Research Tissue-Specific Pathways for Estrogen Regulation of Ovarian Cancer...

Tum

TissCan

MoniqJoshu

Abst

Intro

Estrpreserof posbecauectomuterustransfpausalseventwomethan 5betweTwo aplasticthat Eoccultfacts tlatingbreast

AuthorSectionEndocrAnschu

Note:Resear

CorresAnschuMail St2066; F

doi: 10

©2010

www.a

Published OnlineFirst October 19, 2010; DOI: 10.1158/0008-5472.CAN-10-1238
Published OnlineFirst on October 19, 2010 as 10.1158/0008-5472.CAN-10-1238
Canceresearch

or and Stem Cell Biology

ue-Specific Pathways for Estrogen Regulation of Ovarian

R

cer Growth and Metastasis

ue A. Spillman1,2, Nicole G. Manning2, Wendy W. Dye2, Carol A. Sartorius2, Miriam D. Post3,
a Chuck Harrell2, Britta M. Jacobsen2, and Kathryn B. Horwitz2,3
ractMen

thelialdiseasxenogquantreceptLaser-E2-reg

levels by 1cancers

s' Affiliatiof Gynecoinology, antz Medical

Supplemench Online (h

ponding Atz Medical Cop B198-4,ax: 303-724

.1158/0008-

American A

acrjourna

Downlo

opausal estrogen (E2) replacement therapy increases the risk of estrogen receptor (ER)–positive epi-ovarian cancers (EOC). Whether E2 is tumorigenic or promotes expansion of undiagnosed preexistinge is unknown. To determine E2 effects on tumor promotion, we developed an intraperitoneal mouseraft model using ZsGreen fluorescent ER− 2008 and ER+ PEO4 human EOC cells. Tumor growth wasified by in vivo fluorescent imaging. In ER+ tumors, E2 significantly increased size, induced progesteroneors, and promoted lymph node metastasis, confirming that ERs are functional and foster aggressiveness.captured human EOC cells from ER− and ER+ xenografted tumors were profiled for expression ofulated genes. Three classes of E2-regulated EOC genes were defined, but <10% were shared withulated breast cancer genes. Because breast cancer selective ER modulators (SERM) are therapeutically
E2-reg
ineffective in EOC, we suggest that our EOC-specific E2-regulated genes can assist pharmacologic discoveryof ovarian-targeted SERM. Cancer Res; 70(21); 8927–36. ©2010 AACR.

prolonistingreceptE2 stimalthouteronethat EHer

promofluoreengrafsite-spmodeERT slymphthrougwe idecers, athat t

duction

ogen replacement therapy (ERT) was hailed as thever of femininity, and before 1980, it represented >67%tmenopausal hormone prescriptions (1). However,se unopposed ERT increased uterine cancer risk, hyster-y became a prerequisite. This procedure removed the, leaving the ovaries intact and susceptible to neoplasticormation. Epithelial ovarian cancer (EOC) is a meno-disease whose relative risk (RR) rises in the sixth andh decades of life from basal levels (RR, 1.0) in untreatedn to RR of 1.22 to 1.72 in women prescribed ERT moreyears (2–6). Although epidemiologic data postulate a linken extendedERT and EOC, causality has not been proven.lternative hypotheses exist: that ERT directly causes neo-transformation of ovarian surface epithelium (OSE), orRT promotes proliferation and metastasis of preexisting,disease. Evidence for the first hypothesis is found in thehat estrogen (E2) levels in ovarian tissues exceed circu-

00-fold, and E2 has direct genotoxic effects in(7). The second hypothesis postulates that

and inER+ o

Mate

Cell cPEO

(WashwereTX). Cdem rSequeand MCO2 in

ons: 1Department of Obstetrics and Gynecology,logic Oncology, 2Department of Medicine, Division ofd 3Department of Pathology, University of ColoradoCampus, Aurora, Colorado

tary data for this article are available at Cancerttp://cancerres.aacrjournals.org/).

uthor: Monique A. Spillman, University of Coloradoampus, Academic Office 1, 12631 East 17th Avenue,P.O. Box 6511, Aurora, CO 80045. Phone: 303-724--2053; E-mail: [email protected].

5472.CAN-10-1238

ssociation for Cancer Research.

ls.org

Research. on June 15, 201cancerres.aacrjournals.org aded from

ged ERT promotes proliferation and expansion of preex-malignancy. In fact, 38% to 62% of EOCs are estrogenor-α (ER) positive (8, 9), making them good targets forulation. Ovarian tumors from older women (≥65 years),ghmore likely to be ER+, are less likely to express proges-receptors (PR; ref. 10). This paucity of PR may suggestR signaling becomes deficient with aging.e, we test aspects of the second hypothesis that ERTtes expansion of preexisting malignancy. Using highlyscent ER− and ER+ human ovarian cancer cell linested in the peritoneum of nude mice, we developed aecific, rigorously controlled, quantitative xenograftl of ERT-stimulated ovarian cancer. We show thatubstantially increases proliferation and risk of distantnode (LN) metastasis only in ER+ tumors, possiblyh E2 regulation of cellular motility genes. Additionally,ntify E2-regulated genes specific to ER+ ovarian can-nd not shared with ER+ breast cancers. We proposehese genes can be used for pharmacologic discoveryterventions with E2 antagonists that uniquely targetvarian cancers.

rials and Methods

ulture4 cells (11) were a gift from Dr. Anne Bowcockington University, St. Louis, MO), and 2008 cells (12)obtained from Southwestern Medical School (Dallas,ell lines were authenticated by polymorphic short tan-epeat analysis at the University of Colorado DNAncing and Analysis Core before xenografting (July 2008
ay 2010). Cells were routinely cultured at 37°C and 5%RPMI 1640 with 10% fetal bovine serum. Cells were
8927

8. © 2010 American Association for Cancer

http://cancerres.aacrjournals.org/

taggedroutin

MousStud

Care aathymDawle107 ZstaininsubcuweeklyLightuteri wand flfrozenexpresdehydimmu

IVIS iBiw

with ISciencImagequantFluoredata wcosmi(ROI)racicanalystrol anexactbling tderiveated in

HistolHis

accordfollowPR (1:dine (BTi micsoftwatumorfor ER

ProlifAfte

at 10 mBrdUrchemithe nunucleifields/were c

GeneIntr

zen amice.roundDevicMolecAmpliOvatiowas creplicawere hgenomCanceusedhormobetweacrossnificanwere fhighes+E2 trversusperforand aGX7.3Springobtainbeennumbacc.cg

Resu

FluormousTwo

werewhich(11), wducedmaneovarienous Eand tushowstumordiseascanceomentease wtumortumorovarieenviroshow

Spillman et al.

Cance8928


with ZsGreen by retroviral transduction (13) andely tested for absence of Mycoplasma.

e xenograftsies were University of Colorado Institutional Animalnd Use Committee approved. Ovariectomized femaleic nude mice (5–6 weeks of age; Harlan Sprague-y) were anesthetized with Avertin and injected i.p. withGreen-tagged ovarian cancer cells. A silastic pellet con-g 2 mg cellulose or 17β-estradiol (E2) was implantedtaneously (14). Mice were weighed and monitoredfor fluorescent tumor burden using Illumatool BrightSystem LT-9500 (Lightools Research). At necropsy,ere weighed to document appropriate estrogenizationuorescent tumors were resected. Tumor samples werein optimal cutting temperature compound for genesion profiling or fixed overnight in 4% paraformal-e at 4°C and paraffin embedded for pathology andnohistochemical analyses.

magingeekly fluorescent quantitative tumor imaging was doneVIS 200 Optical Imaging System (Xenogen, Caliper Lifees) in live mice under isoflurane anesthesia. Living2.60.1 (Caliper Life Sciences) software was used for

itative analysis at 0.5-, 1-, and 5-second time points.scent background subtraction was performed, andere displayed in log scale photons, flat fielded, withc ray correction. Each abdominal region of interestquantitation was 3.05 cm by 4.12 cm; a separate tho-background ROI was 2.23 cm by 1.79 cm. Statisticalis used Student's t test for the difference between con-d E2-treated groups at each time point, and Fisher'stest for differences in categorical variables. Cell dou-imes were calculated from exponential growth curvesd from the fluorescent imaging data. Graphs were cre-GraphPad Prism 5.

ogy and immunohistochemistrytology and immunohistochemistry were conducteding to standard protocols (15). Antibodies used were ass: ER (1:20, SP1 clone RM9101; LabVision/NeoMarkers),500, PgR1294 clone M3568; Dako), and bromodeoxyuri-rdUrd; 1:50, 347580; Becton Dickinson). A Nikon Eclipseroscope, Ds-Fi1 color camera, and NIS Elements AR 3.1re were used for digital photography. Human ovariantissue arrays were from BioChain Institute and stainedand PR.

erationr 10 to 12weeks of tumor growth, 200 μL BrdUrd (Sigma)g/mL in PBS was injected i.p. 3 hours before sacrifice.

d incorporation into DNAwas assessed by immunohisto-stry. The proliferative fraction was calculated by dividingmber of BrdUrd-positive nuclei by the number of totalper high-power field. A minimum of three high-power
tumor and a minimum of three independent tumorsounted for each cell line and condition.
high-gular n

r Res; 70(21) November 1, 2010

Research. on June 15, 201cancerres.aacrjournals.org Downloaded from

expression profilingaperitoneal tumors from 2008 and PEO4 cells were fro-t necropsy from quadruplicate −E2- and +E2-treatedHuman tumor cells were laser dissected from sur-ing mouse tissues using an Autopix 100e (Moleculares). RNA was isolated (Picopure RNA Isolation kit,ular Devices), amplified (WT-Ovation FFPE RNAfication System V2, NuGEN), and biotin labeled (FL-n cDNA Biotin Module V2, NuGEN). RNA integrityonfirmed (Bioanalyzer 2100, Agilent). One −E2 PEO4te was degraded and excluded from analyses. Samplesybridized to U133 Plus 2.0 (Affymetrix) whole humane expression arrays in the University of Colorador Center Microarray Core. Gene expression analysisPartek software. An ANOVA analysis was done withne and cell type as fixed factors and an interactionen hormone and cell type included. Genes absentall cell types were filtered out, and genes with a sig-t interaction (P < 0.05) between hormone and cell typeiltered in. Statistically significant genes were sorted byt-fold expression change between PEO4 −E2 versuseatments, and with no significant change in 2008 −E2+E2 treatment. Supervised clustering analysis was

med using standard correlation as similarity measuren average linkage clustering algorithm (GeneSpring.1). Overlapping gene lists were generated using Gene-(GX7.3.1). E2-regulated breast cancer gene lists wereed from published data (16–18). Microarray data havesubmitted to Gene Expression Omnibus (accessioner GSE22600; http://www.ncbi.nlm.nih.gov/geo/query/i?acc=GSE22600).

lts

escent ovarian cancer intraperitoneal xenografte modeldifferent human serous ovarian carcinoma cell lines

engrafted in immunodeficient mice: 2008 cells (12),are ER− and PR−, served as controls for PEO4 cellshich are ER+ and PR+. Both cell lines were stably trans-with a ZsGreen expressing retrovirus for bright per-nt green fluorescence. Female athymic nude mice,ctomized 4 to 6 weeks after birth to remove endoge-2 and estrous cycling, were injected i.p. with 107 cells,mors were allowed to grow for 60 to 70 days. Figure 1Afluorescent imaging at necropsy of ER− 2008 ovariansmall-bowel implants. Figure 1B shows ER+ PEO4

e, which exhibits the hallmarks of advanced ovarianr—peritoneal studding (arrow), bowel implants, andal implants (asterisk). At necropsy, intraperitoneal dis-as visible in 67% (42 of 62) of mice bearing ER− 2008s and in 69% (50 of 72) of mice bearing ER+ PEO4s. Concordance between the two indicates that thectomized mouse peritoneum provides a hospitablenment for tumor growth. Both 2008 and PEO4 tumorssimilar histology, morphologically consistent with
rade EOC. The ER− 2008 implants (Fig. 1C) have irreg-uclei, prominent nucleoli, loose cell association, and
Cancer Research



modershownuclei

IntrapincreaE2

cyclinplemesubcuconfirweighdeprivmice,Afte

suppleQualit2008 teffectand Fthe in(SuppvolumtumorFigureFluorei.p. inE2-trenine cthat th

63 dayimplanER+ PE2. FiPEO4(left)controdiseas(P = 0tial grthat 2whichcells (a doudays f

ProloPro

advanextrapFigurelocatiFigurebearinfluoreFigure(arrowagains

Figurehistopahumanimplanti.p. withPEO4 (intraperphotogHallmacancerbowel sarrow,(black acakingenlargenodules2008 (CimplantH&E, shadenocpleiomonucleoldistinctindicatiScale bars, 10 μm.

Estrogen Promotes Ovarian Cancer Growth and Metastasis

www.a


ate mitotic activity. The ER+ PEO4 nodules (Fig. 1D)cohesive cells with distinct cell membranes, prominent, large nucleoli, and moderate mitotic activity.

eritoneal ovarian tumor burden is significantlysed by ERTfalls from 128 pg/mL physiologic levels in the normalg mouse to 31 pg/mL in ovariectomized mice. E2 sup-ntation with 17β-estradiol–releasing pellets implantedtaneously restores blood levels to 139 pg/mL (14). Wemed restoration of physiologic E2 by measuring uterinets. Mean control uterine weights were 0.06 g in E2-ed ovariectomized mice and 0.18 g in E2-supplementeda statistically significant increase (P = 0.002).r establishing the physiologic effectiveness of E2mentation, we assessed E2 effects on tumor volumes.ative fluorescent examination of intraperitoneal ER−

umors at necropsy failed to reveal a significant E2on disease volume (Supplementary Fig. S1A, B, E,). In contrast, prolonged E2 significantly increasedtraperitoneal disease volume in ER+ PEO4 implantslementary Fig. S1C, D, G, and H). Differences in tumore in response to E2 were confirmed by quantifyingfluorescence by photon flux IVIS imaging (Fig. 2).2A shows serial quantitation of ER− 2008 implants.scent images were obtained 22, 49, and 63 days afterjection of tumor cells in five control (left) and fiveated (right) mice. Data plotted as mean and SEM for
ontrol and nine E2-treated mice (Fig. 2B) clearly showe points overlap (P = 0.42, 0.83, and 0.69 for 22, 49, and
staticFigs. 1

acrjournals.org


s, respectively), indicating that ER− 2008 ovarian tumorts are not E2 responsive. In contrast, quantitation ofEO4 tumor volume showed a significant increase withgure 2C illustrates 10 ovariectomized mice bearingimplants 22, 49, and 63 days after i.p. injection withoutor with (right) E2 supplementation. At 22 days, thel and ERT points overlap (P = 0.46), but by 49 days,e volume in the ERT mice was significantly increased.033) and rose further at 63 days (P = 0.013). Exponen-owth curves plotted from the fluorescent data show008 control cells have a doubling time of 38.74 days,is similar to the 35.31 day doubling time for 2008 E2not shown). On the other hand, PEO4 control cells havebling time of 93.25 days, which is accelerated to 18.08or PEO4 E2 cells (not shown).

nged ERT promotes LN metastasesliferation of intraperitoneal tumors is a hallmark ofced-stage EOC. Additionally, 10% of patients exhibiteritoneal LN metastases, an ominous finding (19).3 shows our ability to model ovarian tumor cell trans-

on from the abdominal cavity to extraperitoneal LN.3A shows in situ liver, heart, and lungs of a mouseg ER+ PEO4 tumors. Paired bilateral LNs are clearlyscent (arrows), indicating the presence of tumor cells.3B shows histologic confirmation of PEO4 tumor cells) infiltrating under the nodal capsule and pushingt lymphoid tissues. Histologically, PEO4 cells meta-

1. Fluorescent imaging andthology of intraperitonealovarian cancer xenografts. Nude mice were injected107 ER− 2008 (A) or ER+

B) cells, and fluorescentitoneal disease wasraphed at necropsy.rks of advanced ovarianare shown: mesenterictudding (black solidA), peritoneal studdingsterisk, B), and omental(black arrow, B). Notement of PEO4 tumor. Paraffin sections of) and PEO4 (D) tumors were stained withowing high-gradearcinomas withrphic nuclei and prominenti. Cell borders are morein the PEO4 implant,ng a more cohesive tumor.

to LN resemble PEO4 cells in the peritoneum (compareC and 3B). LN metastases were present in 6% (2 of 33)

Cancer Res; 70(21) November 1, 2010 8929



of PEOtomize(10 ofmice wated wER+ P

In ERThi

The qsuppleent ananalys

the abLN mtumortomizlevels(Fig. 4in the(Fig. 4indicaWe

hetero

Figurecells i.plive micmice/codiseaseat 22, 49, and 63 d. D, data for 10 mice/cohort were graphed. ERT significantly stimulated disease at 49 and 63 d (red star). (One mouse showed novisible from th

Spillman et al.

Cance8930


4 and in 0% (0 of 26) of 2008 tumor-bearing, ovariec-d mice without E2 supplementation. This rose to 26%39) of ER+ PEO4 mice and to 9% (3 of 33) of ER− 2008ith E2 supplementation. In this model, ERT is associ-ith a significant (P = 0.03) increase in metastases inEO4 mice.

+ tumors, prolonged ERT induces PRsrty-six percent of EOCs express both ER+ and PR+ (8).uantitative increase in PEO4 tumor burden in E2-mented mice (Fig. 2D) suggested that ERs were pres-

intraperitoneal disease at 22 and 49 d, died from infection, and is absent

d functional. Figure 4 is an immunohistochemicalis of ER and PR in 2008 and PEO4 tumor implants in

havingsimila



sence of E2, and of PEO4 intraperitoneal implants andetastases in the presence of E2. In contrast to 2008s that lack ER (Fig. 4A), PEO4 implants from ovariec-ed mice without E2 supplementation express high ER(Fig. 4B). PRs are absent in both types of tumorsE and F). With E2, high levels of ER were maintainedPEO4 intraperitoneal implants and LN metastasesC and D). Additionally, E2 treatment induced PR,ting that the ERs were functional (Fig. 4G and H).noted that ER and especially PR expression wasgeneous, with some cells strongly positive and others

e Panel A day 63 +E2 cohort.)

2. ERT significantly increases ER+ intraperitoneal disease volume. A, 10 ovariectomized nude mice received 107 ER− ZsGreen-tagged 2008. (−E2 or +E2 supplementation). The five −E2 mice are on the left; the five +E2 mice are on the right. IVIS imaging heat maps were obtained ine at 22, 49, and 63 d after i.p. injection. Disease burden ranges from lowest (blue) to highest (red). B, quantitation of fluorescent photon flux in ninehort show the mean and SEM. Statistical comparison of the means was done by a two-tailed t test. ERT does not change the volume of ER−

. C, 10 ovariectomized nude mice received 107 ZsGreen-tagged PEO4 cells, −E2 (left) or +E2 (right), and disease was quantified by IVIS imaging

lower or negative receptor levels. We asked whetherr heterogeneity existed in a human ovarian serous

Cancer Research



adeno(SuppwerePR−. Tanalog

E2 prHav

we nein endble inwere irationchemi

stateswithouof posin theincorpare uncells (aggresthe p(23.9%ER− 2the ag

Figuremice, pin the p

Figurereceivemetastasis identified by ZsGreen fluorescence was resected, paraffin embedded, sectioned, and stained with H&E. PEO4 tumor cells (arrow) invadethe sub


www.a


carcinoma array by ER and PR immunohistochemistrylementary Fig. S2). Overall, 17 of 66 (27%) of tumorsER+/PR+, whereas another 10 of 66 (15%) were ER+/hey showed heterogeneity of receptor expressionous to the xenografts.

omotes proliferation of ER+ ovarian tumorsing shown that ERs are functional in PEO4 tumors,xt asked whether the known mitogenic effects of E2ometrial and breast cancers were similarly demonstra-ovarian cancers (20). To that end, tumor-bearing micenjected with BrdUrd before necropsy, and its incorpo-

capsular space above smaller, darker lymphocytic cells.

into S-phase DNA was quantified by immunohisto-stry (Fig. 5). Proliferating cells were present at basal

data mfluore

araffin embedded, and subjected to immunohistochemistry for ER and PR. 2008resence or absence of E2, but PRs require E2 for induction. ER and PR distribu

acrjournals.org


in both 2008 and PEO4 tumors of ovariectomized micet E2 supplementation (Fig. 5A). With E2, the numberitive nuclei increased in the PEO4 implants but not2008 implants (Fig. 5A). Quantitation of the BrdUrdoration (Fig. 5B) shows that without E2, more cellsdergoing mitosis in 2008 cells (10.7%) than in PEO45.5%), suggesting that, on average, 2008 cells are moresive. However, E2 significantly (P = 0.0001) increasedroliferative fraction of ER+ PEO4 tumor implants) while having nonsignificant (P = 0.188) effects on008 tumors (13.6%). This suggests that E2 increasesgressiveness of ER+ disease. The BrdUrd proliferation
irror the growth doubling times calculated from the
scent data in Fig. 2A and C. PEO4 control cells had

4. PRs are induced by ERT only in ER+ PEO4 tumors. A to H, ER+ PEO4 and ER− 2008 tumor nodules were excised from 10-wk −E2 or +E2+

3. ERT promotes distant LN metastasis. A, a representative mouse showing distant paratracheal LN metastases (white arrows). This moused 107 ZsGreen PEO4 cells i.p., +E2 (11 wk). The abdominal and thoracic cavities with the heart and lungs in situ are shown. B, a PEO4 LN

tumors lack ER and PR. PEO4 tumors and metastases are ERte heterogeneously. Scale bars, 10 μm.




the sloed byunaffe

E2-rethoseTo

ovariafrom oexpresanalyzcapturmousethe fohumanand Mter anPEO4or dowEach ca uniqregulacriteriwell aThe

one ofIn Fig.left, wtheirgene bter rangenes

and inlow inin ERincludscriptinto tER+ PvalidaimmuE2-regestrog(22). SCla

(red)ment.absengenestheirpathwcells. SovariaCD44genesE2 regCla

E2-depSixteeulatioulumgene

Figureinjectedimmunothe pro ere no

Spillman et al.

Cance8932


west doubling time, which was significantly accelerat-E2. The intermediate doubling times of 2008 cells werected by E2.

gulated genes of ER+ ovarian cancer differ fromof ER+ breast cancerdefine transcriptional pathways regulated by ERT inn cancers, ER+ PEO4 and ER− 2008 tumor implantsvariectomized control and E2-treated mice were genesion profiled. To insure that only human cells wereed, quadruplicate sets of human cells were first lasered from tumors to separate them from surroundingcells. RNA was obtained from human tumor cells of

ur experimental conditions, which were hybridized tomicroarrays and gene expression profiled (Materials

ethods). Fig. 6 shows the supervised hierarchical clus-alysis of genes from untreated (−) and E2-treated (+)and 2008 tumors that are relatively upregulated (red)nregulated (blue) compared with unchanged genes.olumn represents an individual tumor, and each rowue gene probe set. To be considered significantlyted, each gene had to pass strict statistical ANOVAa for cell line and hormone interaction (P < 0.05) ass >1.5-fold expression changes.heat map indicates that E2-responsive genes fall intothree categories arbitrarily defined as class A, B, and C.6A, genes from ER+ PEO4 +E2 tumors cluster to the farhereas genes from the −E2 control tumors cluster toright. The two tumor types originate from differentranches (top). In contrast, the ER− 2008 samples clus-

liferative index of PEO4 cells from 5.5% to 23.9%; effects on 2008 cells w

domly with regard to control or +E2 status. Class A(Fig. 6A, top) are ones whose expression is low (blue)

(26). Rmore



dependent of E2 in ER− 2008 tumor implants, similarly−E2 ER+ PEO4 implants, but upregulated (red) by +E2+ PEO4 implants. Class A genes were expected toe classic E2-inducible transcripts. In fact, PR tran-s called PGR, well known to be E2 regulated (21), fellhis category and were significantly induced by E2 inEO4 tumors. E2 induction of PR transcript expressiontes the statistical analysis and confirms the PR proteinnohistochemical data (Fig. 4G). A second knownulated transcript of interest, called gene regulated byen in breast cancer 1 (GREB1), was also identifiedupplementary Table S1 lists the class A genes.ss B genes (Fig. 6A, bottom) are constitutively highin ER− 2008 tumor implants independent of E2 treat-In ER+ PEO4 tumors, their expression is low in thece of E2 (blue), but they are E2 inducible (red). Class Bhave lost regulation by E2 in ER− ovarian cancer cells—upregulation having been hijacked by other signalingays—while retaining E2 regulation in ER+ ovarian cancereveral class B genes previously defined as having roles inn cancer metastasis and stem cell signatures included(23). Palladin (24) and fascin 1 (25) are cellular motilitywhose expression was not previously suspected as beingulated. Supplementary Table S2 lists class B genes.ss C genes (Fig. 6B) are constitutively high (red) inrived ER+ PEO4 cells but downregulated by E2 (blue).n class C genes were identified with known E2 reg-n, including KDEL (Lys-Asp-Glu-Leu) endoplasmic retic-protein retention receptor 3 (KDELR3), an interestinglinked to chemotherapy response in ovarian cancers

t significant.

5. ERT increases cell proliferation in ER+ PEO4 tumors. A, mice (−E2 or +E2) bearing ER− 2008 (a and b) or ER+ PEO4 (c and d) tumors werewith BrdUrd 3 h before necropsy. Tumors were paraffin embedded and sectioned, and nuclei incorporating BrdUrd (brown) were identified byhistochemistry. B, the proliferative index for each tumor type and hormonal condition was quantified. ERT significantly (P < 0.0001) increased

OPN1 is a cancer-associated testis antigen, which ishighly expressed in ER− than in ER+ breast cancers

Cancer Research



(27). SPCR cA, B, aselectediffereOur

genessuppleour E2breast(NEDDGLDCA andly pubingly, oB ovarcancerSupple

Discu

Epidshowof E2

tion opromothrougtest ththat ain respcence,and qlivingBy u

the spmodelthe rawithincreaPEO4

Figureor ER+

Tumorsarrays.changerow a tclass Aare dow


www.a


upplementary Table S3 lists class C genes. Quantitativeonfirmed the E2-regulated expression of a subset of classnd C genes (Supplementary Fig. S3). Six genes wered for quantitative PCR, and each one was statisticallyntially expressed in each class for a 100% validation rate.laboratory has previously published E2-regulatedin T47D human breast cancer cells using a similar E2-mented nudemouse xenograftmodel (17). We compared-regulated ovarian cancer genes to the previously definedcancer genes and found that only three class A genes9, FBXL7, and CA12) and three class C genes (B3GNT3,, and PPAP2B) were shared. Next, we compared the classB E2-regulated ovarian cancer genes with other previous-lished E2-regulated breast cancer genes (16–18). Surpris-nly 6.8% (10 of 146) of class A and 9.9% (15 of 152) of classian cancer genes overlapped with E2-regulated breast
genes (16–18). Shared class A and B genes are listed in E2-ind
growtAs expER− 2induceease. Mthe na

mentary Table S4.

ssion

emiologic meta-analyses assessing menopausal ERT
a combined RR of 1.22 of developing EOC after 5 years impor
and B genes that are upregulated by E2 only in ER cells; they are constitutivelnregulated by E2 from high basal levels in ER+ cells; they are constitutively low

acrjournals.org


f EOC is unclear. We hypothesized that prolonged ERTtes growth and metastasis of preexisting ER+ EOC cellsh activation of specific gene expression programs. Tois, we developed a mouse model of preexisting diseasellows quantitative assessment of ovarian tumor burdenonse to E2 exposure. Our model uses ZsGreen fluores-allowing external serial visualization of tumor implantsuantitation by emission fluorescent photon flux, inmice.sing E2 supplementation in an ovariectomized mouse,ecific effects of estrogenization can be analyzed. Thisproduces serum E2 levels of 125 pg/mL (14), similar tonge obtained in women (28). We find that comparedovariectomized mice, continuous ERT significantlysed fluorescence signal intensity in mice bearing ER+

tumor implants (Fig. 2D). The increase was due touced cell proliferation as quantified from exponentialh curves (Fig. 2D) and by BrdUrd incorporation (Fig. 5B).ected, ERT had no effect on the proliferative index of008 tumors. This confirmed our hypothesis that ERT-d growth of ovarian cancer cells is limited to ER+ dis-odeling the action of ERT in ovariectomized mice attural site of ovarian cancer implantation is critically
tant. Previously, Langdon and colleagues (29) implanted
use (3). The role of E2 in the development and promo- PEO4 cells for 4 weeks in the flank of intact cycling nude

6. Supervised hierarchical clustering of genes expressed in ovarian cancers identifies three ERT-regulated subgroups in ER+ disease. ER− 2008PEO4 ovarian tumor cells were grown i.p. in ovariectomized nude mice supplemented without (−) or with (+) E2-releasing pellets for ∼11 wk.were harvested and human cells were laser captured from frozen sections. RNA was hybridized to U133 Plus 2.0 whole human genome expressionStatistical analyses were performed as described in Materials and Methods. Statistically significant genes were sorted by highest-fold expressionbetween PEO4 −E2 versus +E2, and with no significant change in 2008 −E2 versus +E2. Each column represents a separate tumor, and eachranscript. Red (high) and blue (low) indicate relative transcript expression levels. A, genes regulated in ER+ PEO4 versus ER− 2008 showing

+ −
y low (class A) or high (class B) in ER cells. B, class C genesin ER− cells.



mice binhibittal diffby LainfluenindepealonementTwo

ovariastimuupregucanceupregugulatewhosedownrE2-uprwith psecondepitheidentifum cethe tw19 andOur

ovariacontroclass Bbut E2criticadencelity” gthe clas E2The

cancethat mopedoccurrincreametasin thein ER−

the holung mlines (PR,

E2 regas anexpresby imfoundtumorare nowhichwithinovaria

relatio(38–40cancegivescancemostThe

estroggens hEOC ttumorrate. IEOC (with Elized 5responzole, w12 weexpressubgroCur

fen anand bmay ring diovariaationgenestargettheseand spOur

womecautioworseof ERthat Ethe stsion eInte

for mprotec(46). Tin thein genexprespublis(16–1sharedwas psizes tcologioverlaregulain cDmay h

Spillman et al.

Cance8934


efore ERT supplementation. Paradoxically, growth wased in comparison with controls. Significant experimen-erences between our models include use of cycling micengdon and colleagues (29), in which progestationalces are retained (30). Our ovariectomized model allowsndent future assessment of the role of progesteroneor when combined with E2 as in EPRT hormone replace-therapies.previously published reports of E2-regulated genes in

n cancer cell lines were in vitro and used short-term E2lation. O'Donnell and colleagues (31) identified 5 E2-lated and 23 E2-downregulated genes in PEO1 ovarianr cells. We compared our class A genes with theirlated genes, and our class C genes with their downre-d genes, and only two overlapped. One of these, Cyr61,protein product promotes cell adhesion, was E2

egulated in the study by O'Donnell and colleagues, butegulated class B in ours. E2 upregulation of Cyr61 agreesrevious reports in MCF-7 breast cancer cells (32). Thestudy (33) used transformed human ovarian surface

lial cells and ovarian cancer cell lines. The authorsied 155 E2-regulated genes in ovarian surface epitheli-lls, and 315 in cell lines, of which 19 overlapped betweeno models. There were no exact matches between thoseour E2-regulated genes.study is the first to identify E2-regulated genes in ER+

n cancer xenografts using ER− tumors as negativels. This comparison was essential to identification ofgenes, which are constitutively high in ER− tumors,regulated in ER+ tumors. Class B may represent genesl to the transition from E2 sensitivity to E2 indepen-. Intriguingly, 15 of the class B genes were “cell moti-enes, including CD44 (34) and fascin 1 (35). None ofass B motility genes have been previously identifiedregulated.importance of E2 regulation of cell motility in ovarianr is further supported by our striking observationice treated with prolonged ERT preferentially devel-distant LN metastases (Fig. 3). Although metastasesed in both ER+ and ER− tumors, they were significantlysed by ERT only in ER+ tumors. ERT may also promotetasis through direct action on the metastatic nichehost compartment explaining its more subtle effectsdisease. Previously, E2 has been shown to regulatest compartment of ovariectomized mice to promoteetastasis of ER− lung and mammary carcinoma cell36).the canonical E2-regulated gene in the breast, is alsoulated in PEO4 cells (37). E2 induction of PR servedinternal control for our models and was a highlysed class A gene. E2 regulation of PR was confirmedmunohistochemistry in PEO4 tumors (Fig. 4) and wasto be heterogeneous, analogous to human ovarians (Supplementary Fig. S2). Areas of high PR expressiont exclusively restricted to areas of high ER expression,may indicate that ERs are downregulated in some cells
tumors. Heterogeneity of ER and PR expression withinn tumors may explain conflicting results on the cor-
In scontro



n between receptor expression and survival advantage). Maintenance of functional ER in younger ovarianr patients, as evidenced by coexpression of PR (10),support to our hypothesis that a subset of ovarianrs is E2 regulated. Our gene expression profiles areapplicable to this subgroup of patients.presence of E2-upregulated genes in EOC makes anti-en therapy an attractive premise. Although antiestro-ave been a mainstay of breast cancer treatment, inhey have been reserved for chemotherapy-refractorys, where they have a relatively low overall responsen a Gynecologic Oncology Group study of recurrent41), tamoxifen elicited an 18% response rate, correlatedR expression. The pure antiestrogen fulvestrant stabi-0% of disease, but few tumors regressed (42). A similarse rate was seen with the aromatase inhibitor letro-here 20% of patients had stable disease for at least

eks on therapy (43). The response correlated with highsion of ER and PR, defining an “endocrine-sensitive”up (43).rent selective ER modulators (SERM) such as tamoxi-d fulvestrant were designed to regulate genes in breastone. The lack of significant response to SERM in EOCeflect poor tissue specificity of current agents, includ-fferences in the tissue repertoire of ER coregulators inn versus breast tumors, rather than bona fide attenu-of ERα signaling. Our ERT-regulated ovarian cancerrepresent the largest collection of potential therapeutics in the literature. SERM could be screened againstgenes to identify compounds likely to have efficacyecificity in EOC.data would suggest that administration of ERT ton who have ovarian cancer should be viewed withn, but there are no clinical trials to suggest that thisns prognosis (44). One small prospective clinical trialT administered to ovarian cancer patients suggestedRT was associated with prolonged survival. However,udy failed to stratify the patients by tumor ER expres-ither at enrollment or at time of recurrence (45).restingly, addition of a progestin to estradiol (EPRT)enopausal hormone replacement therapy seems to betive against EOC, but deleterious in breast cancerhe differential tissue specificity of ERT versus EPRTbreast and ovary is likely to be mediated by differencese expression patterns. We have compared the genesion profile of our PEO4 EOC tumors with previouslyhed E2-regulated gene data sets in breast cancer8). Less than 10% of the ovarian cancer genes arewith breast cancer genes. No single signaling pathway

redominant in the shared cohort, a finding that empha-he importance of tissue-specific targeting of pharma-c therapy. Although we have attributed the paucity ofp between our ovarian E2-regulated genes and E2-ted breast cancer genes to tissue specificity, differencesNA arrays, data analysis, and time of E2 stimulationave also contributed to the perceived discrepancy.
ummary, we have developed a site-specific, rigorouslylled, quantitative mouse model of E2-stimulated human
Cancer Research



ovariadistanlarmogenespharmtarget

Disclo

No p

Grant

Gyne(M.A. SSpillma

Refe1. He

no19

2. Zhan64

3. Peassa p

4. RoC.Ep

5. Lathe

6. Berep17

7. HoRe

8. Racan

9. KoA.de317

10. Liutorlary

11. Lannin616

12. DiSto

13. Habreag20

14. Sadifftum

15. Hamome

16. Creinxen

17. Halatewit

18. FraBShumunEn


www.a


n cancer growth. ERT substantially increased the risk oft LN metastasis, possibly through E2 regulation of cellu-tility genes. Additionally, we have identified E2-regulatedspecific to ovarian cancers. These can be used for
acologic discovery and interventions that specifically
5K12HDHorwitz(K.B. Ho

Theof pageaccorda

ER+ ovarian cancers.

sure of Potential Conflicts of Interest

an breast cancer cells: insights into gene networks and pathwaysderlying estrogenic control and proliferation and cell phenotype.docrinology 2003;144:4562–74.

19. TarespaCa

20. Staest

21. HorecofEn

22. GhregCa

23. ZhofRe

24. Goorg

25. HanoBio

26. ChSVtio20

27. Yamipa

28. Grho

29. Laesan

30. Armmo29

31. O'Drecin20

32. Safames20

33. Syestna81

34. Bointtoan

35. HuC.

acrjournals.org


Support

cologic Cancer Foundation/Susan G. Komen Career Development Awardpillman), Ovarian Cancer Research Fund/Liz Tilberis Scholars (M.A.n), NIH Women's Reproductive Health Research Scholar grant00127-10 (M.A. Spillman), NIH National Cancer Institute CA026869 (K.B.), Avon Foundation (K.B. Horwitz), Breast Cancer Research Foundationrwitz), and National Foundation for Cancer Research (K.B. Horwitz).costs of publication of this article were defrayed in part by the paymentcharges. This article must therefore be hereby marked advertisement innce with 18 U.S.C. Section 1734 solely to indicate this fact.

otential conflicts of interest were disclosed.Received 04/13/2010; revised 08/19/2010; accepted 08/30/2010; published

OnlineFirst 10/19/2010.

rencesmminki E, Kennedy D, Baum C, McKinlay S. Prescribing ofncontraceptive estrogens and progestins in the United States,74-86. Am J Public Health 1988;78:1479–81.ou B, Sun Q, Cong R, et al. Hormone replacement therapyd ovarian cancer risk: a meta-analysis. Gynecol Oncol 2008;108:1–51.arce C, Chung K, Pike M, Wu A. Increased ovarian cancer riskociated with menopausal estrogen therapy is reduced by addingrogestin. Cancer 2009;115:531–9.driguez C, Calle E, Coates R, Miracle-McMahill H, Thun M, HeathEstrogen replacement therapy and fatal ovarian cancer. Am Jidemiol 1995;141:828–35.cey JJ, Mink P, Lubin J. Menopausal hormone replacementrapy and the risk of ovarian cancer. JAMA 2002;288:334–41.ral V, Bull D, Green J, Reeves G. Ovarian cancer and hormonelacement therapy in the Million Women Study. Lancet 2007;369:03–10.S-M. Estrogen, progesterone, and epithelial ovarian cancer.

prod Biol Endocrinol 2003;1:73–8.o B, Slotman B. Endocrine factors in common epithelial ovariancer. Endocr Rev 1991;12:14–26.mmoss F, Pfisterer J, Thome M, Schafer W, Suerbrei W, PfleidererSteroid receptors in ovarian carcinoma: immunohistochemicaltermination may lead to new aspects. Gynecol Oncol 1992;47:–32.J, Hirsch M, Lee H, Matulonis U. Prognosis and hormone recep-status in older and younger patients with advanced-stage papil-serous ovarian carcinoma. Gynecol Oncol 2009;115:401–6.gdon S, Lawrie S, Hay F, et al. Characterization and properties ofe human ovarian adenocarcinoma cell lines. Cancer Res 1988;48:6–72.aia P, Sinkovics J, Rutledge R, Smith J. Cell-mediated immunityhuman malignant cells. Am J Obstet Gynecol 1972;114:979–89.rrell JC, Dye WW, Allred DC, et al. Estrogen receptor positiveast cancer metastasis: altered hormonal sensitivity and tumorgressiveness in lymphatic vessels and lymph nodes. Cancer Res06;66:9308–15.rtorius CA, Shen T, Horwitz KB. Progesterone receptors A and Berentially affect the growth of estrogen-dependent human breastor xenografts. Breast Cancer Res Treat 2003;79:287–99.rrell JC, Dye WW, Harvell DM, et al. Estrogen insensitivity in adel of estrogen receptor-positive breast cancer lymph nodetastasis. Cancer Res 2007;67:10582–91.ighton C, Cordero K, Larios J, et al. Genes regulated by estrogenbreast tumor cells in vitro are similarly regulated in vivo in tumorografts and human breast tumors. Genome Biol 2006;7:R28.rvell D, Richer J, Allred D, Sartorius C, Horwitz K. Estradiol regu-s different genes in human breast tumor xenografts comparedh the identical cells in culture. Endocrinology 2005;147:700–13.sor J, Danes JM, KommB, Chang KC, Lyttle CR, Katzenellenbogen. Profiling estrogen up- and down-regulated gene expression in

ngjitgamol S, Levenback C, Beller U, Kavanagh J. Role of surgicalection for lung, liver, and central nervous system metastases intients with gynecologic cancer: a literature review. Int J Gynecolncer 2004;14:399–422.ck G, Korach K, Gorski J. Relative mitogenic activities of variousrogens and antiestrogens. Steroids 1989;54:277–43.rwitz K, Aiginger P, Kuttenn F, McGuire WL. Nuclear estrogeneptor release from antiestrogen suppression: amplified inductionprogesterone receptor in MCF-7 human breast cancer cells.docrinology 1981;108:1703–9.osh M, Thompson D, Weigel R. PDZK1 and GREB1 are estrogen-ulated genes expressed in hormone-responsive breast cancer.ncer Res 2000;60:6367–75.ang S, Balch C, Chan MW, et al. Identification and characterizationovarian cancer-initiating cells from primary human tumors. Cancers 2008;68:4311–20.icoechea S, Arneman D, Otey C. The role of palladin in actinanization and cell motility. Eur J Cell Biol 2008;87:517–25.shimoto Y, Skacel M, Adams J. Roles of fascin in human carci-ma motility and signaling: prospects for a novel biomarker? Int Jchem Cell Biol 2005;37:1787–804.en Y, Han B, Zhu L, Wang W, Li L, Lancaster JM. A supervisedD approach to ovarian cancer chemotherapy response predic-n with across factor normalization. Int Conf Bioinf Biomed Eng09:1–4.ng F, Foekens J, Sieuwerts A, et al. Laser microdissection andcroarray analysis of breast tumors reveal ER-a related genes andthways. Oncogene 2006;25:1413–9.ow D. Metabolism of endogenous and exogenous reproductivermones. Obstet Gynecol Clin North Am 2002;29:425–36.ngdon S, Ritchie A, Young K, et al. Contrasting effects of 17β-tradiol on the growth of human ovarian carcinoma cells in vitrod in vivo. Int J Cancer 1993;55:459–64.aiz-Pena GN, Mangala LS, Spannuth WA, et al. Estrous cycledulates ovarian carcinoma growth. Clin Cancer Res 2009;15:71–7.onnell A, Macleod K, Burns D, Smyth J, Langdon S. Estrogeneptor-α mediates gene expression changes and growth responseovarian cancer cells exposed to estrogen. Endocr Relat Cancer05;12:851–66.mpath D, Winneker R, Zhang Z. Cyr61, a member of the CCNily, is required for MCF-7 cell proliferation: regulation by 17β-

tradiol and overexpression in human breast cancer. Endocrinology01;142:2540–8.ed V, Zhang X, Lau K-M, Cheng R, Mukherjee K, Ho S-M. Profilingrogen-regulated gene expression changes in normal and malig-nt human ovarian surface epithelial cells. Oncogene 2005;24:28–43.urguignon L, Gilad E, Rothman K, Peyrollier K. Hyaluronan-CD44eraction with IQCAP1 promotes Cdc42 and ERK signaling, leadingactin binding, Elk-1/estrogen receptor transcriptional activation,
d ovarian cancer progression. J Biol Chem 2005;280:11961–72.W, McCrea P, Deavers M, Kavanagh J, Kudelka A, VerschraegenIncreased expression of fascin, motility associated protein, in cell



culma

36. BaElico

37. LanterCa

38. Hetwein ode

39. Arirecma

40. TanTanep

41. HapaOnCa

42. Argin tco

43. Bociazosu

44. Miaft

45. Guca

Spillman et al.

Cance8936


tures derived from ovarian cancer and in borderline and carcino-tous ovarian tumors. Clin Exp Metastasis 2000;18:83–8.nka CL, Lund CV, Nguyen MTN, Pakchoian AJ, Mueller BM,ceiri BP. Estrogen induces lung metastasis through a hostmpartment-specific response. Cancer Res 2006;66:3667–72.gdon S, Gabra H, Bartlett JM, et al. Functionality of the proges-one receptor in ovarian cancer and its regulation by estrogen. Clinncer Res 1998;4:2246–51.cht J, Kotsopoulos J, Hankinson S, Tworoger S. Relationship be-en epidemiologic risk factors and hormone receptor expressionvarian cancer: results from the Nurses' Health Study. Cancer Epi-miol Biomarkers Prev 2009;18:1624–30.as-Pulido H, Smith H, Joste N, et al. Estrogen and progesteroneeptor status and outcome in epithelial ovarian cancers and lowlignant potential tumors. Gynecol Oncol 2009;114:480–5.
gjitgamol S, Manusirivithaya S, Khunnarong J, Jesadapatarakul S,wanich S. Expressions of estrogen and progesterone receptors in
ithelial ovarian cancer. Int J Gynecol Cancer 2009;19:620–7.

46. Invanwo



tch K, Beecham J, Blessing J, Creasman W. Responsiveness oftients with advanced ovarian cancer to tamoxifen. A Gynecologiccology Group study of second-line therapy in 105 patients.ncer 1991;68:269–71.enta P, Thomas S, Judson P, et al. A phase II study of fulvestranthe treatment of multiply-recurrent epithelial ovarian cancer. Gyne-l Oncol 2009;113:205–9.wman A, Gabra H, Langdon S, et al. CA125 response is asso-ted with estrogen receptor expression in a phase II trial of letro-le in ovarian cancer: identification of an endocrine-sensitivebgroup. Clin Cancer Res 2002;8:2233–9.chaelson-Cohen R, Beller U. Managing menopausal symptomser gynecologic cancer. Curr Opin Oncol 2009;21:407–11.idozzi F, Daponte A. Estrogen replacement therapy for ovarianrcinoma survivors. Cancer 1999;86:1013–8.estigators Writing Group for the Women's Health Initiative. Risks
d benefits of estrogen plus progestin in healthy postmenopausalmen. JAMA 2002;288:321–33.
Cancer Research



Published OnlineFirst October 19, 2010.Cancer Res Monique A. Spillman, Nicole G. Manning, Wendy W. Dye, et al. Cancer Growth and MetastasisTissue-Specific Pathways for Estrogen Regulation of Ovarian

Updated version

10.1158/0008-5472.CAN-10-1238doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2010/10/18/0008-5472.CAN-10-1238.DC1

Access the most recent supplemental material at:

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

Subscriptions

Reprints and

[email protected] at

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

Permissions

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

.http://cancerres.aacrjournals.org/content/early/2010/10/15/0008-5472.CAN-10-1238To request permission to re-use all or part of this article, use this link

Research. on June 15, 2018. © 2010 American Association for Cancercancerres.aacrjournals.org Downloaded from


http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-10-1238

http://cancerres.aacrjournals.org/content/suppl/2010/10/18/0008-5472.CAN-10-1238.DC1

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/early/2010/10/15/0008-5472.CAN-10-1238


Research Tissue-Specific Pathways for Estrogen Regulation of Ovarian Cancer...

Documents

Transcript of Research Tissue-Specific Pathways for Estrogen Regulation of Ovarian Cancer...