E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r. com/ loca te /yexc r

Research Article

Conditional inactivation of Brca1 in the mouse ovarian surfaceepithelium results in an increase in preneoplastic changes

Katherine V. Clark-Knowlesa,c,⁎, Kenneth Garsonc, Jos Jonkersd, Barbara C. Vanderhydena,c

aDepartment of Cellular and Molecular Medicine, University of Ottawa, 501 Smyth Rd., Box 926, Ottawa, Ontario, Canada K1H 8L6bDepartment of Obstetrics and Gynecology, University of Ottawa, 501 Smyth Rd., Box 926, Ottawa, Ontario, Canada K1H 8L6cOttawa Health Research Institute, 501 Smyth Rd., Box 926, Ottawa, Ontario, Canada K1H 8L6dDivision of Molecular Biology, Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands

A R T I C L E I N F O R M A T I O N

⁎ Corresponding author. Ottawa Health ReseaE-mail address: kclar075@uottawa.ca (K.V

0014-4827/$ – see front matter © 2006 Elsevidoi:10.1016/j.yexcr.2006.09.026

A B S T R A C T

Article Chronology:Received 22 June 2006Revised version received25 September 2006Accepted 25 September 2006Available online 3 October 2006

Epithelial ovarian cancer (EOC) is thought to arise from the ovarian surface epithelium (OSE);however, the molecular events underlying this transformation are poorly understood.Germline mutations in the BRCA1 tumor suppressor gene result in a significantly increasedrisk of developing EOC and a large proportion of sporadic EOCs display some sort of BRCA1dysfunction. Using mice with conditional expression of Brca1, we inactivated Brca1 in themurine OSE and demonstrate that this inactivation results in the development ofpreneoplastic changes, such as hyperplasia, epithelial invaginations, and inclusion cysts,which arise earlier and aremore numerous than in control ovaries. These changes resemblethe premalignant lesions that have been reported in human prophylactic oophorectomyspecimens from women with BRCA1 germline mutation. We also report that inactivation ofBrca1 in primary cultures of murine OSE cells leads to a suppression of proliferation due toincreased apoptosis that can be rescued by concomitant inactivation of p53. Theseobservations, along with our finding that these cells display an increased sensitivity tothe DNA-damaging agent cisplatin, indicate that loss of function of Brca1 in OSE cellsimpacts both cellular growth control and DNA-damage repair which results in altered cellbehavior manifested as morphological changes in vivo that arise earlier and are morenumerous than what can be attributed to ageing.

© 2006 Elsevier Inc. All rights reserved.

Keywords:Brca1Ovarian cancerPreneoplasticMouse modelEpithelium

Introduction

Approximately 90% of ovarian cancers are epithelial in natureand are thought to arise from the ovarian surface epithelium(OSE) [1]. The OSE is a single layer of squamous to cuboidalcells that forms the outer layer of the ovary and is separatedfrom the underlying ovarian stroma by a basement mem-brane. It is a simple mesothelium that is embryonicallyderived from the coelomic epithelium, and has a very plastic

rch Institute, 501 Smyth R. Clark-Knowles).

er Inc. All rights reserved

phenotype which retains the ability to undergo epithelio-mesenchymal transition [2]. There are several lines ofevidence that indicate that the OSE is the tissue from whichovarian epithelial carcinomas develop. Firstly, with age andrepeated ovulatory cycles, the OSE tends to assume a moreirregular contour and forms invaginations or clefts into thestroma. These clefts may pinch off completely and formepithelial inclusion cystswithin the ovary. Histological studieshave shown that the OSE lining these clefts and cysts has

d., Box 926, Ottawa, Ontario, Canada K1H 8L6. Fax: +1 613 247 3524.

.

134 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

often begun to undergo early metaplastic changes, such asassuming amore columnar cell shape, and expressing some ofthe markers of ovarian tumors, such as E-cadherin [3] and thetumor antigen CA125 [1]. It is postulated that these morpho-logical changes may constitute premalignant lesions thatcould eventually give rise to ovarian neoplasia. Histologicalexamination of ovaries removed prophylactically fromwomen considered at high risk for developing ovarian cancer,either due to a strong family history of the disease or to thepresence of a germline mutation in the BRCA1 gene, revealedmore of these morphological changes in the OSE than inovaries removed incidentally during a total abdominal hyster-ectomy [4–6]. In one study, 6% of prophylactically removedovaries from known BRCA1 germline mutation carriers werefound to harbor microscopic ovarian carcinomas [7].

Five to fifteen percent of ovarian cancers are thought to bedue to hereditary factors, and 90% of these can be attributed togermline mutations in the BRCA1 gene [8,9]. BRCA1, known asthe breast/ovarian cancer susceptibility gene and located onchromosome 17q21 in humans, is a 220 kDa, 1863 amino acidnuclear phosphoprotein [10,11]. It is composed of 24 exons,with exon 11 comprising 60% of the gene [12]. Germlinemutations in this gene inherited in an autosomal dominantfashion and generally resulting in a truncated protein product,significantly increase the lifetime risk of breast and ovariancancer, leading to the classification of BRCA1 as a tumorsuppressor gene. These mutations confer up to a 60% lifetimerisk of ovarian cancer, compared to approximately 2% in thegeneral population [13,14]. Though somatic mutations inBRCA1 are rare, reduced or absent protein expression hasbeen seen in 90% of sporadic ovarian tumors examinedindicating that epigenetic factors, mainly promoter hyper-methylation, are also involved in its regulation [15–17]. TheBRCA1 gene has been implicated in a wide variety of cellularprocesses, including maintenance of genome integrity [18],DNA damage recognition and repair [19,20], cell cycle check-point control [21,22], and apoptosis [23].

There currently exists a small number of mouse models ofovarian carcinoma based on the alteration of specific geneticpathways [24–27]; however, none of these have targeted theBrca1 gene. While it has been extensively studied in mousemodels of mammary epithelial tumorigenesis [28,29], and onestudy has examined loss of Brca1 in murine ovarian granulosacells [30], the role of Brca1 in the transformation of the OSEremains unclear. The latency ofmammary tumor formation inmice in which Brca1 was inactivated in the mammaryepithelium could be shortened by the concomitant inactiva-tion of the p53 tumor suppressor gene [29,31], indicating thatloss of Brca1 may be an early event that renders a cell moresusceptible to transformation via further adverse geneticevents.

In the present study, a conditional and site-directed stra-tegy was employed to inactivate Brca1 in the OSE, in order tocircumvent the embryonic lethality that is a consequence ofgermline deletion of this gene [32]. Targeted inactivation ofBrca1was achieved via adenoviral delivery of Cre recombinaseunder the ovarian bursal membrane in transgenic micebearing loxP sites in introns 4 and 13 of the Brca1 gene. Weexamined the ovaries of these mice at various time pointsfollowing the inactivation of Brca1 in the OSE, to determine if

there was a progressive alteration in epithelial morphology oreven tumor formation as has been observed in humanprophylactic oophorectomy specimens. In conjunction withthe in vivo work, we also established primary cultures of OSEcells from these mice to determine the impact of conditionalinactivation of Brca1 on their rate of proliferation andsensitivity to cisplatin.

Materials and methods

Experimental animals

Brca1loxP/loxP [FVB;129-Brca1tm2Brn] mice, bearing loxP sites inintrons 4 and 13 of the Brca1 gene, and p53loxP/loxP [FVB;129-Trp53tm1Brn] [33]mice bearing loxP in introns 2 and 10 of the p53gene, were obtained from the Mouse Models of HumanCancers Consortium Mouse Repository (National CancerInstitute, Rockville, MD, USA). We intercrossed these micethrough multiple generations to create a colony of homo-zygous Brca1loxP/loxP/p53loxP/loxP double conditional knockoutmice. All animal experiments described in this study wereperformed according to the Guidelines for the Care and Use ofAnimals established by the Canadian Council on Animal Care.

Genotyping

Brca1loxP/loxP mice were genotyped using primers Brca1int4fwd(5′ TAT CAC CAC TGA ATC TCT ACC G 3′) and Brca1int4rev (5′GAC CTC AAA CTC TGA GAT CCA C 3′) or Brca1int13fwd (5′TAT TCT TAC TTC GTG GCA CAT C 3′) and Brca1int13rev (5′TCC ATA GCA TCT CCT TCT AAA C 3′). Amplification withprimers Brca1int4fwd and Brca1int4rev yields a 461 bp and391 bp fragment for floxed and wild-type sequences, respec-tively, while primers Brca1int13fwd and Brca1int13rev yield a562 bp fragment for the floxed sequence and a 492bp fragmentfor the wild-type sequence. Cre-mediated excision of exons 5–13 (hereby designated Brca1Δ5–13) was detected by PCR ampli-fication using primers Brca1int4fwd and Brca1int13rev to yielda 600 bp product. Detection of any remaining unrecombinedsequence following Cre exposure was performed utilizingprimers for exon 11 of Brca1, yielding a 592 bp product [34].

DNA extraction

DNA extraction buffer (50 mM KCl, 10 mM Tris–HCl, 2 mMMgCl2, 0.1 mg/ml gelatin, 0.45% Nonidet, and 0.45% Tween-20)containing 40 μg/ml proteinase K (Roche, Mississauga, ON,Canada) was added to cells or tissue and incubated at 58°Covernight. DNA was precipitated using a standard protocolusing saturated NaCl and isopropanol. The DNA pellet waswashed with 75% ethanol, air dried, and resuspended in Tris–EDTA (50 mM, pH 6.8).

Primary culture of mouse OSE (MOSE) cells

Prior to MOSE cell collection, mice were superovulated byadministration of 5 I.U. pregnant mare serum gonadotropin(PMSG; Intervet, Whitby, ON, Canada) intraperitoneally (IP),followed 46–48 h later by IP administration of 5 I.U. human

135E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

chorionic gonadotropin (hCG; Sigma-Aldrich, Oakville, ON,Canada). Ovaries were individually dissected 72 h followinghCG administration. The ovaries were washed twice withphosphate-buffered saline (PBS, Hyclone, Logan, UT, USA),incubated in 0.25% Trypsin/PBS (1 ml/ovary; Invitrogen,Burlington, ON, Canada) in a 15 ml Falcon tube at 37°C for1 h. Alpha minimum essential medium (α-MEM; Gibco BRL,Burlington, ON, Canada) containing 4% heat-inactivated fetalbovine serum (HI-FBS, Cansera, Etobicoke, ON, Canada), 5 U/ml penicillin/0.005 mg/ml streptomycin solution (Sigma-Aldrich), 0.1 μg/ml gentamicin (Gibco-BRL), and 1 μg/mlinsulin–transferrin–sodium–selenite solution (ITSS; Roche)was added to inactivate the trypsin and the tube was agitatedgently by hand to remove the OSE cells from the ovary. Thesupernatant containing the OSE cells was centrifuged at1000 rpm for 10 min. The cells were resuspended in theabove medium and plated in 24-well tissue culture plates(Becton-Dickinson, Oakville, ON, Canada) in the above med-ium and passaged upon reaching confluence. Following thefirst passage the cells were grown in α-MEM containing 10%HI-FBS.

Adenovirus administration

Recombinant adenoviruses Ad5CMVeGFP (AdGFP), Ad5CMVLacZ(AdLacZ), or Ad5CMVCre (AdCre) (Vector Development Labora-tory, Houston, TX, USA) were delivered to the OSE in vivo viaintrabursal injection. Six to eight week-old virgin animals wereanesthetized via intraperitoneal injection of Avertin (2.5% v/v in0.85% NaCl, 0.015–0.017 ml/g body weight; Sigma-Aldrich) andthe ovaries were individually accessed via dorsal incision.Approximately 10 μl of adenovirus (4×107 pfu/μl in PBS) wasdelivered using a 1 cm3 syringe and a 30-gauge beveled needleat the oviductal–bursal junction. Both the left and right ovarieswere injected in each animal. The animals were not bred overthe course of the experiment.

For in vitro infections, 5.0×105 MOSE cells were infectedfollowing an established protocol [25]. All experiments wereinitiated following replating of the cells 72 h after adenoviralinfection. In order to examine LacZ expression in whole ovaries,the ovaries were fixed in 0.2% gluteraldehyde in 0.1M PBS andstained with X-Gal (Bio-WORLD, Dublin, OH, USA). eGFP expres-sion was visualized with a Leica MZFLIII fluorescence stereo-microscope (Leica Microsystems, Wetzlar, Germany).

Histological analysis

Animalswere sacrificed via CO2 asphyxiation at 60, 120, 180, or240 days post-injection. The ovaries (with the ovarian fat padand bursa intact) were removed individually along with theattached oviduct and a portion of the uterus, fixed in formalinand paraffin-embedded. Five micrometer serial sections werecut either for hematoxylin and eosin (H&E) staining orimmunohistochemistry. Morphological changes to the OSEwere assessed by examining five non-consecutive H&E sec-tions, spaced at 20–25 μm intervals, from the middle of eachovary at 200× magnification using an Olympus BX50 micro-scope (Olympus, Melville, NY, USA). This number and spacingof sections were deemed to be a sufficiently representativesampling of the whole ovary based on previous analysis of a

subset of serial sections of whole ovaries. Samples wereblinded prior to the histological analysis. Sections wereevaluated for the number of areas of columnar cells, areas ofhyperplasia, and invaginations of the OSE, as well as epithelialinclusion cysts. Hyperplasia was defined as cells that had lostapical/basal polarity and exhibited layeringwith the formationof papillations of the OSE. Areas of columnar cells orhyperplasia were defined as a segment of epithelium thatconstituted these morphological changes flanked on eitherside by morphologically normal epithelium. Epithelial invagi-nations were defined as structures where the OSE distinctlyinvaginated into the ovarian stroma at or close to a right angleto the normal OSE such that the structure was flanked byovarian stroma on both sides. Inclusion cystswere defined as aspherical structure (as assessed by examining serial sections,to ensure that these structures were not merely invaginationssectioned tangentially) within the ovarian stroma with adiscernible epithelium-lined lumen. The epithelial nature ofthe lining of inclusion cysts was confirmed via immunohis-tochemistry for cytokeratin 19. Proliferation in the OSE in vivowas assessed by counting the number of Ki67-positive cells inthe OSE in five non-consecutive sections per ovary.

Immunohistochemistry

Paraffin sections were deparaffinized in xylene and rehy-drated in ethanol according to standard protocol. Hightemperature antigen retrieval was performed using sodiumcitrate buffer (pH 6.0) and endogenous peroxidase activitywas blocked using 3% hydrogen peroxide in Stockholm PBS(S-PBS). All samples were blocked using an avidin/biotinblocking kit (DAKO, Cytomation, Carpentaria, CA, USA).Primary antibodies were diluted in S-PBS at the followingconcentrations: Cytokeratin 19 (CK19), 1:100 (TROMA-1,Developmental Studies Hybridoma Bank, University ofIowa, USA); Ki67, 1:25 (DAKO); p53, 1:50 (Santa CruzBiotechnology; Santa Cruz, CA, USA); E-cadherin, 1:100(Santa Cruz); and Collagen IV, 1:100 (Novus Biologicals,Littleton, CO, USA). Sections were incubated with primaryantibody at room temperature overnight. Following three 5-min washes in S-PBS, sections were stained with an anti-rabbit or rat secondary antibody (1:200, DAKO) for 20 minfollowed by three 5-min washes and incubation with astreptavidin/horseradish peroxidase solution (1:200, DAKO)for 20 min. Developing was performed with diaminobenzi-dine (DAB) as a substrate (0.2% DAB, 0.001% H2O2; Sigma-Aldrich). Slides were counterstained with hematoxylin,dehydrated, and mounted on glass slides.

In vitro proliferation assay

MOSE cells that had been infected 72-h earlier with eitherAdCre recombinase (Δ) or AdGFP (loxP/loxP) were plated in 24-well tissue culture plates at a density of 2.0×104 cells/well.Cells were trypsinized with 0.05% trypsin (Hyclone) andcounted using a Coulter Counter (Beckman Coulter, Missis-sauga, ON, Canada) at 24, 48, 72, or 96-h time points. Growthmedium was replaced every 48 h. Experiments were per-formed three times in triplicate, with a separate infection foreach replicate.

136 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

Cisplatin sensitivity assay

MOSE cells that had been infected 72-h earlier with eitherAdCre or AdGFP were plated in 12-well plates at a density of8.0×104 cells/well. The following day the normal growthmedium was replaced with either fresh media or mediacontaining 1 μM or 5 μM cisplatin (1 mg/ml stock, FauldingPharmaceutical Co, Paramus, NJ, USA). Remaining adherentcells were trypsinized and counted using a Coulter Counter at24 and 48 h following the addition of cisplatin. Experimentswere performed three times in triplicate, with a separateinfection for each replicate.

Flow cytometry

Cells were trypsinized andwashed twicewith PBS, followed byfixation in 70% ethanol at −20°C for a minimum of 1 h. Bothfloating and adherent cells were collected for analysis. Cellswere then washed twice with PBS containing 1% serum andresuspended in propidium iodide (Sigma, 20 mg/ml in PBS)containing RNAse A (40 mg/ml, MBI Fermentas, Burlington,ON, Canada). Samples were analyzed using a BD LSR flowcytometer with CellQuest software (BD Biosciences, Missis-sauga, ON, Canada).

Western blotting

Protein samples were collected in RIPA buffer containingprotease and phosphatase inhibitors and protein content wasmeasured using a commercially available protein assay (Bio-Rad Protein Assay Kit, Bio-Rad Laboratories, Mississauga, ON,Canada). Samples were separated on a 4–12% Bis–Tris pre-castpolyacrylamide gel (Invitrogen) and transferred to a nitrocel-lulose membrane (Hybond C Extra, Amersham, Oakville, ON,Canada). Blocking was carried out with 5% milk in Tris-buffered phosphatewith Tween-20 (TBS-T). For all subsequentimmunoblotting, antibodies were diluted to the appropriateconcentration in 5% milk in TBS-T. Blots were incubated withthe following primary antibodies: rabbit anti-p53, 1:500 (SantaCruz) andmouse anti-GAPDH, 1:2000 (AbCam, Cambridge, MA,USA). Visualization of protein bands was performed using theECL Plus Western blotting detection system (Amersham) andGeneSnap image acquisition system (Syngene, Frederick, MD,USA).

Statistical analyses

Counts of morphological features in the OSE are expressed asthe mean±SEM (standard error of the mean) of the number ofmorphological features per section in five non-consecutiveovarian sections for n ovaries, where n is the number ofovaries examined. In vitro cell counts are expressed as themean±SEM of three independent experiments performed intriplicate. The probability of significant differences uponcomparison of only two groups was determined by Student'st test. When multiple groups were analyzed, statisticalcomparisons where made by analysis of variance (ANOVA).Bonferroni's posttest was used to determine significancebetween specific groups when whole group differences weredetected by ANOVA. P-values are two-sided. Analysis was

performed using Graphpad Prism statistical software (Graph-pad Software, San Diego, CA, USA).

Results

The OSE can be efficiently infected with adenoviral vectorsboth in vivo and in vitro

To ensure that intrabursal administration of adenoviralvectors resulted in infection confined to the OSE cells, withoutsignificant leakage from the bursal space, animals wereinjected with either AdGFP or AdLacZ and the ovaries wereremoved and examined 1 week post-surgery. Intrabursaladministration of adenoviral vectors results in a high rate ofinfectivity that is confined to the OSE, without penetratinginto the ovarian stroma below (Fig. 1B). Surrounding tissues,such as the ovarian fat pad, oviduct, and uterus, do notdemonstrate signs of infection (Fig. 1A). Administration ofadenoviral vectors to MOSE cells in vitro also results in theinfection of the majority of cells (Fig. 1C). No overt phenotypicchanges were observed in vitro in cells in which Brca1 wasconditionally inactivated.

Detection of recombination at loxP sites following adenoviralCre recombinase infection

Recombination at loxP sites following infection with AdCrewas confirmed by collecting Cre-infected (Brca1Δ5–13) andAdGFP-infected (Brca1loxP/loxP) ovaries at 180 days post-intra-bursal injection and removing the OSE cells as described inMaterials and methods. Recombination at loxP sites is evidentonly in the AdCre infected cells (Fig. 2A). Genomic DNA wasalso extracted fromMOSE cells infected in vitro. Recombinationat loxP sites could be detected in all of the samplesweanalyzedwith little if any unrecombined fraction remaining (Fig. 2B).

Conditional inactivation of Brca1 in murine OSE results in anincrease in morphological changes

To assess the consequences of inactivation of the Brca1 tumorsuppressor gene in the OSE, mice bearing loxP sites flanking asegment of their Brca1 gene were injected under the ovarianbursal membrane with either adenoviral Cre recombinase toconditionally inactivate Brca1 (Brca1Δ5–13) or AdGFP as a controladenovirus (Brca1loxP/loxP). Five non-consecutive 5 μm sectionsat 20–25 μm intervals from each ovary were assessed formorphological changes to the OSE defined as areas ofcolumnar cells, areas of hyperplasia, epithelial invaginations,and epithelial inclusion cysts. Examples of thesemorphologiesare shown in Fig. 3. There were no significant differencesbetween Brca1loxP/loxP and Brca1Δ5–13 ovaries in terms of thedimensions of age-matched ovaries in the sections evaluated.

The number ofmorphological changes in the OSE increasedoverall in both the Brca1Δ5–13 and Brca1loxP/loxP ovaries over time(Table 1). Significant increases in the number of changes wereevident after 180 days in the Brca1Δ5–13 group (P<0.001),whereas a significant difference was not evident until the240-day time point in the Brca1loxP/loxP group (P<0.01). Thispattern was also seen in terms of increases in the number of

137E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

areas of columnar epithelium over time. There was asignificant increase in the number of invaginations of theOSE in the Brca1Δ5–13 group between the 60 and 240-day timepoints (P<0.05),whereas therewasno significant change in thenumber of invaginations in the Brca1loxP/loxP group during thecourse of the study.

Not only the time course of morphological changes wasaffected by Brca1 inactivation, but there was also a moredramatic increase in the number of epithelial changes in theOSE of the Brca1Δ5–13 mice vs. Brca1loxP/loxP mice over time. Thetotal number of morphologic changes in the surface epithe-lium increased 6-fold between 60-day and 240-day post-intrabursal injection in the Brca1Δ5–13 mice, whereas only a 2-fold increase was observed in the Brca1loxP/loxP mice over thesame time period.

Fig. 1 – The OSE can be infected efficiently with adenoviralvectors both in vivo and in vitro. (A) One week followingintrabursal injection of AdGFP, eGFP expression (seen in theright panel using a GFP microscope) is confined to the ovarybeneath the bursa (arrow), and expression cannot bedetected in surrounding tissues such as the ovarian fat pad(fp), oviduct (ov), or uterus (u) (labeled in the left panel). Thesestructures can be seen clearly in the left panel, which is thesame ovary shown under white light. The arrowhead in theleft panel indicates a photographic lighting artifact. (B) Oneweek following intrabursal injection of AdLacZ, ovaries wereremoved, stained with X-Gal, and forceps were used toremove a portion of the OSE (strip between arrows) todemonstrate that infection is confined to the OSE and doesnot penetrate into the underlying ovarian stroma.(C) Cultured Brca1loxP/loxP MOSE cells were infected withAdLacZ with an MOI of 200 and stained with X-Gal after 24 h.Scale bars represent 1.6 mm (B) or 50 μm (C).

Fig. 2 – Detection of recombination at loxP sites followingadenoviral Cre recombinase infection in vivo and in vitro.(A) PCR of genomic DNA from OSE cells collected fromBrca1loxP/loxP mice 180 days following intrabursal injection ofAdCre (Brca1Δ5–13) or AdGFP (Brca1loxP/loxP ). (B) PCR of genomicDNA collected from cultured Brca1loxP/loxP OSE cells 72 hfollowing in vitro infection with AdCre (Brca1Δ5–13) or AdGFP(Brca1loxP/loxP ).

Overall, there were significantly more surface epithelialchangesobserved inovaries inwhichBrca1hadbeen inactivatedin the OSE than in control ovaries after 240 days (12.96±1.21 vs.9.42±0.76, P<0.05). Significantly more areas of columnarepithelium were observed in the Brca1loxP/loxP ovaries at the 60-day time point (0.99±0.21 vs. 2.26±0.28, P<0.05); however, thiswasnot sustainedat later timepoints. Therewereno significantdifferences in the number of areas of hyperplasia between theBrca1loxP/loxP and Brca1Δ5–13 ovaries at any time point examined.Therewere 4-foldmore epithelial invaginations in theBrca1Δ5–13

ovaries as compared to Brca1loxP/loxP at 240 days post-intrabursalinjection (1.44±0.70 vs. 0.34±0.14, P<0.05). Epithelial inclusioncystswere only observed in theBrca1Δ5–13 ovaries and only at the180 and 240-day time points.

An increase in morphological changes in the Brca1Δ5–13 OSE isnot due to an increase in proliferation

Very few (≤5) Ki67-positive cells were found in the OSE in any ofthe sections examined (Fig. 3H). There was no significantdifference in thenumber ofKi67-positive cells in theOSEbetweenthe Brca1loxP/loxP and Brca1Δ5–13 group at any of the time pointswhere a difference in the number of morphological changes wasobserved (data not shown). No association was seen betweenmorphological features of the OSE such as hyperplasia, invagina-tions, or inclusion cysts and the presence of Ki67-positive cells.

Fig. 3 – Morphological changes observed in the murine OSE. (A) Representative ovary from a Brca1loxP/loxP mouse 180 daysfollowing intrabursal injection of AdGFP (H&E, 25X). (B) Representative ovary from a Brca1Δ5–13 mouse 180 days followingintrabursal injection of AdCre (H&E, 25× arrows indicate areas of hyperplasia). (C) Normal cuboidal OSE cells (H&E, 200×).(D) Columnar OSE cells (H&E, 200×). (E) Hyperplasia (H&E, 200 ×). (F) Epithelial invagination (arrowhead, H&E, 100×). (G) Epithelialinclusion cyst (arrow, H&E, 100×). The arrowhead indicates what is in fact an invagination in tangential section. (H) Very fewKi67-positive cells were found in the OSE (arrow, immunohistochemistry for Ki67, 100×). Scale bars are equal to 50μm for 200×,0.1 mm for 100×, or 0.4 mm for 25×.

138 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

Expression of proteins involved in ovarian cancer initiationand progression

Immunohistochemical analysis was performed to determinethe expression of p53, E-cadherin, and Collagen IV, proteinswhich have all been shown to be involved in ovarian cancerinitiation and progression in humans, oftenwith alterations inexpression patterns seen in prophylactic oophorectomy speci-mens [35–37] (Fig. 4). p53 expression was not detected in theOSE of any of the ovaries examined in our study, nor was itfound in the epithelial invaginations or inclusion cysts,

although it was expressed strongly in murine SV40 Tag-induced ovarian tumors. E-cadherin expression is absent inthe normal humanOSE, with overexpression seen in inclusioncysts and tumors. Rodents normally do express E-cadherin inthe OSE, thus we hypothesized that perhaps in mice expres-sion would be reduced or lost with morphologic changesconsistent with preneoplasia. Expression was detected uni-formly in the OSE of all the mouse ovaries we examined, withno observable difference in expression levels in normal vs.morphologically altered OSE. Consistent Collagen IV expres-sion was observed in the basement membrane beneath the

Table 1 – Distribution of morphological features in the OSE over time following conditional inactivation of Brca1

Epithelialmorphology

Time (days)

60 120 180 240

Columnar cells Brca1Δ5–13 0.99±021a,* (n=9) 2.50±0.52a (n=7) 4.23±0.62b (n=9) 5.22±0.38b (n=10)Brca1loxP/loxP 2.26±0.28a (n=10) 2.73±0.26a (n=8) 3.50±0.21a (n=8) 3.61±0.71b (n=10)

Hyperplasia Brca1Δ5–13 1.16±0.47a 2.96±1.58a 4.54±0.88a 4.94±0.89b

Brca1loxP/loxP 1.80±0.44a 1.65±0.50a 2.68±0.42a,b 5.46±0.67b

Invaginations Brca1Δ5–13 0a 0.26±0.26a 0.60±0.33a 1.44±0.70b,*Brca1loxP/loxP 0.08±0.06a 0.05±0.05a 0.08±0.05a 0.34±0.14a

Inclusion cysts Brca1Δ5–13 0 0 0.19±0.19 0.80±0.00Brca1loxP/loxP 0 0 0 0

Total changes Brca1Δ5–13 2.14±0.54a 5.71±1.67a,b 9.56±1.69b 12.96±1.21b,*Brca1loxP/loxP 4.14±0.63a 4.43±0.73a 6.25±0.58a,b 9.42±0.76b

Numbers represent the mean±SEM number of morphological changes per section over five non-consecutive sections in n ovaries. Time is thenumber of days after intrabursal injection of AdCre (Brca1Δ5–13) or AdGFP (Brca1loxP/loxP). Letters are used to denote a significant differencebetween time points for that treatment group. * Indicates a significant difference between the Brca1Δ5–13 group and the Brca1loxP/loxP group at thattime point (P<0.05).

139E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

OSE in all of the mouse ovaries irrespective of morphologicalchanges in the OSE.

Inactivation of Brca1 in MOSE cells in vitro results in asuppression of proliferation with increased apoptosis that canbe rescued by concomitant inactivation of p53

Proliferation assays were performed on the adherent cells ofprimary cultures of Brca1loxP/loxP MOSE cells infectedwith eitherAdCre (Brca1Δ5–13) or AdGFP (Brca1loxP/loxP). The doubling time ofthese cells is quite long (approximately 48 h) and there were nodifferences in the proliferation rates of the Brca1Δ5–13 orBrca1loxP/loxP cells at 24, 48, or 72 h after plating; however, by96 h the Brca1loxP/loxP cells outgrew the Brca1Δ5–13 cells (Fig. 5A).A later subset of experiments in which the proliferation curvewas extended to 120 h demonstrated that this inhibition ofproliferation in the Brca1Δ5–13 cells was also significant at thislater time point, with a decrease in cell number of 40% ascompared to Brca1loxP/loxP cells (data not shown). Flow cyto-metric analysis of both adherent and floating cells demon-strated a sub-G1 fraction in the Brca1Δ5–13 cells that was nearlydouble that of the Brca1loxP/loxP cells after 96 h (31.00%±5.63% vs.17.44%±3.87%, P<0.05, Fig. 5B). The Brca1Δ5–13/Trp53Δ2–10 cells inwhich both Brca1 and p53 had been conditionally inactivatedvia infection with adenoviral Cre recombinase displayed asignificantly greater rate of proliferation than Brca1loxP/loxP/p53loxP/loxP cells at both the 72 (P<0.01) and 96-h (P<0.001) timepoints (Fig. 5C). Conditional inactivation of p53 alone inp53loxP/loxP cells did not result in a significant enhancement ofproliferation (Fig. 5D). Therewas no significant difference in thesub G1 fraction between Trp53Δ2–10 and p53loxP/loxP cells orbetween Brca1Δ5–13/p53Δ2–10 and Brca1loxP/loxP/p53loxP/loxP cells(data not shown).

Brca1Δ5–13 OSE cells have an increased sensitivity to cisplatin

Primary cultures of Brca1loxP/loxP and Brca1Δ5–13 MOSE cellswere treated with either 1 μM or 5 μM cisplatin for 24 or

48 h. At the 24-h time point at either concentration, therewas no difference between the cell types in cisplatinsensitivity, as measured by the number of adherent cellsin treated groups compared to cells not exposed to cisplatin(data not shown). By 48-h, however, the number ofadherent Brca1Δ5–13 cells remaining following treatmentwith 5 μM of cisplatin was significantly reduced comparedto Brca1loxP/loxP cells (47.35%±3.04% vs. 66.18%±6.10%,P<0.05, results expressed as a percentage of untreatedcells, Fig. 6A). Both the Brca1loxP/loxP cells and the Brca1Δ5–13

cells displayed a significant increase in the sub G1 fractionfollowing treatment with cisplatin; however, there was agreater increase seen in the Brca1Δ5–13 cells (Brca1loxP/loxP:6.72%±1.22% at 0 μM vs. 18.76%±2.95% at 5 μM, P<0.05,Brca1Δ5–13: 12.68%±1.49% at 0 μM vs. 29.35%±4.65% at 5 μM,P<0.01, Fig. 6B). Further cell cycle analysis revealed a trendtowards a greater proportion of Brca1Δ5–13 cells in the G2/Mphase of the cell cycle than Brca1loxP/loxP cells following 48 hof cisplatin treatment; however, these results were notsignificant (26.89%±5.58% vs. 12.19%±2.91%, P=0.08). Thenumber of surviving Trp53Δ2–10 cells following the samecisplatin treatment was also significantly reduced as comparedto p53loxP/loxP cells (31.18%±2.13% vs. 39.68%±2.81%, P<0.05, Fig.6C). There was no significant difference, however, in thenumber of surviving Brca1Δ5–13/Trp53Δ2–10 cells as compared tothe Brca1loxP/loxP/p53loxP/loxP cells following cisplatin treatment(59.96%±3.52% vs. 70.46%±5.90%, Fig. 6D). As assessed byWestern blotting, Brca1Δ5–13 cells displayed a greater increasein p53 protein expression after both 24 h and 48 h oftreatment with 5 μM cisplatin than was observed in theBrca1loxP/loxP cells (Fig. 6E).

Discussion

This study is the first to demonstrate that inactivation of theBrca1 tumor suppressor gene in the mouse OSE leads to anincrease in morphological changes in that tissue may

Fig. 4 – Immunohistochemical analysis of expression patterns of proteins known to be involved in ovarian tumorigenesis.Expression of E-cadherin (A–C; brown stain), Collagen IV (D–F; reddish-brown stain), and p53 (G–I; nuclear stain) in the OSE ofcontrol (AdGFP-injected) Brca1loxP/loxP mice (B, E, H) and OSE of (AdCre-injected) Brca1Δ5–13 mice (C, F, I). Murine uterineepithelium (A), murine liver (D), and a murine ovarian tumor (G) were used as positive controls. Scale bars are equal to 25 μm(A, D, G) or 50 μm (B, C, E, F, H, I).

140 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

constitute premalignant lesions similar to those that havebeen reported in human prophylactic oophorectomy speci-mens from BRCA1 germline mutation carriers. The numberand scope of morphological changes in the OSE increased inboth the ovaries in which Brca1 had been conditionallyinactivated and in control ovaries, which was not unexpected,as an increase in alterations in surface epithelial morphologywith agehas beenpreviously reported inmice [38] andhumans[4,39]. Significant increases in these changes arose earlier inthe Brca1Δ5–13 OSE, however, and there were significantlymoreepithelial changes overall observed in the ovaries of theseanimals than in controls at 240 days following intrabursalinjection. These data indicate that loss of function of the Brca1tumor suppressor gene in this tissue may accelerate the deve-lopment of morphologic alterations, resulting in the earlierappearance of some of the more complex epithelial changes,such as invaginations and inclusion cysts.

Epithelial invaginations and inclusions cysts have longbeenproposed to be theprecursor lesionsof ovarian carcinoma[40], representing the most advanced morphologic change inthe OSE. Age at diagnosis of ovarian cancer is typically muchearlier in BRCA1mutation carriers (54 years vs. 63 years in non-

carriers) [41], and as our data and human prophylacticoophorectomy specimens would suggest, preneoplasticchanges in the ovarian epithelium also appear earlier. Therewere significantly more areas of columnar epithelial cellsnoted in the ovaries of control mice at the earliest time point;however, this difference was not sustained through the latertime points. It is possible that this observation was due theexpression of, or immune reaction to, eGFP that was not seenin adenoviral Cre recombinase infected cells. We found nodifference in expression levels or patterns of several proteinsknown to be markers of human ovarian tumorigenesisbetween the Brca1Δ5–13 and control groups, which may pointto a species-related difference ormay suggest that alteration inthe expression of these proteins is not a direct consequence ofinactivation of Brca1 and is related to genetic events that occurlater in transformation.

We evaluated the proliferative capacity of the OSE in whichBrca1 had been inactivated in vivo by staining ovarian sectionsfor the proliferation marker Ki67. Studies of prophylacticoophorectomy specimens have found conflicting results inthis area, with some studies demonstrating an increase inproliferation in the OSE of BRCA1 mutation carriers [5], while

Fig. 5 – Conditional inactivation of Brca1 in cultured MOSE cells results in a suppression of proliferation that can be rescued byconcomitant inactivation of p53. (A) Proliferation of Brca1loxP/loxP MOSE following infection with AdCre (Brca1Δ5–13) to inactivateBrca1 or with AdGFP (Brca1loxP/loxP). (B) Percentage of cells in the sub G1 fraction at the 96 h time point following infection withAdCre (Brca1Δ5–13) or AdGFP (Brca1loxP/loxP). (C) Proliferation following concomitant inactivation of Brca1 and p53 in Brca1loxP/loxP/p53loxP/loxP MOSE after infection with AdCre (Brca1Δ5–13/Trp53Δ2–10) or AdGFP (Brca1loxP/loxP/p53loxP/loxP). (D) Proliferation ofp53loxP/loxP MOSE following infection with AdCre (Trp53Δ2–10) to inactivate p53 or AdGFP (p53loxP/loxP). * Indicates a significantdifference between at that time point where P<0.05, ** indicates a significant difference where P<0.01, and *** indicates asignificant difference where P<0.001. Error bars indicate SEM. All proliferation experiments were performed three times intriplicate. Flow cytometric analysis was performed on three independent experiments.

141E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

others found no difference [42]. Our results indicate that anincrease in the number of areas of altered epithelial morphol-ogy in the OSE due to inactivation of Brca1 is not accompaniedby a notable change in proliferation, thus the morphologicalchanges observed are not necessarily the result of the loss ofthe growth suppressor activity of Brca1. Our analysis of theeffect of inactivation of Brca1 in MOSE cells in vitro also foundthat the Brca1-deficient cells did not proliferate faster thancontrols and in fact loss of Brca1 function appeared to suppresstheir proliferation, which was in fact due to a nearly 2-foldincrease in basal levels of apoptosis. These results may seemcounterintuitive, given the tumor suppressor status of Brca1;however Brca1 null mice die embryonically due to majorproliferation defects and increased apoptosis [32,43] andcultures ofmouse embryonic fibroblasts (MEFs)with a deletionof exon 11 of Brca1display a senescence-like growthdefect [44].This phenomenonhasnot been investigated extensively in theovary; though one study found that human OSE cells fromwomen with a known or suspected germline mutation inBRCA1proliferated at a slower rate than theOSEof their normalcounterparts [45]. Our study is the first to demonstrate thiseffect in a conditional model of Brca1 inactivation in themurine OSE, which thereby suggests that this model may besuitable for more thorough investigations of the role of BRCA1in the behavior of the ovarian surface epithelium.

It has been postulated that loss of function of Brca1 results ina destabilized genome that is more sensitive to DNA-damagingassaults. Thus, in the absence of the caretaker function ofBrca1, other repair pathways, such as that of p53, are activatedresulting in slowed proliferation as damage is repaired. It maybe that complete loss of function of Brca1 in normal adulttissues simply creates a genetically unstable environment, theimpact of which is not wholly evident until further geneticalterations take place. We propose that an unstable genome inOSE cells, resulting from loss of function of Brca1, results inaltered cell behaviormanifested asmorphological changes thatarise earlier and are more numerous than what can simply beattributed to ageing. Our in vitro data would suggest that OSEcells in which Brca1 has been inactivated are more sensitive toDNA damaging agents, indicating a possible repair defect. Celldeath in theOSE due to an inability to recognize and repair DNAdamage could possibly result in tissue remodeling that wouldproduce the morphologic alterations reported here; however,the exact mechanism by which these cellular events translateto these more global tissue changes remains unknown.

There is considerable evidence, fromboth knockoutmodelsand mammary tumor models, that the p53 tumor suppressorgene plays a significant role in both Brca1-associated growthcontrol and tumorigenesis [28,46–48]. p53 dysfunction iscommonly associated with ovarian carcinoma, particularly in

Fig. 6 – Brca1Δ5–13 MOSE cells display an increased sensitivity to the DNA-damaging agent cisplatin in vitro associated with anincrease in p53 protein expression. (A) Brca1loxP/loxP MOSE cells infected with either AdCre (Brca1Δ5–13) or AdGFP (Brca1loxP/loxP)and treated with 5 μM cisplatin for 48 h. Values are the number adherent cells presented as a proportion of similar cellsnot exposed to cisplatin (untreated cells). (B) Percentage of cells in the sub G1 fraction following 48 h of treatment with 0 μM or5 μM cisplatin. (C) p53loxP/loxP MOSE cells infected with either AdCre (Trp53Δ2–10) or AdGFP (p53loxP/loxP) and treated with 5 μMcisplatin for 48 h. Values are the number adherent cells presented as a proportion of similar cells not exposed to cisplatin(untreated cells). (D) Brca1loxP/loxP/p53loxP/loxP MOSE cells infected with either AdCre (Brca1Δ5–13/Trp53Δ2–10) or AdGFP(Brca1loxP/loxP/p53loxP/loxP) and treated with 5 μM cisplatin for 48 h. Values are the number adherent cells presented as aproportion of similar cells not exposed to cisplatin (untreated cells). (E) Protein was collected from Brca1Δ5–13 and Brca1loxP/loxP

cells after 0, 24, and 48 h of treatment with 5 μM cisplatin and subjected toWestern blotting with an anti-p53 antibody. GAPDHexpression was used as a loading control. Error bars indicate SEM. * indicates a significant difference where P<0.05, ** indicatesa significant difference where P<0.01. All proliferation experiments were performed three times in triplicate. Flow cytometricanalysis was performed on three independent experiments.

142 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

tumorsassociatedwitha germlineBRCA1mutation [49,50], andprophylactic oophorectomy specimens from BRCA1 heterozy-goteshavedemonstrated increasedp53 immunostaining in theOSE [5]. We immunostained all of our ovarian sections for p53expression and found none in any of the samples examined. Itcouldbe that suchaneventoccurs later than the timepointsweexamined or was simply not detectable by this method.

While upregulation of p53 expression was not observed inMOSE cells in which Brca1 had simply been inactivated,increased protein levels were seen in these cells when theywere challenged with the DNA-damaging agent cisplatin, achemotherapeutic agent commonly used to treat ovariancancer. These cells were also more sensitive to treatmentwith cisplatin, as measured by cell survival, providing further

evidence for the hypothesis that loss of Brca1 in theOSE leads toa DNA-damage repair defect and hence destabilized genomethat, in the presence of intact p53 function, results in acommitment to cell death. Cisplatin treatment did result in anincrease in apoptosis, as determined by sub G1 fraction, whichwasmore pronounced in the Brca1Δ5–13 cells. This observation isin keepingwith the finding that ovarian cancer patients bearinggermlineBRCA1mutations display an improved initial responseto cisplatin treatment and slightly improved survival as a result[51]. The p53 status of these cells also modulated theirsensitivity to cisplatin treatment. While Trp53Δ2–10 cells weremore sensitive to cisplatin than p53loxP/loxPMOSE cells, therewasno difference in cisplatin sensitivity seen between theBrca1Δ5–13/Trp53Δ2–10 MOSE and Brca1loxP/loxP/p53loxP/loxP MOSE.

143E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

This result disagrees with a study of Brca1 inactivation in p53-nullmouse cells that found that cells deficient in both Brca1 andp53 were increasingly sensitive to carboplatin [52]. This is not aphenomenon that has been extensively studied in normal cells,and thus these results warrant further investigation.

Our study also found, when OSE cell proliferation wasexamined in vitro, that concomitant inactivation of p53 in MOSEcells rescued the growth suppression that was seen withinactivation of Brca1 alone and leads to a substantially in-creased growth rate as compared to control cells. Loss of p53function has been demonstrated to partially rescue theembryonic lethality of complete deletion of Brca1 and appearsto play a critical role in mammary tumorigenesis in mice [53].Ovarian tumor formation was not observed in any of the micein which Brca1 alone was inactivated in the OSE, even morethan 1 year after inactivation.Wehypothesize that inactivationof p53, or another genetic “hit”,may also beanecessary event inBrca1-associated ovarian tumorigenesis and experiments arecurrently underway to inactivate both genes in tandem viaintrabursal injection in Brca1loxP/loxP/p53loxP/loxP mice.

Chodankar et al. [30] found that Brca1 inactivation in thegranulosa cells of the murine ovarian follicle led to thedevelopment of benign cystadenomas; however, no carcino-mas were seen. They proposed that, since these tumorspossessed normal Brca1 alleles, Brca1 might influence tumordevelopment indirectly via an effector secreted by granulosacells. Our study found that inactivation of Brca1 in the OSE ledto the development of preneoplastic epithelial changes similarto those observed in the ovaries of human BRCA1 mutationcarriers; however, it was not sufficient for the formation ofcarcinomas. It may be that inactivation of Brca1 in bothepithelial and non-epithelial tissues of the ovary is necessaryfor a cooperative effect that would lead to carcinogenesis.Placing conditional inactivation of Brca1 under the control of aubiquitous ovarian promoter, such as that of MISIIR, wouldenable the investigation of this possibility.

Given the recent finding that as many as 15% of ovariancarcinomas are related to germline mutations in BRCA1 [9],coupled with the high rates of BRCA1 dysfunction seen in spo-radic ovarian tumors, it is imperative that we further our un-derstanding of the role of this gene in transformation in theovary. Thismousemodel of conditional inactivation of Brca1 inthe OSE will provide a useful tool for studying not only theimpact of loss of function of Brca1 in these cells, but also forexamining the formation of early preneoplastic lesions in theovary.

Acknowledgments

We gratefully acknowledge Dr. Anton Berns and Dr. JosJonkers for providing the Brca1loxP/loxP ahead of publicationand for providing the p53loxP/loxPmice. The authors thank KerriCourville and Colleen Crane for their technical assistance withinitiating the intrabursal injections and with histology andimmunohistochemistry. This research was supported byfunds from the National Cancer Institute of Canada and theCanadian Institutes of Health Research (B.C.V.) and by OntarioGraduate Scholarships and an Ontario Graduate Scholarshipin Science and Technology (K.V.C.).

R E F E R E N C E S

[1] N. Auersperg, M.I. Edelson, S.C. Mok, S.W. Johnson, T.C.Hamilton, The biology of ovarian cancer, Semin.Oncol. 25(1998) 281–304.

[2] N. Auersperg, A.S. Wong, K.C. Choi, S.K. Kang, P.C. Leung,Ovarian surface epithelium: biology, endocrinology, andpathology, Endocr. Rev. 22 (2001) 255–288.

[3] K. Sundfeldt, Y. Piontkewitz, K. Ivarsson, O. Nilsson,P. Hellberg, M. Brannstrom, P.O. Janson, S. Enerback, L. Hedin,E-cadherin expression in human epithelial ovarian cancerand normal ovary, Int. J. Cancer 74 (1997) 275–280.

[4] H. Salazar, A.K. Godwin, M.B. Daly, P.B. Laub, W.M. Hogan,N. Rosenblum, M.P. Boente, H.T. Lynch, T.C. Hamilton,Microscopic benign and invasive malignant neoplasms anda cancer-prone phenotype in prophylactic oophorectomies[see comments], J. Natl. Cancer Inst. 88 (1996) 1810–1820.

[5] P.W. Schlosshauer, C.J. Cohen, F. Penault-Llorca, C.R.Miranda, Y.J. Bignon, J. Dauplat, L. Deligdisch, Prophylacticoophorectomy: a morphologic and immunohistochemicalstudy, Cancer 98 (2003) 2599–2606.

[6] B.A. Werness, A.M. Afify, K.L. Bielat, G.H. Eltabbakh, M.S.Piver, J.M. Paterson, Altered surface and cyst epithelium ofovaries removed prophylactically from women with a familyhistory of ovarian cancer, Hum. Pathol. 30 (1999) 151–157.

[7] A. Finch, P. Shaw, B. Rosen, J. Murphy, S.A. Narod, T.J. Colgan,Clinical and pathologic findings of prophylacticsalpingo-oophorectomies in 159 BRCA1 and BRCA2 carriers,Gynecol. Oncol. 100 (2005) 58–64.

[8] A. Berchuck, K.A. Heron, M.E. Carney, J.M. Lancaster, E.G.Fraser, V.L. Vinson, A.M. Deffenbaugh, A. Miron, J.R. Marks,P.A. Futreal, T.S. Frank, Frequency of germline and somaticBRCA1 mutations in ovarian cancer, Clin. Cancer Res. 4(1998) 2433–2437.

[9] T. Pal, J. Permuth-Wey, J.A. Betts, J.P. Krischer, J. Fiorica,H. Arango, J. Lapolla, M. Hoffman, M.A. Martino, K. Wakeley,G. Wilbanks, S. Nicosia, A. Cantor, R. Sutphen, BRCA1 andBRCA2 mutations account for a large proportion of ovariancarcinoma cases, Cancer 104 (2005) 2807–2816.

[10] Y. Chen, A.A. Farmer, C.F. Chen, D.C. Jones, P.L. Chen, W.H.Lee, BRCA1 is a 220-kDa nuclear phosphoprotein that isexpressed and phosphorylated in a cell cycle-dependentmanner, Cancer Res. 56 (1996) 3168–3172.

[11] H. Ruffner, I.M. Verma, BRCA1 is a cell cycle-regulated nuclearphosphoprotein, Proc. Natl. Acad. Sci. U. S. A. 94 (1997)7138–7143.

[12] Y. Miki, J. Swensen, D. Shattuck-Eidens, P.A. Futreal,K. Harshman, S. Tavtigian, Q. Liu, C. Cochran, L.M. Bennett,W. Ding, R. Bell, J. Rosenthal, C. Hussey, T. Tran, M. McClure,C. Frye, T. Hattier, R. Phelps, A. Haugen-Strano, H. Katcher,K. Yakumo, Z. Gholami, D. Shaffer, S. Stone, A strongcandidate for the breast and ovarian cancer susceptibilitygene BRCA1, Science 266 (1994) 66–71.

[13] B. Aunoble, R. Sanches, E. Didier, Y.J. Bignon, Major oncogenesand tumor suppressor genes involved in epithelial ovariancancer (review), Int. J. Oncol. 16 (2000) 567–576.

[14] M.C. King, J.H. Marks, J.B. Mandell, Breast and ovarian cancerrisks due to inherited mutations in BRCA1 and BRCA2,Science 302 (2003) 643–646.

[15] J.P. Geisler, M.A. Hatterman-Zogg, J.A. Rathe, R.E. Buller,Frequency of BRCA1 dysfunction in ovarian cancer, J. Natl.Cancer Inst. 94 (2002) 61–67.

[16] P.A. Russell, P.D. Pharoah, F.K. De, S.J. Ramus, I. Symmonds,A. Wilson, I. Scott, B.A. Ponder, S.A. Gayther, Frequent loss ofBRCA1 mRNA and protein expression in sporadic ovariancancers, Int. J. Cancer 87 (2000) 317–321.

[17] K.Y. Chan, H. Ozcelik, A.N. Cheung, H.Y. Ngan, U.S. Khoo,Epigenetic factors controlling the BRCA1 and BRCA2 genes

144 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

in sporadic ovarian cancer, Cancer Res. 62 (2002)4151–4156.

[18] V. Yu, Caretaker Brca1: keeping the genome in the straightand narrow, Breast Cancer Res. 2 (2000) 82–85.

[19] J.J. Chen, D. Silver, S. Cantor, D.M. Livingston, R. Scully,BRCA1, BRCA2, and Rad51 operate in a common DNAdamage response pathway, Cancer Res. 59 (1999)1752s–1756s.

[20] K. Yoshida, Y. Miki, Role of BRCA1 and BRCA2 as regulators ofDNA repair, transcription, and cell cycle in response to DNAdamage, Cancer Sci. 95 (2004) 866–871.

[21] J.S. Larson, J.L. Tonkinson, M.T. Lai, A BRCA1 mutant altersG2-M cell cycle control in human mammary epithelial cells,Cancer Res. 57 (1997) 3351–3355.

[22] Y. Yan, R.S. Spieker, M. Kim, S.M. Stoeger, K.H. Cowan,BRCA1-mediated G2/M cell cycle arrest requires ERK1/2kinase activation, Oncogene 24 (2005) 3285–3296.

[23] N.S. Shao, Y.L. Chai, E. Shyam, P. Reddy, N.V. Rao, Induction ofapoptosis by the tumor suppressor protein BRCA1Oncogene 13 (1996) 1–7.

[24] D.C. Connolly, R. Bao, A.Y. Nikitin, K.C. Stephens, T.W. Poole,X. Hua, S.S. Harris, B.C. Vanderhyden, T.C. Hamilton, Femalemice chimeric for expression of the simian virus 40 TAg undercontrol of the MISIIR promoter develop epithelial ovariancancer, Cancer Res. 63 (2003) 1389–1397.

[25] A. Flesken-Nikitin, K.C. Choi, J.P. Eng, E.N. Shmidt,A.Y. Nikitin, Induction of carcinogenesis by concurrentinactivation of p53 and Rb1 in the mouse ovarian surfaceepithelium, Cancer Res. 63 (2003) 3459–3463.

[26] S. Orsulic, Y. Li, R.A. Soslow, L.A. Vitale-Cross, J.S. Gutkind,H.E. Varmus, Induction of ovarian cancer by defined multiplegenetic changes in a mouse model system, Cancer Cell 1(2002) 53–62.

[27] D.M. Dinulescu, T.A. Ince, B.J. Quade, S.A. Shafer, D. Crowley,T. Jacks, Role of K-ras and Pten in the development of mousemodels of endometriosis and endometrioid ovarian cancer,Nat. Med. 11 (2005) 63–70.

[28] S.G. Brodie, X. Xu,W. Qiao, W.M. Li, L. Cao, C.X. Deng, Multiplegenetic changes are associated with mammarytumorigenesis in Brca1 conditional knockout mice, Oncogene20 (2001) 7514–7523.

[29] X. Xu, K.U. Wagner, D. Larson, Z. Weaver, C. Li, T. Ried,L. Hennighausen, A. Wynshaw-Boris, C.X. Deng, Conditionalmutation of Brca1 in mammary epithelial cells results inblunted ductal morphogenesis and tumour formation, Nat.Genet. 22 (1999) 37–43.

[30] R. Chodankar, S. Kwang, F. Sangiorgi, H. Hong, H.Y. Yen,C. Deng, M.C. Pike, C.F. Shuler, R. Maxson, L. Dubeau,Cell-nonautonomous induction of ovarian and uterine serouscystadenomas in mice lacking a functional Brca1 in ovariangranulosa cells, Curr. Biol. 15 (2005) 561–565.

[31] V.L. Cressman, D.C. Backlund, E.M. Hicks, L.C. Gowen,V. Godfrey, B.H. Koller, Mammary tumor formationin p53- and BRCA1-deficient mice, Cell Growth Differ. 10(1999) 1–10.

[32] L.C. Gowen, B.L. Johnson, A.M. Latour, K.K. Sulik, B.H. Koller,Brca1 deficiency results in early embryonic lethalitycharacterized by neuroepithelial abnormalities, Nat. Genet.12 (1996) 191–194.

[33] J. Jonkers, R. Meuwissen, G.H. van der, H. Peterse,D. van, V., A. Berns, Synergistic tumor suppressor activity ofBRCA2 and p53 in a conditional mouse model for breastcancer, Nat. Genet. 29 (2001) 418–425.

[34] M.K. Sgagias, K.U. Wagner, B. Hamik, S. Stoeger, R. Spieker,L.J. Huber, L.A. Chodosh, K.H. Cowan, Brca1-deficient murinemammary epithelial cells have increased sensitivityto CDDP and MMS, Cell Cycle 3 (2004)1451–1456.

[35] R. Hutson, J. Ramsdale, M. Wells, p53 protein expression in

putative precursor lesions of epithelial ovarian cancer,Histopathology 27 (1995) 367–371.

[36] S.L. Maines-Bandiera, N. Auersperg, Increased E-cadherinexpression in ovarian surface epithelium: an early step inmetaplasia and dysplasia? Int. J. Gynecol. Pathol. 16 (1997)250–255.

[37] C.D. Capo-chichi, E.R. Smith, D.H. Yang, I.H. Roland,L. Vanderveer, C. Cohen, T.C. Hamilton, A.K. Godwin,X.X. Xu, Dynamic alterations of the extracellular environmentof ovarian surface epithelial cells in premalignanttransformation, tumorigenicity, and metastasis, Cancer 95(2002) 1802–1815.

[38] O.L. Tan, P.R. Hurst, J.S. Fleming, Location of inclusion cysts inmouse ovaries in relation to age, pregnancy, and totalovulation number: implications for ovarian cancer? J. Pathol.205 (2005) 483–490.

[39] B.A. Werness, A.M. Afify, K.L. Bielat, G.H. Eltabbakh,M.S. Piver, J.M. Paterson, Altered surface and cyst epitheliumof ovaries removed prophylactically from women with afamily history of ovarian cancer, Hum. Pathol. 30 (1999)151–157.

[40] N. Auersperg, S.L. Maines-Bandiera, H.G. Dyck, Ovariancarcinogenesis and the biology of ovarian surface epithelium,J. Cell Physiol. 173 (1997) 261–265.

[41] J. Boyd, Y. Sonoda, M.G. Federici, F. Bogomolniy, E. Rhei,D.L. Maresco, P.E. Saigo, L.A. Almadrones, R.R. Barakat,C.L. Brown, D.S. Chi, J.P. Curtin, E.A. Poynor, W.J. Hoskins,Clinicopathologic features of BRCA-linked and sporadicovarian cancer, JAMA 283 (2000) 2260–2265.

[42] R.R. Barakat, M.G. Federici, P.E. Saigo, M.E. Robson, K. Offit,J. Boyd, Absence of premalignant histologic, molecular,or cell biologic alterations in prophylactic oophorectomyspecimens from BRCA1 heterozygotes, Cancer 89 (2000)383–390.

[43] R. Hakem, J.L. de la Pompa, C. Sirard, R. Mo,M.Woo, A. Hakem,A. Wakeham, J. Potter, A. Reitmair, F. Billia, E. Firpo, C.C. Hui,J. Roberts, J. Rossant, T.W. Mak, The tumor suppressor geneBrca1 is required for embryonic cellular proliferation in themouse, Cell 85 (1996) 1009–1023.

[44] L. Cao,W. Li, S. Kim, S.G. Brodie, C.X. Deng, Senescence, aging,and malignant transformation mediated by p53 in micelacking the Brca1 full-length isoform, Genes Dev. 17 (2003)201–213.

[45] P.A. Kruk, A.K. Godwin, T.C. Hamilton, N. Auersperg,Telomeric instability and reduced proliferative potential inovarian surface epithelial cells from women with afamily history of ovarian cancer, Gynecol. Oncol. 73 (1999)229–236.

[46] V.L. Cressman, D.C. Backlund, E.M. Hicks, L.C. Gowen,V. Godfrey, B.H. Koller, Mammary tumor formation inp53- and BRCA1-deficient mice, Cell Growth Differ. 10 (1999)1–10.

[47] C.X. Deng, Tumorigenesis as a consequence of geneticinstability in Brca1 mutant mice, Mutat. Res. 477 (2001)183–189.

[48] X. Xu, W. Qiao, S.P. Linke, L. Cao, W.M. Li, P.A. Furth,C.C. Harris, C.X. Deng, Genetic interactionsbetween tumor suppressors Brca1 and p53 in apoptosis, cellcycle and tumorigenesis, Nat. Genet. 28 (2001)266–271.

[49] S.R. Lakhani, S. Manek, F. Penault-Llorca, A. Flanagan,L. Arnout, S. Merrett, L. McGuffog, D. Steele, P. Devilee,J.G. Klijn, H. Meijers-Heijboer, P. Radice, S. Pilotti,H. Nevanlinna, R. Butzow, H. Sobol, J. Jacquemier, D.S. Lyonet,S.L. Neuhausen, B.Weber, T.Wagner, R.Winqvist, Y.J. Bignon,F. Monti, F. Schmitt, G. Lenoir, S. Seitz, U. Hamman,P. Pharoah, G. Lane, B. Ponder, D.T. Bishop, D.F. Easton,Pathology of ovarian cancers in BRCA1 and BRCA2 carriers,Clin. Cancer Res. 10 (2004) 2473–2481.

145E X P E R I M E N T A L C E L L R E S E A R C H 3 1 3 ( 2 0 0 7 ) 1 3 3 – 1 4 5

[50] S.J. Ramus, L.G. Bobrow, P.D. Pharoah, D.S. Finnigan,A. Fishman, M. Altaras, P.A. Harrington, S.A. Gayther,B.A. Ponder, L.S. Friedman, Increased frequency of TP53mutations in BRCA1 and BRCA2 ovarian tumours, GenesChromosomes Cancer 25 (1999) 91–96.

[51] I. Cass, R.L. Baldwin, T. Varkey, R. Moslehi, S.A. Narod, B.Y.Karlan, Improved survival in women with BRCA-associatedovarian carcinoma, Cancer 97 (2003) 2187–2195.

[52] A. Fedier, R.A. Steiner, V.A. Schwarz, L. Lenherr, U. Haller,D. Fink, The effect of loss of Brca1 on the sensitivity toanticancer agents in p53-deficient cells, Int. J. Oncol. 22 (2003)1169–1173.

[53] R. Hakem, J.L. de la Pompa, A. Elia, J. Potter, T.W. Mak,Partial rescue of Brca1 (5–6) early embryonic lethality byp53 or p21 null mutation, Nat. Genet. 16 (1997)298–302.