Oncogene (2004) 23, 77227725

& 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00
www.nature.com/onc

SHORT REPORTS

Tumor formation in p53 mutant ovaries transplanted into wild-type female hosts
Chun-Ming Chen1,3, Junn-Liang Chang2 and Richard R Behringer*,1
1 Department of Molecular Genetics, University of Texas, MD Anderson Cancer Center, Houston, TX, USA; 2Department of Pathology, Armed Forced Taoyuan General Hospital, Taoyuan County, Taiwan

p53 gene alterations correlate highly with advanced ovarian carcinoma in women. In mice, p53 deciency predominantly results in the formation of lymphomas and sarcomas. However, ovarian epithelial tumors have not been documented in p53 homozygous mutant (p53/) mice, probably because they die before other tumors can form. To determine whether p53/ ovaries can develop epithelial tumors, they were transplanted into the ovarian bursae of histocompatible wild-type recipient females. The p53/ ovarian grafts formed tumors B1 year posttransplantation. The tumor type was angiosarcoma, suggesting that vascular tissues are predisposed to tumor formation in p53/ovaries. These ndings suggest that p53 deciency alone is not sufcient for ovarian epithelial tumorigenesis in mice. Thus, other genetic lesions are likely required to develop mouse models of human ovarian cancer. Oncogene (2004) 23, 77227725. doi:10.1038/sj.onc.1208037 Published online 30 August 2004 Keywords: ovarian cancer; tumor suppressor; p53; ovary transplantation; hemangiosarcoma

Ovarian cancer is the second most prevalent cause of cancer-related death and the fth leading cancer among American women (Jemal et al., 2003). Approximately 90% of human ovarian cancer is derived from the ovarian epithelium (Auersperg et al., 2001). Genetic alterations of oncogenes and tumor suppressor genes correlate with ovarian cancer formation and progression (Bast and Mills, 2000). Among these genetic alterations, p53 mutations are the most common in advanced ovarian cancer (for review, see Shelling et al., 1995; Schuijer and Berns, 2003). In contrast, p53 gene alterations are infrequent in benign ovarian tumors or the early stages of ovarian cancer. These ndings suggest that p53 mutations play important roles in human ovarian cancer progression.
*Correspondence: RR Behringer, Department of Molecular Genetics, University of Texas, MD Anderson Cancer Center, 1515 Holcombe Blvd, Houston, TX 77030, USA; E-mail: rrb@notes.mdacc.tmc.edu 3 Current address: Faculty of Life Sciences, National Yang-Ming University, Taipei 112, Taiwan Received 13 January 2004; revised 12 May 2004; accepted 20 May 2004; published online 30 August 2004

Several strains of p53 knockout mice have been created by gene targeting in mouse embryonic stem cells (Donehower et al., 1992; Jacks et al., 1994; Purdie et al., 1994). p53/ mice spontaneously develop a spectrum of tumors that depends in part upon genetic background (Venkatachalam et al., 2002). The tumor latency for p53/ mice is generally B68 months of age. Lymphomas and sarcomas are the primary tumor types on C57Bl/6 129 (B6/129) mixed and C57Bl/6 inbred genetic backgrounds (Venkatachalam et al., 2002). Carcinoma incidence in p53/ mice is relatively rare. The incidence of ovarian tumors in p53/ mice has not been reported, most likely because other types of cancer cause lethality before these tumors might arise. Given sufcient time, it seems reasonable to imagine that p53 mutant ovaries should be able to form tumors. Therefore, we used the classic technique of ovary transplantation into histocompatible wild-type female hosts to avoid the lethal tumor-burden that develops in p53/ mice. The ovaries of 3- to 5-week-old p53/ mice on a C57Bl/6J (B6) congenic background (Jackson Laboratory, Bar Harbor, ME, USA) were removed and onehalf of each ovary was transplanted into the ovarian bursa of ovariectomized age-matched histocompatible C57Bl/6J (B6) inbred or B6/129F1 hybrid mice (Figure 1a). p53 / ovaries were transplanted, which served as controls. Subsequently, ovarian tumor formation was monitored by palpating the abdomen of the recipient females twice per week. However, at 11 months post-transplantation, we were only able to identify one case of tumor formation by abdominal palpation. Therefore, we examined p53/ ovarian grafts at 78 and at 1113 months post-transplantation (Table 1). p53/ ovarian grafts were grossly normal in appearance up to 78 months post-transplantation, consistent with the lack of spontaneous ovarian tumor formation in p53/ mice (Figure 1c). However, by 1113 months post-transplantation, all successfully transplanted p53/ ovaries (ve of six transplants) formed tumor masses (Figure 1e and f). The ovarian tumors were hemorrhagic (two of ve), pale (one of ve), or cystic (two of ve) in appearance (Figure 1e and f and not shown). Histologically, the hemorrhagic ovarian tumors exhibited irregular vascular channels lined by pleiomorphic neoplastic endothelial cells (Figure 2b). Focal stromal invasion, cellular atypia with bizarre nuclei, and highly

Tumor formation in p53 mutant ovaries C-M Chen et al

7723

Figure 1 Tumor formation in p53/ ovaries. (a) Ovary transplantation scheme. (b, c) Reproductive tracts of recipient females with p53 / (b) and p53/ (c) ovary grafts at 8 months posttransplantation. (df) Tracts of recipient females with p53 / (d) and p53/ (e, f) ovary grafts at 13 months post-transplantation. p53/ovarian tumors can be identied grossly showing cystic (blue arrows), pale (black arrow), and hemorrhagic (red arrow) appearances. Bar, 1 cm. Methods: B6 p53 / or p53/ ovaries were transplanted into the ovarian bursae of B6 or B6129F1 histocompatible recipient females that had had their endogenous ovaries removed as described (Nagy et al., 2003)

Figure 2 Histological and immunohistochemical analysis of p53/ ovarian tumors. (a) Histological section of control ovary. H&E staining. (b, c) Histological sections of p53/ ovarian tumors with hemorrhagic (b) or pale (c) appearances. (d) CD34 immunostaining of control ovary. The arrows indicate endothelial cells (upper left) and a subset of cells surrounding ovarian follicles (right) showing CD34-positive immunostaining. (e, f) CD34 immunostaining of p53/ ovarian tumors. The sections shown in (e) and (f) were adjacent to those shown in (b) and (c), respectively. (g, i) Histological sections of p53/ ovarian tumors with cystic appearance. H&E staining. (h, j) CD34 immunostaining of p53/ ovarian tumors. The sections shown in (h) and (j) were adjacent to those shown in (g) and (i), respectively. The arrow represents the area of pleiomorphic hyperplastic cells showing CD34-positive immunostaining. Bar, 100 mm. Methods: Sections (4 mm) were cut from formalin-xed, parafn-embedded tissues. H&E staining and immunohistochemistry were performed as described (Chen and Behringer, 2004). Anti-CD34 antibody was used (1 : 50; Cedarlane, Ontario, CA, USA)

Table 1

Summary of p53 heterozygous and homozygous ovarian grafts No. of graftsa Tumors/recovered graftsb 78 (Mo) 1113 (Mo) 0/1 5/5 5/8 10/12 Recovered grafts/ total grafts

p53 genotype

+/ /

8 12

0/4 0/5

a Ovaries were bisected and one-half of each ovary was transferred into a single ovarian bursa of a wild-type histocompatible (B6 or B6/129F1) mouse whose endogenous ovaries were removed. bOvarian grafts were examined at 78 or 1113 months post-transplantation. Mo, months

frequent mitotic gures were observed (Figure 2b). The p53/ ovarian grafts that had a grossly pale appearance (Figure 1e) had a solid growing pattern with hypercellularity and anaplasia microscopically (Figure 2c). The neoplastic cells showed oval, spindle, or irregular shape supported by brovascular stroma. Bizarre nuclei, hyperchromatin, and highly frequent mitotic gures

were observed, indicating the malignant potential of these tumors (Figure 2c). Normal ovarian histology (Figure 2a) was replaced by neoplastic cells (Figure 2b and c). Histological analysis of the malignant tumors indicated angiosarcoma (Turusov and Mohr, 1994; Mohr, 2001). Therefore, CD34 immunohistochemistry was performed to conrm the endothelial identity of vessels in the p53/ ovarian tumors. The p53/ ovarian tumors showed moderate to strong CD34 immunostaining (Figure 2e and f). Interestingly, the two cystic appearing p53/ ovarian grafts had pleiomorphic hyperplastic cells showing CD34-positive immunostaining, suggesting that they were premalignant (Figure 2gj). All of p53/ ovarian grafts developed CD34-positive neoplasms by 13 months in this study. p53 / and p53/ ovarian grafts at 78 months showed CD34 immunostaining specically localized to endothelial cells of vessels and a subset of cells surrounding ovarian follicles (Figure 2d and data not shown). Thus, there is a transition between these two time points when p53/ endothelial cells progress to malignancy. We also performed immunohistochemistry for asmooth muscle actin (a-SMA), cytokeratin 8 (CK8),
Oncogene

Tumor formation in p53 mutant ovaries C-M Chen et al

7724

and anti-Mu llerian hormone (AMH), markers that are expressed in different subsets of ovarian tissues. a-SMA is expressed in vascular smooth muscle cells and myobroblasts of external theca layer of ovarian follicle (Figure 3a). Vascular smooth muscle cells in the tumors showed positive immunostaining for a-SMA (Figure 3b). The majority of the neoplastic cells in the p53/ ovarian tumors were a-SMA negative or had scattered positive staining cells perhaps due to tissue repair or the stromal response to neoplasia (Figure 3b and c). CK8 is expressed specically in the ovarian surface epithelium (Figure 3d). We found that p53/ ovarian tumors were lined by a single cell layer (Figure 3e) or focal invaginations of the ovarian surface epithelium that were CK8 positive (Figure 3f). However, the neoplastic cells were CK8 negative. We also examined cell proliferation using Ki67 immunohistochemistry to examine whether the focal invaginations of the ovarian surface epithelium had malignant potential. We only found very few Ki67-positive cells in the focal invaginated area (Figure 3f, inset) similar to the normal appearing ovarian surface epithelium (data not shown). The most prominent Ki67-positive cells were neoplastic cells with giant or bizarre nuclei. Approximately, 50 70% of the neoplastic cells were Ki67 positive in

Figure 3 Immunohistochemical analysis of p53/ ovarian tumors. (a) a-SMA expression in control ovary. Vascular smooth muscle cells and myobroblasts of external thecal layer of ovarian follicle express a-SMA (arrows). (b, c) a-SMA expression in p53/ ovarian tumors. Vascular smooth muscle showed a-SMA-positive staining (arrow). The majority of neoplastic cells were a-SMA negative. (d) CK8 expression in ovarian surface epithelium (arrow). (e, f) CK8 expression in p53/ ovarian tumors. Normal appearance (e) or focal invasion (f) of ovarian surface epithelium positive for CK8 immunostaining (arrow). The adjacent slide immunostained for Ki67; nuclei are brown (inset). Most of the epithelial cells showing a glandular arrangement were Ki67 negative. (g) AMH expression in control ovarian follicles. (h, i) AMH expression in p53/ ovarian tumors. Only remaining granulosa cells were AMH positive. Bar, 100 mm. Methods: Immunohistochemistry was performed using the following antibodies: a-SMA (1 : 800; Clone 1A4, Sigma, Saint Louis, MI, USA); CK8 (1 : 100; TROMA-1, NICHD supported Developmental Studies Hybridoma Bank, University of Iowa); AMH (1 : 100; MIS C-20; Santa Cruz Biotechnology Inc.); and Ki67 (1 : 1000; NCL-Ki67p; Novacastra Laboratories, UK)
Oncogene

comparison to 510% of the ovarian stromal cells that were Ki67 positive except for granulosa cells that were well labeled by Ki67 (data not shown). Moreover, AMH is expressed in granulosa cells of developing follicles (Figure 3g). The remaining granulosa cells of the ovarian tumors were AMH positive, but the neoplastic cells were AMH negative (Figure 3h and i). Our studies show that B6 p53/ ovarian grafts consistently develop angiosarcoma after B1 year posttransplantation rather than epithelial malignancy. Interestingly, p53 gene mutations are associated with human angiosarcoma (Naka et al., 1997; Zietz et al., 1998). Thus, our ndings provide direct evidence for p53 deciency in the development of angiosarcoma. Ovarian angiosarcomas are extremely rare in humans. Histopatholgical characteristics of ovarian angiosarcomas are similar to angiosarcomas arising from other soft tissues although the histological appearances may vary. These tumors consist of irregular vascular channels lined by anaplastic cells with pleiomorphic nuclei with a high frequency of mitotic gures (Nucci et al., 1998, Platt et al., 1999). Factor VIII, CD34, or CD31 expression in the neoplastic cells identies them as vascular tissue (Nucci et al., 1998; Jylling et al., 1999). Human ovarian angiosarcomas are highly aggressive with poor prognosis. There are numerous transgenic and knockout mice that die at birth or before ovarian tumor formation, hampering the generation of mouse models for ovarian carcinoma. Homotopic or heterotopic transplantation of ovaries from neonates or older females into histocompatible recipients provides a way to circumvent premature mutant lethality (Chen and Behringer, 2004). The ovary transplantation approach used in this study provides fundamental information regarding the latency and tumor spectrum of p53/ ovarian tumors. In agreement with recent reports (Orsulic et al., 2002; Flesken-Nikitin et al., 2003), p53 deciency alone does not seem to be sufcient for ovarian epithelial malignancy. However, p53 deciency with activated Akt, cMyc, and K-ras pathways (Orsulic et al., 2002) or an inactivated retinoblastoma pathway (Flesken-Nikitin et al., 2003) leads to the transformation of the ovarian epithelium in mice. Our studies were performed on a B6 congenic background. It is possible that p53 deciency on a different genetic background may lead to the formation of ovarian epithelial tumors. Interestingly, p53 heterozygous and homozygous mutant mice examined on a BALB/c genetic background develop mammary carcinoma (Kuperwasser et al., 2000; Blackburn et al., 2003). Ovarian cancer incidence in p53-BALB/c mice has not been reported. Taken together, our ndings are consistent with the idea that multiple genetic lesions and genetic modiers coordinate with loss of p53 in the tumorigenesis of the ovarian surface epithelium.
Acknowledgements This work was supported by a Development Grant from the National Cancer Institute Specialized Programs of Research Excellence in Ovarian Cancer (CA83639) and grants from the

Tumor formation in p53 mutant ovaries C-M Chen et al

7725 National Institutes of Health (NIH) (HD30284) and Department of Defense Ovarian Cancer Research Program (OC020187) to RRB. Veterinary resources were supported by the NIH Cancer Center Support (Core) Grant CA16672.

References Auersperg N, Wong AS, Choi KC, Kang SK and Leung PC. (2001). Endocr. Rev., 22, 255288. Bast Jr RC and Mills GB. (2000). Ovarian Cancer: Methods and Protocols Bartlett JMS (ed). Humana Press: New Jersey, pp. 3748. Blackburn AC, Brown JS, Naber SP, Otis CN, Wood JT and Jerry DJ. (2003). Cancer Res., 63, 23642368. Chen CM and Behringer RR. (2004). Genes Dev., 18, 320332. Donehower LA, Harvey M, Slagle BL, McArthur MJ, Montgomery Jr CA, Butel JS and Bradley A. (1992). Nature, 356, 215221. Flesken-Nikitin A, Choi KC, Eng JP, Shmidt EN and Nikitin AY. (2003). Cancer Res., 63, 34593463. Jacks T, Remington L, Williams BO, Schmitt EM, Halachmi S, Bronson RT and Weinberg RA. (1994). Curr. Biol., 4, 17. Jemal A, Murray T, Samuels A, Ghafoor A, Ward E and Thun MJ. (2003). CA Cancer J. Clin., 53, 526. Jylling AM, Jorgensen L and Holund B. (1999). Pathol. Oncol. Res., 5, 318319. Kuperwasser C, Hurlbut GD, Kittrell FS, Dickinson ES, Laucirica R, Medina D, Naber SP and Jerry DJ. (2000). Am. J. Pathol., 157, 21512159. Mohr U (ed). (2001). International Classication of Rodent Tumors: The Mouse. Springer-Verlag: Berlin. Nagy A, Gertsenstein M, Vintersten K and Behringer RR (eds). (2003). Manipulating the Mouse Embryo: A Laboratory Manual. Cold Spring Harbor Laboratory Press: New York. Naka N, Tomita Y, Nakanishi H, Araki N, Hongyo T, Ochi T and Aozasa K. (1997). Int. J. Cancer, 71, 952955. Nucci MR, Krausz T, Lifschitz-Mercer B, Chan JK and Fletcher CD. (1998). Am. J. Surg. Pathol., 22, 620630. Orsulic S, Li Y, Soslow RA, Vitale-Cross LA, Gutkind JS and Varmus HE. (2002). Cancer Cell, 1, 5362. Platt JS, Rogers SJ, Flynn EA and Taylor RR. (1999). Gynecol. Oncol., 73, 443446. Purdie CA, Harrison DJ, Peter A, Dobbie L, White S, Howie SE, Salter DM, Bird CC, Wyllie AH, Hooper ML and Clarke AR. (1994). Oncogene, 9, 603609. Schuijer M and Berns EMJJ. (2003). Hum. Mutat., 21, 285291. Shelling AN, Cooke IE and Ganesan TS. (1995). Br. J. Cancer, 72, 521527. Turusov VS and Mohr U (eds). (1994). Pathology of Tumours in Laboratory Animals. Vol II Tumours of the Mouse 2nd edn, IARC Scientic Publications no. 111. International Agency for Research on Cancer: Lyon, France. Venkatachalam S, Tyner S and Donehower LA. (2002). Tumor Models in Cancer Research Teicher BA (ed). Humana Press: New Jersey, pp. 247261. Zietz C, Ro ssle M, Haas C, Sendelhofert A, Hirschmann A, Stu hrs U. (1998). Am. J. Pathol., 153, rzl M and Lo 14251433.

Oncogene