Background
The stromal/mesenchymal compartment of the endometrium performs a variety of tasks important for uterine physiology, including relaying specific aspects of steroid hormone signaling to the overlying epithelium. An example of such mesenchymal-to-epithelial signaling occurs in response to estradiol (E
2) binding to and activating estrogen receptor (ESR1), inducing the expression of stromal-derived growth factors that stimulate epithelial cell cycle progression, hypertrophy, and initiating secretory functions (reviewed in [
1]). In invasively implanting species, the stroma also undergoes decidualization during early pregnancy following embryo apposition and attachment to the uterine luminal epithelium, a process inherently regulated by progesterone (P
4) following E
2 priming. Here, stromal cells terminally differentiate and contribute to pregnancy by performing placenta-like functions until such time that the embryo develops its own nutrient and gas exchange apparatus, the placenta [
2]. Stromal decidualization is regulated, in part, by cues derived from the epithelium such as Indian hedgehog.
It is thought that ESR1 mediates E
2-initiated signaling in the uterus. However, it is generally understood that E
2-initiated transcriptional and physiological changes occur in two phases [
3]. The first occurs within 2–6 hours, and the second takes place between 24–72 hours. Although many E
2-initiated transcriptional events require binding of ESR1 to estrogen response elements (ERE), many other genes are regulated in an ER-dependent, but ERE-independent fashion [
4]. This suggests that ESR1 interacts with other transcriptional modulators that in turn interact with DNA to regulate gene expression at promoter sites distinct from EREs. Within the uterine epithelium, one such ESR1 interacting molecule is the transcriptional co-activator β-catenin [
5,
6]. The late transcriptional response to E
2 is thought to be mediated, in part, by the ESR1:β-catenin interaction. Equally complex signaling mechanisms likely coordinate P
4 responses, but such pathways are less clearly understood.
β-catenin is best known for its central role in the canonical wingless-type MMTV integration site family member (Wnt) signaling pathways and β-catenin is essential for development, transcription, cell adhesion and tumorigenesis [
7]. In the absence of Wnt signaling, β-catenin is found in the cytoplasm either as a component that binds cadherins to α-catenin and the cytoskeleton at adherens junctions or in a complex with adenomatous polyposis coli (APC), axin, and glycogen synthase kinase 3β (GSK-3β), wherein it is phosphorylated and subject to ubiquitination and proteasomal degradation. Activation of frizzled receptors by Wnt ligands disrupts the APC complex and inhibits GSK-3β activity causing an accumulation of unphosphorylated (
i.e., activated) β-catenin, which promotes its nuclear localization and subsequent regulation of target gene expression [
8]. β-catenin is therefore uniquely situated at a bottleneck in the Wnt signaling pathway.
Much of the focus of steroid hormone signaling studies in the uterus has been directed at the epithelial compartment. In the present study, the function of β-catenin in the stromal compartment was investigated in the contexts of steroid hormone action and stromal cell decidualization. Our findings reveal that conditional inactivation of β-catenin in endometrial stroma results in disrupted progesterone signaling and complete loss of stromal cell decidua-lization, indicating that steroid-dependent and β-catenin signaling pathways intersect to regulate postnatal uterine functions.
Discussion
Adult endometrial functions are temporally regulated by sex steroid hormones that require interplay between the epithelial and underlying stromal compartments. Ovarian-derived E
2 generated during each estrous/menstrual cycle stimulates epithelial cell proliferation. The proliferative epithelial response to E
2 is largely an indirect event that involves stromal release of epithelial mitogens such as IGF-1 [
22,
23]. It was recently established through conditional mutagenesis studies that stromal-derived ESR1 is fundamental for directing epithelial cell proliferation, while epithelial ESR1 is dispensable [
24]. In turn, ovarian P
4 completely abolishes E
2-induced epithelial cell proliferation
in vivo[
25]. Clinically, P
4 is applied prophylactically in some settings to treat estrogen-dependent endometrial cancer and to alleviate potential complications during hormone replacement therapies that can arise due to the unopposed actions of estrogens. These fundamental actions of E
2 and P
4 within the endometrium are further validated in pharmacological studies where steroid hormone actions are attenuated, as well as through the use of mutant mice deficient in expression of ESR1 and PGR.
Amhr2
Cre/+
;Ctnnb1
d/d
mice are infertile, which engenders two previously unappreciated points for consideration. First, that deletion of
Ctnnb1 from the stromal, but not epithelial, compartment results in failed decidualization, suggests that this transcriptional co-activator mediates steroid hormone actions in the endometrium that are critical for fertility. Further investigation is needed to determine if β-catenin interacts in parallel with PGR, forming a complex that in turn regulates expression of genes in stromal tissue whose encoded products contribute to decidualization. Precedence for the convergence of β-catenin and steroid hormone signaling pathways has been established in the uterus. Alternatively, the PGR and β-catenin signaling pathways may work in series where PGR results in activation of another pathway, such as WNTs that in turn utilize β-catenin function. This scenario is supported by recent findings where WNT4 was shown to be a key regulator of normal postnatal uterine development and progesterone signaling during embryo implantation and decidualization [
26]. Additional evidence for a PGR-β-catenin interaction comes from
in vitro decidualization studies using human stromal cells where PGR expression was shown to be essential for nuclear translocation of β-catenin [
27].
The second point for consideration is that stromal β-catenin is necessary for transcriptional regulation of both stromal and epithelial factors that are important for initiating decidualization and embryo attachment. Stromal β-catenin-deficiency results in failed up-regulation of
Ihh in the epithelium, as well as
Ptch1 and
Gli1 in the stroma suggesting that stromal P
4 signaling mediates events not only in the stromal compartment, but also in the overlying epithelium. It is concluded from this investigation that stromal β-catenin is a component of the signaling conduit through which P
4 coordinates events in the overlying epithelium. Recent tissue recombination studies involving the use of wild type and
Pgr-null stroma and/or epithelia support this concept [
28]. Here, Simon
et al. established that neonatal tissue recombinants containing wild type epithelium and PGR-deficient stroma were unable to show elevated levels if
Ihh in the epithelium in response to P
4 treatment [
28].
Some studies have suggested direct inhibitory actions of P
4 on E
2-induced epithelial cell proliferation. During the time of embryo implantation on day 4 of pregnancy in mice the epithelium does not express PGR despite observation of clear progestational response on the epithelium [
29,
30]. How then does P
4 signal in the epithelium in the absence of PGRs? The “progestamedin hypothesis” suggests that P
4-induced paracrine factors secreted from the stromal compartment indirectly regulate P
4 actions on the epithelium [
30]. It was recently established that E
2-induced epithelial proliferation is suppressed by P
4 actions in the stromal compartment involving a HAND2-dependent mechanism [
31]. Progesterone induces the transcriptional inhibitor HAND2, which in turn suppresses specific members of the fibroblast growth factors family in the stromal compartment [
31]. Our study places β-catenin squarely in the middle of P
4-dependent mesenchymal-to-epithelial signaling during the initiation of maternal:embryo interaction.
A number of signaling factors and down-stream transcription factors have been identified through mutant mouse studies as critical components coordinating decidualization. Some of these include IHH, WNT4, HOXA10, HOXA11, Src-kinase, BMP2 and COUP-TFII reviewed in [
32]. Indian hedgehog localizes to the epithelium in response to P
4 at the time of embryo implantation, and tissue restricted deletion of the gene using the
PgrR-Cre mouse model results in failed decidualization [
33‐
35]. From our study it is clear that transcription of members of the IHH pathway is reduced in
Amhr2
Cre/+
;Ctnnb1
d/d
uteri in response to steroid hormones; however, additional functional studies are necessary to determine exactly how β-catenin is linked to the IHH signaling pathway.
Uteri from
Amhr2
Cre/+
;Ctnnb1
d/d
mice are smaller in size than control uteri, which could confound the interpretation of these results. However, four lines of evidence suggest that the failure of
Amhr2
Cre/+
;Ctnnb1
d/d
uteri to decidualize stems from disruption of steroid hormone receptor signaling rather than from altered prenatal or early postnatal uterine development. First, expression studies reveal that uteri from
Amhr2
Cre/+
;Ctnnb1
d/d
mice have the potential to respond normally to E
2 and P
4 in that uterine mRNA and protein levels of ESR1 and PGR do not differ between control and mutant female mice. Second, uteri from control and mutant female mice display a normal response to E
2, at least in terms of epithelial proliferation and stromal imbibition. Since the stromal compartment mediates the proliferative response in the epithelium, our findings indicate that the uterine stromal compartment of
Amhr2
Cre/+
;Ctnnb1
d/d
mice is fully capable of disseminating proliferative signals to the epithelium. Third, the actions of P
4 are not completely ablated in the uteri of
Amhr2
Cre/+
;Ctnnb1
d/d
mice, since several genes previously shown to be targets of P
4 action show the expected pattern of expression. For instance,
Hmga2 (high mobility group AT-hook 2),
Cdkl1 (cyclin-dependent kinase-like 1), and
Ldb2 (LIM domain binding 2) were shown to be down-regulated by P
4 treatment
in vivo[
36]. Based on our microarray analysis each of these genes was down-regulated similarly in control
Ctnnb1
flox/flox
and mutant
Amhr2
Cre/+
;Ctnnb1
d/d
uteri
in vivo (data not shown). Conversely,
S100a6 (calcyclin),
Irg-1 (immune responsive gene-1) and
Fst (follistatin), three genes shown to be up-regulated by P
4[
37], were equitably up-regulated in
Ctnnb1
flox/flox
and
Amhr2
Cre/+
;Ctnnb1
d/d
uteri (data not shown). Fourth, indifferent stromal cell proliferation was observed in response to a hormone regimen consistent with early pregnancy. This suggests that the proliferative stromal cell response to P
4 is not dependent upon β-catenin. In sum, these findings indicate that β-catenin deficiency in the stromal compartment of
Amhr2
Cre/+
;Ctnnb1
d/d
uteri results in aberrant gene expression of a specific cassette of P
4-dependent genes, several of which belong to the IHH signaling cascade, but that other P
4 responses are normal.
Two functional studies were previously published on β-catenin in the uterus. In the first, β-catenin activity was indirectly assessed through the use of
Tcf/Lef-LacZ transgenic mice [
38]. Here, β-galactosidase activity was used to identify coupling of β-catenin with the TCF/LEF transcriptional complex
in situ. Based on this model, β-catenin activity was observed in the luminal epithelium and circular smooth muscle, an event that required the presence of an embryo. It was concluded that β-catenin was no longer active by late DOP5. However, β-catenin activity was defined by its ability to activate the
Tcf/Lef-LacZ transgene, and β-catenin function was not addressed using deletional analysis (
e.g., gene knockdown or mutant mice deficient in β-catenin). Additionally, while we and others [
14] have since demonstrated the presence of active (
i.e., dephosphorylated and nuclear) β-catenin in decidualizing stromal cells, Mohamed
et al. were unable to detect transcriptional activity for the TCF/LEF complex in the stromal compartment, suggesting β-catenin may regulate gene expression within the stromal compartment by a TCF/LEF-independent mechanism.
More recently, Jeong
et al. used the
Pgr-Cre transgenic mouse line to delete β-catenin from PGR-expressing tissues, including all compartments of the uterus [
21]. Using this model system, β-catenin deficiency in the entire uterus resulted in pleiotropic effects leading to infertility, most likely because of the inability of stromal cells to terminally differentiate and E
2-induced morphological defects. The design, and therefore the conclusion, of our study differ to some extent from this previous report. First, while uteri from
Amhr2
Cre/+
;Ctnnb1
d/d
female mice lack expression of β-catenin in the myometrial and stromal compartments, as with the
Pgr-Cre model, expression of β-catenin was retained in luminal and glandular epithelia using the
Amhr2
Cre
mouse line. Second, deletion of β-catenin in all compartments of the uterus resulted in metaplastic formation of the luminal epithelium in the intact mouse [
21]. Analysis of the ovaries indicated that ovarian function was preserved. Jeong
et al., concluded that β-catenin deficiency in the epithelium was the source of the metaplastic phenotype. This conclusion is well justified in that mutations in the human
Ctnnb1 gene are commonly associated with endometrial hyperplasia. Although our data does not rule out control of epithelial metaplasia by epithelial β-catenin, they indicate that, since
Amhr2
Cre/+
;Ctnnb1
d/d
mutant uteri also develop metaplasia,
albeit with reduced severity and incidence, the lack of β-catenin in stroma alone can dictate formation of epithelial metaplasia. As with β-catenin, deletion of APC, a component of the β-catenin signaling pathway, from the uterine stromal compartment results in a more severe phenotype where endometrial hyperplasia and carcinogenesis are observed [
39].
Competing interests
None of the authors have competing interests.
Authors’ contributions
LZ completed most of the IHC and RT-PCR and some of the decidualization and histology experiments. AP completed experiments centered on progesterone and estrogen signaling, proliferative responses, receptor expression analyses, as well as some of the decidualization experiments. She also participated in drafting the manuscript. LZ assisted with IHC and animal husbandry. JT provided mice for these experiments and contributed to the experimental design and writing. JP was involved in all aspects of these studies and drafted the manuscript. All authors participated in editing and revising the manuscript. All authors read and approved the final manuscript.