Abstract

Introduction

Endometrial cancer is the most common type of gynecological cancer. In the United States, approximately 41 200 cases are diagnosed and approximately 7350 women die from the disease each year (Jemal et al., 2007). The majority of endometrial cancers (~90%) are adenocarcinomas, which originate in the uterine epithelial cells. All of the endometrial cancers can be further delineated into two types (Deligdisch and Holinka, 1987; Di Cristofano and Ellenson, 2007). Type I endometrial cancers are estrogen (E2)-dependent as they are associated with conditions that result in elevated E2 levels and appear mostly in pre- and peri-menopausal women. Frequently, these cancers show mutations in DNA-mismatch repair genes (MLH1, MSH2 and MSH6), PTEN, K-ras and β-catenin (Kong et al., 1997; Risinger et al., 1997, 1998). In contrast, type II endometrial cancers are E2-independent, diagnosed mostly in post-menopausal women, in thin and fertile women or in women with normal menstrual cycles.

β-catenin (Ctnnb1) was identified initially as a protein associated with the cytoplasmic region of E-cadherin, which is a transmembrane protein involved in cell adhesion. β-catenin is a component of the Wnt signaling pathway, which is involved in tissue differentiation during embryonic development. In the absence of a Wnt signal in normal cells, β-catenin forms a complex, which includes glycogen synthase kinase 3β (GSK-3β) and the product of the gene that is the genetic determinant of adenomatous polyposis coli. GSK-3β phosphorylates β-catenin targeting it for ubiquitin-dependent degradation by the proteasome, thereby maintaining a low level of free cytoplasmic β-catenin (Aberle et al., 1997). The most common molecular alterations in tumor cells leading to disruption of β-catenin degradation are mutations that inactivate adenomatous polyposis coli or activate β-catenin itself (Morin et al., 1997). These alterations produce an accumulation of cytoplasmic β-catenin that translocates into the nucleus, where it interacts with members of the lymphoid enhancer factor-1/T-cell factor (Lef-1/Tcf) families (Korinek et al., 1997) to activate the transcription of various genes including cyclin D1 (Tetsu and McCormick, 1999) and MYC (He et al., 1998). In endometrial carcinomas, mutations in β-catenin occur in approximately 17% of cases (Fukuchi et al., 1998; Kobayashi et al., 1999; Mirabelli-Primdahl et al., 1999; Ikeda et al., 2000; Saegusa and Okayasu, 2001; Saegusa et al., 2001).

In addition to its role in endometrial cancer, β-catenin has been shown to be an important component of normal uterine function. The process of implantation consists of tightly regulated reactions including apposition of the blastocyst, attachment to and invasion into the uterine epithelium, and decidualization of the uterine stroma (Lee et al., 2007a). Active β-catenin is initially located only in the uterine epithelium at the time of implantation, but then progressively increases in the decidualizing stroma cells (Mohamed et al., 2005; Jha et al., 2006; Herington et al., 2007).

Here, we have generated murine models of dominant stabilized β-catenin (PR^cre/+Ctnnb1^f(Ex3)/+) and ablated β-catenin (PR^cre/+Ctnnb1^f/f) in the uterus to investigate the role of β-catenin in endometrial function and tumorigenesis. The dysregulation of β-catenin resulted in defects in glandular formation, decidualization and fertility. Stabilized β-catenin induces endometrial glandular hyperplasia, whereas ablation of β-catenin results in squamous cell metaplasia. These analyses demonstrate that β-catenin has an important role in normal uterine function as well as in tumorigenesis.

Results

Generation of mice with β-catenin conditionally stabilized and conditionally ablated

In various cancers, mutations of the phosphorylation sites in exon 3 of β-catenin have been identified which prevent its degradation (Fukuchi et al., 1998; Garcia-Rostan et al., 1999; Samowitz et al., 1999). To examine the role of this mutant form of β-catenin, the GSK-3β phosphorylation sites in exon 3 were flanked by loxP sites (Ctnnb1^f(Ex3)) (Harada et al., 1999). Since mutations in β-catenin have been identified in endometrial cancers, Ctnnb1^f(Ex3)/+ mice were crossed with the PR^Cre mouse to generate tissue specific stabilization of β-catenin (PR^cre/+Ctnnb1^f(Ex3)/+) in PR-expressing cells (Soyal et al., 2005). Western blot analysis demonstrated the presence of both full-length and exon 3-deleted β-catenin in the PR^cre/+Ctnnb1^f(Ex3)/+ mice validating successful generation of these mice (Figure 1a). Immunohistochemical analysis for total β-catenin demonstrated that it is expressed primarily in the luminal epithelium at the cellular junctions, confirming earlier findings (Hou et al., 2004). In the uteri of the PR^cre/+Ctnnb1^f(Ex3)/+ mice, we observed increased levels of β-catenin and increased nuclear localization in the epithelium compared with controls demonstrating the increased activation of β-catenin further validating this model (Figure 1b). Although increased β-catenin in the endometrial stroma would be expected using the PR^Cre model, we were not able to discern an increase in this compartment possibly owing to the sensitivity of the immunohistochemistry or an alteration in the epithelial-stroma paracrine communication that altered β-catenin expression.

Open in a separate window

Figure 1

Analysis of conditionally dominant stabilized and conditionally ablated β-catenin in the murine uterus. Eight-week-old PR^cre/+, PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice were sacrificed. Portions of the uterus were saved for protein isolation and fixed in 4% paraformaldehyde for immunohistochemistry. (a) Western blot analysis of β-catenin in whole uterine extracts from PR^cre/+, PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice (b) Immunohistochemical analysis for β-catenin in PR^cre/+, PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri. The arrows indicate nuclear β-catenin. These experiments demonstrate that β-catenin is conditionally activated or ablated in the uteri of these mice.

To complement the investigations of the overexpression of β-catenin, mice with tissue specific ablation of β-catenin in the uterus were also generated. Owing to the embryonic lethality of β-catenin knockout mice, it was necessary to conditionally ablate β-catenin in the uterus (Haegel et al., 1995; Huelsken et al., 2000). β-catenin floxed mice (Ctnnb1^f/f) were crossed to the PR^Cre mouse to conditionally ablate β-catenin in PR-expressing tissue (PR^cre/+Ctnnb1^f/f) (Brault et al., 2001; Soyal et al., 2005). Western blot analysis confirmed the ablation of β-catenin in these mice (Figure 1a). Immunohistochemical analysis demonstrated an absence of β-catenin expression in the uteri of the PR^cre/+Ctnnb1^f/f mice, further validating successful generation of these mice (Figure 1b).

Both PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice exhibit fertility defects and fail to undergo the decidualization reaction

To assess their role in uterine function, female PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice were mated to wild-type male mice for 6 months. PR^cre/+Ctnnb1^f(Ex3)/+ mice were severely sub fertile with only one of seven mice having one litter compared with twenty-nine litters from the seven control mice (Table 1). The litter size was comparable to that of control mice. PR^cre/+Ctnnb1^f/f mice were completely infertile (Table 1). Therefore, altering β-catenin expression in PR-expressing cells alters murine fertility.

Table 1

Fertility defect of PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice

Genotype	No. of mice tested	No. of litters	No. of pups	Mean of litter/mouse	Mean of pups/litter
Ctnnb1f(Ex3)/+	7	29	193	4.14 ± 0.39	6.66 ± 0.37
PRcre/+Ctnnb1f(Ex3)/+	7	1	7	0.17	7.00
Ctnnb1f/f	7	31	197	4.43 ± 0.45	6.35 ± 0.37
PRcre/+Ctnnb1f/f	7	0	0	0.0	0.0

Open in a separate window

As the PR^Cre mouse shows Cre recombinase activity in the pituitaty, ovary, uterus and mammary gland, the cause of infertility in these mice maybe because of a defect in any of these tissues (Soyal et al., 2005). To test for an ovarian phenotype, female PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice were examined for their ability to ovulate normally in response to a superovulatory regimen of gonadotropins. PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f yielded 16.50 ± 4.50 and 22.67 ± 5.36 oocytes, respectively, which did not differ significantly from control mice (18.50 ± 2.50). Also, histological analysis of the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+ Ctnnb1^f/f 6 weeks ovary did not show any alterations in ovarian morphology(data not shown). Finally, random cycling female PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice exhibited normal levels of P4 (6.25 ± 0.02 ng/ml and 6.40 ± 0.45 ng/ml, respectively) and E2 (25.76 ± 3.55 pg/ml and 27.03 ± 3.52 pg/ml, respectively) compared with controls (6.41 ± 0.97 ng/ml P4 and 24.18 ± 2.89 pg/ml E2). Although these data do not rule out a pituitary defect, it shows that ovarian morphology, steroidogenesis and function were not affected in the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+ Ctnnb1^f/f females, which suggests that the fertility defect is primarily because of a uterine defect.

To determine if the infertility was in part owing to loss of the ability to support implantation, we investigated the ability of the uterus to undergo a decidual reaction. Upon embryo attachment to the uterine epithelium, the stroma cells around the implantation site become highly proliferative and transform into a more epitheloid shape with increased vascualrization. This decidual reaction can be artificially induced by administration of a decidual trauma following a specific hormone regime (Finn and Hinchliffe, 1964). The unstimulated contralateral horn serves as a control. The extent of the decidual reaction can be assayed by the increase in weight gain of the stimulated horn compared with the control horn. Ovariectomized female PR^cre/+Ctnnb1^f(Ex3)/+, PR^cre/+Ctnnb1^f/f and PR^cre/+ mice (N=3) were treated with hormones and the uterus was mechanically stimulated to mimic the signaling of the embryo at implantation and to induce decidualization (see Materials and Methods section). Gross anatomy of the decidual horn showed lack of an increase in size for both the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice compared with the control mice (Figure 2a). The ratio of stimulated to unstimulated horn weight in both the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice confirmed the diminished decidual reaction (Figure 2b). These results demonstrate that β-catenin is tightly regulated during decidualization as both stabilization and ablation of β-catenin in mice obliterate the decidual response.

Open in a separate window

Figure 2

PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice are unable to undergo the hormonally induced decidual reaction. Six-week-old mice were ovariectomized and 2 weeks later were subjected to a hormone regime and a decidual stimulus. (a) Gross anatomy of the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri show lack of an increase in size of the decidual horn compared with controls. (b) Stimulated horn to unstimulated horn weight ratio was decreased in the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri. ***P<0.001, one-way ANOVA followed by Tukey’s post hoc multiple range test. ANOVA, analysis of variance.

PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice showed altered endometrial epithelial differentiation

Examination of the uteri of the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice showed a morphological defect. Uterine shape and size were examined at 1–8 weeks of age. At 4 weeks of age, there was a significant decrease in uterine weight of the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+ Ctnnb1^f/f mice compared with controls that persisted until 8 weeks of age (Figure 3a). PR^cre/+Ctnnb1^f(Ex3)/+ uteri were smaller in length than control mice, whereas PR^cre/+Ctnnb1^f/f mice were similar in length, but were thinner than the control mice (Figure 3b). The uteri of the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice were examined for uterine morphology(Figure 4). As early as 2 weeks of age, differences in morphology can be noted. PR^cre/+Ctnnb1^f(Ex3)/+ mice have enlarged glands, whereas the PR^cre/+Ctnnb1^f/f mice have no significant difference compared with control mice. At 6 weeks of age, PR^cre/+ Ctnnb1^f(Ex3)/+ mice exhibit signs of endometrial hyperplasia. The glands are increased in size and number and they show hyperchromatic nuclei and pseudostratification of the epithelial cells. In contrast, the PR^cre/+ Ctnnb1^f/f mice have a decreased number of glands. In addition, the luminal epithelium exhibits squamous cell metaplasia. These results support the role of β-catenin in the regulation of endometrial epithelial differentiation.

Open in a separate window

Figure 3

PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri exhibit developmental defects. (a) Uterine weight was determined for females 1–8 weeks old. Decreased uterine weight was observed for PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice after 4 weeks of age. (b) Gross anatomy of the uteri showed that PR^cre/+Ctnnb1^f(Ex3)/+ uteri are smaller in length, but of the same thickness as control uteri, whereas PR^cre/+Ctnnb1^f/f uteri are the same length, but thinner than control uteri. *P<0.05, **P<0.01, one-way ANOVA followed by Tukey’s post hoc multiple range test. ANOVA, analysis of variance.

Open in a separate window

Figure 4

Histology of PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri 2, 6 and 8-week-old PR^cre/+, PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+ Ctnnb1^f/f mice were sacrificed and the uteri were fixed in 4% paraformaldehyde for hematoxylin & eosin staining. PR^cre/+Ctnnb1^f(Ex3)/+ uteri demonstrated enlarged glands as early as 2 weeks of age. At 6 weeks of age, these mice exhibit endometrial hyperplasia consisting of an increased size and number of glands that have hyperchromatic nuclei and pseudostratification of the epithelial cells. PR^cre/+Ctnnb1^f/f uteri have no significant difference in morphology compared with controls until 6 weeks of age. At 6 weeks of age, there are decreased number of glands and squamous cell metaplasia in the luminal epithelium.

To determine if the loss of β-catenin resulted in squamous cell metaplasia, the expression of cytokeratin 14 and p63 was investigated. Cytokeratin 14 is a basal epithelial cell marker that appears only upon the initiation of stratification of the epithelium (Hong et al., 2004; Koster et al., 2007). p63 is a transcription factor of the p53 family and is a marker of squamous cell carcinomas (Lin et al., 2006). It has been shown to playa role in the activation of β-catenin signaling in the regulation of epithelial cell differentiation and tumorigenesis (Trink et al., 2007). There was no change in cytokeratin 14 or p63 expression in the stroma or epithelium of 8-week-old PR^cre/+Ctnnb1^f(Ex3)/+ mice compared with controls (Figure 5). In the PR^cre/+ Ctnnb1^f/f mice, cytokeratin 14 and p63 expression was unchanged in the stroma, but was significantly increased in epithelial cells compared with the control uteri (Figure 5). These results demonstrate that the epithelium of the PR^cre/+Ctnnb1^f/f mice transform into a squamous cell type, whereas the epithelium of the PR^cre/+ Ctnnb1^f(Ex3)/+ mice do not.

Open in a separate window

Figure 5

Cytokeratin 14 and p63 were increased in the endometrial epithelium of PR^cre/+Ctnnb1^f/f mice. (a) Immunohistochemical analysis of cytokeratin 14 demonstrates increased expression in the epithelium of PR^cre/+Ctnnb1^f/f uteri, but not PR^cre/+Ctnnb1^f(Ex3)/+ uteri compared with controls. (b) Immunohistochemical analysis of p63 shows that it is also increased in the epithelium of PR^cre/+ Ctnnb1^f/f uteri, but not PR^cre/+Ctnnb1^f(Ex3)/+ uteri compared with controls.

PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice have increased proliferation

To determine if β-catenin signaling is altered in these mice, known β-catenin targets were examined. Cyclin D1 is a known target of β-catenin, such that overexpression of β-catenin leads to unregulated entry into the cell cycle and increased proliferation (He et al., 1998; Shtutman et al., 1999). In the PR^cre/+Ctnnb1^f(Ex3)/+ mice, there is an increase in cyclin D1 expression (Figure 6a). The increased expression was localized to the glandular epithelium as the stroma exhibited no change in cyclin D1 levels (Figure 6b). In the PR^cre/+Ctnnb1^f/f mice, there was no change in cyclin D1 expression by western blot analysis (Figure 6a). However, immunohistochemical analysis of cyclin D1 showed increased expression in basal epithelial cells with no change in the stroma. As uterine growth is regulated byE2 and unopposed E2 action is a hallmark of endometrial cancer, we next examined the expression of estrogen receptor-α (ERα) in the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri. In the PR^cre/+Ctnnb1^f(Ex3)/+ mice, there was increased ERα expression in the epithelium and reduced levels in the stroma. In the PR^cre/+Ctnnb1^f/f mice, there was increased ERα expression in the basal epithelial cells with no change in the columnar epithelium or stroma (Figure 6b).

Open in a separate window

Figure 6

Cyclin D1 and ERα is dysregulated in the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri. Eight-week-old female mice were sacrificed. Portions of the uterus were saved for protein isolation and fixed in 4% paraformaldehyde for immunohistochemistry. (a) Western blot analysis of whole uterine extracts for cyclin D1. PR^cre/+Ctnnb1^f(Ex3)/+ uteri showed increased levels of cyclin D1, whereas PR^cre/+Ctnnb1^f/f uteri showed no change in cyclin D1 levels. (b) Immunohistochemical analysis of uteri from PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice. In PR^cre/+Ctnnb1^f(Ex3)/+ uteri, both cyclin D1 and ERα was increased in the epithelium. In PR^cre/+Ctnnb1^f/f uteri, both cyclin D1 and ERα was increased in the basal epithelial cells. ERα, estrogen receptor-α.

To determine if the uteri of PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f exhibited an altered response to E2, PR^cre/+Ctnnb1^f(Ex3)/+, PR^cre/+Ctnnb1^f/f and PR^cre/+ mice were ovariectomized, and treated with vehicle (sesame oil) or E2 (0.1 μg per mouse in 100 μl sesame oil) for 4 h or 3 days. PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri exhibited no difference in weight gain or expression of the epithelial ERα target genes lactotransferrin (Ltf) and complement component 3 (C3) when the mice were treated with vehicle or E2 for 4 h (Figures 7a and b). However, upon E2 treatment for 3 days, PR^cre/+ Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice did not show the increase in uterine weight usually seen with this treatment (Figure 7a). Furthermore, the expression of Ltf and C3 was decreased in these mice compared with E2-treated PR^cre/+ uteri (Figure 7b). These results suggest that PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice have a decrease in some ERα target genes upon induction byE2.

Open in a separate window

Figure 7

PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri exhibit a defect in the E2 response. (a) Ovariectomized PR^cre/+Ctnnb1^f(Ex3)/+, PR^cre/+Ctnnb1^f/f and PR^cre/+ mice were injected with vehicle (sesame oil) or E2 (0.1 μg per mouse) for 4 h or 3 days. Uterine weight was determined. Decreased uterine weight was observed for PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice treated with E2 for 3 days compared with PR^cre/+ mice. (b) ERα regulated gene expression in uteri of PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice. Real-time RT–PCR analysis of Ltf and C3 was performed in uteri of PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice. The results represent the mean ± s.e.m. of three independent RNA sets. **P<0.01; ***P<0.001, one-way ANOVA followed by Tukey’s post hoc multiple range test. ANOVA, analysis of variance; ERα, estrogen receptor-α.

Discussion

To study the role of the Wnt signaling pathway in the adult uterus, we generated mice in which β-catenin was stabilized (PR^cre/+Ctnnb1^f(Ex3)/+) or ablated (PR^cre/+ Ctnnb1^f/f) in the reproductive tract using previously generated mice with modified β-catenin alleles and the PR^Cre mouse (Harada et al., 1999; Brault et al., 2001; Soyal et al., 2005). These mice were found to have severe fertility defects most likely because of altered uterine morphology and function as seen by lack of a decidual response. Furthermore, the uterus undergoes developmental defects including a reduction in overall weight and a deregulation of glandular formation. Interestingly, the epithelium also undergoes differentiation. In the case of the stabilization of β-catenin, the epithelium takes on characteristics of endometrial hyperplasia. When β-catenin is ablated, squamous cell metaplasia develops as shown by its positive staining for p63 and cytokeratin 14. We have examined these mice up to 1 year of age and these mice were able to survive without evidence of further progression of this pathology suggesting that alterations in β-catenin are not sufficient for the development of endometrial cancer. These results suggest a role for β-catenin in both endometrial function and dysfunction.

Deregulation of β-catenin resulted in female mice with impaired fertility and an absence of a decidual response. These results reflect the expression data as well as information gleamed from other mouse models on the importance of Wnt signaling during implantation and decidualization. Both Lif^−/− and Hoxa10^−/− mice are unable to undergo embryo implantation and decidualization of the stroma (Stewart et al., 1992; Lim et al., 1999). At day5 of pregnancy, Sfrp4 expression is downregulated in the stroma of the Lif^−/− mice (Daikoku et al., 2004). In the Hoxa10^−/− mice, Wnt4 expression is completely absent during early pregnancy, whereas Sfrp4 expression is more widespread in the stroma. Furthermore, conditional ablation of Bmp2 in the uterus resulted in mice that were unable to undergo the decidualization reaction even though embryo attachment occurred (Lee et al., 2007b). Both Wnt4 and Wnt6 expression was greatly reduced and Sfrp4 expression was increased during decidualization in these mice. When isolated mouse stromal cells were treated with hormones and Bmp2 to induce decidualization, Wnt4 was identified as a Bmp2 target during this in vitro decidualization, but Wnt3a, Wnt6, Wnt7a and Wnt10a were not (Li et al., 2007). Furthermore, knockdown of Wnt4 in these cells prevented them from undergoing the in vitro decidualization reaction. From these mouse models, it has been demonstrated that tight regulation of Wnt signaling is necessary for a successful decidualization reaction to occur in the uterus.

The PR^Cre mouse model recombines genes in all compartments of the uterus, as well as the pituitary, mammary gland and preovulatory granulos a cells. No ovarian phenotype was detected in these mice and the inability of the uterus to undergo a hormonally induced decidual response indicates that the fertility defect is primarily because of a uterine defect although a pituitary contribution to this phenotype cannot be ruled out. The PR^Cre mouse shows Cre activity in both endometrial epithelial and stromal compartments, as well as in the myometrium. Therefore, it is not clear if the alteration in the epithelial cell phenotype owing to β-catenin is a direct or an indirect effect on the epithelial compartment. Insight into this question is addressed by mice in which β-catenin was ablated in the myometrium and stroma using the MISII-Cre mouse model (Arango et al., 2005; Deutscher and Hung-Chang Yao, 2007). This mouse exhibited a normal epithelium suggesting that the phenotype seen in our mouse model is because of a direct effect on the epithelium by β-catenin. However, a direct investigation of the phenotype of the expression of a stabilized β-catenin in the endometrial epithelium will have to wait until a uterine epithelial Cre model is developed.

Three members of the Wnt family(Wnt4, Wnt5a and Wnt7a) have been shown to be critical for uterine development. Female Wnt4^−/− mice exhibit sex reversal in that they lack a Müllerian duct, begin to develop the epididymis, and have a gonad associated with a fat pad (Vainio et al., 1999). Wnt7a^−/− uteri are smaller, lack glands, contain a stratified epithelium, thin stroma and two smooth muscle layers (Miller and Sassoon, 1998; Parr and McMahon, 1998; Carta and Sassoon, 2004). The posterior end of the uterus fails to develop in female Wnt5a^−/− mice (Mericskay et al., 2004). These mice lack a cervix and vagina, and the uterus is 60–90% shorter. When these uterine horns were grafted into kidney capsules, the uteri failed to develop glands demonstrating that Wnt5a is necessary for gland formation. These results all show that Wnt signaling is critical for uterine development and specifically for the formation of glands. By comparing to our model, it appears that the phenotype of Wnt5a and Wnt7a is mediated in part through β-catenin.

In addition to the glandular phenotype, both the PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f uteri exhibit an altered epithelium. The uteri of the PR^cre/+ Ctnnb1^f(Ex3)/+ mice develop signs of endometrial hyperplasia such as an increase in glandular size and number, hyperchromatic nuclei and pseudostratification of the epithelial cells. Endometrial hyperplasia is often an early event in endometrial cancer (Di Cristofano and Ellenson, 2007). In one sampling of endometrial hyperplasias and carcinomas, mutations in β-catenin resulting in its nuclear accumulation were found in 70% of hyperplasia cases, but only56% of carcinoma cases suggesting that mutations in β-catenin are an early event in the development of endometrial carcinoma (Nei et al., 1999). Mutations in β-catenin resulting in its nuclear accumulation are found in 13–69% of endometriod endometrial cancers (Fukuchi et al., 1998; Nei et al., 1999; Kim et al., 2002; Moreno-Bueno et al., 2002; Catasus et al., 2004; Wappenschmidt et al., 2004; Irving et al., 2005; Saegusa et al., 2005). These mutations commonly occur as missense mutations in exon 3, where by the Ser/Thr phosphorylation sites of GSK-3β are mutated. Thus, β-catenin is prevented from being degraded resulting in increased β-catenin levels. Several studies have indicated that overexpressed β-catenin can be imported into the nucleus (Staal et al., 1999; Mao et al., 2001). Activation of β-catenin signaling through its nuclear localization increases the expression of c-myc and cyclin D1, resulting in the promotion of tumor progression by stimulating cell proliferation (He et al., 1998; Shtutman et al., 1999; Tetsu and McCormick, 1999). We observed the nuclear accumulation of β-catenin and an increase in cyclin D1 expression in the epithelial cells of PR^cre/+Ctnnb1^f(Ex3)/+ mice in which β-catenin was stabilized through deletion of the GSK-3β phosphorylation sites. As most cancers begin in the epithelium, these results on the nuclear accumulation of β-catenin in the epithelium provide evidence as to the ontogenetic role of β-catenin in endometrial cancer.

The uteri of the PR^cre/+Ctnnb1^f/f mice exhibit squamous cell metaplasia of the luminal epithelium. The mechanism of squamous cell metaplasia of the benign endometrium in humans remains unknown; however, it is known to be associated with various conditions including adenocarcinoma, senile endometrium, chronic endometritis, radiation, submucosal myoma, endometrial hyperplasia and lesions resulting from use of an intrauterine device (Anderson et al., 2002). p63 is not only a basal cell marker, but also a squamous cell metaplasia marker in the uterus. In the skin, p63 has been shown to be necessary for the stratification of the epithelium (Koster et al., 2007). Since p63 is highly expressed in the squamous cell metaplasia region of the PR^cre/+Ctnnb1^f/f mice, the loss of β-catenin may be the mutation that activates p63 leading to squamous cell metaplasia. p63 binds as part of an enhancer complex to the promoter of cytokeratin 14 leading to its expression, which is why cytokeratin 14 is found in cells expressing p63 (Romano et al., 2007). These results suggest a possible role for Wnt signaling in the development of squamous cell metaplasia in the uterus.

One hallmark of endometrial cancers is unopposed estrogen signaling (Di Cristofano and Ellenson, 2007). E2 stimulates proliferation of epithelial cells in the mouse uterus (Martin et al., 1973; Huet-Hudson et al., 1989). In contrast, P4 is inhibitory to E2-mediated proliferation of the luminal and glandular epithelial cells. The ability of ovarian steroids to regulate uterine cell proliferation depends upon the ability of hormonal stimulation to regulate growth factor communication networks between stroma and epithelium in the uterus. In the PR^cre/+Ctnnb1^f(Ex3)/+ mice, ERα expression was increased in the epithelium, but was reduced in the stroma. Likewise, in the PR^cre/+Ctnnb1^f/f mice, there was increased ERα expression in the basal epithelial cells. However, analysis of the expression of epithelial E2 target genes Ltf (Pentecost et al., 1988) and C3 (Sundstrom et al., 1989), as well as E2 stimulated uterine weight gain showed no alteration in these models after a 4-h E2 treatment, but lack of an E2 response after a 3-dayE2 treatment. The lack of an increase in the expression of these E2 target genes maybe because of the fact that their full induction requires both stromal and epithelial ERα (Buchanan et al., 1999; Cunha et al., 2004), and stroma ERα was not altered in this model. Alternatively, the differentiation of the epithelium owing to the stabilization or loss of β-catenin may have altered the abundance of cofactors, or epigenetically modified these particular target genes to render them unresponsive to steroid stimulation. A large-scale genomic analysis of the altered expression of the uterine epithelium of these models will be necessary to determine the full impact of the altered phenotype on steroid hormone responsiveness. These results suggest that β-catenin may mediate the response of the uterus to E2 signaling. As such, the Wnt signaling pathway may be an important target of the P4 regulated pathways that inhibit E2-induced proliferation, an inhibition that is absent in endometrial cancer.

In conclusion, we have demonstrated a role for β-catenin in both endometrial function and dysfunction using PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mouse models. PR^cre/+Ctnnb1^f(Ex3)/+ and PR^cre/+Ctnnb1^f/f mice exhibit both developmental and fertility defects demonstrating how its regulation is important to adult uterine function. Furthermore, these mice develop endometrial hyperplasia and squamous cell metaplasia, respectively, elucidating a role for β-catenin in endometrial dysfunction. The results of this investigation provide significant insights into our understanding of the importance of β-catenin in female reproduction and endometrial cancer. By studying the role of the Wnt signaling pathway, further insights into the hormone regulation of the uterus and the ways it is altered in uterine dysfunction can be uncovered.

Materials and methods

Acknowledgments

References