Requirement for β-Catenin in Anterior-Posterior Axis Formation in Mice

The anterior-posterior axis of the mouse embryo becomes explicit morphologically at embryonic day (E) 6.5, when the first mesoderm forms in the primitive streak region at the posterior side of the embryo. However, recent experiments show that anterior-posterior polarity is established at least 1 d earlier (for reviews see Tam and Behringer 1997; Beddington and Robertson 1999): first signs of anterior-posterior polarity are detectable before primitive streak formation by the asymmetric expression of cerberus-like, Hex, and other markers in the prospective anterior portion of the visceral endoderm (Rosenquist and Martin 1995; Belo et al. 1997; Thomas et al. 1998). Accordingly, genetic evidence in the mouse demonstrates that genes expressed in the visceral endoderm, like Otx2, Lim1, and nodal, are required for anterior development (Shawlot and Behringer 1995; Varlet et al. 1997; Acampora et al. 1998; Rhinn et al. 1998; Shawlot et al. 1998). In addition, genes expressed in the embryonic ectoderm, such as cripto, are also necessary for the establishment of anterior-posterior polarity (Ding et al. 1998). In Xenopus and zebrafish, components of the Wnt signaling pathway have been implicated in the induction of embryonic body axis (Harland and Gerhart 1997; Heasman 1997; Moon and Kimelman 1998; Solnica-Krezel 1999). β-Catenin is a central player of the Wnt signaling pathway (Behrens et al. 1996; Molenaar et al. 1996; Cavallo et al. 1997), and in Xenopus, accumulation of β-catenin on the dorso-anterior side of the embryo is the earliest sign of axis formation (Schneider et al. 1996; Larabell et al. 1997). Accordingly, overexpression of β-catenin in Xenopus embryos induces formation of an additional embryonic axis (Heasman et al. 1994; Funayama et al. 1995).

Protein stability of β-catenin is controlled through Wnt/wingless signaling. Wnt/wingless activates frizzled receptors, and through dishevelled, induces an increase in cytoplasmic β-catenin by preventing its degradation in proteasomes (for review see Cadigan and Nusse 1997). The proteins axin and/or conductin, in cooperation with the tumor suppressor gene product adenomatous polyposis coli, are involved in the control of β-catenin degradation, which depends on serine-threonine phosphorylation of β-catenin by GSK3β and subsequent ubiquitination (Rubinfeld et al. 1996; Yost et al. 1996; Aberle et al. 1997; Zeng et al. 1997; Behrens et al. 1998; Ikeda et al. 1998; Jiang and Struhl 1998). The increased levels of β-catenin allow interaction with transcription factors of the lymphoid enhancer factor (LEF)/T-cell factor (TCF) family and activation of gene expression (Behrens et al. 1996; Huber et al. 1996; Molenaar et al. 1996; He et al. 1998; Tetsu and McCormick 1999). In Xenopus, target genes of this pathway such as Siamois, Twin, and goosecoid have been identified, which play a role in axis formation (Heasman et al. 1994; Brannon et al. 1997; Fan and Sokol 1997; Laurent et al. 1997). In addition, inactivating mutations of adenomatous polyposis coli or activating mutations of β-catenin, which result in constitutive nuclear signaling and gene activation, have been identified in tumor progression (Morin et al. 1997; Rubinfeld et al. 1997; He et al. 1998; Tetsu and McCormick 1999; for review see Polakis 1999).

β-Catenin contains repeated elements, the armadillo repeats, that are flanked by unique NH₂- and COOH-terminal domains (McCrea et al. 1991; Peifer 1993; Huelsken et al. 1994a). The central armadillo repeats of β-catenin bind directly to the LEF/TCF transcription factors, and the COOH-terminal domain of β-catenin largely mediates transcriptional activation (Behrens et al. 1996; van de Wetering et al. 1997; Vleminckx et al. 1999). β-Catenin is also located in adherens junctions, where it binds cadherins through the armadillo repeats and establishes a link to the cytoskeleton via NH₂-terminally bound α-catenin (Butz and Kemler 1994; Hinck et al. 1994; Huelsken et al. 1994b; Aberle et al. 1996; Nieset et al. 1997). Plakoglobin, the closest relative of β-catenin in vertebrates, can also bind cadherins and α-catenin, and in addition, mediates the interaction between desmosomal cadherins and the intermediate filament system (Huelsken et al. 1994b; Troyanovsky et al. 1994; Sacco et al. 1995; Kowalczyk et al. 1997). Ablation of mouse genes that encode components of the cadherin–catenin system results in adhesion defects during embryogenesis: in E-cadherin– and α-catenin–deficient embryos, the epithelium of the trophectoderm disintegrates at the blastocyst stage (Larue et al. 1994; Riethmacher et al. 1995; Torres et al. 1997). Plakoglobin-deficient embryos exhibit abnormal adhesion of myocardial cells that result in heart rupture at midgestation (Ruiz et al. 1996). At late gestation, adhesion defects are also observed in the skin of plakoglobin-deficient mice (Bierkamp et al. 1996).

We report here the effect of β-catenin–null mutations in mice. In embryos that lack β-catenin, we observe a block in anterior-posterior axis formation at E6.0. At this stage, cerberus-like and Lim1 expressing cells are mislocated in the distal visceral endoderm, and other anterior markers like Hex and Hesx1 are not induced. Subsequently, no mesoderm and head structures are formed, and the mutant embryos retain the differentiation of the early egg cylinder stage but continue to grow. Using chimeric embryos, we show that β-catenin function is required in the embryonic ectoderm. β-Catenin–deficient embryos contain well-developed adherens junctions, in which plakoglobin substitutes for β-catenin. In analogy to Xenopus and zebrafish, the observed block in axis formation of β-catenin–deficient mouse embryos appears to reflect a signaling function of β-catenin.

For construction of the targeting vectors, genomic fragments isolated from a λGEM-12 129/Ola library were introduced into the pTV0 vector (Riethmacher et al. 1995). The β-cat^del vector contained at the 5′ arm genomic sequence from the end of the first intron to the beginning of the third exon, i.e., encodes only for the first 14 amino acids of β-catenin. The 3′ arm contained exons 5 to 16. Targeting vector β-cat^lacZ contained a β-galactosidase cDNA with a nuclear localization signal fused to the ATG translation initiation codon of β-catenin. The 3′ arm contained sequences from the end of the sixth intron to exon 16. In the vector β-cat^lacZ(hyg), neomycin was replaced by a hygromycin resistance cassette. Homozygous mutant embryonic stem (ES) cells were generated by the consecutive electroporation of the two targeting vector variants, β-cat^lacZ(neo) and β-cat^lacZ(hyg). Two independent, heterozygous ES cell clones (derived from E14 ES cell line) for the loci β-cat^del and β-cat^lacZ(neo) were used to generate chimeric mice by blastocyst injection as described (Riethmacher et al. 1995). Heterozygous mutant animals were bred on a mixed 129×C57Bl6 background, and showed no developmental abnormalities. PCR genotyping was performed using the primers TGG CTT CTT CAG GTA GCA TTT TCA GTT C, CAT TCA TAA AGG ACT TGG GAG GTG T, and GCC TTC TAT CGC CTT CTT GAC G (β-cat^del), or CAT GGA CAG GGG TGG CCT GA, TGT TTT TCG AGC TTC AAG GTT CAT, and AGA ATC ACG GTG ACC TGG GTT AAA (β-cat^lacZ).

β-Galactosidase staining of embryos, immunofluorescence analysis, and EM were performed as described (Hogan et al. 1994; Huelsken et al. 1994b; Riethmacher et al. 1995; Varlet et al. 1997). Embedding in paraffin or plastic was performed according to the manufacturer's protocols (Paraplast, Sherwood Medical; Technovit 7100, Kulzer Heraeus). Sections were stained with hematoxylin/eosin or 0.1% pyroninG (for lacZ-stained embryos). Size and proliferation of embryos between E5.5 and E7.5 were quantified by counting total numbers of embryonic cells and metaphases on consecutive 4,6-diamidino-2-phenylindole–stained sections. In situ hybridizations with digoxygenin-labeled RNA probes (Ding et al. 1998; Liu et al. 1999) were performed according to the manufacturer's protocols (Boehringer Mannheim) followed by PCR genotyping. For immunogold labeling, embryos were embedded in Unicryl (British BioCell Int.), and semithin sections were used for orientation, followed by immunocytochemical labeling of ultrathin sections with antiplakoglobin antibody and 12 nm colloidal gold goat anti–rabbit IgG (Jackson ImmunoResearch Inc.). Sections were contrasted with uranyl acetate and lead citrate.

Aggregation chimeras were generated from wild-type embryos and embryos obtained from crossing heterozygous β-cat^lacZ and β-cat^del mutant mice as described (Hogan et al. 1994). Injection chimeras were produced either by injection of wild-type blastocysts with heterozygous or homozygous β-cat^lacZ ES cells or by injection of blastocysts obtained from crossing heterozygous β-cat^lacZ and β-cat^del mice with wild-type ES cells (Riethmacher et al. 1995). For high or low ES cell contribution, ∼20 or 5 ES cells were injected, respectively. 50 embryos were produced by either method, and three of the most advanced chimeric embryos between E8.0 and E9.5 were analyzed in detail by sectioning. Mutant cells were detected by β-galactosidase staining, followed by PCR genotyping of blue cells for the additional presence of the β-cat^del allele.

We generated β-catenin–deficient mice by homologous recombination in ES cells using two targeting vectors, β-cat^del and β-cat^lacZ (Fig. 1, a–c). No truncated β-catenin mRNA was detectable in the targeted ES cells (Fig. 1 d), and Western blotting with an antibody against an epitope of the COOH terminus of β-catenin revealed absence of truncated protein (data not shown). Homozygous β-cat^del and β-cat^lacZ mutant mouse embryos showed identical phenotypes: a Mendelian ratio of β-catenin–deficient embryos was observed up to E7.5 of development, but the ratio was reduced to 9% at E8.5, and no homozygous mutant embryos were found at E9.5 (Fig. 1 e, and data not shown). Homozygous mutant embryos lacked expression of β-catenin mRNA (as revealed by reverse transcription PCR, data not shown) and protein (Fig. 2, a and b) and were morphologically indistinguishable from wild-type and of identical size at E5.5 of development (Fig. 2, c–f). All embryonic cell layers of the early egg cylinder stage, i.e., embryonic and extraembryonic ectoderm as well as visceral and parietal endoderm were present. At subsequent stages, wild-type and mutant embryos displayed different morphology: at E6.5, wild-type embryos were stretched and had an extended lumen (Fig. 2 g). Mutant embryos were more compact, but nevertheless exhibited well-structured epithelia (Fig. 2 h, see below). At E7.5, wild-type embryos had formed mesoderm at the posterior side and were in the process of gastrulation (Fig. 2 i). Mesoderm formation and gastrulation were not observed in β-catenin–deficient embryos (Fig. 2 k), and the early mesoderm and primitive streak markers Brachyury and goosecoid were not detectable (Fig. 3, a, b, e, and f). The embryonic and extraembryonic ectoderm of the mutant embryo was found to express Oct4 and bone morphogenetic protein (BMP) 4, respectively, i.e., proximal-distal polarity of the embryo was established correctly (Fig. 3c, Fig. d, Fig. g, and Fig. h). Mutant embryos continued to grow between E5.5 and E7.5: cell numbers increased from 1,500 to 10,000 cells, which corresponds to a 2.5-fold reduction when compared with wild-type embryos (Fig. 3i and Fig. k). After E7.5, apoptosis became prominent in the embryonic ectoderm of β-catenin mutant embryos (as determined by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling [TUNEL]-staining) (data not shown).

In wild-type embryos at E5.5, the homeobox gene Hex is expressed in the visceral endoderm at the tip of the egg cylinder, and the expression domain moves unilaterally to the prospective anterior side at E6.0 (Thomas et al. 1998). Remarkably, Hex is not expressed in β-catenin–deficient embryos at E6.0 (Fig. 4, a and b). The secreted factor cerberus-like is expressed distally and anteriorly in the visceral endoderm of wild-type embryos at E5.5, and is restricted to the anterior end at E6.0 (Belo et al. 1997). We found that in β-catenin–deficient embryos, cerberus-like expression is restricted to the distal tip and does not extend anteriorly (Fig. 4c and Fig. d). In situ hybridizations on sections of mutant embryos confirmed symmetrical expression of cerberus-like in the distal visceral endoderm (Fig. 4m and Fig. p; compare with the asymmetric distribution in wild-type controls in Fig. 4l and Fig. n). Lim1 is a marker for anterior-posterior polarity, which is expressed in the wild-type embryo both in the anterior visceral endoderm and in the posterior mesodermal wings at E6.75 (Shawlot and Behringer 1995; Liu et al. 1999). Lim1 is located distally in β-catenin–deficient embryo (Fig. 4e and Fig. f). Anterior differentiation of mutant embryos could not be observed: Hesx1 is a marker for anterior differentiation in the wild-type embryo, which is initially expressed in the anterior visceral endoderm and prechordal plate precursor at E7.0, and later in the adjacent ectoderm, which will give rise to the ventral prosencephalon (Hermesz et al. 1996; Dattani et al. 1998). Hesx1 is not expressed in the β-catenin–deficient embryos (Fig. 4g and Fig. h). We also tested the expression of Otx2, which becomes restricted to the anterior side of the wild-type embryo at E7.0 (Fig. 4 i) (Acampora et al. 1998). Otx2 was not expressed in β-catenin–deficient embryos at this stage (Fig. 4 k). In addition, the midbrain–hindbrain marker Engrailed1 was not expressed in the mutant embryos (data not shown).

We also observed a lack of proximal-distal regionalization of the visceral endoderm by ultrastructural analysis: in wild-type embryos, the visceral endoderm of the extraembryonic region is composed of columnar, highly vacuolized cells with dense microvilli, whereas cells in the embryonic region are squamous with low numbers of microvilli (Fig. 5, a and c). This regionalization of the visceral endoderm was not observed in β-catenin–deficient embryos; instead, the cells displayed an intermediate morphology with fewer vacuoles and microvilli (Fig. 5b and Fig. d). No morphological differences could be observed in the embryonic or extraembryonic ectoderm between wild-type and β-catenin–deficient embryos. These data indicate that a change in the visceral endoderm accompanies defective anterior-posterior axis formation in β-catenin–deficient embryos.

Homozygous mutant, β-galactosidase–expressing ES cells that lack β-catenin mRNA and protein were generated by the consecutive electroporation of the two targeting vectors β-cat^lacZ(neo) and β-cat^lacZ(hyg) (Fig. 1c and Fig. d; data not shown). Aggregation chimeras between wild-type and homozygous mutant embryos and injection chimeras with high numbers of homozygous mutant ES cells did not develop beyond the egg cylinder stage, i.e., resembled β-catenin–deficient embryos (Fig. 6 b; Fig. 6 a, control; data not shown). Thus, a high contribution of β-catenin–deficient cells is not compatible with normal development beyond E6. Moreover, these experiments show that β-catenin function is required in the embryonic ectoderm, since it is known that injected ES cells contribute preferentially to this tissue in chimeric embryos (Beddington and Robertson 1989; Varlet et al. 1997). However, when low numbers of homozygous mutant ES cells were injected, chimeras continued development beyond the egg cylinder stage, and β-catenin–deficient cells contributed to anterior and mesodermal structures (e.g., headfolds and paraxial mesoderm; Fig. 6c and Fig. d). Thus, β-catenin is not required in a cell-autonomous manner for the formation of anterior or mesodermal structures in the early mouse embryo. Reciprocally, chimeric embryos were generated by injecting high numbers of wild-type ES cells into β-catenin–deficient blastocysts, i.e., extraembryonic tissues like visceral endoderm are produced from β-catenin–deficient cells. In these chimeras, anterior and mesodermal differentiation was observed, as demonstrated by the development of headfolds and heart (Fig. 6e and Fig. f). The extent of rescue in the chimeras correlated with the amount of wild-type cells in the embryonic ectoderm. Taken together, these data indicate that β-catenin function is necessary in the embryonic ectoderm but not in extraembryonic tissues like the visceral endoderm during early postimplantation development of the mouse.

In confocal immunofluorescence analysis, identical staining patterns of E-cadherin and α-catenin were observed in wild-type and mutant embryos at E7.0 (Fig. 7, a and b; data not shown). On an ultrastructural level, intact adherens junctions and desmosomes were detected (Fig. 7e and Fig. g; data not shown). In wild-type embryos, plakoglobin, the closest relative of β-catenin in the armadillo family, is localized preferentially in desmosomes and distributed in a spotty pattern along the membrane (Fig. 7 c; data not shown). In contrast, in homozygous mutant embryos, plakoglobin was detected in increased amounts and was distributed uniformly along the membrane (Fig. 7 d). Immunogold localization confirmed redistribution of plakoglobin to lateral membranes in the mutant embryos (Fig. 7f and Fig. h). Thus, plakoglobin appears to substitute for β-catenin in adherens junctions of mutant embryos. In vitro data had shown earlier that plakoglobin can mediate the interaction between E-cadherin and α-catenin (Huelsken et al. 1994b).

Here we show that anterior-posterior axis does not form appropriately in β-catenin–deficient mouse embryos. Axis formation in the mutant embryos is blocked, and the prospective anterior visceral endoderm is mislocated, as judged by the expression of the markers cerberus-like and Lim1 at the distal tip. Subsequently, no mesoderm and head structures form, and markers of posterior, mesodermal differentiation like Brachyury and goosecoid, as well as markers of anterior differentiation like Hex, Hesx1, Otx2, and Engrailed1, are not expressed in the mutant embryos. Intercellular adhesion is maintained in β-catenin–deficient embryos, since plakoglobin substitutes for β-catenin. Our data implicate the signaling capacity of β-catenin, and thus the Wnt pathway, in early axis formation of mammalian embryos.

In Xenopus, the role of β-catenin and the Wnt pathway in body axis formation and dorso-anterior specification has been well-established (Funayama et al. 1995; Harland and Gerhart 1997; Heasman 1997; Moon and Kimelman 1998). Axis formation precedes gastrulation as indicated by the enrichment of endogenous β-catenin in nuclei on the prospective dorso-anterior side of the blastula (Schneider et al. 1996; Larabell et al. 1997), and a block of β-catenin expression results in failure of axis development (Heasman et al. 1994). β-Catenin is required as a signaling molecule in this process, since a fusion protein consisting of the DNA-binding domain of the transcription factor LEF-1 and the COOH-terminal transcriptional activation domain of β-catenin is sufficient to induce an additional axis in the frog (Vleminckx et al. 1999). Also in zebrafish, signaling mediated by β-catenin and the Wnt pathway is essential for embryonic axis formation (Nasevicius et al. 1998; Peleari and Maischein 1998; Sumoy et al. 1999). Our ablation of the β-catenin gene in the mouse produced a defect in anterior-posterior axis formation at E6, i.e., earlier than gastrulation. Moreover, goosecoid, a target gene of β-catenin–mediated signaling in Xenopus (Laurent et al. 1997; Peleari and Maischein 1998; Roeser et al. 1999), is not expressed in β-catenin–deficient mouse embryos. Therefore, we suggest that in the mouse β-catenin is also required as a signaling molecule for anterior-posterior axis formation. Interestingly, Engrailed2 has been also characterized as a target gene of β-catenin–mediated signaling in Xenopus (McGrew et al. 1999), and we found that the related gene Engrailed1 is not expressed in β-catenin–deficient embryos. β-Catenin function at the egg cylinder stage may depend on interaction with high mobility group box transcription factors such as LEF-1 or TCF3, which are expressed at this stage (Roose 1999). A genetic analysis of the mouse TCF3 function has not been reported; other LEF/TCF family members function during later developmental stages (van Genderen et al. 1994; Verbeek et al. 1995; Korinek et al. 1998). Another mutation of the β-catenin gene in mice was reported previously (Haegel et al. 1995), but anterior-posterior axis formation was not examined in these mutants.

The phenotype of β-catenin–deficient embryos is distinct from other mutant mice that display a defective anterior-posterior axis. Mutations in Smad2, Smad4, or ActRIB, which block signaling of members of the transforming growth factor (TGF) β family, affect embryonic differentiation at the egg cylinder stage before gastrulation (Gu et al. 1998; Sirard et al. 1998; Waldrip et al. 1998; Weinstein et al. 1998). In the absence of Smad2, cerberus-like is not expressed, and the epiblast exclusively forms extraembryonic mesoderm (Waldrip et al. 1998), indicating that Smad2 is required for the initial generation of anterior-posterior organizing centers. This phenotype is more severe than the β-catenin mutant, and suggests that signaling of members of the TGFβ/BMP family may precede or cooperate with the β-catenin–dependent pathway. In cripto^−/− embryos, the anterior-posterior axis is initially formed but mislocated, i.e., both Hex and cerberus-like are expressed distally, whereas Lim1 is expressed proximally. Anterior neuroectoderm forms at the distal end of the embryo, as assessed by the expression of Hesx1 (Ding et al. 1998). This is a less severe phenotype than that observed in β-catenin–deficient embryos, which also show mislocalization of anterior visceral endoderm markers but completely lack subsequent anterior or posterior differentiation. Hex and Hesx1 are not expressed in β-catenin–deficient embryos, suggesting that induction of these genes may depend on β-catenin (Zorn et al. 1999). These data indicate that β-catenin may operate earlier than or cooperate with EGF–cripto, Frl-1, and cryptic (CFC) molecules. Consistent with this, Wnt and β-catenin signaling does not rescue an oep/cripto mutant phenotype in zebrafish (Gritsman et al. 1999). Moreover, gene ablation has shown that Wnt3 is essential for formation or maintenance of the primitive streak (Liu et al. 1999). Wnt3 is not essential for the formation of the anterior organizing center, since anterior visceral endoderm markers are localized correctly in Wnt3-deficient embryos, but no further anterior or posterior differentiation was observed. Hypomorphic mutations of axin, which functions as a negative regulator in the Wnt pathway by reducing β-catenin stability (Zeng et al. 1997; Behrens et al. 1998), and overexpression of chicken Wnt8 in the mouse cause duplication of the posterior axis, i.e., a second primitive streak (Zeng et al. 1997; Poepperl et al. 1997). Taken together, these data indicate that signaling by members of the Wnt pathway plays a role for at least two different steps during axis formation in the mouse: (a) for the establishment of initial anterior-posterior polarity (as detected in our β-catenin mutants); and (b) for the formation of posterior structures such as the primitive streak (as found in the Wnt3 mutants) (Liu et al. 1999). β-Catenin and Wnt3 mutants both lack mesoderm and anterior neural ectoderm, probably due to an essential function of β-catenin in Wnt3-dependent signaling. It is not known which Wnt genes require β-catenin during the initial formation of polarity in the mouse; several Wnt genes are expressed during early stages of mouse embryogenesis (Gavin et al. 1990; McMahon et al. 1992; Bouillet et al. 1996), and might take over important, possibly partially redundant functions. However, β-catenin–mediated signaling is not required for the general specification of all embryonic axes, since markers for the proximal-distal axis like Oct4 and BMP4 are unchanged in the mutant embryos.

We have shown that high contribution of β-catenin–deficient cells to the epiblast of chimeric embryos leads to developmental arrest. In contrast, absence of β-catenin in extraembryonic tissues like the visceral endoderm allowed anterior and mesodermal differentiation. This indicates that β-catenin function is required in the embryonic ectoderm for axis formation, and that signals from the embryonic ectoderm may contribute to the patterning of the underlying endoderm. To further localize the cells that depend on β-catenin, we carried out two types of experiments. First, we used a mouse reporter strain that carries a lacZ transgene under the control of multiple LEF/TCF binding sites (Roose 1999) (a generous gift of Dr. H. Clevers, University of Utrecht, Utrecht, The Netherlands). Whereas β-galactosidase activity was detected at E6.5 (Roose 1999), we could not locate activity at E5.5 and E6.0. However, it should be noted that this promoter appears not to respond to all β-catenin–mediated signals in vivo or in cell culture. Second, we attempted to identify nuclear β-catenin at the egg cylinder stage by confocal immunofluorescence, but could not detect such a signal. Moreover, experiments with chimeric embryos that have a low contribution of β-catenin–deficient cells show that β-catenin is not required cell-autonomously for mesodermal differentiation. This is in accordance with experiments in Xenopus that demonstrate that β-catenin–deficient marginal zones can be instructed by β-catenin–overexpressing animal caps to form dorsal mesoderm (Wylie et al. 1996).

We showed that cellular adhesion of epithelial cells in the early mouse embryo is not grossly disturbed in the absence of β-catenin. Epithelia in the mutant embryos are well-developed, and the cells are connected by well-defined adherens junctions and desmosomes. Furthermore, β-catenin–deficient cells in chimeric embryos contributed to various epithelia such as head and limb bud ectoderm. α-Catenin was found to be located along lateral cell membranes in mutant embryos, although β-catenin is normally required to connect α-catenin to classical cadherins. Instead, the protein level of plakoglobin was found to be enhanced, and plakoglobin was redistributed to adherens junctions in β-catenin–deficient embryos. Apparently, plakoglobin can take over the function of β-catenin in cell adhesion of mutant embryos, a function which has been investigated by in vitro experiments (Huelsken et al. 1994b). This prevents the early disintegration of epithelia due to defective adhesion that is observed at the blastocyst stage in mice mutant for the E-cadherin or α-catenin genes (Larue et al. 1994; Riethmacher et al. 1995; Torres et al. 1997). Cell detachment from the ectodermal cell layer at E7 was reported in the previously generated mutation of the β-catenin gene (Haegel et al. 1995). It was not rigorously shown that this earlier mutation corresponds to a null allele. Given the structure of the used targeting vector in Haegel et al. 1995, it is possible that an NH₂-terminally truncated β-catenin protein is produced from the mutant allele described previously, which could bind to cadherins but not to α-catenin. Such a molecule would act in a transdominant manner and disturb adhesion, in contrast to the rescue of adhesion we observe in our null-mutants. It should be noted that no β-catenin mRNA or protein is produced by both our β-cat^del or β-cat^lacZ alleles, which therefore represent null mutations of the β-catenin gene. Thus, our data suggest that plakoglobin can substitute for the adhesive function of β-catenin at the egg cylinder stage. Apparently, plakoglobin cannot compensate for the proposed signaling function of β-catenin in axis formation. This is in accordance with previous findings in Xenopus, which indicate that plakoglobin does not significantly participate in Wnt signaling (Kofron et al. 1997; Miller and Moon 1997; Ben-Ze'ev and Geiger 1998).

The early postimplantation lethality caused by null mutations in β-catenin precludes the analysis of its function in many tissues and events at later developmental stages that depend on Wnt signaling. A function of β-catenin in hair formation and in skin tumors such as pilomatricomas has been identified by the overexpression of an activated form of β-catenin in the skin of mice (Gat et al. 1998). Conditional gene ablation will also allow to study functions of β-catenin at later developmental stage.

We thank Dieter Riethmacher (Max-Delbrueck-Center, Berlin, Germany and Centre for Molecular Neurobiology, Hamburg, Germany) for advice in generating mutant mice; Martin Blum (Institute of Genetics, Forschungszentrum, Karlsruhe, Germany) for the introduction to early mouse embryogenesis and in situ hybridization probes; Kurt Herrenknecht (EISAI Company, London) for providing the antiserum against plakoglobin; and Rosa Beddington (Medical Research Council's National Institute for Medical Research, Mill Hill, London), Tewes Bouwemeester (European Molecular Biology Laboratory, Heidelberg, Germany), and Maki Wakamiya and Richard Behringer (M.D. Anderson Cancer Center, University of Texas at Houston, Houston, TX) for providing further probes for in situ hybridization.

This work was supported in part by a grant of the Volkswagen-Stiftung to J. Huelsken and W. Birchmeier.