Progesterone Inhibition of Wnt/β-Catenin Signaling in Normal Endometrium and Endometrial Cancer Free

Purpose. Wnt signaling regulates the fine balance between stemness and differentiation. Here, the role of Wnt signaling to maintain the balance between estrogen-induced proliferation and progesterone-induced differentiation during the menstrual cycle, as well as during the induction of hyperplasia and carcinogenesis of the endometrium, was investigated.

Experimental Design: Endometrial gene expression profiles from estradiol (E₂) and E₂ + medroxyprogesterone acetate–treated postmenopausal patients were combined with profiles obtained during the menstrual cycle (PubMed; GEO DataSets). Ishikawa cells were transfected with progesterone receptors and Wnt inhibitors dickkopf homologue 1 (DKK1) and forkhead box O1 (FOXO1), measuring Wnt activation. Expression of DKK1 and FOXO1 was inhibited by use of sequence-specific short hairpins. Furthermore, patient samples (hormone-treated endometria, hyperplasia, and endometrial cancer) were stained for Wnt activation using nuclear β-catenin and CD44.

Results: In vivo, targets and components of the Wnt signaling pathway (among them DKK1 and FOXO1) are regulated by E₂ and progesterone. In Wnt-activated Ishikawa cells, progesterone inhibits Wnt signaling by induction of DKK1 and FOXO1. Furthermore, using siRNA-mediated knockdown of both DKK1 and FOXO1, progesterone inhibition of Wnt signaling was partly circumvented. Subsequently, immunohistochemical analysis of the Wnt target gene CD44 showed that progesterone acted as an inhibitor of Wnt signaling in hyperplasia and in well-differentiated endometrial cancer.

Conclusion: Progesterone induction of DKK1 and FOXO1 results in inhibition of Wnt signaling in the human endometrium. This Wnt inhibitory effect of progesterone is likely to play a rate-limiting role in the maintenance of endometrial homeostasis and, on its loss, in tumor onset and progression toward malignancy. (Clin Cancer Res 2009;15(18):5784–93)

The female sex hormones estradiol (E₂) and progesterone play rate-limiting roles in the cyclical renewal of the inner layer of the uterus (endometrium). In the first half of the regular menstrual cycle, the proliferation phase, E₂ is required to expand the endometrial layer by inducing cell proliferation. In the second half of the menstrual cycle, the secretory phase, progesterone levels increase, which antagonizes the proliferative activity of E₂ by inducing differentiation of epithelial and stromal cells of the endometrium (1). Thus, inhibition of E₂-induced proliferation by progesterone is crucial for the maintenance of homeostasis in the endometrium.

Increased estrogen signaling often underlies endometrial hyperplasia and is a well-established risk factor for endometrial cancer (2). Because progesterone inhibits estrogen-induced endometrial proliferation, progesterone has been used in its synthetic form [i.e., medroxyprogesterone acetate (MPA)] in palliative treatment of advanced and recurrent endometrial cancer with modest though significant response rates (15-25%; ref. 3). Progesterone has also been used as a primary treatment for endometrial carcinoma confined to the endometrial layer of the uterus, for example, in premenopausal women determined to preserve fertility. Response rates in these women can be up to 60% (4, 5), indicating that progesterone signaling in well-differentiated endometrial cancer is a potent inhibitor of endometrial carcinogenesis.

Wnt/β-catenin signaling has a central function in the maintenance and control of stem cell compartments where it regulates the fine balance between stemness (Wnt-On) and differentiation (Wnt-Off). This central role in homeostasis of adult stem cell niches is reflected by the frequent association of Wnt/β-catenin signaling defects with different cancer types (e.g., breast, colon, stomach, liver, ovary, uterine, skin, etc.). Central in canonical Wnt signaling is the so called “destruction complex.” This complex consists of the scaffold proteins adenomatosis polyposis coli (APC), AXIN1 or AXIN2 (conductin), β-catenin (CTNNB1), and the two kinases, casein kinase I and glycogen synthase kinase 3β. In the absence of Wnt, ligand formation of the destruction complex induces phosphorylation of β-catenin, which results in its degradation. On Wnt signaling, the destruction complex is not formed and, consequently, intracellular β-catenin accumulates and translocates to the nucleus. In the nucleus, β-catenin interacts with members of the T-cell factor/lymphoid enhancer binding factor (TCF/LEF) family of transcription factors, thus modulating the expression of a broad spectrum of Wnt downstream target genes (6). The latter include genes encoding for proteins with central roles in proliferation and survival, such as cyclin D1 (CCND1) and IGF-I (insulin-like growth factor 1), as well as others relevant for migration and invasion (e.g., CD44). Constitutive activation of the Wnt pathway was first observed in familial adenomatous polyposis where individuals carrying a germ-line APC mutation developed intestinal polyps on somatic inactivation of the wild-type APC allele. Notably, in the vast majority of sporadic colorectal cancers, Wnt signaling is constitutively activated as the result of somatic APC mutations. In the remaining cases, either oncogenic β-catenin mutations or alterations in genes encoding secreted proteins that are able to inhibit Wnt signaling such as dickkopf homologue 1 (DKK1; ref. 7) are found.

In endometrial cancer, gene mutations leading to constitutive activation of canonical Wnt/β-catenin signaling are an early event (8). Well-differentiated endometrioid adenocarcinomas show nuclear β-catenin accumulation in at least 30% of the cases (31%: ref. 9; 85%: ref. 10). Accordingly, oncogenic β-catenin mutations have been identified in 15% to 40% of endometrial carcinomas (11, 12), whereas loss of heterozygozity of APC was reported in 24% of cases with nuclear β-catenin staining (13). In addition, the APC A1 promoter was found to be hypermethylated in 47% of endometrial cancers with nuclear β-catenin (13) often in correlation with microsatellite instability (14). APC mutation analysis revealed truncating mutations in 10% of all endometrial cancers (15).

In the endometrium, Wnt/β-catenin signaling is involved in a number of developmental processes. For example, it has been shown that Wnt4 is required for Mullerian duct initiation (16), Wnt7A for subsequent differentiation (17), and Wnt5A for posterior outgrowth of the female reproductive track (18). Furthermore, Wnt/β-catenin signaling has also been implicated in regulation of the normal menstrual cycle (19, 20). Nuclear β-catenin staining, for example, is observed during the proliferative phase of the menstrual cycle (21). Likewise, Nei et al. (21) showed the absence of nuclear β-catenin staining during the second half (differentiation phase) of the menstrual cycle. Hence, inhibition of Wnt/β-catenin signaling may correlate with the inhibition of E₂-induced proliferation. Hou et al. (22) showed that estrogen-induced endometrial proliferation was effectively inhibited by the Wnt/β-catenin inhibitor SFRP2, and Jeong et al. (23) showed that correct β-catenin expression is vital for normal uterine function. As progesterone counteracts the proliferative effects of E₂ during the menstrual cycle, it may do so by inhibiting E₂-induced Wnt/β-catenin signaling.

Based on the above data, we postulate that progesterone counteracts the proliferative effects of E₂ during the normal menstrual cycle, during hyperplasia, and during early endometrial carcinogenesis by inhibiting Wnt/β-catenin signaling. According to this hypothesis, sex hormones may modulate Wnt/β-catenin signaling in the normal and aberrant endometrium to maintain the balance between proliferation (Wnt-On) and differentiation (Wnt-Off).

A description of the inclusion and exclusion criteria and of the histologic and molecular findings in the endometrium was documented earlier (1, 24). The study groups were control group (8 subjects, no hormonal treatment), E₂ group (7 subjects, 2 mg of E₂ administered orally everyday, starting 21 d before surgery), and E₂ + MPA group (6 subjects, 2 mg E₂ + 5 mg MPA administered orally everyday, starting 21 d before surgery). After treatment, endometrial tissue was dissected out, RNA was isolated and processed, and gene expression was measured using the Affymetrix U133plus2 GeneChips containing 54,614 probe sets, representing ∼47,000 transcripts (Affymetrix).

A description of the inclusion and exclusion criteria and of the histologic and molecular findings in the endometrium was documented earlier (25). The study groups were PE, proliferative endometrium (4 subjects); ESE, early-secretory endometrium (3 subjects); MSE, mid-secretory endometrium (8 subjects); and LSE, late-secretory endometrium (6 subjects). Endometrial tissue was dissected as described above, and isolated RNA was used on the same microarrays as described above (Affymetrix U133plus2 GeneChip).

In total, 10 well-differentiated tumor samples were stained for progesterone receptor and CD44. From one representative sample, hyperplasia as well as tumor was available. Paraffin-embedded samples were obtained from the Erasmus University Medical Center Department of Pathology. The histopathologic diagnoses of all samples were reviewed by our pathologist, Patricia C Ewing, M.D., Ph.D. Descriptions of the samples are given in Results and in the legends to the figures.

All raw gene expression data have been posted at the GEO DataSets option in PubMed and are freely available to the scientific community. Raw data of both studies (24, 25) were normalized as a group using robust multiarray analysis normalization (26). From the Talbi data (25), we only used gene expression data from well-characterized tissue samples (Table 1 from ref. 25, with the exception of samples 455 and 562, which were not made available to the GEO DataSets option in PubMed). Statistical analysis of microarray (SAM) was done to identify significant differentially regulated genes between different groups (E₂, E₂ + MPA, PE, ESE, MSE, and LSE; ref. 27). The median false discovery rate was set to 1% for these comparisons. Genes with differential expression due to difference in tissue source, rather than hormonal effects, were identified and excluded from subsequent analysis. Cluster analysis was done as described by Hanifi-Moghaddam et al. (24). Pathway analyses were conducted using Ingenuity Pathway software.⁶

Ishikawa cell lines were maintained as described earlier (28). During the transfection, infection, and hormone stimulation experiments, the cells were cultured in DMEM/F12 supplemented with 5% dextran-coated charcoal-treated FCS in the presence of penicillin and streptomycin. Transfection was done 24 h after seeding of the cells (5,000 per well in a 24-well plate); 4 h after transfection, hormones were added to the cells (concentrations are indicated in the figures). At 48 h after transfection, the cells were harvested to perform reporter assays. The stably progesterone receptor–transfected cell lines used in the short hairpin RNA (shRNA) experiments were generated as described earlier (28); transient transfections were done as described earlier (29). The TOPflash and FOPflash vectors were obtained from Millipore (Upstate). DKK1 expression vector (cloned in pcDNA3.1) was obtained from Dr. X. He (Department of Neurology, Division of Neuroscience, Harvard Medical School, Boston, MA; ref. 30); forkhead box O1 (FOXO1) expression vector (cloned in pcDNA3.1) was obtained from Dr. J.J. Kim (Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, Northwestern University, Chicago, IL; ref. 31); and dnTCF4 expression vector (cloned in pcDNA3.1) was a gift from Dr. H. Clevers (Hubrecht Institute, Developmental Biology and Stem Cell Research, University Medical Center, Utrecht, the Netherlands; ref. 32).

The production of lentivirus and the generation of stable cell line were done as described before (33). The shRNAs specific for hDKK1 (n = 5) and hFOXO1 (n = 6), nontarget shRNA control vector, and TurboGFP control vector were obtained from The MISSION TRC shRNA libraries from Sigma-Aldrich. The packaging plasmids were provided by the Naldini lab, Vita-Salute San Raffaele Univerity, Milan, Italy.

Immunohistochemistry and Western blotting were done essentially as described previously (1, 29). The dilutions of antibody used in immunohistochemistry were rat monoclonal CD44 antibody, 1:1,000 (BD Pharmingen); rabbit monoclonal β-catenin antibody, 1:4,000 (Epitomics); and mouse progesterone receptor antibody, 1:50 (Ab-8 cocktail, Neomarkers). The dilutions in Western blotting were goat anti-hDKK1 antibody, 1:500 (AF1096, R&D Systems); rabbit polyclonal anti-hFOXO1 antibody, 1:5,000 (A300-297A, Bethyl Laboratories); and rabbit anti–β-tubulin, 1:1,000 (Ab6046, Abcam).

SAM analysis between the endometrial gene expression profiles obtained from untreated postmenopausal women in comparison with women treated for 21 days with E₂ or E₂ + MPA indicated that 5,932 probe sets (representing ∼4,500 genes) were significantly up-regulated or down-regulated (P < 0.00015).

Cluster analysis of these significantly up-regulated or down-regulated genes reveals three distinct dendrogram branches: one cluster encompassed all endometrial gene expression profiles from E₂-treated women; the second contained profiles from untreated controls; and the third contained profiles from both E₂ + MPA–treated women and untreated controls (Fig. 1A). Ten genes were investigated using quantitative real-time reverse transcription-PCR in a previous study to validate differential expression (24). In Fig. 1B, highly E₂-regulated genes are shown to be counteracted (compensated for) by the addition of MPA. In total, 438 genes were identified as being significantly E₂ regulated (i.e., >3-fold up-regulated or down-regulated; Fig. 1B, filled columns). For 377 (233 + 144) of these genes, this E₂ regulation was counteracted (compensated for) either in part or fully by simultaneous MPA administration (Fig 1B, open columns). These analyses show that treatment with E₂ alone has a profound effect on endometrial gene expression, and that this effect is greatly diminished on addition of MPA to the E₂ treatment.

We next integrated our profiling results with previously published gene expression data obtained during the main phases of the normal menstrual cycle (25). SAM analysis revealed a total of 11,866 probe sets (representing ∼9,000 genes) significantly regulated between any of the six indicated groups [the E₂- and E₂ + MPA–treated endometria from the present study and four different phases of the menstrual cycle: proliferative endometria (PE), early-secretory endometria (ESE), mid-secretory endometria (MSE), and late-secretory endometria (LSE); ref. 25]. These differentially expressed genes were used for cluster analysis (Fig. 1C). Three clusters were recognized: cluster 1, the samples from E₂-treated patients and those from women in proliferative phase; cluster 2, early- and mid-secretory phase specimens; and cluster 3, the gene expression files from the E₂ + MPA–treated patients and from women in late-secretory phase. These results indicate that E₂ signaling is a very important factor during the proliferative phase of the menstrual cycle (Fig. 1C, Cluster 1). Furthermore, the addition of MPA to the E₂ treatment creates an endometrium with obvious similarities to the late-secretory phase of the menstrual cycle (Fig. 1C, Cluster 3).

The gene expression data from Fig. 1C were subsequently used in a pathway analysis.⁷

It was observed that a number of pathways were significantly regulated by one of the treatments or in different phases of the menstrual cycle. For example, when we compared the proliferative endometrium to the early- or mid-secretory endometrium, “cancer,” “reproductive system disease,” and “gastrointestinal disease” were all significantly regulated (P < 0.006). It is interesting to note that gastrointestinal disease and cancer were both found to be significantly regulated, which could point to involvement of Wnt/β-catenin signaling. On assessing the involvement of “canonical pathways,” it was found that “Wnt/β-catenin signaling” was also significantly regulated (P < 0.003) in both the early-secretory and the mid-secretory endometrium compared with the proliferative endometrium.

Because of our interest in the Wnt/β-catenin signaling pathway, we decided to perform a more thorough pathway analysis for either downstream targets or integral parts of the Wnt/β-catenin signaling pathway⁸

(refs. 34, 35; Supplementary Table S1). On performing this analysis, a large number of differentially expressed genes (n = 228) were recognized. Because our investigations center on the mechanism through which progesterone counteracts the proliferative effects of E₂ during the menstrual cycle and during early endometrial carcinogenesis, we have focused on the progesterone-induced Wnt/β-catenin signal inhibitor DKK1 (36) and the alleged inhibitor FOXO1 (refs. 37, 38; Fig. 2).

To investigate the molecular mechanisms underlying progesterone-driven inhibition of Wnt/β-catenin signaling, we used the well-differentiated Ishikawa endometrial cancer cell line (39). As shown in Fig. 3A, TOP/FOPflash reporter assay analysis indicates that Wnt/β-catenin signaling is constitutively active in Ishikawa cells. Furthermore, we could clearly show that Wnt/β-catenin signaling is inhibited on MPA treatment (Fig. 3A). The concentration of MPA necessary to fully inhibit Wnt/β-catenin signaling is ∼0.1 nmol/L, and this is close to the reported dissociation constant for both progesterone receptor isoforms (40). Using this ligand concentration (0.1 nmol/L), molecular excess of the antiprogestagen Org-31489 could fully reverse Wnt/β-catenin signaling inhibition. Furthermore, it was shown that both activated progesterone receptors (progesterone receptor A and progesterone receptor B) could inhibit Wnt/β-catenin signaling, although progesterone receptor B did so at lower MPA concentrations (Supplementary Fig. S1).

As shown in Fig. 2, the expression of the Wnt/β-catenin signaling inhibitors DKK1 and FOXO1 is elevated during the early-secretory phase of the menstrual cycle, peaks during the mid-secretory phase, and declines during the late-secretory phase (Fig. 2). Furthermore, for DKK1, direct progesterone receptor activation of the promoter region was measured (Supplementary Fig. S2). These findings could indicate that progesterone inhibits Wnt signaling by inducing the expression of DKK1 and FOXO1 during the early- and the mid-secretory phase of the menstrual cycle. As shown in Fig. 3B, transfection of Ishikawa cells with increasing amounts of DKK1 or FOXO1 expression vector (30, 100, and 300 ng/well) results in the progressive inhibition of Wnt/β-catenin signaling. When cells were transfected with DKK1 or FOXO1 in the presence of MPA (Fig. 3C), or with both DKK1 and FOXO1 together, Wnt/β-catenin signaling inhibition was even more pronounced. Notably, inhibition of Wnt/β-catenin signaling by DKK1 and FOXO1 together was comparable to inhibition by the strong dominant-negative expression vector dnTCF1 (32).

To further substantiate that induction of DKK1 and FOXO1 by progesterone is central to inhibition of Wnt/β-catenin signaling, expression of both DKK1 and FOXO1 was reduced by the use of lentiviral sequence-specific shRNAs. Five DKK1 and six FOXO1 targeting lentiviruses were used to infect Ishikawa cells. As shown in Fig. 4A, partial inhibition of progesterone induction of DKK1 and FOXO1 expression was achieved by infection with combinations of all available specific DKK1 or FOXO1 shRNAs. Based on these results, Ishikawa cells infected with either DKK1- or FOXO1-specific shRNAs were then used to measure progesterone inhibition of Wnt signaling. MPA inhibition of TOPflash Wnt/β-catenin reporter system was partly circumvented by the lentiviral shRNAs directed against DKK1 and FOXO1 (Fig. 4B). Furthermore, prevention of MPA activity was observed when DKK1 and FOXO1 shRNAs were combined. In conclusion, progesterone-induced up-regulation of DKK1 and FOXO1 is responsible, in part, for Wnt/β-catenin signaling inhibition in the endometrium.

It has been reported that at least 30% of early estrogen-associated tumors (type I endometrioid) display the hallmark of Wnt signaling activation, namely, nuclear β-catenin accumulation (9, 13). We performed immunohistochemistry on histologic sections from 10 well-differentiated endometrium carcinomas, and 3 of these showed nuclear β-catenin staining. To confirm that intracellular β-catenin accumulation was accompanied by up-regulation of known Wnt/β-catenin downstream targets, CD44 immunohistochemical analysis was also done on these three nuclear β-catenin–positive tumors (41). In Fig. 5A, squamous morules are seen staining strongly for nuclear β-catenin, and CD44 is shown in the same areas. Squamous morules occur commonly in endometrial carcinoma and have been documented to be positive for nuclear β-catenin as well as its upstream stimulating transcription factor, TCF4 (42–44). Moreover, in regions of the tissue where β-catenin staining is present but not nuclear, CD44 staining is low (Fig. 5A, arrowheads). The other two endometrial cancer samples studied also showed high CD44 staining in those areas of the adenocarcinoma showing nuclear β-catenin accumulation (data not shown).

In agreement with our hypothesis that E₂ induces Wnt/β-catenin signaling to trigger proliferation of the endometrium, it was observed that expression of the Wnt target gene CD44 in the endometrium of the seven E₂-treated women was higher than that in the six E₂ + MPA–treated women (Fig. 5B). Furthermore, when CD44 expression was measured during hyperplasia, CD44 expression became markedly reduced on treatment with progestagens (Supplementary Fig. S3).

This particular patient underwent hysterectomy because of the development of a well-differentiated endometrial carcinoma. Immunohistochemical staining for CD44 in this tumor showed a heterogeneous staining pattern. In some fields, it was clear that staining for CD44 was positive in areas of carcinoma, whereas adjacent areas of hyperplasia were negative (Fig. 6A). Endometrial tumors often progress to become progestagen independent, and therefore progesterone receptor expression was also assessed by immunohistochemistry and compared with CD44 staining (Fig. 6B). Epithelial CD44 staining was minimal in areas where expression of the progesterone receptor was positive (Fig 6B, filled arrows). Conversely, in regions devoid of progesterone receptor signaling, CD44 levels were clearly enhanced (Fig 6B, open arrows). To further substantiate our findings, we reviewed 10 cases of well-differentiated endometrial cancer. In all these cases, when the progesterone receptor was present, CD44 levels were low or undetectable, and in those regions of the tumor where the progesterone receptor was lost, CD44 expression was high (Supplementary Fig. S4). These data suggest that Wnt/β-catenin signaling can be inhibited by progesterone in estrogen-associated endometrial hyperplasia and cancer, but only in the presence of progesterone receptors (Figs. 5 and 6).

Sustained or unopposed estrogen signaling in the endometrium results in hyperplasia (45), which is followed in ∼25% of cases by tumor formation (2). Two groups of women are known to be at high risk for endometrial cancer because of sustained or unopposed estrogen-like activity: those who are significantly overweight (ref. 46; relative risk >4) and those receiving tamoxifen (a partial estrogen antagonist) for breast cancer treatment (ref. 47; 5-year usage relative risk >7). It has been estimated that up to 40% of all endometrial cancers may be related to obesity, a figure that is increasing as the prevalence of obesity increases.

It has been reported that at least 30% of these early estrogen-associated tumors (type I endometrioid) display the hallmark of Wnt/β-catenin signaling activation, namely, nuclear β-catenin accumulation (9, 13), suggesting a cause-effect relationship between sustained estrogen signaling and aberrant Wnt/β-catenin signaling. Accordingly, we found a significant number of targets and components of the Wnt/β-catenin signaling pathway among the genes differentially regulated in the endometrium of E₂-treated women (Supplementary Table S1). This is indicative of the effect of estrogen signaling on this pathway. For example, the Wnt/β-catenin signaling target IGF-I, regarded as one of the most important endometrial growth factors (48), was found to be strongly up-regulated (12-fold) by E₂ and inhibited after the addition of progesterone. In literature, there are also a number of reports implicating progesterone and E₂ in regulation of Wnt/β-catenin signaling in the endometrium. Cloke et al. (49), for example, showed that knockdown of the progesterone receptor resulted in activation of Wnt/β-catenin signaling in differentiating human endometrial stromal cells. In agreement with this, Satterfield et al. (50) could show that progestagens, administered to pregnant sheep, induced a transient decline in Wnt/β-catenin signaling activity. Seemingly in contrast to these finding, Rider et al. (51) showed that estrogens activated Wnt/β-catenin signaling in stromal cells, but only after initiation by progestagens. Finally, Katayama et al. (52) showed that administration of an estrogen agonist to immature female rats resulted in down-regulation of Wnt7A and up-regulation of Wnt4 in the uterus, whereas Catalano et al. (53) showed up-regulation of Wnt5A in the endometrium by antiprogestagen treatment in healthy volunteers. These reports and our own observations led us to hypothesize that Wnt/β-catenin signaling is stimulated by estrogens and inhibited by progestagens during the menstrual cycle.

To investigate steroid regulation of Wnt signaling during the menstrual cycle, the possible involvement of the progesterone-induced Wnt/β-catenin signaling inhibitors DKK1 and FOXO1 was investigated by use of the well-differentiated endometrial cancer cell line Ishikawa. For DKK1, there is good evidence that this protein can bind to the Wnt coreceptors LRP5 and LRP6, thus inhibiting Wnt/β-catenin signaling (54). For FOXO1, the evidence is scarcer. Essers et al. (55) reported that FOXO1 directly binds to β-catenin, and Almelda et al. (38) could show that this binding results in inhibition of Wnt/β-catenin signaling. In the endometrial cancer cell line Ishikawa, where we could show that Wnt/β-catenin signaling was constitutively active, progesterone effectively inhibited Wnt/β-catenin reporter activity (Fig. 3A). Furthermore, on progesterone treatment, there was up-regulation of DKK1 and FOXO1 expression (Fig. 4; ref. 31), an observation in keeping with the profound up-regulation of the expression of both genes during the secretory phase of the menstrual cycle (Fig. 2; ref. 20). Interestingly, Tulac et al. (20), as well as Kane et al. (56), show that progesterone induced DKK1 expression specifically in stromal cells of the endometrium. In analogy to Wnt5A, which is produced by stromal cells and acts on uterine glands (18), it is possible that in vivo stroma-produced DKK1 is specifically important in inhibiting Wnt/β-catenin signaling in glandular and luminal epithelial cells of the endometrium.

On transfection of Ishikawa cells with vectors encoding either DKK1 or FOXO1, Wnt/β-catenin signaling was clearly inhibited, thus indicating that both proteins can serve as Wnt/β-catenin signal inhibitors in response to progesterone. Indeed, inhibition of Wnt/β-catenin signaling by progesterone treatment was partly circumvented by shRNA-driven down-regulation of DKK1 or FOXO1 expression.

By using the Wnt/β-catenin downstream target CD44 as a marker for Wnt signaling, it was shown that E₂ treatment seems to result in enhanced Wnt/β-catenin signaling. CD44, hyaluronic acid receptor, in this respect is an interesting marker because it has been implicated in the progression and metastasis of a number of different cancers (57). E₂-induced CD44 expression was counteracted (compensated for) by MPA treatment (Fig. 6B). Furthermore, estrogen-induced endometrial hyperplasia (45) also displayed increased CD44 expression, which was shown to be inhibited by progestagen treatment (Supplementary Fig. S3). Notably, in an endometrial carcinoma arising during treatment of endometrial hyperplasia with progestagen, areas with enhanced Wnt/β-catenin signaling (CD44 positive) generally corresponded to areas of carcinoma, whereas regions that displayed reduced Wnt/β-catenin signaling (CD44 negative) tended to correspond to areas with the morphology of hyperplasia. Staining the same regions for progesterone receptor expression indicated that the presence of progesterone receptors correlated with a decrease in Wnt/β-catenin signaling, whereas its absence coincided with enhanced Wnt/β-catenin signaling. Similar observations were reported in other well-differentiated endometrial tumors (Supplementary Fig. S4) and several years ago by Hanekamp et al. (58), who could also show a clear correlation between the absence of progesterone receptors and the presence of CD44.

In endometrial hyperplasia and cancer, we have thus observed that when progesterone receptor signaling is intact, progesterone indeed seems to be able to inhibit Wnt signaling (inhibit CD44 expression). This inhibitory effect of progesterone most likely limits cancer progression, which is observed in 15% to 25% of patients with metastatic endometrial cancer who respond to progesterone therapy (3, 59). Progesterone treatment, however, only renders a temporary relief, and the disease often becomes progesterone insensitive (59). Based on the findings of the present study, it seems warranted to consider treatment of endometrial cancer patients whose disease has become progesterone resistant with inhibitors of the Wnt/β-catenin signaling pathway (60) that are acting downstream of progesterone signaling. Endostatin has been shown to inhibit Wnt/β-catenin signaling through inhibition of LEF1 (61). LEF1 is an important part of the Wnt/β-catenin pathway, which we have shown to be up-regulated by estrogens during the proliferative phase of the menstrual cycle and down-regulated by progestagens during the secretory phase of the menstrual cycle (ref. 25; Supplementary Table S1). Endostatin may thus be a potential therapeutic option in progesterone-insensitive metastatic endometrial cancer.

In summary, our results provide support for the hypothesis that during the proliferative phase of the menstrual cycle, increased E₂ levels induce Wnt/β-catenin signaling to enhance proliferation, whereas during the secretory phase, progesterone levels inhibit Wnt/β-catenin signaling thereby counterbalancing E₂-induced proliferation and enhancing differentiation. Furthermore, progesterone also seems to be able to inhibit Wnt/β-catenin signaling during endometrial carcinogenesis, thus inhibiting the disease. In addition, we provide mechanistic evidence that the Wnt/β-catenin signaling inhibitors DKK1 and FOXO1 are acting downstream of the progesterone receptor to trigger inhibition of Wnt/β-catenin signaling.

No potential conflicts of interest were disclosed.

We thank Drs. X. He and H. Clevers for providing us with the DKK1 and dnTCF expression vectors, and the group of Dr. L.C. Guidice for sharing their gene expression data with the scientific community through PubMed Geo DataSet.