|Home | About | Journals | Submit | Contact Us | Français|
It has been two decades since investigators discovered the link between the Drosophila wingless (Wg) gene and the vertebrate oncogene int-1, thus establishing the family of signaling proteins known as Wnts. Since the inception of the Wnt signaling field, there have been 19 Wnt isoforms identified in humans. These secreted glycoproteins can activate at least two distinct signaling pathways in vertebrate cells, leading to cellular changes that regulate a vast array of biological processes, including embryonic development, cell fate, cell proliferation, cell migration, stem cell maintenance, tumor suppression, and oncogenesis. In certain contexts, one subset of Wnt isoforms activates the canonical Wnt/β-catenin pathway that is characterized by the activation of certain β-catenin-responsive target genes in response to the binding of Wnt ligand to its cognate receptors. Similarly, a second subset of Wnt isoforms activates β-catenin-independent pathways, including the Wnt/ calcium (Wnt/Ca) pathway and the Wnt/planar cell polarity (Wnt/PCP) pathway, in certain cellular contexts. In addition, research has identified several secreted proteins known to regulate Wnt signaling, including the Dickkopf (DKK) family, secreted Frizzled-related proteins (sFRPs), and Wnt inhibitory factor-1 (WIF-1). The advent of technologies that can provide genome-wide expression data continues to implicate Wnts and proteins that regulate Wnt signaling pathways in a growing number of disease processes. The aim of this review is to provide a context on the Wnt field that will facilitate the interpretation and study of Wnt signaling in the context of human disease.
Advances in both microarray-based transcriptional profiling and epigenetics have played important roles in identifying diseases where changes in Wnt signaling may be involved. Today, it seems that almost any type of genome-wide profiling will result in the observations of altered regulation or silencing of at least one or more proteins implicated in Wnt signaling. Our aim is to place the current state of knowledge regarding Wnt signaling in a context that will be helpful for investigators studying diseases where different Wnt proteins may be implicated by microarray or epigenetic studies. By capitalizing on experimental observations of Wnt function in model systems, researchers can develop predictions and hypotheses regarding the role of Wnts in disease models that can be tested with some of the tools and techniques developed over the past 20 years. This review will trace the development of our understanding of Wnt signaling over the past two decades, along with the role of Wnt isoforms, receptor context, secreted Wnt inhibitors, and the techniques used to evaluate Wnt signaling in disease states.
Wnt signaling has been implicated in almost all types of fundamental disease processes relevant to the skin, including inflammation, cancer, wound repair, stem cell biology, and aging (Moon et al., 2004; Brack et al., 2007; Chien and Moon, 2007; Liu et al., 2007; Manolagas and Almeida, 2007). Furthermore, every major cell type in the adult skin is dependent on regulation by Wnt signaling, which accounts for the large body of literature surrounding the central role of this pathway in skin biology. Seminal studies revealed that Wnts are critical for determining the fate of epidermal stem cells, solidifying their role as one of the master regulators of skin development, skin maintenance, and skin repair and regeneration (Kishimoto et al., 2000; Huelsken et al., 2001; Merrill et al., 2001; DasGupta et al., 2002). In the case of hair follicles, where the role of Wnt signaling has been defined as well as in any adult tissue niche, Wnts regulate hair follicle development, the entry of follicle cells into the active growth phase, and even the regeneration of hair follicles after wounding (Andl et al., 2002; Van Mater et al., 2003; Ito et al., 2007; Narhi et al., 2008; Zhang et al., 2008). Aberrant activation of this pathway is seen in the majority of pilomatricomas, further highlighting the role of this pathway in promoting follicular cell fate (Chan et al., 1999). In addition, Wnt signaling is critical in cutaneous oncogenesis as both a regulator of pathological Hedgehog pathway responses (Hoseong Yang et al., 2008) and as an essential factor for maintaining cancer stem cells in models of squamous cell carcinoma (Malanchi et al., 2008). Wnts play a critical role in the development of epidermal melanocytes, and are one of only three factors (in addition to endothelin and stem cell factor) that are necessary to differentiate a human embryonic stem cell into a fully functional melanocyte (Fang et al., 2006). In the case of malignant melanoma, Wnt signaling has been linked to metastatic progression and patient prognosis (Bittner et al., 2000; Kageshita et al., 2001; Weeraratna et al., 2002; Carr et al., 2003; Maelandsmo et al., 2003; Weeraratna, 2005; Hoek et al., 2006). The recent discovery that the X-linked genodermatosis known as focal dermal hypoplasia (Goltz’s syndrome) is directly linked to Wnt signaling further highlights the importance of this pathway in both normal human development and skin biology (Grzeschik et al., 2007; Wang et al., 2007).
The history of Wnt research highlights the value of developmental models such as Drosophila, Xenopus, and zebrafish, which have been instrumental in the seminal studies that defined this signaling pathway and shaped the canonical model of Wnt/β-catenin signaling that we know today. Twenty years ago, the realization that the Drosophila wingless (Wg) gene and the vertebrate int-1 oncogene are orthologs heralded the birth of the Wnt signaling field (Rijsewijk et al., 1987). At that time, some of the early understanding of the biochemical and molecular events involved in oncogenesis was being developed, and this initial finding indicated that a pathway used for the development of an organism could participate in the process of oncogenesis. The discovery that ectopic expression of the vertebrate int-1/ Wnt-1 oncogene in Xenopus embryos results in axis duplication and two-headed frogs further solidified the role of Wnt as an important regulator of both embryonic development and oncogenesis (McMahon and Moon, 1989). In the intervening two decades, an abundance of research has clarified the mechanisms by which Wnt and other developmental pathways can be aberrantly regulated during oncogenesis and other disease processes.
The link between Wnt and β-catenin was first made in 1990, when a study in Drosophila made a seminal observation linking Wnt/Wg to post-translational regulation of the segment polarity gene armadillo, the Drosophila homolog for vertebrate β-catenin (Riggleman et al., 1990). In the next few years, data emerged linking armadillo/β-catenin to adherens junctions and also to a protein known as adenomatous polyposis coli, which was already implicated in familial adenomatous polyposis (McCrea et al., 1991; Rubinfeld et al., 1993). Together, these studies implicated β-catenin in both cell adhesion and tumor formation. In 1995, investigators showed that overexpression of certain domains of β-catenin in Xenopus led to axis duplication, which was similar to the initial findings with Wnt-1 overexpression, implying that β-catenin had roles in signaling that were independent of its role in cell adhesion (Funayama et al., 1995). As these results solidified the role of Wnt in oncogenesis, researchers began earnestly characterizing the ability of Wnt isoforms to induce the transformation of cancer cells.
Intense interest in the developing field of Wnt signaling combined with innovations in molecular biology led to rapid identification of progressively more vertebrate Wnt isoforms, each displaying seemingly unique patterns of expression during vertebrate development. Indeed, developmental models continue to provide pioneering insight into the molecular mechanisms underlying Wnt signaling. In 1994, genetic studies identified a critical role for a protein known as Dishevelled (Dsh/Dvl) in propagating the Wnt/Wg signal (Klingensmith et al., 1994; Noordermeer et al., 1994). The same year genetic studies showed that the Drosophila zeste-white 3/Shaggy protein, a homolog of the vertebrate enzyme glycogen synthase kinase 3 (GSK3), regulated cellular armadillo/β-catenin in a manner that opposed a Wnt/Wg signal (Peifer et al., 1994). The requirement for zw3/Shaggy/ GSK3 in mediating Wnt/Wg signaling in the developing fly leg further highlighted the importance of this enzyme (Diaz-Benjumea and Cohen, 1994). A key paper that established the principal regulation of the pathway came, interestingly, from studies in frog embryos (Yost et al., 1996). This paper was the first to show that GSK3 directly phosphorylates β-catenin, that this phosphorylation resulted in the degradation of β-catenin, and that mutating the GSK3 phosphorylation sites resulted in a stable, hyperactive β-catenin. Moreover, this was the first paper to show that β-catenin was found in the nucleus, and that this nuclear accumulation was directly controlled by the regulated post-translational stability of β-catenin. Following the demonstration that β-catenin levels increased upon binding of Wnt ligand to Fzd receptors in Drosophila cells, these collective data formed the foundation for our current model (Figure 1) of Wnt/β-catenin signaling (Bhanot et al., 1996; Yang-Snyder et al., 1996).
The current model of Wnt/β-catenin signaling appears to be quite well conserved in both vertebrates and invertebrates, although many fundamental molecular events remain unclear despite two decades of intense research (Figure 1). For example, although it is widely recognized that binding of Wnt ligand to Fzd and low-density lipoprotein receptor-related protein 5/6 (LRP5/6) co-receptors leads to the activation of Dvl, our understanding of the molecular mechanisms underlying Dvl activation remains incomplete. Furthermore, the exact sequence of events that allows Dvl to inhibit the β-catenin destruction complex is still unsettled, as is the role of nuclear localization of Dvl. Recent studies also suggest that the regulation of Dvl degradation can play a critical role in modulating the signaling pathway (Angers et al., 2006; Chan et al., 2006). Although it is well-accepted that the phosphorylation of β-catenin by GSK3 leads to ubiquitination and subsequent proteasomal degradation, recent studies continue to identify new members involved in modulating the β-catenin destruction complex, including the first X-linked tumor suppressor gene WTX (Major et al., 2007). Although the activation of transcriptional changes by nuclear β-catenin in both developmental and cancer models has focused primarily on interactions with transcription factors of the T-cell factor (TCF) and lymphoid-enhancing factor (LEF) families, recent studies suggest that there may be still other transcription factors involved in β-catenin signaling that are yet to be identified (Olson et al., 2006). Most models of Wnt signaling generally end with the activation of target genes by β-catenin, whereas the mechanisms involved in turning off the pathway remain unresolved. Nevertheless, the current model whereby Wnt binding leads to cellular phenotypes through the β-catenin-mediated transcriptional regulation of target genes has provided the basis for many studies that have lent insight into the role of this pathway in both embryonic development and the biology of cancer and other human diseases (Moon et al., 2004; Chien and Moon, 2007).
As more Wnt isoforms were identified, it soon became clear that the Wnt/β-catenin model was not sufficient to explain the actions of all identified Wnt isoforms. Early studies with Xenopus Wnt-5a showed that overexpression of this isoform led to effects that were distinct from the axis duplication and neural malformations seen with Xenopus Wnt-1, Wnt-3, or Wnt-8 (Moon et al., 1993). Thus, in Xenopus, this class of Wnts appeared to affect cell movements rather than cell fate, which was different than what was observed with Wnt-1, Wnt-3, and Wnt-8. The classification of Wnt signaling was further refined based on studies in other systems into two main pathways: (1) the transforming, dorsalizing, or axis-inducing class and (2) the non-transforming class (see Table 1). The class of transforming, dorsalizing, or axis-inducing Wnts displayed phenotypes in animal models and cellular models that were similar to β-catenin over-expression, and today we would categorize these Wnts as predominantly activators of the Wnt/β-catenin pathway. The observed β-catenin-independent regulation of polarity, asymmetric cell division, and morphogenetic movements during vertebrate gastrulation by these non-transforming Wnts led to their initial classification as “non-canonical” Wnts, although the term “β-catenin-independent Wnts” is now preferred (Veeman et al., 2003). It should be noted that the dogmatic classification of Wnt ligands as either activators of Wnt/β-catenin signaling or regulators of β-catenin-independent Wnt signaling is difficult, as it is clear that the pathway activated by a Wnt isoform can be highly dependent on cellular context and factors such as the complement of cell surface Wnt receptors (Mikels and Nusse, 2006). Nevertheless, several well-studied isoforms appear to consistently activate one of the two pathways across numerous cellular and developmental models, and are often referred to in the literature as a “canonical Wnt” or a “non-canonical Wnt”.
Several models for β-catenin-independent Wnt pathways have been proposed. Genetic studies identified several putative pathway members that have been implicated in regulating planar cell polarity (PCP) during Drosophila development, and these proteins are collectively known as the Wnt/PCP pathway (Veeman et al., 2003). The observation that certain β-catenin-independent Wnts can lead to increases in intracellular calcium (Ca) levels has led to a model for Wnt/Ca signaling in both developmental biology models and cultured human cells (see Table 2) (Kohn and Moon, 2005). The implication of certain proteins in both Wnt/PCP and Wnt/Ca pathways suggests that they may be components of the same signaling network rather than two distinct pathways, although this hypothesis awaits further experimental confirmation. In any event, the current data overwhelmingly support the existence of both β-catenin-dependent and β-catenin-independent Wnt pathways in vertebrate cells.
In vertebrates, exogenous perturbation of β-catenin-independent Wnt signaling results in gastrulation phenotypes called convergent extension defects, which are characterized by embryos that are shortened along the body axis and widened in the lateral axis because of defective cell migration (Veeman et al., 2003). Certain proteins implicated in PCP in Drosophila exhibit convergent extension defects that are similar to what is seen upon gain of function or loss of function of certain β-catenin-independent Wnt isoforms, supporting the hypothesis that they may be involved in a common pathway (Veeman et al., 2003). Although diverse downstream effectors—intracellular Ca, c-Jun N-terminal kinase, protein kinase C, CaMK, nuclear factor of activated T-cells (NFAT), and small GTPases such as Rho and Rac— have been implicated, there is as yet no reliable assay for measuring β-catenin-independent Wnt activation across different cell types (see Figure 1). As a consequence, many studies have relied on developmental models such as mice, zebrafish, and Xenopus. The phenotypes observed in developmental models led to the hypothesis that β-catenin-independent Wnts were involved in the regulation of cell motility and/or cell adhesion, and studies of β-catenin-independent Wnt signaling in the context of human cancer cell lines seem to support this model (Jonsson and Andersson, 2001; Weeraratna et al., 2002; Kurayoshi et al., 2006; Nishita et al., 2006; Safholm et al., 2006).
Further studies discovered that β-catenin-independent Wnt pathways can also inhibit Wnt/β-catenin signaling (Topol et al., 2003; Weidinger and Moon, 2003). Initial studies showing that overexpression of Wnt-5a in Xenopus could inhibit the effects of overexpression of the canonical Wnt-1 isoform provided the first hints that these two pathways could have antagonistic effects (Torres et al., 1996). Subsequent studies have supported the hypothesis that antagonism of Wnt/β-catenin signaling plays a relevant role in multiple contexts, including limb regeneration (Stoick-Cooper et al., 2007). Continued studies may eventually reveal whether inhibition of Wnt/β-catenin signaling by β-catenin-independent Wnt isoforms plays a major role in disease pathogenesis, and also whether activation of β-catenin-independent Wnt pathways can provide a potential therapeutic avenue for inhibiting Wnt/β-catenin signaling in the context of certain disease states.
The activation of either the Wnt/β-catenin pathway or β-catenin-independent Wnt pathway appears to be dictated primarily by the context in which these Wnts are studied. In large part, these contexts have included their observed phenotypes in gain-of-function and loss-of-function studies in vertebrate developmental models, their ability to stabilize and increase cytosolic and nuclear levels of β-catenin, their ability to activate β-catenin-responsive reporters or gene targets in different cell lines or organisms, and their ability to elicit transformation of the C57MG breast cell line (see Table 1). Together, these types of experimental observations have provided a guideline for determining whether different Wnt isoforms have the ability to activate Wnt/β-catenin and β-catenin-independent Wnt pathways in different models. Although the specific Wnt isoform often predicts the downstream signaling pathway (see Table 1), there are notable exceptions that are likely due to changes in receptor context. The importance of receptor context on Wnt signaling was shown in human embryonic kidney cells, where Wnt-5a can either activate or inhibit Wnt/β-catenin signaling depending on the expression of different receptors (Mikels and Nusse, 2006).
Developmental models such as the zebrafish, Xenopus, and mouse have provided the basis for many of the defining studies in the field. In these models, loss-of-function studies from different mutations (such as the Wnt-5b pipetail mutant or the Wnt-11 silberblick mutant in zebrafish) or from the targeted knockdown of expression using Wnt-specific mor-pholinos that block translation or RNA processing have been particularly revealing regarding the roles and actions of different Wnt isoforms during embryonic development. Although gain-of-function studies have been used to show the potential actions of Wnts in this context, these studies also lead to non-physiological effects of Wnts in tissues where they are not normally expressed, and thus require caution with the interpretation of results. These models provide a rapid experimental assay for testing the potential function of Wnt isoforms and Wnt pathway proteins in the well-characterized context of embryonic development.
It is tempting to try and broadly classify Wnts as either β-catenin-dependent or β-catenin-independent isoforms, but the incomplete overlap between observed measures of the pathway, such as β-catenin stabilization and C57MG transformation, suggests that other context-specific factors may be important. For example, some Wnts, such as Wnt-5b and Wnt-11, display phenotypes associated with β-catenin-independent Wnt signaling in developmental models, while also displaying weak transformation properties that more closely resemble what would be expected with activation of the Wnt/β-catenin pathway in C57MG breast cells (see Table 1). There are at least 10 different Fzd isoforms that have been identified to date, along with a cadre of candidate β-catenin-independent Wnt receptors, adding further complexity to the equation (Cheyette, 2004; Chien and Moon, 2007). Furthermore, Fzd isoforms form both homo-oligomers and hetero-oligomers, further confounding efforts to classify receptor-ligand-binding partners in a broadly reliable manner (Kaykas et al., 2004). In addition, other properties of Wnt proteins, such as their secretion, stability, and even their interaction with other environmental factors such as the extracellular matrix, may all depend heavily on a cell context (Bradley and Brown, 1990; Burrus and McMahon, 1995).
So why would vertebrates evolve to have so many different Wnt isoforms when extensive research has uncovered only two to three distinct pathways? During development, the presence of multiple Wnts that activate a shared β-catenin-dependent or β-catenin-independent pathway may merely reflect tissue-specific expression patterns rather than actual functional differences between Wnt isoforms. Wnts are themselves subjected to post-translational modification through glycosylation and lipid modifications including palmitoylation, and these moieties could play an important role in determining the distinct properties of different Wnt isoforms (Takada et al., 2006; Komekado et al., 2007; Kurayoshi et al., 2007). Some of the limitations in studying different Wnt isoforms derive from the fact that the purification of these proteins in robustly active forms was only recently achieved, and as techniques improve, our understanding of the biological distinctions between isoforms will undoubtedly be refined (Willert et al., 2003).
As if combinations between 19 Wnt ligands and 10 Fzd receptor isoforms were not enough to foster Wnt signaling diversity between cell types, there are other non-Wnt secreted proteins that can bind Fzd receptors to activate Wnt/β-catenin signaling (Hendrickx and Leyns, 2008). The Norrie disease protein (NDP, also called “Norrin,” encoded by the NDPH gene) was identified as a high-affinity ligand for FZD4 based on the astute observation that Ndph mutant mice exhibit similar vascular defects in the eyes and ears to the mice mutant in Fzd4 (Xu et al., 2004). Similar to Wnts, NDP binds both Fzd4 and LRP5/6 co-receptors to stabilize β-catenin and activate Wnt/β-catenin signaling. In humans, mutations of NDPH, FZD4, and LRP5 have all been observed in patients with familial exudative vitreoretinopathy, confirming the important role of these three proteins in regulating vascular development of the eye (Chen et al., 1993; Qin et al., 2005).
Another class of proteins in the thrombospondin type-1 repeat superfamily called R-Spondins can also activate Wnt/ β-catenin signaling through FZD and LRP co-receptors (Kamata et al., 2004; Kazanskaya et al., 2004; Nam et al., 2006). Spondins also exhibit synergy with Wnt ligands in activating Wnt/β-catenin signaling, and interestingly are downregulated with loss of Wnt1 or Wnt3A in mice, suggesting a possible role in feed-forward signaling (Kamata et al., 2004). In humans, R-Spondin (encoded by the gene RSPO1) regulates skin differentiation, sex determination, and predisposition to squamous cell carcinoma (Parma et al., 2006). Without these developmental studies, neither Spon-dins nor NDP exhibit homology to Wnts that would predict their function as activators of Wnt/β-catenin signaling. Consequently, it is likely that other as-yet unidentified proteins may be able to activate Wnt/β-catenin signaling, either through engagement of FZD and LRP5/6 receptors or through novel mechanisms.
The expression of Wnts in different disease conditions is not itself sufficient to conclude that Wnt signaling is occurring, as certain secreted inhibitors can inhibit Wnt signaling at or before the level of receptor binding (see Figure 1). Recent studies implicating Wnts in aging suggest that humoral levels of Wnt are physiologically relevant, and secreted inhibitors might also act as important modulators of Wnt signaling both locally and within an organism’s circulation (Brack et al., 2007; Liu et al., 2007). During the second decade of the Wnt field, developmental biologists identified a protein called Dickkopf-1 (DKK-1) as an antagonist of Wnt signaling (Glinka et al., 1998). Soon after, a family of DKK proteins was identified, and of the four known human isoforms, DKK-1 and DKK-4 were shown to inhibit the axis duplication induced by Wnt/β-catenin activation in Xenopus embryos (Krupnik et al., 1999). Epistasis experiments with DVL and FZD indicated that DKK-1 and DKK-4 likely acted upstream of FZD and DVL in the pathway, which was soon confirmed by studies demonstrating that DKK-1 potently inhibited Wnt signaling by binding to the Wnt co-receptor LRP5/6 (Bafico et al., 2001; Mao et al., 2001; Semenov et al., 2001). Further studies identified the Kremen family of proteins as a second class of DKK receptors (Niehrs, 2006). Although studies indicate that DKK-1, DKK-2, and DKK-4 regulate Wnt/β-catenin signaling in vertebrate cells, DKK-3 appears to be a more divergent form of the family both based on its functional effects and by sequence analysis (Niehrs, 2006). Although all DKK family members contain a unique DKK-N motif and a putative co-lipase fold region, DKK-3 also shares a unique SGY domain with a protein known as DKK-like-1 or Soggy. Little is known about Soggy outside of a potential role in spermatogenesis and its homology to DKK-3 (Kaneko and DePamphilis, 2000; Kohn et al., 2005).
A growing number of studies have implicated DKK family members in human disease based on transcriptional profiling and epigenetic studies (Niehrs, 2006). In particular, alterations of DKK expression have been observed in a number of cancer models, which is not surprising given the importance of Wnt signaling in cancer biology. Modulation of DKK-1 expression in disease models, including mesothelioma, Alzheimer’s, and ischemic neuronal injury, has shown some therapeutic promise (Chien and Moon, 2007). The role of DKK expression changes as either contributory or compensatory responses is a frequent question that arises during the interpretation of transcriptional profiling experiments. For example, in the case of colorectal carcinoma, where the Wnt/ β-catenin activation is clearly an integral player in carcinogenesis, the observed epigenetic silencing of DKK-1 and other DKK family genes is likely to contribute further to the activation of the Wnt/β-catenin pathway (Hoffman et al., 2004; Kuhnert et al., 2004; Aguilera et al., 2006; Sato et al., 2007). DKK-1 expression itself can be directly upregulated by Wnt/β-catenin signaling in a TCF-dependent manner, but this mechanism appears to be lost through silencing in colon cancer (Gonzalez-Sancho et al., 2005). By contrast, although the large majority of hepatoblastoma tumors exhibit nuclear β-catenin suggestive of the Wnt/β-catenin pathway activation, over 80% of tumors display increased DKK-1 transcript relative to normal fetal liver, suggesting that elevated DKK-1 may be potentially be a negative feedback response in response to the activation of Wnt/β-catenin signaling (Wirths et al., 2003).
Another potentially interesting twist arises in the case of DKK-2, which has been shown in Xenopus studies to activate the Wnt/β-catenin pathway in synergy with overexpression of certain Fzd receptor isoforms (Wu et al., 2000). This observed activation of Wnt/β-catenin signaling by DKK-2 is antagonized by DKK-1, but as these assays rely in large part on overexpression, the issue of whether DKK-2 functions as a true activator of Wnt/β-catenin signaling under normal physiological conditions remains unresolved (Wu et al., 2000). In the case of the DKK-2 knockout mouse, the loss of DKK-2 leads to constitutive Wnt/β-catenin signaling in the cornea that results in blindness because of the development of a keratinized epithelium and skin appendages, suggesting that DKK-2 functions as an inhibitor of Wnt/β-catenin signaling in the context of this tissue (Mukhopadhyay et al., 2006). Again, receptor context may play a large role in determining the action of a secreted ligand, as the ability of DKK-2 to act as both an activator and an inhibitor of Wnt/β-catenin signaling could be determined by the expression levels of different DKK receptors such as LRP5/6 and the Kremen family of receptors (Niehrs, 2006).
Another class of proteins called secreted Frizzled-related proteins (sFRPs) were first identified around 1996 and postulated to be potential modulators of Wnt signaling (Hoang et al., 1996; Rattner et al., 1997). Currently, there are five members of the sFRP family in humans, although sFRP-3 is many times designated as FrzB (Frizzled motif associated with bone development), reflecting its original name before the reclassification of these proteins as one family. Soon after their discovery, studies on these proteins supported the notion that they could bind Wnts and thus act as modulators of Wnt/β-catenin signaling (Lin et al., 1997; Wang et al., 1997a, b; Hoang et al., 1998; Bafico et al., 1999). Although the sFRPs contain an N-terminal cysteine-rich domain homologous to the Wnt-binding domain of the Frizzled receptor, the exact role of this domain in the binding of Wnt ligand by sFRPs is still unclear (Kawano and Kypta, 2003). Interestingly, this cysteine-rich domain in sFRPs can dimerize with other sFRPs and also bind Frizzled itself, suggesting that inhibition of Wnt signaling by sFRPs could potentially occur either through the binding of soluble Wnt ligand or the formation of non-functional receptor complexes with cell-surface Frizzled receptors (Bafico et al., 1999).
Studies from developmental models support the idea that sFRPs inhibit Wnts that activate either Wnt/β-catenin signaling or β-catenin-independent pathways, which is not entirely surprising as Wnt isoforms of both classes bind to Frizzled receptors. However, interpretation of the role of sFRPs in different tissues and disease states is complicated by the fact that different sFRP family members can have different, and even opposing, effects in certain contexts. In breast cancer cells, sFRP-1 appeared to increased β-catenin levels, whereas sFRP-2 appeared to have no change or even increased β-catenin levels (Melkonyan et al., 1997). In the developing kidney, it appears that sFRP-1 can downregulate Wnt/β-catenin signaling, whereas sFRP-2 may function as a competitive inhibitor of sFRP-1 without directly affecting Wnt/β-catenin signaling (Yoshino et al., 2001). Coupled with studies suggesting that sFRP-1 can act as an activator of Wnt/ β-catenin signaling at lower concentrations, it has been suggested that the sFRPs may act in a biphasic manner, acting as antagonists when there are lower relative concentrations of Frizzled receptor, while serving as an activator of Wnt signaling in the presence of higher relative concentrations of Frizzled receptor, potentially by augmenting the transport of Wnt ligand to these sites (Uren et al., 2000; Kawano and Kypta, 2003). Consequently, the question of how sFRPs may act in the context of different disease states may not be easily resolved by simple transcriptional profiling, as the amount of protein for sFRPs, Frizzled receptors, and Wnt ligand is likely all relevant to the functional outcome for the cell. Nevertheless, the increasing reports of epigenetic inactivation of sFRP genes in different cancer models implicate them as important regulators of oncogenesis (Caldwell et al., 2004; Lee et al., 2004; Suzuki et al., 2004; Fukui et al., 2005; He et al., 2005; Marsit et al., 2005, 2006; Zou et al., 2005; Aguilera et al., 2006; Liu et al., 2006; Qi et al., 2006; Urakami et al., 2006a; Nojima et al., 2007; Sato et al., 2007).
Wnt-inhibitory factor 1 (WIF-1) is a mysterious protein that binds Wnt ligands through a unique WIF domain that does not exhibit homology to the cysteine-rich Wnt-binding regions of sFRP or Frizzled proteins (Hsieh et al., 1999). Interestingly, the atypical receptor tyrosine kinase receptor RYK also contains an extracellular WIF domain, and this receptor is one of several proteins implicated as a cellular receptor for the β-catenin-independent Wnt-5a isoform (Cheyette, 2004; Lu et al., 2004; Keeble et al., 2006; Harris and Beckendorf, 2007). The exact mechanisms underlying the regulation of Wnt signals by WIF-1 are not completely understood, but the observed silencing of WIF-1 in different cancers along with the effects of WIF-1 gain of function and loss of function in different cancer models suggests that like DKKs and sFRPs, this protein likely plays an important role in oncogenesis and other biological processes related to Wnt signaling (Wissmann et al., 2003; Mazieres et al., 2004; Taniguchi et al., 2005; Aguilera et al., 2006; Batra et al., 2006; Lin et al., 2006, 2007; Tang et al., 2006; Urakami et al., 2006b).
Wnt signaling plays an integral role in maintaining cellular homeostasis, defined as a cell’s ability to maintain a certain state of differentiation, and the loss or disruption of this homeostasis can have deleterious consequences that manifest as both acute and chronic disease states. As a result, the question of whether Wnt signaling is active in different diseases has potential diagnostic, prognostic, and therapeutic implications. Alterations of Wnt signaling in disease can occur through changes in the expression of Wnt isoforms or secreted Wnt modulators, and also through mutations that either activate or inactivate critical components of the Wnt signaling pathway. The study of Wnt signaling in disease is further complicated by the complex interaction of this pathway with other signaling networks (Figure 1), making it difficult to draw conclusions on the state of Wnt signaling pathways through transcriptional profiling studies alone. Consequently, investigators have utilized other experimental approaches to provide independent corroboration of Wnt signaling activity. In the context of human disease models, the following three types of studies are used to substantiate the involvement of Wnt/β-catenin signaling: (1) staining for nuclear β-catenin in patient samples; (2) the use of optimized Wnt-responsive reporter assays to assess pathway activation; and (3) validation of changes in known endogenous Wnt target genes.
In the current model of Wnt/β-catenin signaling, the destruction complex rapidly targets cytosolic β-catenin for proteasome-mediated degradation in the absence of a Wnt signal; hence, the presence of nuclear β-catenin would not be expected without activation of the Wnt/β-catenin pathway by either ligand binding or perturbation of the pathway downstream of the receptor (that is, with a gene mutation or deletion) resulting in constitutive activation. Consequently, the presence of nuclear β-catenin has been used as a surrogate marker of Wnt activation in histopathology studies of patient samples. Studies in animal models and cell culture models using other corroborating experimental measures of Wnt signaling have supported the fact that the Wnt/β-catenin pathway activation is indeed accompanied by the nuclear translocation of β-catenin, and in the case of archived patient tissues where extraction of proteins and nucleic acids can be technically limited, staining for nuclear β-catenin may be the only available method for determining Wnt/β-catenin pathway status.
There are caveats that investigators should recognize when trying to interpret the meaning of nuclear β-catenin in these situations. Although the presence of nuclear β-catenin likely implies, at the minimum, some prior activation of the Wnt/β-catenin pathway, there are several nuclear inhibitors of β-catenin signaling that may function to offset the presence of nuclear β-catenin in certain contexts (Daniels and Weis, 2002; Takemaru et al., 2003; Gottardi and Gumbiner, 2004). In addition, the lack of observed nuclear β-catenin staining may reflect technical limitations in detection related to tissue preservation or antibody sensitivity rather than a definitive lack of Wnt/β-catenin signaling, as in theory only a few molecules of nuclear β-catenin may be needed to regulate critical gene targets after a Wnt signal. A recent report documenting Smad3-dependent nuclear localization of β-catenin after activation of the TGFβ signaling pathway in mesenchymal stem cells also suggests that nuclear localization of β-catenin may not be entirely limited to Wnt/β-catenin signaling (Jian et al., 2006). Nevertheless, testing for nuclear β-catenin can provide important data regarding the role of Wnt/β-catenin signaling, but should always be interpreted in the context of other independent corroborative approaches.
Another valuable tool for the study of Wnt/β-catenin signaling has been the development of optimized TCF-responsive Wnt/β-catenin reporters that have been successfully used in several vertebrate systems, in both transgenic organisms and cultured cells (Barolo, 2006). The promoters for these reporters exploit the known regulation of Wnt/β-catenin-responsive genes by members of the TCF and LEF family of transcription factors, utilizing multimerized TCF/ LEF-binding sites that are upstream of reporters such as β-galactosidase, green fluorescent protein , or a luciferase enzyme to facilitate the measurement of Wnt/β-catenin signaling by visual or rapid enzymatic assays (Figure 2). Again, the use of these reporters should be accompanied by an understanding of their possible limitations. The latest iterations of these reporters all contain at least 8–12 optimized TCF/LEF-binding sites, which is more than the 1–2 TCF/LEF sites often found in endogenous gene targets (Barolo, 2006). These reporters will not provide information in cases where Wnt activation may lead to TCF-independent transcriptional changes (Olson et al., 2006). Conceivably, these reporters would also not be able to distinguish Wnt-mediated activation of TCF-dependent signaling from the activation of TCF by other mechanisms.
Despite these potential caveats, TCF-responsive Wnt reporters have proven to be a tremendously reliable and valuable tool for assessing the response of different cell types to Wnt signals both in vitro and in vivo while also providing an easy assay for investigating the role of other proteins in modulating Wnt/β-catenin signaling. The availability of control reporters with mutated TCF/LEF-binding sites provides an additional tool to confirm the specificity of a suspected Wnt/β-catenin response. In cultured cells, the use of these Wnt reporter systems allows quantitative comparison of Wnt/ β-catenin activation under different conditions, including drug treatments and the targeted manipulation of protein expression in gain-of-function or loss-of-function experiments. In addition to providing a rapid and fairly specific readout for activation of Wnt/β-catenin signaling across a wide range of disease models, the high fidelity of these reporters in cell-based assays has also facilitated high-throughput screening strategies as a means for further identifying molecular regulators of this pathway (Major et al., 2008).
In transgenic models, these reporters can provide valuable information on the temporal and spatial activation of Wnt/β-catenin signaling, although quantitative comparisons in these types of experiments are more difficult (see also Figure 2). To illustrate some of the limitations of transgenic Wnt reporter systems, earlier studies have found discrepancies in the temporal and spatial activation of Wnt reporters while using different transgenic models to study embryonic development (reviewed in Barolo, 2006). Our own group has observed differences between the TOPGAL and BATGAL transgenic reporter mice with regard to the identification of Wnt/β-catenin signaling in various adult tissues, which often necessitates the examination of both models for experimental studies. The observed differences between various transgenic lines involve factors such as the design of the promoter region and transgenic construct, as well as the integration of the reporter construct within the genome. Together, these limitations highlight the need to further perform additional assays validating the activation of Wnt/β-catenin signaling in different contexts, but do not diminish the role of these animal reporters as invaluable tools for studying Wnt/β-catenin signaling in vivo.
Perhaps the most convincing evidence for activation of Wnt/β-catenin is demonstration of the Wnt-dependent regulation of an endogenous Wnt target gene. A growing number of endogenous Wnt target genes have been identified, and are available online at the curated Wnt homepage (http://www.stanford.edu/~rnusse/wntwindow.html). These gene targets have been identified in different tissue contexts, so the universal applicability of most of these genes as markers of Wnt signaling is still unclear. However, one particular Wnt target gene, Axin2 (also known by its aliases of conductin or Axil), has been regarded as perhaps the most reliable endogenous target gene for measuring Wnt/β-catenin pathway activation (Jho et al., 2002). Interestingly, the Axin2 promoter contains eight TCF/LEF-binding sites that are very similar to the optimized sites used in multimerized reporters, and the presence of so many TCF/LEF sites may explain why Axin2 has proven to be a ubiquitous and vigorous endogenous gene marker of Wnt/β-catenin pathway activation (Jho et al., 2002; Lustig et al., 2002). The use of techniques such as real-time PCR, microarray-based transcriptional profiling, and even immunohistochemistry can be used to analyze the relative expression levels of Wnt target genes, facilitating the quantitative comparison of Wnt/β-catenin pathway activation under different experimental conditions in both cultured cell models and whole organisms and tissue.
These approaches all leverage the transcriptional regulation of genes by nuclear β-catenin and TCF/LEF transcription factors to provide experimental evidence of Wnt/β-catenin signaling, and the use of multiple approaches can provide valuable corroborating data when interrogating the status of Wnt/β-catenin signaling in different disease models. The availability of these techniques has contributed greatly to advancing our understanding of the molecular events involved in Wnt/β-catenin signaling. Conversely, the lack of similar assays has made the study of β-catenin-independent Wnt signaling pathways more difficult. As both technology and our understanding of Wnt signaling continue to advance, the development of new experimental methods for measuring Wnt signaling will enhance our understanding of the events that occur downstream of Wnt binding to its receptor.
Wnt signaling will undoubtedly be implicated in a growing number of disease conditions, and identifying the link between Wnt and disease will rely on the sophisticated synthesis and interpretation of different lines of experimental evidence. Given the large numbers of proteins that have been identified so far as regulators of Wnt signaling, it will be difficult to assess the involvement of this pathway in a disease process based on cellular profiling studies alone. However, available tools including cell-based reporters, transgenic reporter animals, and an ever-expanding list of Wnt-regulated target genes offer powerful tools for studying Wnt signaling in model systems.
Even after 20 years of intense study, there are still many unanswered questions remaining in Wnt signaling, particularly with regard to its role in human disease. Developmental biology models that have provided the foundation for this field will probably further generate valuable and essential insight into mechanisms underlying Wnt signaling. Ongoing efforts to study the link between Wnts and a growing number of disease conditions will uncover details that further contribute to our understanding of these complex signaling pathways. Together, these ongoing efforts ensure that a Wnt review 20 years from now may look quite different, but the foundation of this field will always be based on a series of elegant experiments in developmental model systems that alerted the world of science to the importance of this pathway in both development and disease.
R.T.M. is supported as an investigator of the Howard Hughes Medical Institute. A.J.C. is supported by a K08 Career Development Award from the National Institutes of Health and has also been supported by career development awards from the Dermatology Foundation and the American Skin Association. We are indebted to these agencies for their continued support of our ongoing research. As with any review of the Wnt field, this effort is an attempt to give a current overview of a field that continues to advance at a rapid and prolific pace. We therefore apologize for any unintended oversights or omissions. We are grateful to Dr Vera Veronina and Allison Adams for their contributions of previously unpublished data to Figure 2.
CONFLICT OF INTEREST
The authors state no conflict of interest.