Vertebrates have two Armadillo-like proteins, β-catenin and plakoglobin. Mutant forms of β-catenin with oncogenic activity are found in many human tumors, but plakoglobin mutations are not commonly found. In fact, plakoglobin has been proposed to suppress tumorigenesis. To assess differences between β-catenin and plakoglobin, we compared several of their biochemical properties. After transient transfection of 293T cells with an expression vector encoding either of the two proteins, soluble wild type β-catenin does not significantly accumulate, whereas soluble wild type plakoglobin is readily detected. As anticipated, β-catenin is stabilized by the oncogenic mutation S37A; however, the analogous mutation in plakoglobin (S28A) does not alter its half-life. S37A-β-catenin activates a TCF/LEF-dependent reporter 20-fold more potently than wild type β-catenin, and ∼5-fold more potently than wild type or S28A plakoglobin. These differences may be attributable to an enhanced affinity of S37A β-catenin for LEF1 and TCF4, as observed here by immunoprecipitation assays. We show that the carboxyl-terminal domain is largely responsible for the difference in signaling and that the Armadillo repeats account for the remainder of the difference. The relatively weak signaling by plakoglobin and the failure of the S28A mutation to enhance its stability, may explain why plakoglobin mutations are infrequent in malignancies.
Wnts were originally identified as key regulators of early development in Drosophila and as proto-oncogenes in mammals (Nusse and Varmus, 1992). Wnts are secreted polypeptides that bind to members of the Frizzled receptor family, triggering a series of cytosolic events that increase the stability of the armadillo protein (Drosophila) or its homologs in higher vertebrates (β-catenin and plakoglobin, or Armadillo-like proteins [ALPs]), ultimately leading to changes in gene expression (reviewed in Dale, 1998). Subsequent studies have begun to resolve the molecular mechanism underlying regulation of these ALP's by Wnt signaling. Much of what is known about their regulation is based on studies with β-catenin. When not bound to cadherins (Aberle et al., 1996), β-catenin is targeted for degradation via phosphorylation by the constitutively active glycogen synthase kinase 3 (GSK3) (reviewed in Dale, 1998). GSK3 forms a complex with the product of the tumor suppressor gene adenomatous polyposis coli (APC) and with axin, thereby targeting β-catenin for ubiquitin-dependent proteolysis (Behrens et al., 1998; Farr et al., 2000; Hart et al., 1998; Ikeda et al., 1998; Sakanaka et al., 1998; Yamamoto et al., 1998; Zeng et al., 1997), mediated by the F-box protein Slimb/ β-TrCP (Hart et al., 1999; Jiang and Struhl, 1998; Kitagawa et al., 1999; Latres et al., 1999; Liu et al., 1999; Margottin et al., 1998; Marikawa and Elinson, 1998; Winston et al., 1999). In the presence of a Wnt signal, GSK3 activity is inhibited, and cytosolic levels of β-catenin rise (Cook et al., 1996; Dale, 1998). Cytosolic β-catenin enters the nucleus where it can heterodimerize with members of the T-Cell Factor/Lymphoid Enhancer Factor-1 (TCF/LEF) subfamily of high mobility group (HMG) box DNA binding proteins (Behrens et al., 1996; Huber et al., 1996; Molenaar et al., 1996; van de Wetering et al., 1997). Binding of β-catenin alters the activity of TCFs, thereby affecting target gene transcription (reviewed in Barker et al., 2000). A number of transcriptional targets for Wnt signaling have been identified, including the dorsalizing genes Siamois (Brannon et al., 1997; Carnac et al., 1996; Fan et al., 1998; Kessler, 1997), Twin (Laurent et al., 1997), Xnr3 (McKendry et al., 1997); the proto-oncogenes c-myc (He et al., 1998) and cyclin D1 (Tetsu and McCormick, 1999); fibronectin (Gradl et al., 1999); and the gene encoding the matrix metalloprotease matrilysin (Crawford et al., 1999).
In addition to their interactions with TCF/LEFs, ALPs also serve to anchor actin filaments to adherens junctions by binding simultaneously to classical cadherins and to the actin-binding protein α-catenin (reviewed in Aberle et al., 1996). Vertebrates contain a second type of cadherin-based adherens junction, the desmosome, which serves to anchor the intermediate filament network to the plasma membrane (Kowalczyk et al., 1999). Plakoglobin is specialized to bind to desmosomal cadherins (Gelderloos et al., 1997; Wahl et al., 1996; Witcher et al., 1996); β-catenin normally does not bind to these proteins, but can if plakoglobin is absent, as is the case in mice engineered to lack plakoglobin (Bierkamp et al., 1999). Binding of plakoglobin to desmosomal cadherins blocks plakoglobin's ability to bind to α-catenin, thereby inhibiting the interaction of actin filaments (Gelderloos et al., 1997). Drosophila does not appear to have cytoplasmic intermediate filaments, and therefore has only a single type of cadherin-based adherens junction and a single β-catenin/plakoglobin ortholog, Armadillo (Gelderloos et al., 1997). Both plakoglobin and β-catenin have been shown to rescue the adhesion defects associated with mutations in Armadillo (White et al., 1998). In contrast, neither β-catenin nor plakoglobin can completely rescue the defects in Wnt signaling associated with Armadillo mutations in flies; in fact, both mimic the effects of dominant-negative Armadillo mutations in some tissues (White et al., 1998).
The amino terminal domains of β-catenin and plakoglobin contain a highly conserved consensus sequence for phosphorylation by GSK3 (Figure 1b), which is thought to regulate their post-translational stability (reviewed in Polakis, 1999). Phosphorylation of multiple sites is believed to target ALPs for ubiquitin-dependent proteolytic degradation. Hsu et al. (1998) also identified an apparent transcriptional transactivator in the N-terminal region of β-catenin. ALPs contain a central ‘Armadillo repeat’ region composed of 12 repeats of a ∼42 amino acids motif. This region forms a positively charged groove (Huber et al., 1997) and mediates heterotypic interactions, including binding to cadherins (Aberle et al., 1996; Barth et al., 1997), Axin/Axil (Behrens et al., 1998; Ikeda et al., 1998; Kikuchi, 1999a,b), APC (Rubinfeld et al., 1993; Su et al., 1993), Fascin (Tao et al., 1996), and SOXs (Zorn et al., 1999). Its structural similarity to the nuclear localization sequence receptors, the karyopherins/importins, presumably explains why β-catenin can enter the nucleus in a karyopherin/importin- and RAN-independent manner (Fagotto et al., 1998; Yokoya et al., 1999). The carboxyl terminus of the ALPs has been identified as a transactivation domain and is critical for Armadillo signaling (Hsu et al., 1998; Orsulic and Peifer, 1996; Simcha et al., 1998).
Numerous reports have suggested a role for β-catenin in human tumorigenesis (reviewed in Polakis, 1999). Approximately 85% of colon carcinoma cell lines are null for APC and contain elevated levels of cytosolic β-catenin (Korinek et al., 1997); of the colon tumors that are wild type for APC, approximately 50% have mutations in β-catenin (Sparks et al., 1998). In addition, β-catenin mutations are found in ∼25% of melanoma cell lines (Rubinfeld et al., 1997), ∼30% of hepatocellular carcinomas and cell lines (Huang et al., 1999; Kondo et al., 1999; Legoix et al., 1999; Miyoshi et al., 1998; Nhieu et al., 1999; Ogawa et al., 1999), ∼75% of pilomatricomas (Chan et al., 1999), and smaller percentages of several other tumor types (reviewed in Polakis, 1999). With few exceptions, these are point mutations in β-catenin's putative GSK3 phosphorylation domain, and lead to increased stability and accumulation of cytoplasmic and nuclear β-catenin (Polakis, 1999). In contrast, only one report of an analogous mutation in plakoglobin has been reported in a human tumor (Caca et al., 1999). Moreover, loss of heterozygosity at the plakoglobin locus has been observed in human breast and ovarian cancer (Aberle et al., 1995), and it has been proposed that plakoglobin can act as a tumor suppressor (Simcha et al., 1996). Consistent with this proposal, reduced levels of plakoglobin RNA have been found in cervical carcinoma (Denk et al., 1997) and are linked to unfavorable prognosis in patients with non-small cell lung carcinoma (Pantel et al., 1998).
To gain insight as to why β-catenin mutations are much more commonly encountered than mutations in plakoglobin in human tumors, we undertook a variety of studies to compare the activities of these two ALP's in normal and mutant forms. We found differences in stability of the two proteins in human 293T cells, in their affinities for the LEF-1 and TCF-4 DNA binding proteins, and in their ability to transactivate a LEF/TCF-responsive reporter (OT). We have used chimeric constructs to map the activities of different regions of the polypeptides. Together these studies provide a plausible model that explains the differences in the oncogenic potentials of these two similar proteins.
Expression of exogenous β-catenin and plakoglobin reveals differences in protein stability
To compare the properties of β-catenin and plakoglobin, cDNAs for the wild type versions of both genes were cloned into the same expression vector and transfected into human 293T cells. Since Wnt signaling is believed to occur through regulation of the levels of Armadillo-related proteins by an ubiquitin-dependent, phosphorylation-sensitive process, we first measured the accumulation of soluble forms of the two proteins by Western blot analysis. In conditions under which we observe abundant amounts of plakoglobin, we detected little or no increase in β-catenin over endogenous levels (Figure 2a), suggesting either an unanticipated difference in the rates of synthesis from the two closely related expression plasmids or, more likely, a higher rate of β-catenin degradation.
We next compared the effect of an oncogenic point mutation in β-catenin (S37A β-catenin) and an analogous mutation in plakoglobin (S28A Plakoglobin) (see Figure 1b) on protein levels. As previously reported for similar mutants in β-catenin (Morin et al., 1997; Porfiri et al., 1997; Rubinfeld et al., 1997; Yost et al., 1996), immunoblot analysis of lysates derived from cells transfected with S37A β-catenin-expressing DNA displayed a large increase in the amount of soluble β-catenin (Figure 2a). In contrast, expression of the S28A mutant form of plakoglobin did not increase soluble plakoglobin above the amounts seen in cells transfected with plasmid encoding wild type plakoglobin (Figure 2a). Moreover, expression of plakoglobin, either the wild type or the S28A mutant form, increased soluble endogenous β-catenin (data not shown), consistent with other reports (Miller and Moon, 1997, Simcha et al., 1998).
To determine whether the differences between the levels of soluble β-catenin and plakoglobin reflected differences in the stabilities of the two proteins, we measured protein half lives by pulse-chase methods in transiently transfected 293T cells (Figure 2b). We could not detect soluble β-catenin in cells transfected with a wild type β-catenin expression plasmid, even immediately after the labeling period (data not shown). In contrast, β-catenin was readily detected in cells transfected with the S37A β-catenin expression plasmid, and the mutant exhibited a half-life of greater than six hours (Figure 2b). Wild type or S28A plakoglobin both showed intermediate stability, with a half-life of approximately two hours. Thus, we conclude that wild type β-catenin is more susceptible than wild type plakoglobin to proteolytic degradation, and mutations in the amino terminally portions have dramatically different effects on stability, at least in 293T cells.
Expression of exogenous β-catenin and plakoglobin reveals differences in signaling activity
Excessive transcription of the genes regulated by β-catenin-TCF/LEF heteroduplexes is thought to underlie β-catenin's role in oncogenesis (reviewed in Barker et al., 2000). To compare the transcriptional activities of β-catenin and plakoglobin, constructs encoding various forms of the two proteins were transfected along with a TCF/LEF-dependent luciferase reporter construct (OT) into 293T cells. Transfection of wild type β-catenin into 293T cells had minimal effects on the activity of the OT reporter–only threefold induction when 4 μg of plasmid were introduced (Figure 2c). In contrast, transfection of wild type plakoglobin resulted in larger increases in transactivation–up to 17-fold when 4 μg of wild type plakoglobin expression plasmid were introduced (Figure 2c).
Endogenous or exogenous expression of serine/threonine point mutants in β-catenin is known to increase transcription at TCF/LEF target gene promoters (Morin et al., 1997; Porfiri et al., 1997; Rubinfeld et al., 1997). Introduction of as little as 0.2 μg (data not shown) or as much as 3.0 μg of S37A β-catenin plasmid DNA into 293T cells potently transactivated the OT reporter, roughly 80-fold over baseline (Figure 2c). Although the S37A mutant form of β-catenin signaled ∼25-fold more strongly than wild type β-catenin, S28A Plakoglobin signaled no better than wild type Plakoglobin. For either form of Plakoglobin, 15–20-fold transactivation was the largest observed at any dose, merely a quarter of that observed for oncogenic, S37A β-catenin (Figure 2c).
Both the carboxyl and amino termini of β-catenin are necessary for full transcriptional transactivation
To assess the contribution of the ALP domains to the overall signaling capacity of β-catenin, we tested constructs in which portions of the amino or carboxyl terminus of β-catenin were deleted. Munemitsu et al. (1996) created and characterized a mutant of β-catenin lacking the amino terminal 89 amino acids (N89, Figure 3a), and we tested this construct in the TCF-dependent reporter gene assay. The N89 mutant β-catenin transactivated the OT reporter approximately 15–20-fold over baseline levels–one-fifth as active as the full length S37A form of β-catenin (Figure 3b). The difference between the N-terminal deletion and point mutant forms of β-catenin cannot be ascribed to protein stability, as they accumulate to similar levels based on Western blots (Figure 3c). Thus, like Hsu et al. (1998) and Caca et al. (1999), we conclude that at least part of the transactivation potential resides in the amino terminal domain.
Two deletions of the β-catenin carboxyl terminus were generated, removing the final 72 (C72) or 14 (C14) amino acids (Figure 3a). Transfection of either carboxyl terminal deletion construct into 293T cells results in mild transactivation of the OT reporter–about 10–20% of that induced by similar amounts of full length S37A β-catenin (Figure 3c and data not shown). Again, differences in transactivation could not be ascribed to differences in protein expression between the wild type and the C14 mutants since they were expressed at similar levels (Figure 3c). We failed to detect the C72 mutant, which we believe indicates that the epitope recognized by the β-catenin antibody used in these studies lies between 72 and 14 amino acids from the carboxyl terminus. Thus, we conclude that residues in both the amino terminal 89 and carboxyl terminal 14 residues of β-catenin are necessary either for its full transcriptional activity or its normal entry into the nucleus.
Creation of chimeric proteins to identify functional domains of β-catenin and plakoglobin
To determine the relative contribution of ALP domains to the differential signaling capacities of β-catenin and plakoglobin, we generated chimeric β-catenin/plakoglobin fusion constructs. As described above, β-catenin and plakoglobin can be conceptually divided into three portions: the amino terminal region, the Armadillo repeats, and the carboxyl terminal domain (Figure 1a). Chimeric proteins containing several combinations of domains were created from the parental S37A β-catenin and S28A plakoglobin constructs. These fusions are represented as three letter acronyms that correspond to the β-catenin (B) or plakoglobin (P) source for the amino terminus, Armadillo repeats, and carboxyl terminus, respectively (Figure 4a). At least two independently derived constructs were tested for each chimera and each construct was tested a minimum of five times. The data shown are from a representative experiment. Constructs in which the amino terminus of S37A β-catenin replaced the amino terminus of plakoglobin (B-P-P) did not transactivate OT to any greater degree than S28A plakoglobin (Figure 4b). For unclear reasons, the reciprocal construct (P-B-B) did not accumulate to usual levels in 293T cells (data not shown). Chimeric constructs in which the Armadillo repeats of the two proteins were exchanged indicate that roughly 25% of β-catenin's enhanced signaling capacity relative to that of plakoglobin localizes to the Armadillo repeats (compare S37A β-catenin to B-P-B in Figure 4b). Constructs in which the carboxyl termini of β-catenin and plakoglobin were swapped (P-P-B and B-B-P) suggest that as much as 70% of β-catenin's relative signaling potency stems from differences in the carboxyl termini. Western analysis of protein extracts from cells transiently transfected with the chimeras, using antibodies specific for the carboxyl termini of β-catenin and plakoglobin, confirmed that each (except for B-P-P- data not shown) was well expressed (Figure 4c,d). Western analysis using antibodies to the amino terminus of plakoglobin confirmed that the full length plakoglobin and the P-P-B chimeric constructs were equally expressed (data not shown).
Importantly, we detected a small but consistent increase in the cytoplasmic levels of endogenous β-catenin with transfection of either S28A plakoglobin or chimeric constructs containing the plakoglobin C-terminus into 293T cells (Figure 4c). Such an increase in endogenous β-catenin protein levels in the cytosol is consistent with the proposal by Miller and Moon that introduction of exogenous β-catenin or plakoglobin can induce accumulation of endogenous β-catenin, presumably by overwhelming the degradative machinery (Miller and Moon, 1997).
β-catenin and plakoglobin bind differentially to regulators and effectors of Wnt signaling
Our work suggests that, at least in 293T cells, there are differences in the regulation and signaling activity of β-catenin and plakoglobin. To gain further insight into these differences we examined the relative ability of the two proteins to interact with known binding partners. Axin is one of two related suppressors of Wnt signaling that are implicated in the formation of a complex that also contains GSK3, APC, catenins, dishevelled, and other proteins (Behrens et al., 1998; Fagotto et al., 1999; Farr et al., 2000; Hart et al., 1998; Ikeda et al., 1998; Smalley et al., 1999; Yamamoto et al., 1998). This large complex is believed to regulate the phosphorylation, stability, and function of β-catenin. To compare the axin binding activity of the two armadillo homologues, we constructed plasmids encoding forms of both proteins tagged with the V5 epitope (see Materials and methods) and introduced each of them into 293T cells in the company of a plasmid encoding a myc-tagged form of mouse axin. Anti-V5 immunoprecipitates from the cell lysates showed that both β-catenin and plakoglobin interacted efficiently with axin (Figure 5a).
Neither β-catenin nor plakoglobin contain DNA binding domains, but form heterodimers with TCF/LEF DNA binding proteins to regulate target promoter activity. We reasoned that differential affinities for TCF/LEF may underlie the different abilities of these proteins to transactivate the OT reporter used in these studies. To this end, we compared the abilities of β-catenin and plakoglobin to interact with LEF1 and TCF4 (5b,c). As previously reported (reviewed in Barker et al., 2000), β-catenin binds to both LEF1 and TCF4, although, under the conditions used in this experiment, the interaction with TCF4 was consistently stronger. Plakoglobin also bound to TCF4, although the interaction appeared to be weaker than the TCF4/β-catenin interaction. An interaction between plakoglobin and Lef1 was barely detectable under the conditions used in this assay. Thus, β-catenin seems to interact more strongly with LEF1 and TCF4 than does plakoglobin.
β-catenin mutations occur in numerous human tumors (reviewed in Polakis, 1999). Plakoglobin is a gene highly related to β-catenin (Figure 1a). It encodes a protein containing similar consensus motifs for phosphorylation by GSK3 (Figure 1b) and is postulated to associate with similar proteins in the cell (Gelderloos et al., 1997), yet only one case of a plakoglobin mutation in a human tumor has been reported (Caca et al., 1999). To gain insight into why mutations in β-catenin, but not plakoglobin, are commonly found in human tumors, we transfected plasmids directing the expression of wild type versions of both proteins into 293T cells. Whole cell lysates from these transfections showed that transfection of a wild type plakoglobin expression vector results in increased plakoglobin levels in the cytoplasm. In contrast, introduction of an equivalent amount of wild type β-catenin expression vector was not associated with increased β-catenin levels in 293T cells. This was due to a shorter half-life of the β-catenin protein as shown by pulse/chase analysis (Figure 2c and data not shown). Thus, we conclude that, at least in 293T cells, the levels of β-catenin and plakoglobin proteins in the cytoplasm are differentially regulated. Consistent with this, transient transfection of a Wnt1 expression plasmid into 293T cells leads to an increased amount of cytoplasmic β-catenin while not inducing a similar accumulation of cytoplasmic plakoglobin (data not shown).
The protein levels of β-catenin, and potentially plakoglobin, are regulated by a multiprotein complex formed on the scaffolding protein axin (Behrens et al., 1998; Fagotto et al., 1999; Hart et al., 1998; Ikeda et al., 1998; Kodama et al., 1999; Sakanaka et al., 1998; Smalley et al., 1999; Yamamoto et al., 1998). To determine if affinity for axin could account for the differential protein stabilities observed, we examined the ability of the two proteins to associate with axin and found that both could be efficiently coprecipitated with axin in extracts from 293T cells (Figure 5a). Another possibility is that interactions of β-catenin and plakoglobin with components of the ubiquitin-dependent proteolytic system, following phosphorylation by GSK3, could account for their differences in stability. Following phosphorylation by GSK3, the proteins are targeted for degradation via interaction with the vertebrate homolog (β-TrCP) of the Drosophila slimb protein (Hart et al., 1999; Jiang and Struhl, 1998; Kitagawa et al., 1999; Latres et al., 1999; Liu et al., 1999; Margottin et al., 1998; Marikawa and Elinson, 1998; Winston et al., 1999). Sadot et al. (2000) recently reported that both β-catenin and plakoglobin interact with β-TrCP in 293 cell extracts. They further speculate that plakoglobin may have a stronger affinity for the cadherin-based cytoskeletal system and that this may render it less susceptible to the proteolytic machinery. We agree that differences in protein stability are likely to be cell type-specific (Papkoff et al., 1996), depending on the relative expression levels of a variety of interacting proteins including cadherins.
To further compare the two proteins, a mutation in β-catenin found in human tumors (S37A) (Polakis, 1999) and the analogous mutation in plakoglobin (S28A) were created and cloned into expression vectors for transfection of 293T cells. Western blots of whole cell lysates showed that, consistent with many previous reports of mutations in the GSK3 consensus sites of β-catenin, the S37A point mutation resulted in a dramatic increase the cytoplasmic levels of β-catenin protein. Consistent with another report (Caca et al., 1999), transfection of an expression plasmid encoding the analogous mutation in plakoglobin did not result in increased levels of plakoglobin protein in whole cell lysates. Pulse/chase analysis shows that there is no detectable difference in the half-lives of S28A plakoglobin and the wild type plakoglobin protein in 293T cells (Figure 2c).
To further compare the activities of the two proteins, we assayed the abilities of the wild type and point mutated constructs to transactivate a LEF/TCF responsive reporter (OT). As seen by others (Caca et al., 1999; Korinek et al., 1997; Morin et al., 1997; Porfiri et al., 1997) and consistent with the increased levels of β-catenin protein discussed previously (Figure 2a), transfection of cells with the S37A version of β-catenin resulted in a dramatic increase in the transactivation of the reporter relative to wild type β-catenin expression plasmid. In contrast, the wild type and S28A plakoglobin expression plasmids were similar in their ability to transactivate the reporter. Although Caca et al. (1999) previously noted a small increase in the ability of a similarly mutated (S28L) plakoglobin molecule to transactivate a LEF/TCF responsive reporter, both reports are consistent with the idea that the mutation in the β-catenin protein results in more potent transactivation than a similar mutation in plakoglobin. We feel that these observations may help explain why point mutations in β-catenin are more commonly associated with tumorigenesis than similar mutations in plakoglobin. That is, mutations in β-catenin cause a dramatic increase in the cytoplasmic levels of the protein and its subsequent signaling activity, whereas analogous mutations in plakoglobin do not confer extra activity relative to the wild type version. It should be noted that in some rare situations this might not be the case. In the BT549 breast cancer cell line, which lacks E-cadherin expression, an analogous point mutation in plakoglobin does result in an approximately fivefold increase in transactivation relative to the introduction of a plasmid expressing the wild type version of plakoglobin (Caca et al., 1999).
β-catenin and plakoglobin lack DNA binding domains and must bind to HMG box-containing, DNA binding proteins in order to regulate transcription. To assess how such interactions may contribute to differences in transactivation potential, we performed immunoprecipitation analysis on 293T cells cotransfected with β-catenin or plakoglobin and either of two members of the LEF/TCF family, LEF1 and TCF4. As assessed by immunoprecipitation, β-catenin could interact with TCF4 and LEF1. Plakoglobin could also interact with TCF4, but at a reduced efficiency compared to β-catenin. Very little if any LEF1 was found to co-immunoprecipitate with plakoglobin. This is consistent with the demonstration of Simcha et al. (1998) who reported that β-catenin-containing nuclear structures were more efficient at recruiting LEF1 than plakoglobin-containing nuclear structures. However, others have reported that β-catenin and plakoglobin were both effective in immunoprecipitating LEF1 (Zhurinsky et al., 2000; Huber et al., 1996). Perhaps the use of different buffers for the immunoprecipitation reactions could account for these differences.
In light of the signaling discrepancy between S37A β-catenin and S28A plakoglobin, we attempted to identify regions of these homologs that account for this difference. Chimeric versions of the proteins were created by dividing them into three conceptually defined regions: the central region of the proteins containing the Armadillo repeats (Arm), and the regions on the amino (NT) and carboxyl (CT) sides of the armadillo repeats (Figure 1). These constructs were assayed for their ability to transactivate the OT reporter, revealing that most of the difference between the signaling activities of these homologs maps to the carboxyl terminal region. Analysis of other chimeric constructs also implicated regions within the armadillo repeats in the differential signaling of β-catenin and plakoglobin. The fact that the differences between the two proteins map most significantly to the carboxyl terminus is somewhat surprising based on previous observations (Hecht et al., 1999; Hsu et al., 1998) that the carboxyl termini of both proteins strongly transactivates a heterologous reporter when fused to the relevant DNA binding domain (either Gal4 or LexA). The reason for this difference is not clear. It is notable, however, that the carboxyl terminal amino acid sequence is the most divergent region between β-catenin and plakoglobin (Figure 1a).
One question that arises from these studies is whether plakoglobin has any intrinsic signaling activity. It has been postulated that the signaling activity observed when plakoglobin is overexpressed in tissue culture cells and in the developing Xenopus embryo is due to its ability to interact with elements of the β-catenin–ubiquitin-dependent proteolytic degradation system and thereby cause an elevation in the levels of endogenous β-catenin (Miller and Moon, 1997). The fact that Western analysis of lysates of cells into which plakoglobin constructs have been transfected shows increases in cytosolic β-catenin levels (Figure 4c) is consistent with this hypothesis. However, the carboxyl terminal region of plakoglobin has been shown to have intrinsic signaling activity (Simcha et al., 1998). In addition, plakoglobin, but not β-catenin, has been shown to transactivate a reporter gene downstream of the myc promoter (Kolligs et al., 2000). Moreover, a version of plakoglobin deleted for the amino terminal domain and first Arm repeat fails to stabilize endogenous β-catenin, yet can still induce axis duplication when expressed in Xenopus (Klymkowsky et al., 1999; Rubenstein et al., 1997). In addition, if plakoglobin were acting to signal only in an indirect manner, it is difficult to understand how altering its carboxyl terminus would enhance transactivation (Figure 4b). However, a possible explanation to this could be that the carboxyl terminal region of plakoglobin serves to inhibit plakoglobin/LEF1/DNA complex formation (Zhurinsky et al., 2000). An answer to this question may come from future studies in which plakoglobin constructs are introduced into β-catenin-deficient cells (Haegel et al., 1995; Huelsken et al., 2000).
In conclusion, we have found several differences that may explain why mutations in β-catenin, but not plakoglobin, are commonly found in human tumors. One is that the observed point mutations in β-catenin, but not analogous mutations in plakoglobin, confer added protein stability relative to the wild type version of the protein. Secondly, such ‘activated’ versions of β-catenin are 4–5 times more potent than plakoglobin in the signaling assays we have described. In addition, β-catenin seems to bind with higher affinity to two members of the LEF/TCF family of proteins, which mediate the transcriptional effects of β-catenin and possibly plakoglobin.
Materials and methods
Original plasmid sources
Vectors containing cDNAs encoding the proteins used in this study were obtained from the following sources: β-catenin (P Robbins), plakoglobin (W Franke), LEF1 (R Grosschedl), TCF4 (H Clevers), and Axin (F Constantini). The expression vectors pCDNA3, pCDNA3.1, and pCDNA6 were purchased from Invitrogen. A LEF/TCF responsive reporter (OT, an optimized version of TOPFLASH) was obtained from B Vogelstein. The amino terminal deletion form of β-catenin was obtained from P Polakis.
Creation of constructs used in this study
All PCR reactions were performed using Pfu polymerase (Stratagene). For each construct, at least two independently derived clones were tested and found to be similar in their characteristics. Wild type human β-catenin sequence was amplified using primers to create a 5′ Asp718 site upstream of an optimized sequence for translational initiation and a 3′ BamHI site immediately after the stop codon. This fragment was cloned into the Asp718 and BamHI sites of pCDNA3. Wild type human plakoglobin sequence was amplified using primers to create a 5′ EcoRI site upstream of an optimized Kozak sequence and ATG and a 3′ XhoI site after the stop codon. The fragment was cloned into the EcoRI and XhoI sites of pCDNA3. S37A β-catenin and S28A plakoglobin point mutants were created by site-directed mutagenesis of the respective wild type constructs using a Quick-Change Mutagenesis Kit (Stratagene). In both cases, mutant clones were identified by the presence of an NgoMI site (introduced by the mutagenic oligonucleotides) using restriction digest analysis.
To create chimeric constructs with portions of β-catenin and plakoglobin, plakoglobin cDNA was mutagenized to introduce new restriction sites near the 5′ (Afl3) and 3′ (NdeI) ends of the sequences encoding the Armadillo repeats. The resulting changes yielded silent mutations. These sites are already present in the analogous regions of the β-catenin cDNA. The creation of the Afl3 site in the plakoglobin cDNA allowed us to exchange the amino terminal portions of the molecules, while the creation of the NdeI site in the plakoglobin cDNA facilitated the exchange of sequences following the Armadillo repeats. It should be noted that neither swap changed amino acids in the Armadillo repeats, except for an MET to ILE change in the latter part of the armadillo repeat in swaps which exchanged the carboxyl termini of β-catenin and plakoglobin.
Versions of β-catenin and plakoglobin tagged at the carboxyl terminus with both V5 and His6 were created by amplifying the coding sequences of each molecule to create fragments with 5′ and 3′ restriction sites but without stop codons. These fragments were cloned in-frame into the epitope tagging vector pCDNA6 (Invitrogen). Carboxyl terminal deletions of β-catenin were made using site-directed mutagenesis, creating stop codons at various sites in the construct. Further details of all constructions are available upon request.
Cell culture and transient transfections
For each experiment, 6 cm plates were seeded with equal amounts of 293T cells on the night prior to transfection. Cells were transfected by CaPO4 precipitation (Stratagene). For OT reporter assays, 1 μg of OT and 1 μg of pcDNA 3.1-lacZ were cotransfected with 2 μg of expression plasmid encoding various forms of β-catenin, plakoglobin, or β-catenin-plakoglobin chimera. Empty vector DNA was added to normalize the total amount of DNA in every transfection and each transfection was performed in duplicate. Thirty-six hours post-transfection, cells were harvested, extracts were prepared, and luciferase activity was determined (Promega Luciferase Assay System). Readings were normalized for transfection efficiency by measuring β-galactosidase activity (as in Coso et al., 1995).
Immunoprecipitation and Western blots
Transfected 293T cells were harvested on ice with 1 ml of lysis buffer (50 mM Tris-HCl (pH 7.5), 150 mM NaCl 0.5% NP-40, and protease inhibitors). Lysates were microcentrifuged at 14 000 r.p.m. for 10 min at 4°C. For immunoprecipitations, lysates were normalized for total protein and incubated for 2 h at 4°C, with 2 μg of anti-β-catenin (Transduction Laboratories), anti-plakoglobin (Transduction Laboratories), or anti-V5 (Invitrogen) monoclonal antibodies. Forty μl of anti-mouse IgG agarose beads (Sigma) were added, and samples were incubated for an additional hour at 4°C. Agarose beads were washed three times with 1 ml of 0.2× NP-40 lysis buffer and bound proteins were eluted in SDS sample buffer.
Western blots were incubated at room temperature for 1 h in blocking solution (0.1% Tween-20 in Tris buffered saline and 5% dry milk). Blots were immunostained for B-catenin and plakoglobin with the following antibodies (Transduction Labs Anti-β-catenin (Catalog #C19220) and Anti-plakoglobin (Catalog #C26220); Anti-plakoglobin (Zymed Clone PG-11E4)). For Anti-V5 Westerns, blots were incubated for 1 h in anti-V5-HRP antibody (Invitrogen) diluted 1 : 2000 in blocking solution, washed three times in TBS-Tween, subjected to ECL, and exposed to film. For Anti-myc Westerns, blots were incubated for 1 h in mouse anti-myc monoclonal antibody (Santa Cruz) diluted 1 : 1000 in blocking solution, washed three times in TBS-Tween, incubated in anti-mouse IgG-peroxidase (Boehringer Mannheim) diluted 1 : 10 000 in blocking solution for 1 h at room temperature, washed three times in TBS-Tween, exposed to ECL (Amersham), and exposed to film.
293T cells were transfected (as previously detailed) with V5 epitope tagged β-catenin or plakoglobin plasmid DNA, and 36 h after transfection were grown in DMEM without cysteine or methionine (supplemented with 10% dialysed serum and glutamine) for 30 min. Cells were incubated in media containing 0.3 mCi ml of 35S translabel (ICN) for 2 h, labeling media was aspirated, and cells were grown in normal growth media supplemented with 10 times the normal concentration of cysteine and methionine (Sigma). At the indicated chase times, plates were lysed on ice using 1 ml of 0.5% NP-40 lysis buffer. Lysates were frozen at −80°C, thawed, and spun at top speed in a microfuge for 10 min. Each supernatant was removed to a new tube, 2 μg of anti-V5 antibody was added, and each sample was rocked at 4°C for 2 h. Forty μl of a 50% slurry of anti-mouse IgG agarose (Sigma) was added to each tube, and samples were rocked for an additional hour at 4°C. Pellets were washed three times with 1 ml of chilled NP-40 lysis buffer. Thirty μl of SDS loading buffer was added to each pellet, tubes were boiled, and samples were run out on a 10% Tris Glycine denaturing polyacrylamide mini-gel (Novex). Gels were Coomasie stained and destained using a solution of 20% methanol and 5% acetic acid to confirm equal quantities of precipitated antibody in each sample. Gels were incubated in Enlightening enhancement solution for 30 min, dried onto Whatman paper, and exposed to film.
We would like to thank Mario Chamorro for advice on the pulse/chase analysis and other members of the Varmus laboratory for helpful discussions. BO Williams was a post-doctoral fellow of the Damon Runyon-Walter Winchell Cancer Research Fund. GD Barish was a Howard Hughes Medical Institute-NIH Research Scholar. MW Klymkowsky was supported by NIH grant GM54001.