before cisplatin proves to be the optimalsequence to maximize the antitumor effects of chemotherapyin ovarian carcinoma.

The c-Myc oncogene has been implicated in malignantprogression in a variety of human tumors. In many cases,amplification or elevated expression of the c-Myc gene hasbeen associated with poor prognosis or decreased survival rate

[31]. However, a Gynecologic Oncology Group study ofwomen with suboptimally resected, advanced stage epithelialovarian cancer showed that c-Myc amplification was notassociated with disease status following platinum-based com-bination chemotherapy [32] and that a decrease in the amountof c-Myc mRNA [33] or protein [34] was seen in humanovarian carcinoma cells after cisplatin treatment. These resultsare consistent with our findings that showed a consistent ordecreased c-Myc protein expression after administration of15 μM cisplatin (Fig. 4b, lanes 3, 7, and 11), whereas the cellproliferation and apoptosis after cisplatin treatment were in-dependent of c-Myc knockdown or overexpression (Fig. 4d, e,g, h). Thus, these data suggest that the positive correlationbetween efficacy of the sequential combination treatment andhigh expression of c-Myc is dependent on the inhibition ofAurora B by AZD1152 and that ovarian carcinoma driven bythe Myc family genes may be suitable for treatment usingAurora B inhibitors.

In conclusion, the results presented here provide usefulinsights to design sequential combinations strategies, to targetovarian carcinoma cells, and support the targeting of Aurora B

Fig. 5 Aurora B-mediated mitotic functions in ovarian carcinoma areregulated by c-Myc. a WB analysis of c-Myc and Aurora B afterknockdown of c-Myc by siRNA interference in OVCAR-8 cells after48 h of transfection. siRNA-NC served as negative control. b Cell cycledistribution of OVCAR-8 cells after knockdown of c-Myc, with orwithout 24-h treatment of AZD1152, was analyzed by flow cytometry.cQuantification of the percentage of polyploidy in b. dWB analysis of c-Myc and Aurora B expression after lentiviral transduction of pCCL-c-Myc-GFP or pCCL-GFP vector in SKOV3 cells. eCell cycle distribution

of SKOV3 cells after overexpression of c-Myc, with or without 24-htreatment of AZD1152, was analyzed by flow cytometry. fQuantificationof the percentage of polyploidy in e. ***P<0.001. Data are expressed asmean±SD. Results are representative of data generated in at least threeindependent experiments. g WB analysis of histone H1 phosphorylation(p-Histone H1) in cisplatin- and AZD1152-treated OVCAR-8 cells underthree different combination schedules. h WB analysis of p-Histone H1expression after knockdown of c-Myc by siRNA interference inAZD1152-treated OVCAR-8 cells. Ab, antibody

J Mol Med

as an attractive way to increase cell sensitivity to cisplatin,especially in selected group of cisplatin-resistant ovarian car-cinomas that displays a constitutive activation of the c-Myc.This finding may offer a new opportunity for the use ofAZD1152 in combination with cisplatin in preclinical andclinical settings in the management of ovarian cancer patients.

Acknowledgments The work is funded by the Zhejiang ProvincialNatural Science Foundation of China (LY12H16012 and Y2090319),Health Bureau of Zhejiang Province (2011KYA046 and 2013KYA055),National Natural Science Foundation of China (No. 81402132), Funda-mental Research Funds for the Central Universities, Program for ZhejiangLeading Team of S&T Innovation, and Science-Technology Foundationof Zhejiang Province (2012C33056).

Conflict of interest The authors declare no conflict of interest.

References

1. Runnebaum IB, Stickeler E (2001) Epidemiological and molecularaspects of ovarian cancer risk. J Cancer Res Clin Oncol 127(2):73–79

2. Agarwal R, Kaye SB (2003) Ovarian cancer: strategies for overcom-ing resistance to chemotherapy. Nat Rev Cancer 3(7):502–516

3. denHollander J, Rimpi S, Doherty JR, RudeliusM, Buck A, HoelleinA, Kremer M, Graf N, Scheerer M, Hall MA et al (2010) Aurorakinases A and B are up-regulated by Myc and are essential formaintenance of the malignant state. Blood 116(9):1498–1505

4. Zhang L, Zhang S (2011) ZM447439, the Aurora kinase B inhibitor,suppresses the growth of cervical cancer SiHa cells and enhances thechemosensitivity to cisplatin. J Obstet Gynaecol Res 37(6):591–600

5. Qi G, Ogawa I, Kudo Y,MiyauchiM, Siriwardena BS, Shimamoto F,Tatsuka M, Takata T (2007) Aurora-B expression and its correlationwith cell proliferation and metastasis in oral cancer. Virchows Arch450(3):297–302

6. Chen YJ, Chen CM, Twu NF, Yen MS, Lai CR, Wu HH, Wang PH,Yuan CC (2009) Overexpression of Aurora B is associated with poorprognosis in epithelial ovarian cancer patients. Virchows Arch455(5):431–440

7. Gautschi O, Heighway J, Mack PC, Purnell PR, Lara PN Jr, GandaraDR (2008) Aurora kinases as anticancer drug targets. Clin CancerRes 14(6):1639–1648

8. Wilkinson RW, Odedra R, Heaton SP, Wedge SR, Keen NJ, CrafterC, Foster JR, BradyMC,Bigley A, Brown E et al (2007) AZD1152, aselective inhibitor of Aurora B kinase, inhibits human tumor xeno-graft growth by inducing apoptosis. Clin Cancer Res 13(12):3682–3688

9. Ma Y, Weimer J, Fredrik R, Adam-Klages S, Sebens S, Caliebe A,Hilpert F, Eckmann-Scholz C, Arnold N, Schem C (2013) Aurorakinase inhibitor AZD1152 has an additional effect of platinum on asequential application at the human ovarian cancer cell line SKOV3.Arch Gynecol Obstet 288(1):173–182

10. Beussel S, Hasenburg A, Bogatyreva L, Hauschke D, Werner M,Lassmann S (2012) Aurora-B protein expression is linked to initialresponse to taxane-based first-line chemotherapy in stage III ovariancarcinoma. J Clin Pathol 65(1):29–35

11. Mortlock AA, Foote KM, Heron NM, Jung FH, Pasquet G,Lohmann JJ, Warin N, Renaud F, De Savi C, Roberts NJ et al(2007) Discovery, synthesis, and in vivo activity of a new class ofpyrazoloquinazolines as selective inhibitors of aurora B kinase. JMed Chem 50(9):2213–2224

12. Gully CP, Zhang F, Chen J, Yeung JA, Velazquez-Torres G, Wang E,Yeung SC, Lee MH (2010) Antineoplastic effects of an Aurora Bkinase inhibitor in breast cancer. Mol Cancer 9:42

13. Ikezoe T, Takeuchi T, Yang J, Adachi Y, Nishioka C, Furihata M,Koeffler HP, Yokoyama A (2009) Analysis of Aurora B kinase innon-Hodgkin lymphoma. Lab Investig 89(12):1364–1373

14. Ma YX, Li XZ (2013) Effect of aurora kinase B inhibitor AZD1152in the treatment of cisplatin-resistant ovarian carcinoma. ZhonghuaFu Chan Ke Za Zhi 48(1):46–50

15. Zhou X, Fang Y, Jing H, Zhong L, Luo P, Song H, Yang B, He Q(2013) Involvement of mitogen-activated protein kinase in signaltransducer and activator of transcription-1 mediated differentiationinduced by bortezomib in acute myeloid leukemia cells. MolCarcinog 52(1):18–28

16. Chou TC, Talalay P (1984) Quantitative analysis of dose-effectrelationships: the combined effects of multiple drugs or enzymeinhibitors. Adv Enzym Regul 22:27–55

17. Kang MH, Kang YH, Szymanska B, Wilczynska-Kalak U, SheardMA, Harned TM, Lock RB, Reynolds CP (2007) Activity of vincris-tine, L-ASP, and dexamethasone against acute lymphoblastic leuke-mia is enhanced by the BH3-mimetic ABT-737 in vitro and in vivo.Blood 110(6):2057–2066

18. Vader G, Lens SM (2008) The Aurora kinase family in cell divisionand cancer. Biochim Biophys Acta 1786(1):60–72

19. Havelka AM, Berndtsson M, Olofsson MH, Shoshan MC, Linder S(2007) Mechanisms of action of DNA-damaging anticancer drugs intreatment of carcinomas: is acute apoptosis an "off-target" effect?Mini Rev Med Chem 7(10):1035–1039

20. Domcke S, Sinha R, Levine DA, Sander C, Schultz N (2013)Evaluating cell lines as tumour models by comparison of genomicprofiles. Nat Commun 4. doi:Artn 2126 Doi 10.1038/Ncomms3126

21. Boxer LM, Dang CV (2001) Translocations involving c-myc and c-myc function. Oncogene 20(40):5595–5610

22. Hergeth SP, Dundr M, Tropberger P, Zee BM, Garcia BA, Daujat S,Schneider R (2011) Isoform-specific phosphorylation of human link-er histone H1.4 in mitosis by the kinase Aurora B. J Cell Sci 124(Pt10):1623–1628

23. AkiyamaM, Izumi H,WangKY, Yamaguchi T, KumaA, Kitamura N,Harada Y, Oya R, Yamaguchi K, Iwai Y et al (2014) Hypersensitivityto Aurora kinase inhibitors in cells resistant against platinum-containing anticancer agents. Chem, Anticancer Agents Med

24. Azzariti A, Xu JM, Porcelli L, Paradiso A (2004) The schedule-dependent enhanced cytotoxic activity of 7-ethyl-10-hydroxy-camptothecin (SN-38) in combination with Gefitinib (Iressa,ZD1839). Biochem Pharmacol 68(1):135–144

25. Ditchfield C, Johnson VL, Tighe A, Ellston R, Haworth C, JohnsonT, Mortlock A, Keen N, Taylor SS (2003) Aurora B couples chro-mosome alignment with anaphase by targeting BubR1, Mad2, andCenp-E to kinetochores. J Cell Biol 161(2):267–280

26. Nair JS, de Stanchina E, Schwartz GK (2009) The topoisomerase Ipoison CPT-11 enhances the effect of the aurora B kinase inhibitorAZD1152 both in vitro and in vivo. Clin Cancer Res 15(6):2022–2030

27. Ganem NJ, Pellman D (2007) Limiting the proliferation of polyploidcells. Cell 131(3):437–440

28. Nilsson C, Roberg K, Grafstrom RC, Ollinger K (2010) Intrinsicdifferences in cisplatin sensitivity of head and neck cancer cell lines:correlation to lysosomal pH. Head Neck 32(9):1185–1194

29. Kaspers GJ, Smets LA, Pieters R, Van Zantwijk CH, VanWering ER,Veerman AJ (1995) Favorable prognosis of hyperdiploid commonacute lymphoblastic leukemia may be explained by sensitivity toantimetabolites and other drugs: results of an in vitro study. Blood85(3):751–756

30. Kokkinakis DM, Brickner AG,Kirkwood JM, Liu X, Goldwasser JE,Kastrama A, Sander C, Bocangel D, Chada S (2006) Mitotic arrest,apoptosis, and sensitization to chemotherapy of melanomas by me-thionine deprivation stress. Mol Cancer Res 4(8):575–589

J Mol Med

31. Garte SJ (1993) The c-myc oncogene in tumor progression. Crit RevOncog 4(4):435–449

32. Darcy KM, Brady WE, Blancato JK, Dickson RB, Hoskins WJ,McGuire WP, Birrer MJ (2009) Prognostic relevance of c-MYCgene amplification and polysomy for chromosome 8 insuboptimally-resected, advanced stage epithelial ovarian can-cers: a Gynecologic Oncology Group study. Gynecol Oncol114(3):472–479

33. Kashani-Sabet M, Wang W, Scanlon KJ (1990) Cyclosporin A sup-presses cisplatin-induced c-fos gene expression in ovarian carcinomacells. J Biol Chem 265(19):11285–11288

34. Vaskivuo L, Rysa J, Koivupera J, Myllynen P, Vaskivuo T, ChvalovaK, Serpi R, Savolainen ER, Puistola U, Vahakangas K (2006)Azidothymidine and cisplatin increase p14ARF expression inOVCAR-3 ovarian cancer cell line. Toxicol Appl Pharmacol216(1):89–97

J Mol Med