Tumor Repressor Protein 53 and Steroid Hormones Provide a New Paradigm for Ovarian Cancer Metastases

Abstract

The functional status of the tumor repressor protein (TP53 or TRP53) is a defining feature of ovarian cancer. Mutant or null alleles of TP53 are expressed in greater than 90% of all high-grade serous adenocarcinomas. Wild-type TP53 is elevated in low-grade serous adenocarcinomas in women and in our Pten;Kras;Amhr2-Cre mutant mouse model. Disruption of the Trp53 gene in this mouse model did not lead to high-grade ovarian cancer but did increase expression of estrogen receptor α (ESR1) and markedly enhanced the responsiveness of these cells to estrogen. Specifically, when Trp53-positive and Trp53 null mutant mice were treated with estradiol or vehicle, only the Trp53 null and Esr1-positive tumors respond vigorously to estradiol in vivo and exhibit features characteristic of high-grade type ovarian cancer: invasive growth into the ovarian stroma, rampant metastases to the peritoneal cavity, and nuclear atypia. Estrogen promoted and progesterone suppressed the growth of Trp53 null ovarian tumors and tumor cells injected ip, sc, or when grown in matrigel. Exposure of the Trp53 depleted cells to estrogen also has a profound impact on the tumor microenvironment and immune-related events. These results led to the new paradigm that TRP53 status is related to the susceptibility of transformed ovarian surface epithelial cells to estradiol-induced metastases and nuclear atypia via increased levels of estradiol receptor α.

Ovarian cancer is a complex disease that is the fifth most lethal cause of death in women. With more than 14 000 deaths annually in the United States, ovarian cancer is an ominous threat to women's health (1, 2). It is lethal because it is most often detected at a late stage when surgery and current chemotherapies are ineffective in preventing recurrence. This disease is complex because of the diverse cellular origins and histopathologies that are observed in human ovarian cancers (2, 3). However, it is becoming increasingly evident that understanding the functions of wild-type and mutant forms of the tumor repressor protein TP53 (TRP53 or P53) and how they interact with steroid hormones is of particular relevance and paramount importance to the etiology of this disease.

Overwhelming evidence documents a strong link between ovarian cancer subtypes and the status (wild-type, mutant, or null alleles) of TP53. Wild-type TP53 is expressed in low-grade serous adenocarcinomas in mice and women (3), whereas mutant or null alleles of TP53 are present in greater than 90% of all high-grade serous adenocarcinomas, the most prevalent ovarian cancer subtype in women (4). The absence of functional TP53 and genetic instability are the major defining features of high-grade type ovarian cancers (4). Unfortunately, we do not yet understand how TP53 status relates to the etiology or progression of this complex disease in women.

The RAS and PI3K/AKT pathways are mutated, amplified, or deleted in many human cancers including ovarian cancer (4). Although KRAS is not frequently mutated in high-grade human ovarian carcinomas, PTEN/PI3K/AKT and RAS pathways are altered in 45% of high-grade serous ovarian tumors (4); RAS is amplified in 11% of tumors (5), and KRAS mRNA and protein are often overexpressed (6). Thus, the combined activation of PI3K and KRAS signaling pathways likely contributes to ovarian cancer tumorigenesis. We recently generated a Pten/Kras/Amhr2Cre mutant mouse model in which ovarian surface epithelial (OSE) cell transformation is 100% penetrant and occurs at an early age. Furthermore, we discovered that, in the absence of DNA damage, oncogene-induced wild-type TRP53 promotes OSE tumor cell survival in this model of serous ovarian cancer (7). In this context, TRP53 up-regulates genes promoting DNA repair and cell cycle progression (8), the ancestral functions of TRP53 (9). Because TP53 is mutated in the vast majority of human high-grade ovarian cancer, we deleted Trp53 in our mouse model. Rather than displaying more aggressive tumor growth, as would be expected from evidence in high-grade tumors, the Trp53 null mutant OSE cells exhibited a far less aggressive phenotype with small lesions in vivo (7). Quite remarkably, however, the Trp53 depleted cells expressed elevated levels of estradiol receptor α (Esr1) mRNA (7). A unique feature of our mouse model is that ovarian follicular development is severely impaired due to the suppressive effects of KRAS^G12D on granulosa cell proliferation and differentiation (10). As a consequence, ovarian production of steroid hormones is minimal. The absence of ovarian steroids in these mice has allowed us to selectively restore estradiol and progesterone in vivo, as well as in culture, to unveil the potent impact of these steroids on ovarian cancer tumor growth in situ.

Specifically, we show in these studies that estradiol does not augment growth of primary tumors of the Pten/Kras mice that express wild-type TRP53 and low levels of Esr1. However, when the ovarian cancer cells that lack Trp53 and express high levels of estrogen receptor α (ERα) are exposed to estradiol in vivo, the primary tumors exhibit features characteristic of high-grade type ovarian cancer: enhanced, invasive growth into the ovarian stroma, rampant metastases to the peritoneal cavity, and nuclear atypia. Moreover, progesterone can suppress the effects of estradiol. Thus, cancer cells lacking functional Trp53 may be highly susceptible to unopposed estradiol and oncogenic insults at specific stages of tumor development.

The dramatic effects of estradiol and progesterone on tumor growth in the Pten/Kras (Trp53−) mice are highly relevant because increasing evidence indicates that ovarian steroids can impact the incidence and progression of ovarian cancer in women. Specifically, women who take oral contraceptives early in life exhibit a 50% reduction of ovarian cancer later in life. Conversely, postmenopausal women on hormone replacement therapies (11–13) exhibit increased incidence of ovarian cancer. Moreover, the status of TP53 in human ovarian cancer cells may regulate the response of cells to ovarian steroids.

Materials and Methods

Animal procedures

LSL-Kras^G12D;Pten^fl/fl;Amhr2-Cre, mice were derived and genotyped as previously described (10, 14). Trp53^fl/fl mice were obtained from the Mouse Models of Human Cancer Consortium, NCI-Frederick. Trp53^fl/fl;LSL-Kras^G12D;Pten^fl/fl;Amhr2-Cre, mice were derived by crossing the different genotypes as described (7). NOD.CB17.Prkdc^scid/J mice were purchased from The Jackson Laboratory. Animals were housed under a 16-hour light/8-hour dark schedule in the Center for Comparative Medicine at Baylor College of Medicine and provided food and water ad libitum. Animals were treated in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals, as approved by the Animal Care and Use Committee at the Baylor College of Medicine.

Histology and immunohistochemistry

Ovaries and tumors were collected and fixed in 10% formalin, embedded in paraffin, and processed by routine procedures for immunohistochemistry of cytokeratin 8 TROMA-I (Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa, Department of Biology, Iowa City, Iowa) and ERα (SC-542; Santa Cruz Biotechnology). Both ovaries from six mice of each genotype were analyzed for histology. For hormone treatments the recipient mice were treated with one-half of a 0.36 mg estradiol pellet (Innovative Research of America). Tumors were harvested 21 days later and fixed in 10% formalin. To assess tumor growth in each genotype, the extension/length of the papillary structures from the perimeter of the ovary to the tip of the tumor (Figure 1) was measured in ×10 images of five regions per ovary in a minimum of four ovaries per genotype. To assess the number of CK8 and ERα costained cells present in each genotype, four regions per ovary per genotype for a minimum of three ovaries were analyzed.

Figure 1.

Response of Pten/Kras(Trp53+) and Pten/Kras(Trp53−) mutant OSE cells to estradiol. (A) Primary serous adenocarcinomas that exhibit a papillary morphology develop with 100% penetrance and at an early age in the Pten;Kras(Trp53+) mice (upper left panels) (8). The Trp53+ mutant OSE cells are uniformly immunopositive for cytokeratin 8 (CK8) but only a limited number of cells appear to be immunopositive for ERα. By contrast, primary CK8-positive lesions that develop in the Pten;Kras(Trp53−) mice are small and exhibit minimal papillary growth as previously reported (upper right panels) (8). The yellow line indicates the border between the papillary structures and the ovary. (B) When estradiol pellets are inserted into the Pten;Kras(Trp53+) mice at 5 weeks of age, no obvious changes in tumor growth, papillary morphology, or immunostaining of CK8 are observed at 8 weeks (21 d) but immunostaining for ERα is evident in some tumor cells (lower left panels). By contrast, the Trp53− cells not only exhibit CK8 staining but also have uniformly intense immunostaining for ERα (lower right panels). Moreover, when estradiol pellets are inserted into the Pten;Kras(Trp53−) mice at 5 weeks of age, the tumors show remarkable growth, extensive invasion of into the ovarian stroma, and intense ERα staining (lower right panels). (C) To verify further the immunostaining patterns for ERα in the Trp53+ and Trp53− cells, immunofluorescent analyses were done. In the untreated Pten/Kras(Trp53+) tumor cells, ERα is negligible but is evident in some CK8-positive cells in mice exposed to estradiol (left panels). By contrast, untreated and estradiol-treated CK8-positive tumor cells in the Pten/Kras(Trp53−) mice exhibit intense immunostaining for ERα. N = 3 wild-type, 10 Trp53+, and 6 Trp53− mice.

Isolation and culture of primary OSE cells

OSE cells were isolated and cultured as previously described (8).

Inoculation of cells into syngenic mice

Cells (1 × 10⁶) of mutant Trp53+ and Tp53− OSE were injected either ip or sc (along with Matrigel [BD Biosciences] according to the manufacturer's protocol) into at least 10 wild-type mice with the same genetic background for each condition with the exception of the estradiol and progesterone treatment, which was performed on three mice. The recipient mice were either untreated or pretreated for 24 hours, before inoculation of cells, with one-half of a 0.36-mg estradiol pellet and/or a 25-mg progesterone pellet (Innovative Research of America). Tumors were harvested 14 days later and fixed in 10% formalin.

Real-time RT-PCR for mRNAs

Total RNA was isolated using the RNeasy Mini Kit (QIAGEN) and treated with deoxyribonuclease I (DNA-free; Ambion) according to the manufacturer's instructions. cDNA was synthesized with M-MLV Reverse Transcriptase (Invitrogen). Real-time PCR was performed using the SYBR Green JumpStart TaqReadyMix (Sigma Aldrich) on the Rotor-Gene 6000 thermocycler (Qiagen Sciences). Primers (Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org) were used at a concentration of 0.2 μM. Data were normalized to RPL19 using the comparative threshold cycle method. Data were normalized to L19 using the comparative threshold cycle method. Data are presented as the mean ± SEM of at least three experiments performed in triplicate. Differences among groups were analyzed by Student's t test. *P < .05 was considered statistically significant.

Protein expression analysis

Somatic mutations and protein expression data of high-grade serous ovarian carcinomas (HGS) were downloaded from The Cancer Genome Atlas data portal (https://tcga-data.nci.nih.gov/tcga/) on January 10, 2013. A total of 177 HGS with both proteins expression data by reverse phase protein array and TP53 mutation status were extracted for statistical analysis. Normalized protein expression level in the lower 20% was arbitrarily designated as negative expression. Student's t test was used to estimate the significance of difference in ER α protein expression between HGS with wild-type TP53 and HGS with mutant TP53.

Results

Estradiol promotes formation of a high-grade type ovarian cancer, papillary structures, and metastasis of Trp53− OSE tumor cells

The Pten^fl/fl;Kras^G12D;Amhr2-Cre (Pten/Kras) mutant mice that we have generated develop serous OSE cell adenocarcinomas at an early age and with 100% penetrance (10). This remarkable feature has allowed us to examine primary tumor growth in situ and gene expression patterns in a highly reproducible system.

Recent novel observations show that Pten/Kras mutant mouse OSE cells express elevated levels of wild-type tumor repressor protein 53 (TRP53 or P53) and many TRP53 target genes, including p21 (Cdkn1a), miR-34c, and DNA repair genes (7). These results indicated that wild-type TRP53 might be acting to prevent high-grade tumor formation typically associated with Trp53 mutations in women. Despite observations made in other cancer models, disruption of the Trp53 gene in the Pten/Kras mutant mouse strain (Pten^fl/fl;Kras^G12D;Trp53^fl/fl;Amhr2-Cre), designated Pten/Kras (Trp53−), does not generate more aggressive, metastatic tumors. Instead, they exhibit an unexpectedly and strikingly less severe and less invasive epithelial histopathology with small lesions. However, the Pten/Kras(Trp53−) OSE cells express higher levels of the estradiol receptor α (Esr1) mRNA (7). To extend these observations we have analyzed the cell specific expression of estradiol receptor α (ERα) protein in ovarian tumor cells in vivo and have analyzed the response of Trp53+ and Trp53− tumors to estradiol and progesterone in vivo and in matrigel cultures in vitro.

To determine the effects of estradiol on tumor growth in vivo, estradiol pellets were inserted into 3 wild-type, 10 Pten;Kras(Trp53+), and 6 Pten;Kras(Trp53−) mice at 5 weeks of age. All mice were euthanized at 8 weeks of age. OSE cells in the wild-type mice did not show major changes in morphology in response to estradiol (data not shown). Tumors in the Pten;Kras(Trp53+) also did not show obvious changes in morphology in response to estradiol (Figure 1, A and B, left panels). Immunohistochemical and immunofluorescence analyses show that ERα protein is low in tumor cells present in the CK8-positive papillary structures of all Pten/Kras(Trp53+) tumors in the absence of estradiol (Figure 1A, left panels; Figure 1C, upper left panel) but is evident at low levels in cells exposed to estradiol (Figure 1B, left panels; Figure 1C, lower left panel). By contrast, papillary structure formation and tumor cell invasion is markedly enhanced in all Pten;Kras(Trp53−) mice in the presence of estradiol (Figure 1, A–C, right panels). Specifically, the small lesions that are observed in the untreated Pten;Kras(Trp53−) mice (Figure 1A) develop into papillary and invasive structures when the mice are treated with estradiol (Figure 1, B and C). The average extension/growth of the papillary structures on Pten;Kras(Trp53+) ovaries did not change significantly in response to estradiol (8.7 ± 2.5 vs 9.6 ± 3.2 mm). However, the average extension/growth of papillary structures on Pten;Kras(Trp53−) ovaries was significantly increased from 0.8 ± 1.2 to 14.6 ± 4.4 mm after estradiol treatment (P < 4.8 × 10⁻¹³). Furthermore, unlike the Trp53+ cells where only 2 ± 3% of CK8 cells are positive for ERα in the absence of estradiol and 59 ± 1.6% in the presence of estradiol, significantly more of the Trp53 null CK8-positive tumor cells are also immunopositive for ERα in the absence (90 ± 2.6%) and presence of estradiol (92 ± 2.2%) (Figure 1, A–C, right panels). Importantly, the number and intensity of ERα-positive cells are greatest in all Trp53 null tumors that have also acquired an extensive and pronounced papillary and invasive morphology in response to estradiol (Figure 1, B and C, lower right panels).

To characterize further the distinct behaviors of the Trp53+ and Trp53− cells in response to estrogen, the mutant cells were plated in matrigel with or without estradiol (Figure 2A). As observed previously (7), in this three-dimensional, growth factor–rich culture environment, Pten/Kras (Trp53+) OSE cells grew as papillary-like structures compared with the Pten/Kras (Trp53−) OSE cells that grow as spheres (Figure 2A, upper left panel). Similar to the effects of estradiol in vivo (Figure 1, A and B), estradiol has no visible effect on the Pten/Kras (Trp53+) OSE papillary-like structures (Figure 2A, upper right panel). By contrast, the Pten/Kras (Trp53−) OSE cells grew as small clusters of cells in media alone but formed large colonies with papillary structures when grown in the presence of estradiol (Figure 2A, bottom panels), also as observed in vivo (Figure 1, A-C).

Figure 2.

Estradiol promotes metastasis of primary Pten/Kras (Trp53−) OSE cells. (A) To characterize further the distinct behaviors of the Trp53+ and Trp53− mutant OSE cells, they were isolated and plated in matrigel with or without estradiol for 10 days. In this three-dimensional, growth factor–rich culture environment, the Pten/Kras(Trp53+) OSE cells exhibited similar morphology in the absence (left panel) or presence (right panel) of estradiol. In contrast, the Pten/Kras(Trp53−) OSE cells grew as small clusters in media alone (left panel) but formed large colonies with papillary structures when grown in the presence of estradiol (right panel). (B) When Pten/Kras(Trp53+) mice were treated with estradiol, metastases to the peritoneal cavity were not evident at 15 days. In contrast, when Pten;Kras(Trp53−) mice were treated with estradiol pellets for 15 days, ascites fluid was present by 8 weeks of age and metastatic tumors were present on the diaphragm, liver, omentum, and bowel mesentery. (C) Metastatic papillary-like tumors that localize to the omentum (adjacent to the pancreas) in the estradiol-treated Pten/Kras(Trp53+) mice stain positive for CK8 and ERα. Data are representative of at least six different experiments with two to three mice per condition in each experiment.

Primary ovarian tumors in the estradiol-treated Pten;Kras(Trp53+) mice do not exhibit obvious metastases in response to estradiol, whereas the primary ovarian tumors in all estradiol-treated Pten;Kras(Trp53−) mice are highly metastatic (Figure 2B). Multiple metastases were observed throughout the peritoneal cavity (liver, pancreas, omentum, bowel mesentery, and diaphragm) in each estradiol-treated Pten;Kras(Trp53−) mouse and the peritoneal cavities of all mice were filled with ascites fluid (Figure 2B). Similar to the CK8 and ERα-positive cells in papillary extensions on the surface of the ovary, the metastatic tumor cells also express CK8 and ERα (Figure 2C). Primary tumors of the estradiol-treated Pten;Kras(Trp53+) display uniform cell morphology (Figure 3A, upper panels), whereas the tumor cells in the estradiol-treated Pten;Kras(Trp53−) mice have features diagnostic of high-grade carcinomas: a high-degree of nuclear atypia and pleiomorphism (Figure 3A, lower panels). Metastatic Trp53 null tumors on the diaphragm and omentum also exhibit morphological features diagnostic of high-grade serous ovarian cancer: nuclear pleiomorphism and atypia, multiple prominent nucleoli in the nuclei of many cells (Figure 3B). Thus, exposure of the Trp53 null tumor cells to estradiol exerts a specific, rapid, and profound impact on tumor cell growth, cell migration, and morphology.

Figure 3.

Effects of estradiol on tumor cell morphology. (A) Primary ovarian tumors of the Pten;Kras(Trp53−) mice (lower right panel) but not those in the Pten;Kras(Trp53+) mice (upper right panel) exposed to estradiol exhibit nuclear atypia and pleiomorphic nuclei, features that are diagnostic of a high-grade type of serous adenocarcinoma. (B) Metastatic tumors that are derived from primary ovarian tumors of the Pten;Kras(Trp53−) mice and localize to the diaphragm have morphological features diagnostic of high-grade serous ovarian cancer: vacuolated nuclei and nuclear pleiomorphism (arrow). Data are representative of three separate experiments.

Estradiol promotes and progesterone suppresses tumor growth and metastasis of transformed OSE cells

To determine if the effects of estrogen can be modified by progesterone, mice were treated with estrogen and progesterone. The increased formation of papillary extensions on the ovarian surface in estrogen-treated mice was suppressed by progesterone (Figure 4A), indicating that progesterone can block the tumor-promoting effects of estrogen in the Trp53 null cells. Immunofluorescent analyses of CK8 and ERα in the tumor cells of the estradiol- and progesterone-treated mice indicate that ERα levels are severely reduced. The apparent absence of ERα protein in OSE tumor cells from ovaries of mice pretreated (24 h) with estradiol and progesterone suggests that tumor suppression is mediated, in part, by decreased ERα protein (Figure 4B).

Figure 4.

Estradiol promotes and progesterone suppresses tumor growth and metastasis of transformed Pten;Kras(Trp53−) OSE cells in the ovary. (A) Rampant growth of papillary structures is induced by estradiol in ovaries from Pten/Kras(Trp53−) mice treated (upper panels). This response to estradiol is potently blocked by progesterone as shown in mice cotreated with estradiol and progesterone (lower panels). (B) ERα protein is highly expressed in CK8-positive tumor cells in Pten/Kras(Trp53−) mice treated with or without estradiol (Figure 1), but is absent in the small tumors of Pten/Kras(Trp53−) mice treated with estradiol and progesterone. ERα was absent in small tumors of ovaries from three different mice.

Mutant OSE cells from Pten;Kras(Trp53+) and Pten;Kras(Trp53−) mice have been purified and grown in culture as cell lines (7). To determine if the Pten;Kras(Trp53−) cells grow and respond to estradiol in vivo, they were injected either sc or ip into control, estradiol, or estradiol and progesterone-treated syngenic mice. When the Trp53+ tumor cells were injected ip, they attached to many sites and were highly invasive in either the absence or the presence of estradiol (7) (data not shown). These results contrast markedly from apparent absence of rampant metastatic tumor growth of the primary ovarian Trp53+ tumors (7) (Figure 2). By contrast, when the Trp53 null OSE cells were injected ip into untreated mice, tiny tumors were only observed in 1 of 10 animals (Figure 5A, left panel). In sharp contrast, when the Trp53 null OSE cells were injected ip in the presence of estradiol, they attached to many ip sites, grew rapidly, and developed ascites (Figure 5A, middle panel). The pronounced estradiol-induced tumor growth was potently suppressed in mice cotreated with progesterone and estradiol (Figure 5A, right panel). Similar to the metastatic tumors that developed spontaneously from the primary tumor, the tumors that grew from ip injected Trp53− cells expressed cytokeratin 8 and ERα and exhibited a high degree of atypia, and multiple nucleoli (Figure 5B).

Figure 5.

Estradiol promotes and progesterone suppresses tumor growth and metastasis of transformed OSE cells injected ip or sc. (A) When the Pten/Kras(Trp53−) OSE cells were injected ip into syngenic mice, only a few small tumors were observed in the absence of estradiol in 1 of 10 mice (left panel). However, in the presence of estradiol, multiple tumors were observed in all mice and the Trp53− OSE cells attached to the omentum and small bowel (middle panel). By contrast, tumors were not evident when the mice were treated with estradiol and progesterone (right panels). (B) The Trp53− tumors were CK8- and ERα-positive and exhibited a high degree of nuclear atypia as seen by hematoxylin and eosin staining. (C) When the Trp53− tumor cells were injected sc, estradiol significantly increased the size of the tumors within 20 days and this response was blocked by progesterone. Similar but less dramatic increases in growth were observed when Trp53+ tumor cells were injected sc. Data are representative of two to four experiments. *, P < .05.

Estradiol also significantly promoted the growth of sc tumors that developed after subdermal injection of the Trp53− cells and to a lesser degree the growth of Trp53+ cells into syngenic mice (Figure 5C). The estradiol-mediated increase in the size of the Trp53+ and Trp53− sc tumors was also suppressed by cotreatment with progesterone. Of note, the Trp53− cells grew in the sc sites (but not the ip sites), reinforcing the evidence that the tumor microenvironment, possibly including the matrigel used in the sc injections, impacts tumor growth and outcome. Similar responses to estradiol and progesterone are seen in sc xenografts of human ovarian cancer POE4 cells (17). Although sc tumor growth is one index of tumor cell responsiveness, it is important to point out that ovarian cancer cells do not metastasize to these sites; hence, interpretations of these data must be viewed in regard to these physiological limitations. Collectively, growth of tumor cells in ip and sc sites documents that the Trp53 null cells are highly responsive to estradiol and that progesterone can exert potent inhibitory effects on ovarian tumor growth in this context.

Estradiol regulates expression of genes involved in cell proliferation, metastasis, and immune regulation

To analyze the acute effects of hormones on gene expression patterns in pure OSE cells, the Pten;Kras(Trp53−) cells were grown in matrigel. As previously shown, Pten;Kras(Trp53−) OSE cells grow as small clusters in media alone but when cultured in the presence of estradiol they form large colonies (Figures 2A and 6A). Either tamoxifen or progesterone blocked these morphological effects of estradiol. RT-PCR was performed on RNA prepared from these OSE tumor cells grown in matrigel to analyze the expression of estradiol-regulated genes involved in cell cycle progression, metastasis, and immune regulation (Figure 6C). As shown, estradiol alone induced the expression of selected genes (Esr1, Pgr, Mmp9, Cxcl5, Ccl4, Il6, Birc5, Klf4, and Ifitm1). Interestingly, the induction of these genes was abolished when the cells were cotreated with tamoxifen or progesterone (Figure 6C). Thus, either tamoxifen or progesterone blocked the morphological effects of estradiol as well as the expression of estradiol-regulated genes involved in cell cycle progression, metastasis, and immune regulation (Figure 6, A and C). Estradiol did not have a similar effect on the morphology of colonies or gene expression in Pten;Kras(Trp53+) OSE cells grown in matrigel (Figures 2A and 6D). Thus, as in vivo, estradiol promotes rapid tumor growth of the Trp53− but not Trp53+ cells in matrigel. RT-PCR analysis indicates that Esr1 mRNA is present at low levels in Pten;Kras(Trp53+) OSE cells compared with Trp53− OSE cells and that in the Trp53− cells Esr1 mRNA expression increases in response to estradiol similar to increased protein expression in OSE tumor cells in estradiol-treated animals (Figures 1A and 6C). Recent results from The Cancer Genome Atlas database indicate that 93% of human high-grade serous carcinomas express mutant forms of TP53 (4) (Table 1). Moreover, 80.2% of these tumors concurrently express ERα protein (Table 1), suggesting that estradiol and ERα likely impact human ovarian cancer.

Figure 6.

Pten/Kras(Trp53+) and Pten/Kras(Trp53−) OSE cells grow in matrigel and respond to estradiol and progesterone. (A) Trp53− OSE cells were plated in matrigel with or without estradiol, tamoxifen, or progesterone for 24 hours. Pten/Kras(Trp53−) OSE cells grew as small clusters in media alone but formed large colonies when grown in the presence of estradiol. Either tamoxifen or progesterone blocked the effects of estradiol. (B) Quantification of colony size compared with the number of OSE colonies grown in matrigel. (C) RT-PCR analyses document that estradiol induces the expression of genes involved in cell cycle progression, metastases, and immune regulation in Trp53− OSE cells grown in matrigel. Progesterone and tamoxifen blocked the transcriptional effects of estradiol. (D) RT-PCR analyses document that estradiol differentially regulates similar genes in Trp53+ OSE cells grown in matrigel. Data are presented as the mean ± SEM of at least three experiments performed in triplicate. Differences were analyzed by Student's t test. *, P < .05.

Table 1.

Protein Expression of TP53 and ERα in High-Grade Serous Carcinoma Tumors With TP53 Mutation Variants (n = 177)

	No.	% of Total
ERα+
TP53 wild type	8	4.5%
TP53 mutation variant	134	75.7%
ERα−
TP53 wild type	5	2.8%
TP53 mutation variant	30	16.9%

Estradiol requires an intact immune system to promote tumor growth of Trp53− cells

Steroid hormones have well-documented effects on immune cell functions and tumor cell metastasis (18–20). To investigate the potential relevance of immune cell regulation in estrogen-induced ovarian tumor cell metastases, Pten;Kras(Trp53−) OSE cells were injected ip into immune deficient (lacking mature T cells, B cells, and complement) Non-Obese Diabetic scid mice pretreated with estradiol or vehicle control. Tumors were not apparent in the absence of estradiol, as in syngenic wild-type mice. Moreover, in contrast to syngenic immune intact animals, only six very small tumors were found on the bowel mesentery in one of seven estradiol-treated immunocompromised mice (Figure 7). Similar results were obtained using BALB scid mice that lack mature B and T cells but retain an active complement pathway (data not shown). This reduced response of the Trp53 null cells to estradiol in the immune compromised mice is remarkably different than the rampant tumor growth observed with ip injected Pten;Kras(Trp53−) OSE cells in the immune-competent syngenic animals treated with estradiol (Figure 5). These observations, combined with the remarkable expression of immune-related genes in the estradiol treated Pten;Kras(Trp53−) tumors, indicate that an intact immune system contributes in some specific manner to the estradiol-induced growth of Trp53(−) OSE tumor cells in vivo.

Figure 7.

Tumor growth is altered in immune compromised mice. To determine if an intact immune system is required for orthotopic tumor growth, Pten/Kras(Trp53−) OSE cells were injected ip into immunodeficient, ovariectomized Non-Obese Diabetic scid mice pretreated with or without estradiol and harvested at 14 days. Pten/Kras(Trp53−) OSE cells formed only five small visible tumors on the bowel mesentery in one estradiol-treated mouse and exhibited signs of nuclear atypia. No ip tumors were observed in the absence of estradiol. Data are representative of three experiments.

Discussion

Our results show for the first time that TRP53 status is a critical factor controlling the expression of Esr1 and the response of ovarian cancer tumor cells to estradiol. Specifically, estradiol-stimulated ovarian tumor growth, metastases, and features of high-grade carcinoma only occur in the Pten;Kras mutant cells lacking Trp53. This model provides a new paradigm: changes in TRP53 status are related to the susceptibility of transformed OSE cells to estradiol-induced metastases. These results are highly relevant for understanding how TP53 status and steroid hormones control the risk and progression of ovarian cancer in women, an area that is not well defined.

The mechanisms by which estradiol acts to enhance ovarian tumor growth and metastases in theTrp53 null cells appear to involve the regulation of specific genes that control a broad range of cellular functions. As in normal and transformed ERα-positive epithelial cells in other tissues, estradiol increases the expression of the progesterone receptor (Pgr) in the mutant Trp53− OSE cells. Importantly, progesterone potently suppressed the growth-promoting and metastatic effects of estradiol on the Trp53 null cells in culture and in vivo. Similarly, estradiol markedly promoted the expression of Pgr and progesterone suppressed the growth of the human high-grade ovarian cancer cells (POE4) xenografts of adult nude mice (17). Thus, progesterone can act antagonistically to suppress ovarian cancer progression and metastases in these contexts and may be highly relevant to ovarian cancer in women. Specifically, the increase in progesterone receptor may be critical for regulating OSE proliferation in women during their reproductive years because oral contraceptives used for as few as 5 years decrease the risk of ovarian cancer later in life by as much as 50% (13). These data clearly show that steroids impact OSE functions in women who are at risk for high-grade and low-grade ovarian cancer. Although the precise mechanisms by which oral contraceptives are acting are not known, progestins present in oral contraceptives likely counteract the duration and proliferative effects of estradiol. In support of this, progestins added to hormone (estradiol) replacement therapies reduce the incidence of ovarian cancer in postmenopausal women (21) and animal models (22), but this has not been studied extensively. Most ovarian cancer is detected in postmenopausal women, and therefore, it is likely that an indolent tumor can rapidly become aggressive when estradiol action goes unopposed by progesterone or other factors.

Because ERα is elevated in the Trp53− ovarian tumors, TRP53 appears to repress Esr1 transcription in the Pten;Kras(Trp53+) mutant OSE cells. In support of this, TRP53 can block transcription of the Esr1 gene by binding to its promoter (23) or indirectly by increasing KLF4 in breast cancer cells (24). Thus, in Trp53 null OSE cells, the increase in ERα may promote estrogen-mediated metastasis. Recent data suggest that ERα is expressed in a high percentage of high-grade ovarian cancers and therefore may be an important mediator of human ovarian cancer growth and/or metastasis (Table 1). That some ovarian high-grade cancers appear to be ERα-negative indicates that the loss of ERα expression may be a late-stage event in ovarian cancer progression and might be related to mutant forms of Trp53 rather than the complete loss of Trp53.

Our data indicate that estradiol promotes the proliferation and rampant metastases of the Trp53 null tumor cells to multiple sites in the peritoneal cavity, the major site of ovarian cancer metastases in women. In striking contrast, cells of the primary Trp53 null tumors do not metastasize and do not grow as allografts in the peritoneal cavity in the absence of estradiol. Thus, estradiol is an essential mediator of tumor growth and metastasis in Trp53 null OSE tumor cells in our mouse model. Estradiol has also been shown to reduce the latency of onset of tumor growth in a simian virus 40-T antigen mouse model, where presumably TRP53 and RB functions are compromised but were not tested (19). In addition, estradiol has been shown to increase invasion of ovarian cancer epithelial cells in a mouse model of ovarian cancer (15, 25). Of clinical relevance, recent studies indicate that the expression of Esr1 mRNA is a more accurate measure of this receptor in human ovarian cancer cells than are immunostaining results and that elevated levels of Esr1 mRNA (in the absence of progesterone receptor) in high-grade ovarian cancer are associated with poor survival outcome (16). Supporting this, we and others observe high levels of Esr1 mRNA in human (TP53−) SKOV-3 cells (16) and Trp53 mutant POE4 ovarian cancer cell lines (20) that were derived from high-grade tumors and respond to estradiol in culture, in xenografts, and in matrigel. Moreover, progestins can reduce growth of the human ovarian cancer POE4 cells (17). Conversely, steroids appear to have minimal effects in TRP53+ ovarian tumor cells (Figure 1) (15). Thus, there is mounting evidence that steroids impact ovarian cancer growth and progression in women in a TRP53-dependent manner.

The effects of estradiol in our Pten/Kras(Trp53−) mutant mouse model are mediated, in part, by the increased expression of ESR1 in the Trp53 null cells and the induction of specific genes. Genes that are significantly regulated by estradiol were identified by gene ontology analyses of microarray data and verified by growing Trp53− tumor cells in matrigel with or without estradiol, tamoxifen, or progesterone. The categories most highly regulated by estradiol in the Trp53− ovarian cells compared with the control and Trp53+ cells include genes associated with inflammation, immune responses, and chemotaxis. Noteworthy among these genes are cell cytoskeleton genes (Cnn1, Tagln, Tpm1), multiple chemokines (Cxcl1, Cxcl2, Cxcl5), innate immune response factors (C1qb, Il1, Il6, Tnf, Tlr2), as well as the proinflammatory mediators S100a8/S100a9. Thus, the estradiol-responsive ovarian tumor cells in our immune-competent mice appear to produce factors that control the local microenvironment; the microenvironment, in turn, may control or promote tumor development. In support of this, the apparent lack of tumors in estradiol-treated immune-deficient mice indicates that the tumor microenvironment has a significant impact on tumor cell growth and survival. These data warrant future investigation into the modulation of immune effector cell activity as potential therapeutic targets. Future studies will determine estrogen-regulated tumor-host interactions required for tumor growth and metastasis.

In summary, our mouse models provide a powerful system in which 1) to study the interactions of mutant and null forms of Trp53 and the steroid hormones estradiol and progesterone in regulating tumor cell proliferation and metastases in vivo and in culture and 2) to determine the molecular targets of estradiol and progesterone that control ovarian tumor growth. In addition, our studies indicate that estradiol and progesterone may impact the tumor microenvironment and therefore our in vivo model of ovarian cancer has the potential to provide insights into tumor progression and metastasis not possible using human cancer xenografts in immune-compromised mice. Revealing mechanisms by which ERα, TP53, and tumor-microenvironment interact will provide additional therapeutic approaches for women with ovarian cancer.