Members of the catenin family were first isolated complexed with the cytoplasmic domains of cadherins at cell-cell junctions, and this relationship is considered one of the defining aspects of catenins. Their functions at such contacts are multifaceted and remain under active study and discussion. The word catenin (from catena, latin for chain) (Ozawa et al., 1989), reflects the fact that some catenins contribute to the indirect association of cadherins with the underlying actin cytoskeleton, as indicated for β-catenin (and probably γ-catenin/plakoglobin) at adherens junctions (Abe and Takeichi, 2008; Pappas and Rimm, 2006). Cytoskeletal interactions assist, for example, with the execution of cadherin-dependent morphogenic movements by facilitating the application of contractile forces at cell-cell contact zones (Gumbiner, 2005). In addition to adherens junctions, catenins also function at desmosomes, where both plakophilins (catenins) and γ-catenin/plakoglobin are involved with linkages to intermediate filaments (Garrod and Chidgey, 2008; Green and Simpson, 2007; Hatzfeld, 2007; Schmidt and Koch, 2007). Some additional roles of catenins, including the modulation of cadherin endocytosis and small GTPases, are summarized later.
Catenins also act in the nucleus (Daniel, 2007; McCrea et al., 2009; Stepniak et al., 2009) (see the end of this section for references regarding β-catenin). A prominent example is the stabilization of a signaling pool of β-catenin in response to upstream Wnt signals (e.g. Wnt ligands). This occurs via inhibition of a multi-component complex that would otherwise lead to proteasomal destruction of β-catenin (Kimelman and Xu, 2006). Upon entry to the nucleus, β-catenin alters gene activity in a context-dependent manner (Arce et al., 2006; Willert and Jones, 2006). This takes place in association with transcription factors such as those from the T-cell factor (TCF) and lymphoid enhancer-binding factor (LEF) protein family, which directly bind DNA, as well as several transcription and chromatin co-regulators. Varied downstream outcomes are influenced, extending, for example, from stem-cell maintenance to differentiation, and from proliferation to apoptosis (Cadigan and Peifer, 2009; Chien et al., 2009; Clevers, 2006; Grigoryan et al., 2008). Given the wide biological impact of catenins, basic and translational researchers from diverse viewpoints are working to generate insights into the mechanisms that govern and execute their functions.
In this poster article, we present a concise overview of the function of catenin proteins, primarily for the non-specialist. We restrict our attention to those catenins that contain an Armadillo domain [and therefore do not include the structurally unrelated α-catenin (Benjamin and Nelson, 2008; Kobielak and Fuchs, 2004)]. We direct the reader to other brief overviews (Harris and Peifer, 2005; Huelsken and Behrens, 2002; Lien et al., 2008; Macdonald et al., 2007; Reynolds, 2007), or to more in-depth reviews discussing β-catenin (Bienz, 2005; Cadigan and Peifer, 2009; Clevers, 2006; Nelson and Nusse, 2004), or the p120 or plakophilin subfamilies of catenins (Anastasiadis, 2007; Daniel, 2007; Hartsock and Nelson, 2008; Hatzfeld, 2005; Hatzfeld, 2007; Kowalczyk and Reynolds, 2004; Lien et al., 2006; McCrea et al., 2009; McCrea and Park, 2007; Perez-Moreno and Fuchs, 2006; Reynolds and Roczniak-Ferguson, 2004; van Hengel and van Roy, 2007; van Roy and McCrea, 2005; Xiao et al., 2007; Yin and Green, 2004).
Basics of catenin structure and function
With the exception of the structurally unrelated α-catenin (Benjamin and Nelson, 2008; Pokutta and Weis, 2000; Scott and Yap, 2006; Yang et al., 2001), all catenins contain a central Armadillo domain that contains between nine and twelve repeats (each of roughly 40 amino acids) (Choi and Weis, 2005), which fold to produce a super-helix of helices that bears a positively charged groove and crucial binding interfaces (Huber et al., 1997; Shapiro and Weis, 2009). It is worth noting that several proteins contain Armadillo domains but do not associate with cadherins. By definition, these are not catenins; one example is α-importin, which assists in bringing proteins bearing classic nuclear localization signals into the nucleus (Peifer et al., 1994).
Catenin Armadillo regions engage in numerous protein-protein interactions, and promote the association of β-catenin or p120 subfamily members with cadherins (Bienz, 2005; Meng and Takeichi, 2009; Nelson and Nusse, 2004), nuclear gene regulatory factors (Clevers, 2006) or cytoplasmic Rho-family GTPases (Anastasiadis, 2007). The N- and C-terminal domains of catenins exhibit further interesting properties: for instance, the C-terminal domain of β-catenin facilitates the trans-activation of genes in response to Wnt signaling (Arce et al., 2006; Willert and Jones, 2006).
On the basis of primary sequence homology, vertebrate catenins that contain Armadillo domains are divided into three subfamilies referred to by each of three representative members: β-catenin (β-catenin, γ-catenin/plakoglobin); p120 [p120, ARVCF (Armadillo repeat gene deleted in velocardiofacial syndrome), δ-catenin, p0071]; and plakophilin (plakophilin-1 to plakophilin-3). Invertebrates have fewer catenin proteins, namely β-catenin (although variants exist in Caenorhabditis elegans) (Hardin and King, 2008; Phillips and Kimble, 2009), and a single representative of the p120 subfamily that is most similar in sequence homology to vertebrate δ-catenin.
Divisions between the vertebrate catenin subfamilies are reflected at the functional level. For example, plakophilins largely associate with desmosomal cadherins, whereas members of the β-catenin or p120 subfamilies bind to classic cadherins that are present at adherens junctions. For plakophilin subfamily members, it is the N-terminal as opposed to the central Armadillo domain that binds desmosomal cadherins (desmocollins or desmogleins) (Hatzfeld, 2007). Further protein-protein interactions of catenins are modulated by C-terminal PDZ motifs within most p120 subfamily members, or by more extended N- or C-terminal regions (Choi et al., 2006; Gottardi and Gumbiner, 2004; Mo et al., 2009; Solanas et al., 2004).
In several cases, the same catenin might bind more than one cadherin type. For example, γ-catenin/plakoglobin, p120, plakophilins and p0071 reside at both adherens and desmosomal junctions (Borrmann et al., 2006; Calkins et al., 2003; Hatzfeld et al., 2003; Johnson and Boekelheide, 2002; Kanno et al., 2008). Whereas β-catenin does not appear to bind desmosomal cadherins in vivo (Choi et al., 2009), it does associate and functionally interact with other (non-cadherin) transmembrane proteins, such as the epidermal growth factor receptor and MUC1 (Hoschuetzky et al., 1994; Yamamoto et al., 1997). Perhaps a clearer functional distinction is that catenins of the p120 subfamily, and probably also the plakophilins (Hatzfeld, 2007), modulate the activity of Rho-family GTPases (see below) (Anastasiadis, 2007; Hatzfeld, 2007), a property that is lacking in proteins of the β-catenin subfamily.
A further feature of members of the p120 and plakophilin subfamilies is that they are subject to alternative transcript splicing and translational initiation. With regards to p120, the expression of isoform 1 (generated from the most N-terminal translational start site) leads to more robust RhoA inhibition and cell invasion than do shorter isoforms, and consequently might be associated with the progression of certain cancers (Yanagisawa et al., 2008). Phosphorylation of p120 at specific residues has also been found to modulate the relationship of p120 with RhoA (Castano et al., 2007). Even with these and other new findings, much remains to be learned regarding the roles of differentially spliced, translated and phosphorylated isoforms of p120 or of other subfamily members, in cancer and development (Mo and Reynolds, 1996; Reynolds and Roczniak-Ferguson, 2004; van Hengel and van Roy, 2007).
The functions of catenins
Wnt signaling and gene control
As summarized above, β-catenin has broad, conserved functions, contributing to (among other processes) cadherin-cytoskeletal associations (Abe and Takeichi, 2008; Pappas and Rimm, 2006) and canonical Wnt signaling (Bienz, 2005; Chien et al., 2009; Clevers, 2006; Grigoryan et al., 2008; Nelson and Nusse, 2004). More recently, evidence has surfaced in vertebrates that β-catenin is not the only catenin that transduces canonical Wnt signals to the nucleus. For example, in common with β-catenin, isoform 1 of p120 is metabolically stabilized in response to Wnt-pathway stimulation, acting at certain gene promoters in an additive manner with β-catenin to activate transcription (Kim et al., 2004; Park et al., 2006; Park et al., 2005) (J. Y. Hong, J. I. Park, K. Cho, D.G., H. Ji, S. E. Artandi and P.D.M., unpublished results).
Gene activation by vertebrate p120 seems to result from its ability to displace the transcriptional repressor Kaiso from DNA consensus sites in gene control regions (Daniel, 2007). Recently, alternative models of Kaiso function in the context of Wnt signaling have arisen, which support its indirect (TCF-mediated) association with DNA (Iioka et al., 2009; Ruzov et al., 2009a; Ruzov et al., 2009b). Kaiso further binds to methylated sites in DNA that are important in transcriptional repression (Prokhortchouk et al., 2001; Yoon et al., 2003), but no evidence exists that p120 acts to de-repress these gene control regions. Although speculative, several catenins of the p120 subfamily might ultimately prove to participate in transducing vertebrate Wnt or other signals during development or disease progression.
Other nuclear roles?
Beyond their presence at cell-cell contacts (zonula adherens and/or desmosomal junctions), most catenins localize to the nucleus in restricted cellular or developmental contexts. Whereas nuclear binding partners for β-catenin (e.g. TCF/LEF) and p120 (e.g. Kaiso and Glis2) have been described (Arce et al., 2006; Clevers, 2006; Daniel, 2007; Hosking et al., 2007; Willert and Jones, 2006), little is known about the nuclear associations of other p120 or plakophilin subfamily members (Hatzfeld, 2007; McCrea et al., 2009; Schmidt and Jager, 2005). ARVCF-catenin appears to bind the novel (scaffolding?) protein Kazrin (K. Cho, T. Vaught, J. M. Jennings, M. Kloc, D.G., C. Papasakelariou, H. Ji, A. P. Kowalczyk and P.D.M., unpublished results), which shuttles in and out of the nucleus when not at junctional regions or associated with microtubules (Groot et al., 2004; Nachat et al., 2009; Sevilla et al., 2008). δ-catenin associates with Kaiso (Rodova et al., 2004), possibly sharing gene targets with p120. Plakophilin-2 and an isoform of plakophilin-1 exhibit nuclear localization (Chen et al., 2002; Schmidt et al., 1997), which is phosphorylation dependent in the case of plakophilin-2 (Muller et al., 2003), and allows association with RNA polymerase III and perhaps other factors (Mertens et al., 2001). The nuclear localization of catenins thus raises the question of how expansive their functions will ultimately prove to be within this compartment.
For classic cadherins that are present at adherens junctions, such as E-cadherin and N-cadherin, there exist two distinct and conserved catenin-binding sites (Yap et al., 2007). One resides towards the cadherin C-terminus, and binds either β-catenin or γ-catenin/plakoglobin. The other is membrane proximal, and binds a p120 subfamily member (p120, ARVCF, δ-catenin or p0071). Similarly, desmosomal cadherins possess two distinct catenin-binding sites, one associated with γ-catenin/plakoglobin, and the other with any of the three plakophilin proteins (plakophilin-1, plakophilin-2 or plakophilin-3) (Hatzfeld, 2007).
Most catenins were first isolated as part of a complex with various members of the cadherin superfamily. Thus, their roles at adherens or desmosomal junctions, and in cell polarity and motility, were of immediate interest. These roles are still under active study, and although differing models exist (Abe and Takeichi, 2008; Pappas and Rimm, 2006; Weis and Nelson, 2006), β-catenin and plakoglobin probably facilitate indirect interactions between classic cadherins and the actin cytoskeleton at adherens junctions in vivo. When associated with cadherins, p120 subfamily members have been implicated in lateral (cis) clustering of cadherins (Yap et al., 1998), the tethering of signaling or regulatory entities, such as kinases and phosphatases (Lilien and Balsamo, 2005), and the inhibition of cadherin endocytosis (leading to degradation or recycling) (Bryant and Stow, 2004; Erez et al., 2005; Kowalczyk and Reynolds, 2004; Nelson, 2008; Reynolds and Carnahan, 2004; Troyanovsky, 2005; Yap et al., 2007). In summary, the contribution of β-catenin and p120 subfamilies has been shown or is likely to influence cadherin-dependent adhesion, motility and cell polarity, and consequently, larger processes such as development and morphogenesis, tissue homeostasis and disease progression (e.g. cancer).
The plakophilins, which bind desmosomal cadherins (e.g. the desmocollins and/or desmogleins), might share some functional attributes with p120 subfamily members. For example, plakophilins appear to assist in clustering or stabilizing desmosomal cadherins, and might also modulate small GTPases (Hatzfeld, 2007). Similarities with the β-catenin subfamily include plakophilin-mediated interaction of desmosomal cadherins with the cytoskeleton, although the most obvious associations at desmosomes occur with intermediate filaments as opposed to actin microfilaments (Garrod and Chidgey, 2008; Green and Simpson, 2007; Hatzfeld, 2007; Schmidt and Koch, 2007).
Although much remains to be learned with respect to specific functional outcomes, catenins are subject to modification by kinases and/or phosphatases that are enriched at cell-cell contacts (Andl and Rustgi, 2005; Daniel and Reynolds, 1997; Erez et al., 2005; Lilien and Balsamo, 2005; Wheelock and Johnson, 2003). In some instances, phosphorylation promotes catenin release, enabling catenin cytoplasmic or nuclear relocalization and activity, and occurs simultaneously with altered cadherin function. For example, distinct from canonical Wnt-pathway activation, receptor tyrosine kinase stimulation has in some contexts been indicated to produce a signaling pool of β-catenin upon its release from cadherin and α-catenin (Brembeck et al., 2004; Wheelock and Johnson, 2003). Such phosphorylation-induced movement of catenins to the cytoplasm or nucleus suggests one mechanism for functional coordination between cadherin junctions and other cell compartments.
Modulation of small-GTPase activity and the cytoskeleton
As noted, vertebrate p120 subfamily members have certain intriguing properties that β-catenin lacks. One other such property is their modulation of Rho-family GTPases, such as RhoA, Rac1 and Cdc42 (Anastasiadis, 2007; Keil et al., 2007). In their activated GTP-bound forms, small GTPases act in multiple cellular capacities, with a prominent activity being cytoskeletal regulation. This is reflected in structures that are actin or tubulin based, and in myriad coupled processes, such as cell adhesion, polarity, motility and gene regulation. Maintaining small GTPases in active (GTP-bound) versus inactive (GDP-bound) states is influenced by many factors that act in a positive (e.g. GEFs) or negative (e.g. GDIs or GAPs) manner.
When not bound to cadherin, p120 subfamily members associate with GEFs and GAPs to facilitate small-GTPase activation or inhibition (Wildenberg et al., 2006). This occurs, for example, during cytokinesis in the case of p0071 (Wolf et al., 2006). Additionally, p120 subfamily members can bind directly to small GTPases, as has been shown for RhoA (Anastasiadis et al., 2000). Small-GTPase actions are often complex, which is due in part to crosstalk between members, and they are context dependent (e.g. effects vary according to cell type) (Braga and Yap, 2005). Even so, RhoA activation is frequently associated with Rho-kinase activation, and thereby the promotion of stress-fiber formation and contractility in sessile cells. Conversely, activation of Rac1 or Cdc42 often results in heightened lamellipodia and filopodia functions, respectively. Activation of Rac1 and Cdc42 might thus be reflected in cytoskeletal dynamism, and in a significant proportion of contexts is associated with increased cell motility and invasion. Although isoform and context dependent (Yanagisawa et al., 2008), the p120 and plakophilin subfamilies have generally been found to activate Rac1 (and in some cases also Cdc42) and to inhibit RhoA (Anastasiadis, 2007; Hatzfeld, 2007).
An appealing hypothesis has compared p120 family members with rheostats of cell adhesion versus motility (Anastasiadis and Reynolds, 2001; Grosheva et al., 2001; Reynolds and Carnahan, 2004; Reynolds and Roczniak-Ferguson, 2004). That is, when bound to cadherin, p120 protects E-cadherin from endocytosis and degradation, thereby promoting cell-cell adhesion. Conversely, when dissociated from cadherin, p120 exhibits more pronounced GTPase effects (e.g. RhoA inhibition and Rac1 activation), which in some contexts is associated with motile cell states. One can thus envisage physiological scenarios wherein an epithelial-mesenchymal transition (EMT) is facilitated upon release of p120 from cadherin, resulting in both reduced cadherin protein levels (reduced adhesive function) and increased cytoskeletal activity and/or cell motility. In pathological contexts, this might also be a significant factor. For example, promoter methylation resulting in reduced E-cadherin expression might not only have a direct impact upon cell adhesion and thereby cell polarity in cancer, but also upon cytoskeletal activity as a consequence of increased levels of cytoplasmic p120 (or of ARVCF, δ-catenin or p0071).
Finally, evidence exists that p120 interacts with tubulin (Franz and Ridley, 2004; Ichii and Takeichi, 2007; Roczniak-Ferguson and Reynolds, 2003; Yanagisawa et al., 2004), and has an impact upon microtubule stability and dynamics, and thereby on cell motility and directional migration. p120 further binds the plus-end microtubule motor kinesin, apparently relating to the delivery of junctional components to the plasma membrane (Chen et al., 2003; Yanagisawa et al., 2004) and, of further interest, to proteins providing association with the minus-ends of junctional microtubules that are required for organization of the zonula adherens (Ichii and Takeichi, 2007; Meng et al., 2008). The effects of the action of catenins on the microtubule cytoskeleton appear to occur via mechanisms that are not directly attributable to their roles in conjunction with small GTPases. Indeed, even β-catenin, in common with p120 subfamily members (Franz and Ridley, 2004; Ichii and Takeichi, 2007; Myster et al., 2003; Wolf et al., 2006), has been observed at the centrosome and is proposed to contribute to microtubule functions (Huang et al., 2007). In the context of the cadherin-catenin complex at adherens junctions, p120 appears to assist in minus-end capture (Ichii and Takeichi, 2007) and, conceivably, plus-end capture and stabilization of microtubules (Bellett et al., 2009). Taken together, catenins appear to be well poised to assist in the coordination of cell-adhesive, cytoskeletal, motility and gene-regulatory functions.
Development and disease
The key roles of β-catenin in development are well recognized, and include the modulation of cadherin function (Zhurinsky et al., 2000) and the transduction of canonical Wnt signals (see reviews cited below). Many of the functions of β-catenin are shared between vertebrates and invertebrates (with some complexities arising in C. elegans) (Hardin and King, 2008; Phillips and Kimble, 2009), with whole-animal knockouts or knockdowns resulting in embryonic lethality. Given the large number of β-catenin conditional-null and knockdown studies that have been conducted, we will not summarize findings here. Rather, the reader is directed to more complete resources that outline the diverse roles of β-catenin in Wnt signaling, including its roles in maintaining stem-cell compartments, differentiation, apoptosis and proliferation (Cadigan and Peifer, 2009; Chien et al., 2009; Clevers, 2006; Grigoryan et al., 2008; Schneikert and Behrens, 2007; Stepniak et al., 2009) (www.stanford.edu/~rnusse/wntwindow.html; The Wnt Homepage). Numerous findings have also clarified the pathological contributions of β-catenin to cancer. For example, stabilized mutant forms of β-catenin increase transcription from Wnt and β-catenin (TCF/LEF) target genes, as well as disrupt other signaling pathways that are responsive to the nuclear signaling pool of β-catenin.
The removal of p120 also alters vertebrate development, because it is embryonic lethal in the whole animal (Fang et al., 2004) [unpublished results in Davis and Reynolds (Davis and Reynolds, 2006)]. Although the underlying bases for the phenotypes remain under study, evidence exists for effects upon cadherin stabilization, GTPase modulation and nuclear signaling. Furthermore, the targeting of mouse skin revealed a relationship between p120 and nuclear factor-κB (NFκB) signaling, and led to cell hyper-proliferation and neoplasias (Li et al., 2005; Lynch and Hardin, 2009; McCrea and Park, 2007; Perez-Moreno et al., 2006; Perez-Moreno et al., 2008; Stepniak et al., 2009). Because invertebrates lack plakophilins and contain only a single representative of the p120 subfamily (named ‘p120’, but in fact more homologous to δ-catenin), it might be expected that its loss would produce dramatic effects. However, this has not proven to be the case in most, if not all, studies (Magie et al., 2002; McCrea and Park, 2007; Myster et al., 2003; Pacquelet et al., 2003; Pettitt et al., 2003), which has led to the supposition that greater functional prominence arose for p120-like proteins in vertebrates, perhaps owing to the greater diversification of cell-cell junctions and the structural demands that are placed upon them.
When assessing roles of other vertebrate p120 subfamily members such as ARVCF or δ-catenin, results have varied according to the system examined. For example, each catenin studied in the amphibian Xenopus laevis has proved to be essential in early embryogenesis (gastrulation, neural crest migration, etc.) (McCrea and Park, 2007; Stepniak et al., 2009). However, in the mouse, whole-animal removal of ARVCF did not produce a phenotype (Howard Sirotkin and Raju Kucherlapati, personal communication), and removal of δ-catenin led to non-lethal alterations in dendritic spine architecture and to reduced cognitive functions (Israely et al., 2004). The loss of plakophilin family members in mice produces skin and heart phenotypes, probably reflecting effects upon desmosomes, which are enriched in tissues subject to high mechanical stress (Hatzfeld, 2007). Because plakophilins also modulate the activity of Rho-family GTPases and are proposed to have nuclear roles (Hatzfeld, 2007), it will be interesting to examine further their contributions to development and disease.
Catenins have diverse roles in biology, and function in the plasma membrane, cytoplasmic and nuclear compartments. The p120 and plakophilin subfamilies are distinguished from the β-catenin subfamily in that they exhibit alternative splice isoforms or translational initiation, and modulate Rho-family GTPases. Catenins require considerable further study to better understand their nuclear, adhesive and cytoskeletal (among other) roles, and their upstream modulation by signaling pathways and biochemical events (e.g. phosphorylation). Although much has been revealed concerning β-catenin in the context of Wnt signaling, additional important findings will doubtless be forthcoming, such as a deeper insight into its actions in stem cells. Given the multifaceted roles and diversity of vertebrate catenins and their partially overlapping functions, it appears that catenins together form an intricate functional network that might ultimately be best addressed using systems biology approaches.
This work was funded through an NIH RO1 (GM52112), a Texas ARP Grant, and the March of Dimes (1-FY-07-461-01). Assistance with DNA sequencing and other core facilities was provided from a National Cancer Institute Core Grant (CA-16672) to MD Anderson Cancer Center. We apologize that, owing to space limitations, the large number of key original contributions generated from many laboratories have had to be incorporated in the context of citing reviews. Deposited in PMC for release after 12 months.
- © 2010.