|Home | About | Journals | Submit | Contact Us | Français|
Predicting the potential physiological outcome(s) of any given molecular pathway is complex because of crosstalk with other pathways. This is particularly evident in the case of the nuclear hormone receptor and canonical Wnt pathways which regulate cell growth and proliferation, differentiation, apoptosis and metastatic potential in numerous tissues. These pathways are known to intersect at many levels: in the intracellular space, at the membrane, in the cytoplasm and within the nucleus. The outcomes of these interactions are important in the control of stem cell differentiation and maintenance, feedback loops, and regulating oncogenic potential. The aim of this review is to demonstrate the importance of considering pathway crosstalk when predicting functional outcomes of signaling, using nuclear hormone receptor/canonical Wnt pathway crosstalk as an example.
There are 48 known members of the nuclear hormone receptor (NHR) family of transcription factors. They are generally divided into two subtypes: Type I and Type II (Type III and Type IV also exist but they largely composed of orphan receptors, about which little is known). Type I NHRs include the sex steroid receptors such as estrogen receptor α and β (ER-NR3A1 and NR3A2, respectively), progesterone receptor (PR-NR3C3) and androgen receptor (AR-NR3C4), as well as glucocorticoid receptor (GR-NR3C1), among others. These receptors are found in both the cytoplasm and the nucleus and generally bind as homodimers to their DNA recognition sequences. Type II NHRs include the vitamin D receptor (VDR-NR1I1), peroxisome proliferator activativated receptors α, β and γ (PPAR-NR1C1, NR1C2, NR1C3, respectively) and retinoic acid receptor α, β, and γ (RAR-NR1B1, NR1B2, NR1B3, respectively), among others. These receptors are constitutively nuclear and generally bind as heterodimers, most frequently with the NHR retinoid × receptor (RXR-NR2B1, NR2B2, NR2B3 for α, β, and γ, respectively), to their DNA binding elements.
In the absence of ligand the Type II receptors, recruit a large co-repressor complex. On some occasions ligand dependent repression occurs on specific regulatory sequences such as negative VDR binding elements. Once ligand enters the cell it traverses the nuclear membrane and binds to the receptor, inducing a conformational change that allows the sequential recruitment of histone modifying proteins, transcriptional mediator complexes, and, lastly, the basal transcription machinery to activating recognition sequences.
The Wnt pathway is strongly associated with an oncogenic phenotype in many cancers including cancer of the liver, breast and particularly, the colon. In the absence of canonical Wnts, the GSK3β/Axin/APC complex induces the phosphorylation of β-catenin, which targets it for degradation by the proteasome [1–3]. The presence of endocrine and paracrine Wnts induces binding of Frizzled (Fzd) receptor to Low-density lipoprotein receptor related proteins 5 or 6 (LRP5/6), which ultimately leads to the inhibition of the GSK3β/Axin/APC phosphorylation complex and accumulation of unphosphorylated β-catenin.
β-Catenin is a multifunctional protein with important structural and signaling functions. β-Catenin plays a pivotal role in cell-cell adhesion, linking the cytoplasmic domain of the cadherin family of transmembrane proteins to α-catenin and thus connecting the adhesion complex to the actin cytoskeleton. β-Catenin is also an essential component of the Wnt/Wingless signaling pathway that plays important roles in development and pathogenesis. Through this dual role, β-catenin has the potential to allow changes in cell adhesion and junction formation to affect transmembrane signaling and gene expression. The many functions of β-catenin are, derived from three cellular pools of the molecule under strict regulation: a membrane pool of cadherin-associated β-catenin, a cytoplasmic pool and a nuclear pool . β-catenin that is not targeted to the proteasome by the GSK3β/Axin/APC complex can be found in the nucleus, as ‘activated’ β-catenin. ‘Activated’ β-Catenin is a heterodimeric binding partner of LEF/TCF proteins that contain DNA-binding motifs. Together these proteins form a transcription initiation complex and regulate the expression of many genes involved in cell growth[6,7].
Stabilization of β-catenin is likely the key event in the transduction of a Wnt signal and the cell has developed an elaborate system to regulate its steady-state levels. In addition to the GSK3β/Axin/APC destruction complex, the Wnt pathway is also controlled by extracellular antagonists. Secreted Frizzled Related Protein (sFRP) and Wnt Inhibitory Factor 1 (WIF-1) are secreted factors that bind Wnts and prevent them from binding their cognate receptors. Dickkopf (DKK) family members and Wise, a context-dependent secreted protein, both antagonize the Wnt pathway by binding to LRP5/6 and preventing Wnts from binding. Dysregulation of any of these pathway components can lead to the production of over-abundant activated β-catenin and an oncogenic phenotype.
Crosstalk between these two pathways may have differing effects on oncogenic potential. Activation of β-catenin signaling is a key event in colorectal carcinogenesis where mutations in the tumor suppressor gene APC account for up to 80% of cancers  In addition, activating β-catenin mutations have been also reported in melanomas, ovarian cancer, medulloblastomas, and endometrial cancer, underscoring the importance of β-catenin-mediated signaling in oncogenesis [10–12]. The transforming potential of Wnt genes correlates precisely with their ability to induce elevated β-catenin levels, suggesting that β-catenin is instrumental in Wnt-mediated transformation . Although the aberrantly-activated Wnt pathway is almost exclusively associated with a cancer phenotype, aberrantly-activated NHR pathways have varied outcomes with respect to a cancer phenotype. The sex hormones, AR and ER, are thought to promote cancer, while others such as PPARγ and VDR are thought to prevent cancer. While methods of crosstalk are not likely to be exactly the same for each NHR, there are some commonalities. The mechanisms of crosstalk and the reported outcomes are reviewed presently in the context of both development and cancer.
Many NHRs can regulate the transcription of Wnt antagonists. This becomes clear in the context of stem cell differentiation in certain tissues where Wnt expression can drive cell fate. One of the best-characterized processes involving Wnt and NHR signals in stem cell differentiation involves the driving of mesenchymal progenitor cells into an adipocyte or an osteoblast lineage. The commitment of mesenchymal stem cells to an adipogenic or osteogenic cell fate involves an intimate level of cross regulation between the Wnt pathway and several NHRs (Figure 1A). It is generally accepted that an osteogenic cell fate is induced by Wnt/Fzd activity, while an adipogenic cell fate is induced by PPARγ activity. These two key regulators reciprocally inhibit each other on multiple levels. Wnt promotes osteogenesis by blocking the PPARγ and C/EBPα induction that would otherwise occur under adipogenic stimuli [15–17]. Orphan NHR, COUP-TFII, which is a direct target of β-catenin, and, thus, downstream of the Wnt pathway, also inhibits PPARγ expression. It binds to the PPARγ promoter and recruits SMRT, a NHR co-repressor which acts to inhibit the application of active histone marks. This effectively enhances a closed chromatin structure around the PPARγ promoter, preventing transcription. Furthermore, Cyclin D1 and c-myc, two well-documented Wnt targets, bind directly to and thereby inhibit the action of PPARγ.
Wnt-mediated osteogenesis may be controlled by a negative feedback mechanism, as Wnt7a and Fzd9 can increase PPARγ activity and E-cadherin expression in non-small cell lung cancer cells; however, this has not yet been shown in mesenchymal precursors .
Conversely, PPARγ can antagonize Wnt-mediated osteogensis via several mechanisms. In response to adipogenic stimuli, PPARγ expression is increased, and nuclear β-catenin is dramatically decreased. Furthermore, PPARγ overexpression reverses Wnt-1 induced blockade of adipogenesis. This effect may be partially mediated by several mechanisms including a context-dependent decrease in β-catenin transcription, an increase in both GSK3β-dependent and -independent β-catenin degradation, an increase in Wnt antagonist DKK1 and concomitant decrease in the LRP5/6 co-receptors, and a suppression of Wnt expression such as Wnt 10b[17,21–26].
Glucocorticoids (GC), such as dexamethasone, also antagonize the Wnt pathway during differentiation of mesenchymal stem cells and are key components in inducing adipogenic differentiation. GCs increase expression of DKK1, sFRP-1 and Axin-2[27–29] in osteoblastic precursors. In a transgenic mouse model with decreased GC expression targeted to mature osteoblasts, resultant decreased paracrine Wnt signaling from osteoblasts was ameliorated by ectopic Wnt3A expression during cranial bone development. In mammary epithelial cells GC may also antagonize the Wnt pathway via re-routing β-catenin from the nucleus to the cell membrane, possibly by affecting TGFα-dependent, GSK3β-mediated phosphorylation of β-catenin. This GC-dependent modification of differentiation down an adipogenic lineage may underlie the reversible osteoporotic side effects of long-term GC treatment of rheumatoid arthritis.
VDR may also play a role in deciding the fate of mesenchymal stem cells. VDR knockout mice demonstrate a modestly increased induction of PPARγ mRNA in response to adipogenic differentiation conditions. This modest PPARγ induction translates to dramatic derepression of late-stage adipogenic markers. The repression of adipogenesis signals in VDR-sufficient systems may be partially due to the ability of vitamin D to strongly decrease expression Wnt antagonists DKK1 and sFRP2 and as well as activate expression of LRP5 leading to osteogenesis.
Vitamin D modulates secreted Wnt antagonists in situations other than cell differentiation, namely in colorectal cancer cells, where vitamin D is known to be anti-tumorigenic. In contrast to mesenchymal stem cells, DKK1 is increased by vitamin D via an indirect mechanism in colon cancer cells. Surprisingly a vitamin D-mediated decrease in DKK4 expression in both colorectal and breast cancer cell lines was also observed. This is counter-intuitive since DKK4 is thought to antagonize the oncogenic Wnt pathway that is over-active in a vast majority of colorectal cancers. In this study, however, DKK4-enhanced malignant morphology and vascular tubule-like structures in the DLD-1 colon cancer cell line.
Retinoic Acid (RA) is a classic differentiating agent that also modulates extracellular Wnt antagonists. RA-increased DKK1 expression is at least partially responsible for RA-dependent neural differentiation of embryonic stem cells. RA is also a key player in early development. RA-mediated inhibition of Wnt signaling is required for body axis extension in mice, and mice deficient in RA demonstrate abnormal expression of Wnt8A and Wnt3A. RA-mediated disruption of Wnt-3A signaling in the tail bud leads to caudal atresia in mouse embryos. The central nervous system, which also expresses Wnt-3A is unaffected by RA treatment, possibly having to do with differential expression of RARs between the central nervous system and the tail bud. A similar phenomenon is responsible for head formation in Xenopus; Wnt signaling leads to regulation of genes important for dorsal development and RA inhibits this signaling, possibly via a protein-protein interaction between RAR and β-catenin, a mechanism that will be discussed later . The Wnt pathway can also modulate RA signaling and neuronal differentiation, as all-trans-retinoic-acid (ATRA), the ligand for RAR, increased Fzd-4 and Fzd-10 expression in human embryonic tumor cells and caused them to differentiate into neuronal cells. Furthermore, Wnt suppresses CYP26, an enzyme that is responsible for degrading RA into inactive metabolites, possibly functioning in a negative feedback mechanism. RA differentially regulates a host of genes on the oncogenic Wnt-1 background compared to a normal background implying intimate crosstalk between these two pathways. These interactions are depicted in Figure 1B. While the individual roles of Wnt and RA in development are conserved, their crosstalk appears to be restricted to vertebrates, as it remains undetected in amphioxus, a primitive chordate.
Progesterone can also positively modulate DKK1 expression in human endometrial cells . Because progesterone levels reach their peak during this phase, progesterone control of DKK1 expression may play a role in pregnancy and development. Progesterone and estrogen, both ligands for NHRs, are also important in the development of the mammary gland. PR-null animals, whose mammary ducts fail to form side-branches, are rescued by ectopic expression of Wnt-1, indicating that Wnt is acting downstream of progesterone. In the same study, it was shown that estrogen and progesterone induced Wnt-4 expression in vivo in the mammary epithelium. Wnt crosstalk with NHRs in the mammary gland is not only responsible for normal development, but may also contribute to abnormal pathology, as is seen in the example of spontaneous mammary tumorigenesis in ER-knockout mice with ectopic MMTV-driven Wnt-1 expression.
Thyroid hormone (T3) increases β-catenin activity in the colon. This effect, however, was puzzling associated with a concomitant increase in sFRP2 expression, which in conjunction with several Fzds was responsible for increasing β-catenin activity in this system.
Cell-cell adhesion is considered a marker of a differentiated phenotype and its reduction can promote cancer cell metastasis. Thus, it is not entirely surprising that the NHRs that have in anti-cancer properties can affect cadherins, which are key, calcium-sensitive participants in cell-cell adhesion. The first major evidence of this phenomenon comes from RA-induced increase in the strength of adherens junctions in a poorly-adhesive breast cancer cell line[46,47]. In these studies, β-catenin was recruited to the membrane in a calcium-dependent manner, implicating a cadherin as the mediator molecule. This effect is likely partially-responsible for RA- (and to a smaller degree, vitamin D-) induced decreases in β-catenin activity in breast and colon cancer cells, as free β-catenin is recruited to the cell membrane, preventing it from entering the nucleus for transactivation of gene expression. A robust vitamin D-mediated increase in E- cadherin expression does occur in a colon cancer cell-line that expresses high levels of VDR. PPARγ ligands, like RA and vitamin D, also induces differentiation and G1 arrest in pancreatic cancer cells in-part by inducing E-cadherin expression. In contrast, a decrease in E-cadherin expression occurs in hepatocytes treated with PPARγ ligand .
Thyroid hormone (T3) plays divergent roles in the intestine with regards to intersecting with the β-catenin pathway that may be, in part, explained by the three TR isoforms: TRα1, TRβ1, and TRβ2. T3 increased β-catenin activity in the intestinal epithelium of mice by not only regulating sFRP2, but also enhancing β-catenin expression, the effects of which may play a role in development of gut crypts[45,51]. The regulation of both sFRP2 and β-catenin expression involves TRα1 but not TRβ1. When T3 is bound to TRβ1, it dose-dependently suppresses the β-catenin-mediated induction of cyclin D1 expression in SW480 colorectal cancer cells and HEK 293T human embryonic kidney cells. The mechanism of this latter inhibition is likely due to TRβ1 binding to the TCF-response-element in the cyclin D1 promoter. This inhibits bidning of TCF/LEF family members for transactivation of the gene in a β-catenin-dependent manner. In contrast to the colon epithelium, rat pituitary cells that endogenously express all three TRs respond to T3 by decreasing β-catenin mRNA and activity despite a paradoxical increase in proliferation. In this study, the decrease in β-catenin activity is much-prolonged in comparison to the decrease in β-catenin mRNA, and may be at least partially explained by a simultaneous, but longer-lived, increase in Axin2 mRNA and protein expression. The mechanism whereby a decrease in β-catenin activity corresponds with an increase in proliferation in these cells remains unanswered.
GC and PPARγ both modulate β-catenin expression in differentiating mesenchymal stem cells via a number of mechanisms including degradation. ATRA can also induce proteasomal degradation of β-catenin via RXRα in HEK293 and SW480 cells. Upon further exploration, the mechanism of RXR-induced β-catenin degradation seems to involve neither the GSK3β/APC/Axin- nor the p53/Siah pathways, implicating a novel degradation pathway in ATRA-resistant colon cancer cell lines[55,56]. Contrary to these data, RA has also been shown to increaseβ-catenin stabilization in tracheobroncial epithelial cells, however, this does not correlate with increases in β-catenin activity, rather, a decrease, possibly due to the concomitant decrease in nuclear localization.
An interaction between ER and cyclin D1, a primary β-catenin target gene, seems to be unique to this NHR. ER can be directly activated by all three cyclin D isoforms in the absence of ligand[58,59]. Cyclin D is able to form a bridge between ER and the Src family of NHR co-activators via its analogous LxxLL motif . ER has also been shown to inhibit GSK3β in the hippocampus. The PI3K/AKT pathway, which also inhibits GSK3β, can, for this reason, be considered a central component in NHR/Wnt pathway crosstalk, and these interactions are nicely reviewed by Mulholland et al.
Crosstalk between the VDR and Wnt pathways has recently been identified as the root of the hairless phenotype of VDR-null mice (Figure 1C) . Keratinocyte stem cells can differentiate down three lineages: sebocyte, epidermal keratinocyte and follicle keratinocyte (interfollicular epidermis). Defective Wnt signaling drives the stem cells down a sebocyte and epidermal keratinocyte lineage instead of a follicle keratinocyte lineage. This defect causes dermal cysts that are filled with sebocytes and epidermal keratinocytes, a phenotype that is also evident in the abundant lipid-laden dermal cysts seen in VDR-null animals. Although ectopic expression of activated β-catenin targeted to keratinocytes does not rescue the defective hair cycling that results in progressive alopecia in VDR-null mice, the synergistic activation of Lef-1 activity by β-catenin requires the presence of the VDR in the same complex and can occur in the absence of ligand, though it is ligand-enhanced[64,65]. VDR may also act independently of TCF/LEF as a Wnt pathway effector. These interactions drive keratinocyte stem cells to differentiate into follicle keratinocytes. The presence of constitutively-active β-catenin induces tumors in the mouse epidermis, however, the presence or absence of VDR dictates which type of tumor develops. VDR-sufficent mice develop trichofolliculomas, which are similar to human pilomatricoma. In the absence of VDR, however, prolonged β-catenin activation induces basal cell carcinomas, underscoring the potential importance of their interaction in the epidermis.
The best-characterized interaction between NHRs and the canonical Wnt pathway stems from the discovery that RAR binds directly to β-catenin in both breast and colon cancer cells. Since that initial discovery, similar interactions have been discovered for VDR, PPARγ, AR, RXR, LXRα and β and ER, although the last one may be particularly sensitive to the cellular milieu because it has gone undetected in several other studies [17,24,48,49,54,66–73]. Of note, GR, PR, or TR have not yet been shown to interact directly with β-catenin [66,69]. Although this interaction is relatively conserved among NHRs, they differ in some aspects. Some NHRs, such as VDR, PPARγ and AR, are found in a complex with both β-catenin and TCF/LEFs, while some NHRs, such as ER and AR and can be found in a complex with TCF/LEFs in the absence of β-catenin[64,71,74–76]. The functional outcomes of these protein-protein interactions can also differ, depending on the NHR. Generally speaking, β-catenin synergistically activates NHR activity, and the liganded NHR reciprocally deactivates or even represses β-catenin activity. This effect is attributed to a number of mechanisms of action (Figure 2). E-cadherin expression is dramatically up-regulated in response to some NHR agonists which redirects β-catenin from the nucleus to the plasma membrane[20,47–50]. Although this is a latent effect, immediate effects can be accounted for by a competition between TCF/LEFs and NHRs for binding to β-catenin and/or p300[48,49,70]. In addition to trans-repression of Wnt-response genes, active repression is accomplished by liganded NHRs binding directly to TCF/LEF family members and recruiting co-represors such as TLE, NCoR and SMRT. Once these co-repressors are bound, they are not easily displaced by β-catenin, leading to long-term repression.
These interactions are highly dependent on the cellular context and can create divergent effects as we have already summarized for development. For example, PPARγ, whose expression and activity is negatively regulated by Wnts in mesenchymal stem cells, undergoes positive regulation of expression and activity in colon cancer cells in response to a Wnt signal. A similar phenomenon is seen with VDR, which super-activates β-catenin activity in keratinocytes, but attenuates it in colorectal cancer cells[48,49,70].
Many of these mechanisms have been extensively studied in the context of AR and prostate cancer. It is generally accepted that unlike colon cancer in which a majority of cases are attributable to inactivating mutations in APC, no APC mutations are found in prostate cancer . However, the genetic locus of APC is lost in some prostate cancer tissues . Men from families with APC mutations have an increased risk of prostate cancer . APC methylation is commonly found in early stage prostate cancer and correlates with early relapse and poor prognosis [80,81].
Disruption of the β-catenin signaling pathway has been found in prostate cancers . An analysis of primary prostate tumors showed that 5% had activating β-catenin mutations [83,84]. Although this is a low frequency of mutations, the level is comparable with the frequency of β-catenin mutations in colon cancer and one might speculate that another component of the Wnt signaling pathway, besides β-catenin or APC, might be targeted for mutation in prostate cancer.
AR and β-catenin each affect the others’ transcriptional activation of promoters with cognate target sequences that lie downstream in the signaling cascade. β-Catenin enhances AR-mediated transcription and androgen causes translocation of β-catenin to the nucleus in cells that express both proteins [85–88]. The recruitment of β-catenin to AR is ligand-dependent and is favored by ligand-bound conformation of AR [89,90]. This is particularly important because the same conditions that favor AR association with β-catenin also favor association with p160 coactivator molecules that bind not only to AR, but also to β-catenin itself, implying that there may be a three-way interaction in the AR transcriptional complex [91,92]. β-Catenin is affected by a variety of growth and survival pathways including receptor tyrosine kinases like IGF-1 receptor that signal through PI3 kinase (PI3K) and activate GSK3β. PI3K-mediated pathways can affect β-catenin interaction with AR, consistent with the notion that receptor tyrosine kinase activation can affect AR [93,94].
AR, in contrast, has an inhibitory effect on TCF/LEF-mediated transcription and can compete with TCF/LEF molecules for β-catenin binding [95–97]. Moreover, AR interacts with TCF/LEF by binding directly to TCF4 . The interaction of AR and the β-catein/TCF4 complex is also affected by PIN1, a peptidyl-prolyl cis/trans isomerase that stabilizes β-catenin by inhibiting its binding to APC. PIN1 is over expressed in advanced prostate cancer. In cells with intact PTEN, PIN1 may enhance AR signaling by increasing β-catenin levels. In PTEN-deficient cells PIN1 expression may decrease the interaction of AR and β-catenin, thus enhancing TCF4-mediated transcription . We speculate, although it has not been shown, that the putative differentiating and growth attenuation effects of AR in certain cell lines may be due, in part, to inhibition of β-catenin/TCF/LEF transcriptional activation .
The relevance of β-catenin to AR in castration-resistant prostate cancer was expanded by the finding that increased levels of the calpain protease in advanced prostate cancer can cleave the N-terminus of β-catenin and generate constitutively active p75 β-catenin . Not only could activated β-catenin affect AR activity, but also potentially recruit TIF2, a NHR co-activator, to the transcriptional complex so that cancer cells would not need other mechanisms of increasing TIF2 expression . Activation of β-catenin in advanced prostate cancer may be more important to cell proliferation than expression of nuclear β-catenin which is actually inversely correlated with tumor progression and preoperative serum PSA levels and directly related to progression-free survival .
In summary, the interactions between NHRs and the Wnt pathway are myriad, complex, and still under active investigation and may offer therapeutic targets for many endocrine diseases and also help define the etiology and potential preventative measures for these diseases. However, the functional outcome of this crosstalk depends upon cellular context. Understanding the contexts under which NHRs and the Wnt pathway reciprocally affect each other are critical for the exploitation of these therapeutic targets, as can be seen from the examples of PPARγ and VDR, which activate or de-activate the β-catenin pathway in different cellular contexts. Much of the work described here has been done in vitro, but several studies have demonstrated interactions in vivo, which has already lead to a greatly-enhanced understanding of control of stem cell differentiation and development. The intersection of the canonical Wnt pathway with NHRs is surely one of many cases that exemplify the importance of pathway crosstalk. We can look forward to the discovery of new, meaningful interactions that are relevant in vivo, and the refinement of currently recognized interactions that will lead to the understanding of development and the identification promising new drug targets.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.