1.1. Canonical Wnt pathway

The central player in canonical Wnt signalling is β-catenin. In unstimulated cells β-catenin is mainly located to adherens junctions where it is critically involved in maintaining epithelial integrity by binding to E-cadherin and α-catenin. In the absence of Wnt ligands (off-state) excess, cytoplasmic β-catenin is complexed with APC (adenomatous polyposis coli) and Axin both facilitating the phosphorylation of the protein by casein kinase Iα (CKIα) and glycogen synthase kinase 3β (GSK-3β) (Fig. 1). This provokes degradation of β-catenin through the β-TrCP (β-transducin repeat-containing protein) mediated ubiquitin/proteasome pathway resulting in low cytosolic levels [16,17]. Binding of a Wnt ligand (on-state) to the cysteine-rich domain (CRD) of FZD promotes FZD–LRP heterodimerisation triggering a series of events that disrupt the Axin/APC/GSK-3β/CK1α destruction complex [12]. In detail, Wnt stimulation induces recruitment of Dishevelled (Dsh) to the FZD receptor forming a so called signalosome [18]. Moreover Axin, a key negative regulator of β-catenin stability, translocates to the cytoplasmatic tail of LRP catalysed by CK1γ- and GSK-3β-dependent phosphorylation of the FZD co-receptor [19,20]. Sequestration of a critical component (Axin) of the destruction complex and activation of Dsh finally result in impaired degradation and accumulation of β-catenin in the cytosol [21]. Active β-catenin then translocates into the nucleus where it functions as a transcriptional co-regulator [4,12]. It displaces transcriptional inhibitors of the Groucho protein family and histone deacetylases (HDACs) from the T cell-specific factors (TCFs)/lymphoid enhancer-binding factor 1 (LEF-1) and recruits histone acetylases, the Legless family docking proteins (Bcl9) and CBP/p300 thereby converting TFC/LEF-1 into transcriptional activators [12]. Axin, APC and other Wnt components can also enter the nucleus, thereby modulating nuclear trafficking and transcription [22]. For example, APC was suggested to play a critical role in the exchange of co-activator and co-repressor complexes at Wnt target genes [22]. These include genes involved in cell proliferation and migration such as c-myc, c-jun, cyclin D1, CD44, matrilysin, matrix metalloproteinases (MMPs) and urokinase plasminogen activator receptor (uPAR) as well as others summarised at the Wnt homepage (http://www.stanford.edu/~rnusse/wntwindow.html). In addition, TCF/β-catenin dependent expression of Axin-2, FZDs, TCF-1 and other Wnt pathway components was noticed indicating that feedback control is also a feature of Wnt signalling. In addition, the canonical Wnt pathway can be negatively affected by endogenous β-catenin inhibitors such as inhibitor of beta-catenin (ICAT) and Chibby or soluble inhibitors such as secreted frizzled-related protein (sFRPs) or members of the Dickkopf (Dkk) family [12]. The latter bind to LRP and induce receptor internalisation leading to downregulation of Wnt signalling [23].

Open in a separate window

Fig. 1

The canonical Wnt/β-catenin pathway. Off-state; in the absence of a Wnt ligand, β-catenin is bound in a multiprotein degradation complex containing the scaffold protein Axin, the tumour suppressor gene product APC, as well as the kinases CKI and GSK-3β, among others. Upon phosphorylation, β-catenin is ubiquitinated by the β-TrCP–E3-ligase complex and subsequently degraded by the proteasomes. On-state; Wnt ligand associates with FZD and LRP-5/6 co-receptors. This leads to the translocation of Axin to the plasma membrane through direct interaction with LRP-5/6 and Dsh/FZD. β-catenin is released from the multiprotein complex, accumulates in the cytoplasm in a non-phosphorylated form, and subsequently translocates into the nucleus where it promotes transcription of Wnt target genes upon binding to TCF/LEF.

2.1. Uterine development, pre-implantation and decidualisation

In vivo studies using different mouse models support the idea that Wnt signalling is crucially involved in uterine growth and development. Estrogen, for example, induces Wnt4, Wnt5a and FZD2 in the mouse uterus as well as nuclear recruitment of β-catenin in the endometrial epithelium [41]. Inhibition of canonical Wnt signalling trough adenovirus-mediated expression of sFRP2 inhibited estrogen-dependent activation of β-catenin and epithelial cell proliferation suggesting that steroid hormone-dependent uterine cell growth also involves Wnt signalling [41]. Moreover, estrogen was also shown to upregulate the β-catenin-dependent transcription factors TCF-3 and LEF-1 in an estrogen receptor-independent manner [42]. Interestingly, the hormone also provokes physical interaction of ERα with activated LEF-1/TCF-3 and recruits the latter to Wnt/estrogen-dependent target genes suggesting that cross-talk between estrogen and Wnts could be critical for endometrial function [42]. Moreover, progesterone was shown to downregulate GSK-3β in rat uteri, which was suggested as a prerequisite for estrogen-dependent activation of the canonical Wnt pathway [43].

Knock-out studies demonstrated that Wnt4- and Wnt5a-deficient mice display failures in the development of the female reproductive tract [44,45]. Mice lacking Wnt7a do not develop uterine glands and have disorganised uterine smooth muscles [46]. Interestingly, Wnt7a was also identified as a key regulator of female Müllerian duct development which evolves into oviduct, uterus and cervix [47]. In males, Wolffian ducts develop and Sertoli cells of the testes secrete Müllerian-inhibiting substance provoking Müllerian duct regression. In Wnt7a-deficient male mice, however, Müllerian ducts do not regress due to the absence of the receptor for Müllerian-inhibiting substance resulting in generation of pseudohermaphrodites [47]. Wnt7a knock-out mice also lose expression of HoxA10 and HoxA11 in the endometrial stroma, two genes which upon homozygous deletion in mice provoke failures in the decidualisation process resulting in infertility [46].

Canonical Wnt signalling was also shown to be critical for the development of Müllerian duct derivatives. Mice harbouring a conditional deletion of β-catenin in the Müllerian duct mesenchyme display abnormal uteri after birth and largely lack myometrial smooth muscles upon adulthood [48]. Finally, FZD4 knock-out mice were shown to be infertile since formation of the corpus luteum is impaired. Expression of luteal cell-specific genes such as sFRP4 and the LH/hCG receptor was found to be reduced in these mice [49].

With respect to reproduction different FZD receptors (FZD2, FZD4 and FZD6) and Wnt ligands (Wnt4, Wnt5a, Wnt7a, Wnt7b, Wnt11, Wnt16) were shown to be expressed in the mouse uterus before and around the time of implantation as well as during stromal cell differentiation [50]. Expression of most of these genes was highest at implantation (day 5 after fertilisation). Wnt4 and Wnt7b, for example, strongly increase in stromal cells around the implanting blastocyst and in the luminal epithelium, respectively, suggesting specific roles associated with the implantation process [50]. Indeed, Wnt4 has been identified as a critical gene controlling murine as well as human uterine decidualisation since siRNA-mediated silencing of its mRNA impaired the differentiation process [51]. Wnt4 is a target gene of the key regulator BMP2 which is induced upon progesterone treatment of uterine stromal cells [51].

Other Wnts and FZDs, however, also undergo dynamic changes upon implantation and decidualisation and were shown to be regulated by steroid hormones in ovariectomised mice [41,50]. Expression levels of Wnt7a, Wnt7b, Wnt11 and Wnt16 which are significantly higher in uterine implantations sites compared to non-implantation sites decrease upon stromal cell differentiation (days 6 and 7 after fertilisation) [50]. Moreover, non-canonical Wnt receptors such as Ror-2 are expressed in endometrial epithelial and stromal cells of non-pregnant mice [52]. During pregnancy stromal expression of Ror-2 increased and also appeared in uterine NK cells which may suggest a role of the receptor in implantation and regulation of trophoblast invasion, respectively [52].

2.2. Blastocyst development and implantation

Despite the facts that various Wnts, such as Wnt5a and Wnt11 [53] and FZDs are detectable in the pre-implanting embryo, different approaches demonstrated that canonical Wnt signalling is dispensable for blastocyst formation. Gene knock-out studies showed that mutant embryos lacking β-catenin develop into blastocysts [54]. However, maternal delivery of β-catenin might have compensated for lack of the zygotic protein during the pre-implantation period. Hence, mice harbouring a conditional deletion of β-catenin in their oocytes were also utilised demonstrating that lack of both maternal and zygotic β-catenin does not impair blastocyst development [55]. In addition, inhibition of the TCF/β-catenin complex through small molecular inhibitors or overexpression of Dkk1 did not negatively affect blastocyst formation [56]. Hence, it is likely that the diverse zygotic Wnts and FZDs expressed operate through non-canonical signalling such as the Ca²⁺-dependent pathway which was shown to be necessary for pre-implantation development [40].

Whereas β-catenin might be dispensable during the pre-implantation period, blastocyst activation critically involves the canonical Wnt signalling pathway. Upon mating of mice with conditional deletion of β-catenin in oocytes with wildtype male fewer embryos develop as compared to matings with normal mice [55]. In agreement to that, blocking of canonical Wnt signalling through Dkk1 or small molecular inhibitors impairs blastocyst’s competency to implant [56]. Along those lines, a shift from non-canonical signalling in the pre-implantation blastocyst towards canonical signalling in the trophectoderm of activated blastocyst could be demonstrated [56].

However, canonical Wnt signalling is not only required for blastocyst activation but also for successful implantation. Indeed, blastocyst attachment was shown to induce TCF/β-catenin-dependent signalling in circular smooth muscle cells of the myometrium and subsequently in the uterine epithelium at the site of implantation [57]. This might involve the blastocyst-derived Wnt7a which upon uterine intraluminal delivery also activates the pathway in the absence of a zygote [57]. Inhibition of the pathway upon administration of sFRP2 decreased the frequency of implantation [57].

3.1. Endometrium, decidualisation and endometrial diseases

Various investigators identified ligands as well as different Wnt signalling components in the endometrium suggesting that the pathway could be involved in the diverse biological roles of human uterine cell types. Wnt2, 3, 4, 5a, 7a, 8b and FZD1, 4, 6 and 10 mRNAs were detected in endometrial samples and endometrial epithelial/stromal cells using different molecular techniques [69,70]. Hormonal control of Wnt2, 3, 4, 5a upon estrogen or progesterone treatment in vitro could not be detected, however, lack of expression was noticed in different endometrial carcinoma cell lines [69]. Wnt7a, the regulator of HoxA10 and HoxA11 [71], was found to be exclusively expressed on luminal epithelial cells, whereas sFRP4, Dkk1 and FZD were detectable in uterine glands and/or stroma [70].

Microarray analyses aiming to analyse global gene expression during the menstrual cycle also revealed expression of mRNAs of the Wnt pathway as well as their potential regulation through steroid hormones. Dkk1 and Wnt10b mRNAs, for example, were shown to strongly increase between early and mid-luteal phases [72]. In contrast, sFRP1 and sFRP4 were found to be downregulated in endometrial samples taken at the time of the LH surge [73,74], suggesting that the decline of inhibitors of Wnt signalling could be important for human implantation and/or decidual differentiation. Indeed, sFRP4 could be involved in endometrial cell proliferation since elevated levels of the inhibitor were noticed in estrogen-dependent endometrial and breast cancer [75]. Similarly, sFRP1 expression was found to be associated with endometrial cell growth and elevated mRNA levels were detected in endometriotic tissues [76]. Since sFRP1 is known to induce angiogenesis, increased expression could be a critical factor in endometriotic cell proliferation [76]. Others, however, found a decrease of sFRP1 [77] and sFRP4 [78] in cultivated endometriotic stromal cells and endometrial carcinoma cells, respectively, and overexpression of sFRP4 decreased cancer cell proliferation in vitro [78]. Finally, treatment of women with the antiprogestin RU486 and subsequent DNA chip analyses revealed elevated expression of sFRP1, Wnt5a, FZD4, 6, 9, 10, β-catenin and Axin-2 [79]. Since RU486 rapidly provokes endometrial breakdown a role of the Wnt pathway during menstruation and/or endometrial repair has been suggested [79].

Whereas Wnt3 was found to be elevated in the proliferative endometrium, Dkk1 increased in the mid-secretory phase and upon in vitro decidualisation of endometrial stromal cells suggesting a possible role in endometrial differentiation and/or implantation [70,73]. Indeed, Dkk1 was shown to be specifically upregulated by progesterone in human endometrial stromal cells, whereas estrogen or cAMP treatment was not effective [80], and progesterone receptor knock-down decreased expression of the inhibitor [81]. Since progesterone-dependent induction of Dkk1 also inhibited Wnt signalling [82], it is likely that repression of the pathway plays a role in decidualisation. Along those lines, TGF-β1, which could play a role in menstruation, was shown to impair Dkk1 expression in decidualising endometrial stromal cells [83]. In addition, Dkk1 mRNA expression is lower in endometriotic fibroblasts compared to normal endometrial fibroblasts which might be associated with a persistent proliferative potential and impaired decidualisation capacity in endometriosis [77]. Moreover, elevated expression of Wnt7a, Wnt2 and FZD1 was observed in endometriotic tissues [84,85]. Abnormal activation of the Wnt pathway in the mid-secretory endometrium, such as persistent expression of activated β-catenin, may also occur in infertile patients with endometriosis [86].

Recent evidence suggests that estrogen-dependent proliferation may, at least partly, depend on activated Wnt signalling since the hormone was shown to induce Wnt pathway components in endometria of estrogen-treated women [82]. This would be in agreement with the fact that up to 30% of estrogen-associated cancers exhibit nuclear β-catenin expression, the hallmark of canonical Wnt signalling [81,87]. Accordingly, reduced levels of Dkk1 were noticed in endometrial carcinomas and treatment of the cells with recombinant Dkk1 inhibited invasiveness in vitro [88].

Research in the laboratory of M. Knöfler is supported by grant Nr. 12487 of the Jubiläumsfonds of the Austrian National Bank, by grant Nr. P-22687-B13 of the Austrian Science Funds and by a grant (Nr. APP00323OFF) of the Herzfelder’sche Familienstiftung.