β-catenin–mediated cell-adhesion is vital for embryonic forebrain development

Authors


Abstract

Forming a complex structure such as the mammalian brain requires a complex interplay between cells and different signalling cascades during embryonic development. β-catenin plays pivotal roles in these processes by mediating cadherin-based cell adhesion and Wnt signalling. We show for the first time that β-catenin functions predominantly as a mediator of cell adhesion during early development of the mammalian telencephalon. Immunohistochemical analysis demonstrates that β-catenin is localized, together with N-cadherin, to adhesion junctions at the apical lining of the neuroepithelium. The ablation of β-catenin specifically from the forebrain leads to a disruption of apical adherens junctions and a breakdown of neuroepithelial structures. We show that β-catenin–deficient neuroepithelial cells delaminate and undergo apoptosis. Newborn β-catenin mutants lack the entire forebrain and anterior facial structures. Our data also indicate a lack of TCF/LEF-β-catenin–dependent transcriptional activity in the telencephalon of Wnt reporter embryos. Together with the absence of nuclear β-catenin, this finding suggests that canonical Wnt signalling is not active during early telencephalic development. In summary, we demonstrate that β-catenin mediates cell–cell adhesion in the early telencephalon and is vital for maintaining the structural integrity of the neuroepithelium. Developmental Dynamics 233:528–539, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

The neurons and glia of the vertebrate central nervous system (CNS) are derived from precursors within the neuroepithelium (NE). The neural plate, a monolayer of epithelial cells, transforms during primary neurulation into a pseudostratified NE that finally forms the neural tube. During neurulation, the neuroepithelial cells (NPCs) elongate and acquire a bipolar morphology. At the basal side of the neural tube, NPCs form a basement membrane, which separates it from the surrounding tissue. At the apical side, neuroepithelial cells form adherens junctions with each other to form a lining of the neural tube lumen. For all these processes, cell adhesion and cell–cell communication within the neural ectoderm are needed to provide regulatory signals, to form barriers and layers, and to keep the structural integrity and polarity of the forming CNS.

The adhesion molecule N-cadherin is expressed within the NE and forms the apical adherens junctions (Hatta and Takeichi, 1986; Hatta et al., 1987; Redies and Takeichi, 1993; Aaku-Saraste et al., 1996; Nakagawa and Takeichi, 1998). The neural plate gradually switches its expression from E-cadherin to N-cadherin during primary neurulation. After closure of the neural tube, N-cadherin represents the primary calcium-dependent cadherin within the NE. It associates with the adherens junctions between epithelial cells located at the apical side of the NE (Volk and Geiger, 1986; Hatta et al., 1987; Duband et al., 1988; Aaku-Saraste et al., 1996). Like all classic cadherins, N-cadherin shows specific Ca2+-dependent homophilic binding and adhesiveness and is thought to be involved in regulating morphogenesis (Hatta and Takeichi, 1986; Takeichi, 1988; Takeichi et al., 1990; Radice et al., 1997; Redies, 2000). There is strong evidence that N-cadherin–mediated adhesion is a crucial event for the development of the CNS. In Xenopus and zebrafish, overexpression of N-cadherin results in clumping of ectodermal cells and massive cellular disorganization of the neural tube (Detrick et al., 1990; Fujimori et al., 1990). Ectopic overexpression of truncated forms of N-cadherin cause disruption of cell adhesion in the ectoderm and mismigration of neural crest cells (Kintner, 1992; Bitzur et al., 1994; Nakagawa and Takeichi, 1998). Blocking N-cadherin interaction with specific antibodies administered to chicken embryos leads to disorganization of the neural tube accompanied by rosette formation (Ganzler-Odenthal and Redies, 1998). Finally, N-cadherin null-mice died by embryonic day (E) 10 of gestation with massive defects in heart development and undulated neural tubes (Radice et al., 1997).

N-cadherin binds with its C-terminus to the cytosolic protein β-catenin, which interacts with α-catenin that in turn links the cadherin complex to the cytoskeleton (Nagafuchi and Takeichi, 1989; Ozawa et al., 1989; Aberle et al., 1996; Barth et al., 1997; Yap et al., 1997). The removal or the modulation of the binding properties of α- or β-catenin at the adhesion machinery results in a loss or impairment of calcium-dependent cell adhesion (Hirano et al., 1992; Heasman et al., 1994; Watabe et al., 1994; Haegel et al., 1995; Cox et al., 1996).

Beside its function in mediating cell adhesion, β-catenin is known to play an important role as an intracellular mediator of the canonical Wnt signalling pathway (Huelsken and Birchmeier, 2001; Moon et al., 2002). In the absence of Wnt signalling, cytosolic β-catenin is phosphorylated by GSK3β, leading to ubiquitination and rapid degradation through the proteasome pathway (Aberle et al., 1997). Upon Wnt activation of Frz and LRP receptors, GSK3β is inhibited and β-catenin translocates to the nucleus, where it forms a transcriptional activation complex with proteins of the LEF/TCF family (Behrens et al., 1996; Eastman and Grosschedl, 1999; Miller et al., 1999).

In the developing CNS, Wnt signalling plays an important role in many processes such as axis specification, patterning, and growth (Dickinson et al., 1994; Ikeya et al., 1997; Lee et al., 2000; Patapoutian and Reichardt, 2000; Brault et al., 2001; Megason and McMahon, 2002; Moon et al., 2002; Muroyama et al., 2002; Machon et al., 2003). Members of the Wnt-family are expressed in the forebrain, but most are absent from the early telencephalon (Parr et al., 1993; Grove et al., 1998; Lee et al., 2000; Kim et al., 2001). Gain-of-function studies revealed that only Wnt1 and Wnt3a show high mitotic activity in the neural tube (Megason and McMahon, 2002). Wnt1 and Wnt3a display partially overlapping expression patterns; however, both show only very limited expression in the early telencephalon being restricted to the dorsal midline (Parr et al., 1993). The conditional ablation of β-catenin from the Wnt1 expression domain results in a severe loss of midbrain structures that resemble defects observed in Wnt1/Wnt3a knockout mice, thus revealing an important role of canonical Wnt signalling in the midbrain (Ikeya et al., 1997; Brault et al., 2001). In addition, several groups have shown that ectopic expression of Wnt1 or overexpression of stabilized β-catenin, to mimic canonical Wnt signalling, increases mitotic activity in the nervous system; this finding was interpreted as reflecting a function of β-catenin in regulating the expansion of neural precursors (Dickinson et al., 1994; Ikeya et al., 1997; Chenn and Walsh, 2002, 2003; Zechner et al., 2003).

To explore the role of β-catenin during mouse telencephalon development, we performed immunohistochemical analysis and conditional gene ablation of β-catenin in the early mouse forebrain NE by using the Cre-loxP system. We show that β-catenin is highly enriched at the apical endfeet of the neural precursors and colocalizes with N-cadherin at adherens junctions. The ablation of β-catenin in the dorsolateral neural tube NE at E8.5 disrupts apical adhesion junctions and leads to a complete breakdown of neuroepithelial structures in the forebrain. We show that the integrity of the cell polarity and apical endfeet structures is vital for the survival of neuroepithelial cells in the NE. Hence, we conclude that β-catenin plays a critical role in mediating neuroepithelial cell adhesion in early forebrain development.

RESULTS

Localization of β-catenin in the Early Neuroepithelium

To evaluate the role of β-catenin in the developing telencephalon, we first analyzed β-catenin protein expression in the NE. At E9.5 of mouse development, we found β-catenin immunoreactivity preferentially localized to the apical portion of the NE. Immunostaining with anti–β-catenin antibodies revealed intense labeling of ring-like structures at the endfeet of NPCs (Fig. 1A,C–E), which are reminiscent of adherens junctions. Weaker immunostaining was also observed at membranes along the entire length of the neuroepithelial cells (Fig. 1A). However, we did not observe nuclear staining in the telencephalon, neither with polyclonal nor with monoclonal anti–β-catenin antibodies (Fig. 1A,E). To confirm that β-catenin is localized at apical adherens junctions of the NE, we performed double immunofluorescence studies for N-cadherin and β-catenin in the E9.5 telencephalic NE, because it is known that N-cadherin localizes to the lateral plasma membranes and adherens junctions in the NE. In agreement, we found N-cadherin immunostaining localized at the apical side of the NE overlapping with β-catenin immunoreactivity, suggesting that β-catenin is part of the adhesion complex at the apical NE (Fig. 1B–D). In addition, weaker N-cadherin immunostaining was seen along processes and around the cell bodies of neuroepithelial cells, again overlapping with β-catenin immunoreactivity (Fig. 1C,D).

Figure 1.

AC: Double immunolabeling of β-catenin (A) and N-cadherin (B) in the neuroepithelium (NE) of embryonic day (E) 9.5 mouse telencephali revealed colocalization (C) at apical ring-like structures (arrowhead) at endfeet of neuroepithelial cells (NPCs). Overlay with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI, D) shows localization of the nuclei in the NE. E: At higher magnification, labeling with monoclonal anti–β-catenin antibodies shows prominent staining of ring-like structures at the apical surface (arrowheads). No nuclear labeling was observed (open arrowheads). Insert E: β-catenin immunolabeling plus DAPI. F: Immunoprecipitation (IP) of β-catenin from E9.5 and E10.5 telencephali followed by Western blotting against N-cadherin (N-Cad). N-Cadherin protein of approximately 130 kDa (arrowhead) was detected. G: The 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal) staining of LEF/TCF-lacZ reporter mouse at E9.5. No X-Gal staining is visible in the anterior and lateral telencephalon (filled arrowhead). In contrast, strong X-Gal staining is seen in Wnt expression domains of the developing midbrain (open arrowhead). Dashed lines indicate the level of the sections shown in H (telencephalon) and insert H (midbrain). H: Cross-section through the telencephalon of an E9.5 BAT-gal embryo. X-Gal staining is not detectable in the NE of the anterior–lateral telencephalon (filled arrowhead), whereas prominent X-Gal staining was found on cross-sections through the midbrain (insert H, open arrowhead). Scale bars = 10 μm in A (applies to A–D), 5 μm in E and insert E, 100 μm in H and insert H.

Immunoprecipitation of β-catenin from lysates of telencephali from E9.5 and E10.5 embryos followed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting for N-cadherin confirmed an interaction between N-cadherin and β-catenin (Fig. 1F). The converse coimmunoprecipitation experiment using anti–N-cadherin antibodies and detection of β-catenin by Western blotting gave similar results (data not shown).

The strong immunostaining of β-catenin at the apical endfeet of the NE and the colocalization with N-cadherin provided convincing evidence that β-catenin is involved in mediating adhesion properties in these structures. We did not detect nuclear β-catenin immunostaining at E9.5 and E10.5 in the telencephalon. (Fig. 1E) Therefore, to address whether canonical Wnt-signalling is active in the early telencephalon, we analyzed a transgenic TCF reporter mouse line (Maretto et al., 2003). In these mice, nuclear beta-galactosidase (β-Gal) is under the control of a siamois minimal promotor with multiple LEF/TCF binding sites and is expressed upon β-catenin/Wnt signal activation. The 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal) staining of embryos at E9.5 revealed strong β-Gal expression in the mesencephalon and diencephalon. Little staining was detected at the dorsal midline of the posterior telencephalon (Fig. 1G). In contrast to other brain areas, no β-Gal activity was observed in ventral and anterior–lateral regions of the telencephalon (Fig. 1G,H; and Maretto et al., 2003).

Loss of β-catenin in the Early Telencephalon Leads to Severe Malformations

To elucidate in more detail the role of β-catenin in the developing telencephalon, we took advantage of the Cre-loxP system to ablate β-catenin. To do so, we used a mouse line that expresses P1 Cre-recombinase in the early anterior neural tube under the transcriptional control of the FoxG1 promotor to inactivate the β-catenin locus in mice that carried loxP recombination sequences in this gene (Brault et al., 2001). To follow Cre recombination events, we included the ROSA26R reporter allele into the analysis. Cre-mediated recombination of the ROSA26R allele results in constitutive expression of β-Gal in Cre-expressing cells and their progeny. X-Gal staining of embryos carrying FoxG1cre/wt and ROSA26R alleles revealed the expected early Cre-recombination events in the anterior telencephalon and around the otic vesicles starting at E8.75. At E9.5 and E10.5, recombination and reporter activity were seen in the entire NE of the telencephalon, and the otic and optic vesicles (Fig. 2A,B,D; and Dou et al., 1999; Hebert and McConnell, 2000).

Figure 2.

A: Visualization of Cre-recombination events in embryonic day (E) 9.0 embryos monitored by ROSA26R reporter lacZ expression in FoxG1cre/wt/ROSA26R mice. FoxG1-Cre expression is found in the telencephalon (arrowhead), anterior optic vesicle, and otic vesicle. B: E10.5 ROSA26R reporter staining of FoxG1cre/wt/β-cateninfl/wt control embryo. The telencephalon (arrowhead) shows strong 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-Gal) labeling. C: X-Gal ROSA26R reporter staining of E10.5 homozygous FoxG1cre/wt/β-cateninfl/fl mutant embryo. A strong reduction of telencephalic tissue is visible accompanied by a reduction in the number of X-Gal–positive cells (arrowhead). D,E: Horizontal sections of control and homozygous E10.5 mutants combined with X-Gal ROSA26R reporter staining. In control embryos, Cre-recombination is seen in the anterior and lateral telencephalon. Midline structures seen in control embryos (arrow, D) are missing in mutants (E). The dashed lines indicate the tissue lateral to the midline that is lost in the FoxG1/floxed β-catenin mutants. The distance from the anterior midline to the ganglionic eminence (filled arrowheads, D,E) is reduced in mutants due to the loss of forebrain tissue. In β-catenin mutants, the sharp apical border facing the lumen is disorganized in recombined areas and recombined cells enter the lumen (open arrowhead, E). F,G: Newborn control FoxG1cre/wt/β-cateninfl/wt (F) and homozygous FoxG1cre/wt/β-cateninfl/fl mutant (G) embryos. Scale bars = 100 μm in C,D.

Next, we generated embryos carrying FoxG1-Cre and floxed β-catenin alleles. Mutant embryos heterozygous for FoxG1-Cre and homozygous for floxed β-catenin showed severe anatomical abnormalities in forebrain development (Fig. 2C). By E9.5, minor alterations in the size of the telencephalon were observed (data not shown). However, at E10.0 and E10.5, a substantial reduction of forebrain structures was evident in homozygous mutants compared with control embryos (Fig. 2C). E10.5 mutant embryos showed an almost complete loss of the telencephalic vesicles (compare Fig. 2B and C). In addition, the anterior border of the midbrain was clearly visible in mutants, unlike control embryos where it was covered by expanding telencephalic vesicles. Embryos carrying only one floxed β-catenin allele and the FoxG1-Cre allele developed normally and showed strong Cre activity in the developing forebrain (Fig. 2B).

X-Gal staining of horizontal sections of heterozygous β-catenin mutants at E10.5 revealed Cre activity in the NE of the telencephalon (Fig. 2D). In contrast, homozygous mutants showed a severe loss of the anterior and lateral NE in the forebrain and X-Gal–positive cells were found in the lumen of the vesicle. Furthermore, the pseudostratified, compact nature of the NE was abolished and the sharp apical border facing the lumen of the neural tube was lost (Fig. 2E). The dorsal telencephalon showed a medial–lateral regression of the NE (Fig. 2E). Homozygous FoxG1cre/wt/β-cateninfl/fl mutants were born but died within the first hours after birth. They showed severe malformations and were missing forebrain tissue, eyes, and all the structures of snout and upper jaw (Fig. 2G). In contrast, heterozygous FoxG1cre/wt/β-cateninfl/wt mutants did not show alterations in development (Fig. 2F).

β-catenin Is Necessary to Maintain the Apical Adherens Junctions Between Neuroepithelial Cells

To elucidate the cause of the massive malformations seen in the telencephalon of E10 embryos after ablation of β-catenin, we performed immunohistochemical analysis on mutant embryos. Anti–β-catenin immunohistochemistry at E9.5 revealed a loss of apical immunoreactivity in large areas of the anterior telencephalic neuroepithelium where Cre-recombinase activity was found (Fig. 3A, see also Fig. 2E). Typical β-catenin immunoreactivity at the endfeet of NPCs was only observed lateral to the recombined territory of the anterior telencephalon (Fig. 3A). This loss of apical β-catenin staining was never seen in embryos heterozygous for floxed β-catenin (data not shown). Consistent with the findings of β-catenin being colocalized with N-cadherin, N-cadherin immunolabeling was dramatically reduced in areas where β-catenin labeling was absent (Fig. 3B). Overlay pictures of β-catenin and N-cadherin (together with 4′,6-diamidine-2-phenylidole-dihydrochloride [DAPI] to stain the nuclei) showed the complete loss of adherens junctions at the apical side of the NE and in addition, misplaced cells in the lumen of the telencephalon (Fig. 3C). Regions lacking β-catenin immunoreactivity showed a loss of the NE or a breakdown of structural integrity (Fig. 3C, see also Fig. 2E). Figure 3D shows a higher magnification of the region in which the NE is collapsing. Ring-like immunoreactivity of intact adherens junctions was still visible proximal to the perturbed regions. Closer examinations of mutant embryos revealed the first signs of disruption of the telencephalic NE at E9.5. No disruption was seen at E8.75 (data not shown).

Figure 3.

Immunohistochemical analysis of FoxG1cre/wt/β-cateninfl/fl mutants using double labeling. A: β-catenin immunostaining shows a loss of apical β-catenin in the anterior–lateral telencephalon (arrow). Proximal staining reveals the presence of intact endfeet structures (arrowhead). B: N-cadherin immunohistochemistry shows a similar loss of adhesion junctions in the NE in β-catenin–ablated areas. C: Overlay of β-catenin and N-cadherin immunostaining with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI) nuclear labeling. Misplaced cells are found in the lumen of the telencephalon (open arrowhead). D: High-magnification image showing colocalization of β-catenin and N-cadherin at adhesion junctions (arrowhead) at the transition to a proximal area with disrupted apical adherens junctions (arrow). E,F: FoxG1cre/wt/β-cateninfl/fl mutant analyzed with anti-Nestin and anti–β-catenin double immunofluorescence. Nestin immunolabeling (E) shows staining of misplaced cells in the lumen (filled arrowheads) confirming their neuroepithelial origin (F shows corresponding 4′,6-diamidine-2-phenylidole-dihydrochloride [DAPI] staining together with β-catenin labeling). In areas with detectable apical β-catenin immunoreactivity (open arrows in E,F), neuroepithelial cells (NPCs) show an elongated and organized morphology (filled arrows, E,F) in contrast to areas without apical β-catenin staining, where NPCs are disorganized and rounded (open arrowheads, E,F). Insert E: Nestin staining of control embryo showing radial morphology of normal NPCs. Scale bars = 20 μm in A (applies to A,B), in D (applies to C,D), in E (applies to E,F), in insert F.

To address the developmental status of the neuroepithelial cells and the origin of the cells in the lumen of the mutant telencephalic vesicle, we used antibodies against the intermediate filament protein Nestin a marker for neuronal progenitor cells (Hockfield and McKay, 1985; Frederiksen and McKay, 1988). Immunolabeling of Nestin protein in FoxG1cre/wt/β-cateninfl/fl mutants showed intact radial morphology of neuroepithelial cells in areas with apical β-catenin localization (arrow, Fig. 3E,F). In contrast, β-catenin–ablated areas showed a disorganization of the NE and a rounding of Nestin-positive neuroepithelial cells (open arrowheads, Fig. 3E,F). Cells displaced into the lumen of the forebrain also remained Nestin-positive, confirming their neuroepithelial origin (arrowheads, Fig. 3E,F).

To elucidate if displaced cells remain in a polarized state, we used an anti-PKCζ antibody that marks the apical side of NPCs (Izumi et al., 1998; Manabe et al., 2002). Strong PKCζ labeling was seen at the apical side of the NE in nonrecombined areas of the forebrain (Fig. 4B). These same structures were labeled with an anti–N-cadherin antibody (Fig. 4A). In contrast, recombined areas showed a loss of apical adherens junctions and a loss of polarization as observed by redistribution of PKCζ (open arrowheads, Fig. 4A,B). Displaced neuroepithelial cells in the lumen of the forebrain ventricle showed weak PKCζ labeling over the cell membrane similar to N-cadherin. However, no asymmetric localization could be observed, suggesting that the recombined cells lost their polarized character.

Figure 4.

A,B: FoxG1cre/wt/β-cateninfl/fl mutant (embryonic day [E] 10.0) analyzed by anti–N-Cadherin (A) and anti-PKCζ (B) double immunofluorescence. Strong PKCζ immunofluorescence (filled arrowheads in B) demonstrate the polarization of neuroepithelial cells (NPCs) with intact N-cadherin–immunopositive adhesion junctions (filled arrowheads in A). Cells that lose apical adhesion contacts and enter the lumen of the forebrain lose the asymmetric distribution of PKCζ (open arrowhead in B). C,D: In situ hybridization showing N-cadherin (C) and β-catenin (D) on consecutive cryosections of an E10.0 FoxG1cre/wt/β-cateninfl/fl mutant telencephalon. In β-catenin–ablated areas, cells lose their apical adherens junctions, enter the lumen, and retain positive for N-Cadherin mRNA (arrowhead in C). Thus, loss of β-catenin does not directly result in N-cadherin transcriptional down-regulation.

We addressed whether the ablation of β-catenin leads to an indirect loss of adherens junctions by down-regulation of N-cadherin expression at the transcriptional levels. We performed in situ hybridization experiments on consecutive horizontal sections of FoxG1cre/wt/β-cateninfl/fl mutant embryos using N-cadherin and β-catenin riboprobes. In contrast to β-catenin labeling that was severely reduced in displaced NPCs (Fig. 4D), N-cadherin labeling remained positive (Fig. 4C). These results indicate that the loss of adherens junctions is a direct consequence of the loss of β-catenin protein from the N-cadherin–β-catenin adherens complex and not a transcriptional down-regulation of N-cadherin.

To address the differentiation status of the β-catenin–deficient NE, the expression of the neuronal differentiation markers Ngn1 and Ngn2 was analyzed by in situ hybridization. We did not observe changes in expression of these early neuronal markers nor of neurofilament/β-tubulin III compared with control embryos (data not shown).

To visualize ultrastructural changes in the NE after β-catenin ablation, we analyzed FoxG1cre/wt/β-cateninfl/fl mutants by electron microscopy. By E9.5, a visible loss of adherens zones was evident in the anterior–medial telencephalic NE (Fig. 5A). In lateral areas of β-catenin mutants and in the telencephali of control embryos, adherens junctions were clearly visible (arrowheads, Fig. 5B,C). The loss of adherens zones in mutants was associated with cells falling into the lumen of the forebrain (Fig. 5A). These displaced cells looked mostly pyknotic and showed apoptotic bodies. Debris was also visible in the lumen of the neural tube. In addition, neuroepithelial cells in ruptured areas of the NE were loosely packed and had a rounded morphology (open arrowhead, Fig. 4A), whereas cells in intact regions of the NE were elongated and compact (Fig. 5B).

Figure 5.

A: Electron microscopy of the anterior telencephalic neuroepithelium (NE) of FoxG1cre/wt/β-cateninfl/fl mutants showing a lack of adhesion zones. Cells within the NE have a rounded morphology (open arrowhead) and enter the lumen of the telencephalon. Apoptotic bodies are found in the NE and in most cells displaced in the lumen (arrows). B,C: No apoptotic cells are seen in lateral areas of the telencephalon in mutant embryos (B) or the anterior forebrain of control animals (C). In the control and nonrecombined lateral telencephalic neuroepithelium of mutants, adherens junctions are clearly visible (arrowheads). Scale bar = 10 μm in A (applies to A–C).

Reduction in Telencephalic Size in Mutant Embryos Is Due to Increased Apoptosis Rather Than Reduced Mitotic Potential

Mutant embryos deficient in β-catenin in the telencephalon showed a massive loss of forebrain structures. To address the origin of this malformation, we performed whole-mount terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling (TUNEL) analysis of homozygous mutants and heterozygous control embryos. At embryonic day E8.5 of development, no obvious differences in TUNEL labeling were observed between homozygous mutant embryos and controls (not shown). In contrast, by E9.0 and particularly at E9.5, a strong increase in the number of TUNEL-positive cells was observed in the forebrain vesicles of mutants (arrow Fig. 6B and data not shown). Control embryos showed few apoptotic cells, which were mainly localized to the midline and the tip of the anterior telencephalon (Fig. 6A). The distribution of apoptotic cells in β-catenin–ablated mutant embryos coincides with the Cre expression pattern of the FoxG1-Cre line (Fig. 2A; and Hebert and McConnell, 2000). TUNEL labeling on paraffin sections of E10.5 embryos confirmed the strong increase in apoptotic cells in the mutants (Fig. 6D). Most TUNEL-positive cells were found in the lumen of the telencephalon (Fig. 6D). Control embryos revealed only a small number of TUNEL-positive cells within the NE of the forebrain (Fig. 6C).

Figure 6.

A,B: Whole-mount terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling (TUNEL) analysis of embryonic day (E) 9.5 embryos showing an increase in TUNEL labeling in the anterior and lateral telencephalon (arrows) of FoxG1cre/wt/β-cateninfl/fl mutants (B) compared with FoxG1cre/wt/β-cateninfl/wt control embryos (A). C,D: TUNEL labeling on horizontal sections of E10.0 mutant embryos reveals that most cells that entered the lumen of the telencephalon undergo apoptosis (arrow in D). C: Only little TUNEL staining is seen in control embryos. E,F: Nuclear labeling with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI) of sections. G,H: Immunolabeling with phospho-histone3 antibodies (pH3). In the neuroepithelium (NE) of control E10 embryos, prophase nuclei are located at the apical surface of the neuroepithelium. In mutant embryos, prophase cells are found in areas with intact adherens zones (arrow) and in areas with disrupted adhesion junctions (arrowheads) and also in displaced cells (open arrowhead). Scale bar = 50 μm in C (applies to C–H).

We also analyzed the mitotic activity in the NE of E9.5 mutant and control embryos by using anti–phospho-histone3 (pH3) antibodies (Fig. 6G,H). Prophase cells were detected in β-catenin–ablated mutant embryos in the intact NE (arrow, Fig. 6H) but also in the recombined areas with perturbed NE (filled arrowhead, Fig. 6H). Some pH3 labeling was also seen in cells displaced into the lumen of the telencephalon (open arrowhead, Fig. 6H). The total number of pH3-positive cells in the recombined telencephalon was not significantly changed compared with control animals (pH3+ cells in the anterior and lateral area of eight randomly chosen sections; controls, 47.6 ± 7; mutants, 42.0 ± 5; n = 3 embryos). We did not analyze later stages due to the massive loss of neuroepithelial and forebrain structures. Taken together these data provide evidence that the loss of forebrain structures in homozygous β-catenin mutants is caused by apoptosis of NPCs after breakdown of the apical adhesion junctions.

DISCUSSION

Proliferation, differentiation, and cell–cell interactions within the embryonic NE are key steps in the development of the CNS. Our results provide evidence that β-catenin plays a pivotal role in mediating cell adhesion in the early telencephalic NE. β-catenin is localized at apical adherens junctions at the apical side of the NE, and its ablation from the embryonic telencephalon leads to malformation of the forebrain due to a disturbance of adhesion and subsequent cell death of neuroepithelial cells. This finding suggests a central role for adhesion events in the neuroepithelium to maintain development and structural organization in the early telencephalon.

We show β-catenin immunoreactivity in the early telencephalon at E9.0 with a prominent labeling of ring-like structures that resemble previously described adherens junctions identified by F-actin staining in NPCs (Chenn et al., 1998). Colocalization and coimmunoprecipitation of β-catenin and N-cadherin from embryonic telencephali indicate that classic adherens junctions are formed in the mouse telencephalic NE already by E9.0 of development. Similar results were found in the dorsal NE and later in development in the telencephalon in rat, chicken, and mouse (Aaku-Saraste et al., 1996; Chenn et al., 1998; Manabe et al., 2002).

The ablation of β-catenin from early telencephalic structures using the FoxG1-Cre line resulted in massive deformities of the forebrain. Approximately 24 hr after the onset of Cre-mediated recombination, we observed a loosening of the NE and an invasion of neuroepithelial cells into the lumen of the forebrain supporting the hypothesis that β-catenin plays an important role in mediating cell–cell adhesion. Electron microscopic analysis of E9.5 mutant embryos revealed a loss of adhesion zones in the anterior forebrain, in contrast to lateral areas and control embryos where electron dense junctions were clearly visible (Shoukimas and Hinds, 1978; Aaku-Saraste et al., 1996; Ho et al., 2000). Similarly, the loss of adhesion complexes as monitored by β-catenin immunolabeling was only observed in regions where cells entered the lumen of the telencephalon and N-cadherin labeling was abolished. The disruption of the NE progressed in a developmental manner from rostral to caudal, in accordance with the observed expression pattern for FoxG1-Cre. These results show that, soon after ablation of β-catenin, adhesion complexes are lost.

Regions with a loss of apical adherens junctions showed a reduction in localized immunoreactivity for N-cadherin protein. Because we can show N-cadherin mRNA, a transcriptional down-regulation of the N-cadherin gene due to the loss of β-catenin protein can be excluded. Therefore, we conclude that N-cadherin protein is either degraded, down-regulated at the translational level, or redistributed over the NPC surface. Previous studies have shown that the treatment of chicken embryos in ovo with functional blocking anti–N-cadherin antibodies results in a disruption of adherens junctions accompanied by rosette formation in the NE (Ganzler-Odenthal and Redies, 1998). In contrast, and similar to mutant embryos lacking N-cadherin (Radice et al., 1997; Luo et al., 2001), we did not observe rosette formation in the FoxG1-Cre/floxed β-catenin mutants. One explanation is that the loss of β-catenin results in a global and constitutive loss of adherens junctions in the Cre-recombination domain, unlike antibody treatment, which may cause local and reversible alterations in N-cadherin–mediated adhesion.

Canonical Wnt signalling requires nuclear translocation of β-catenin followed by interaction with TCF/LEF transcription factors and activation of Wnt target genes (Funayama et al., 1995; Behrens et al., 1996; Rocheleau et al., 1997; Thorpe et al., 1997). Our analysis of the early telencephalic NE revealed a lack of nuclear β-catenin. This, and the absence of functional Wnt–β-catenin signalling in the anterior telencephalon as assayed using the BAT-Gal reporter line (Maretto et al., 2003), supports the notion that β-catenin is not substantially involved in Wnt signalling in the telencephalic NE at this early time point. Our findings are in agreement with previous reports of an anterior–posterior Wnt signalling gradient that specifies patterning of the forebrain and midbrain (Patapoutian and Reichardt, 2000; Yamaguchi, 2001; Wilson and Houart, 2004), the exclusion of mitogenic Wnt proteins from the telencephalon (Parr et al., 1993; Lee et al., 2000; Kim et al., 2001; Megason and McMahon, 2002), and the presence of numerous Wnt–β-catenin signalling inhibitors in the anterior neural tube (Niehrs, 1999; Kiecker and Niehrs, 2001; Mukhopadhyay et al., 2001; Lagutin et al., 2003; Wilson and Houart, 2004). Furthermore, it has been shown that ectopic activation of Wnt signals in the anterior telencephalon leads to a caudalization of the forebrain (Kiecker and Niehrs, 2001; Nordstrom et al., 2002), indicating that the early anterior telencephalon is a region of low canonical Wnt activity (Maretto et al., 2003; Wilson and Houart, 2004).

We show that, after ablation of β-catenin from the early telencephalic NE in FoxG1-Cre/β-cateninfl/fl mutants, the ability of β-catenin-deficient cells to undergo mitosis is not changed even though the structural integrity of the NE is perturbed. The lack of active LEF/TCF/β-catenin signalling in the anterior–lateral telencephalon and the observations that proliferation was not reduced in the lateral cortex of E13.5 D6-Cre/β-cateninfl/fl mutant mice (Machon et al., 2003) suggest that β-catenin and the canonical Wnt pathway are not major regulators of proliferation in the NE of the early anterior–lateral telencephalon. The ablation of canonical Wnt–β-catenin signalling in Wnt-expression domains of the neural tube leads, in contrast, to severe reductions of neuronal tissue (Brault et al., 2001) and a reduction in proliferation (Machon et al., 2003; Zechner et al., 2003). Conversely, it has been reported that ectopic activation of the Wnt signalling cascade by overexpression of stabilized β-catenin resulted in an expansion of neuronal precursor cells in the embryonic brain and neural tube (Chenn and Walsh, 2002, 2003; Zechner et al., 2003). However, these findings are in contrast to experiments showing that stabilized β-catenin in utero promotes instructive differentiation rather than proliferation of NPCs in a stage-dependent manner (Hirabayashi et al., 2004). These discrepancies remain to be clarified. Hence, β-catenin plays multiple roles in the NE that are region-specific and may depend on developmentally regulated cues. To this end, we propose that endogenous β-catenin in the early anterior telencephalic NE regulates predominantly cell–cell adhesion.

After ablation of β-catenin from the early telencephalon, NPCs loose adhesion, dissociate and enter the forebrain lumen, where they undergo apoptosis. A selective loss of apical contacts has been shown to be an important step in neuronal differentiation of precursor cells in the NE (Chenn et al., 1998; Cayouette and Raff, 2003; Kosodo et al., 2004). Cells that lost their apical adherens junctions remained Nestin-positive but lose their polarized character as shown by the dramatic loss of apical PKCζ labeling. Because PKCζ as well as N-cadherin immunolabeling were still visible on cells that had lost apical adherens junctions but to a much lower extent and were not anymore asymmetrically localized, we conclude that the apical adhesion junctions are important to maintain the polarity in the developing neuroepithelium. However, we did not observe precocious differentiation in β-catenin–deficient embryos as neither early proneural marker genes such as Ngn1/2 nor later markers such as β-tubulin and neurofilament were induced in our mutant mice. The lack of precocious differentiation excludes it as the cause for apoptosis of NPCs observed in our mutants. However, because NPCs in FoxG1-Cre/β-cateninfl/fl mutants undergo cell death after disruption of the apical adhesion contacts, it is possible that the onset of differentiation in our mutants is inhibited by apoptosis.

Induced cell death of NPCs in FoxG1-Cre/β-cateninfl/fl mice is only observed after apical adhesion structures were lost and NPCs became misplaced in the lumen of the forebrain vesicle. This process is reminiscent of the apoptosis mechanism anoikis, or detachment-induced cell death (Frisch and Francis, 1994; Frisch and Screaton, 2001; Grossmann, 2002). Anoikis has been demonstrated for many epithelial cells and shows the pivotal role of cell anchorage for epithelial cell survival (Frisch and Screaton, 2001; Grossmann, 2002). Cadherin-mediated cell adhesion, together with pro- and anti-apoptotic molecules such as AKT, PI-3-kinase, Caspase-9, and Bcl-2 has been shown to regulate anoikis (Li et al., 2001; Grossmann, 2002; Tran et al., 2002). An anoikis-like mechanism may explain apoptosis of NPCs after disruption of adhesion contacts by ablating β-catenin in the NE. However, apoptosis of NPCs is found throughout development of the telencephalon and represents an intrinsic mechanism to regulate the size and shape of the brain that involves numerous extracellular and intracellular molecules (McKay, 1997; Haydar et al., 1999). Therefore, disruption of local trophic support provided by the environment by means of secreted or membrane bound signalling factors after the loss of cell adhesion could also be responsible for the observed induction of apoptosis in our mutants.

Direct cell–cell interactions have been shown to be crucial for fate and proliferation (Temple and Davis, 1994; Ghosh and Greenberg, 1995; Tsai and McKay, 2000). Our findings indicate that direct cell–cell contacts might be necessary for providing trophic support within the NE and that β-catenin–mediated cell–cell adhesion is of critical importance for the developing telencephalon and nervous system formation.

EXPERIMENTAL PROCEDURES

Matings and Embryos

The FoxG1-Cre line (Hebert and McConnell, 2000) was crossed with the floxed β-catenin line (Brault et al., 2001) to generate FoxG1cre/wt/β-cateninfl/wt mice. These were crossed to β-cateninfl/fl/ROSA26Rfl/fl mice that carried the ROSA26R reporter allele (Soriano, 1999). Mutant offspring inheriting FoxG1-Cre and two floxed β-catenin alleles were used for analysis. Offspring carrying FoxG1-Cre and one β-catenin wild-type allele were used as control embryos. Embryos were obtained from timed pregnancies, with plug date defined as embryonic day E0.5.

Histology and β-Galactosidase Staining

For histochemical analysis, embryos were fixed in 4% paraformaldehyde/phosphate buffered saline (PBS) for 20 min at room temperature (RT). After ethanol dehydration and paraffin embedding, horizontal 5-μm microtome sections were rehydrated and incubated in citrate buffer (3.9 g/L citric acid, 10.22 g/L Na2PO42H2O, 1.5% H2O2, 15 min, RT) followed by extensive washings in water. The sections were then incubated in 66 mM Tris/HCl, 50 mM ethylenediaminetetraacetic acid (EDTA), pH 9.0, at 120°C for 20 min and washed in PBS. Sections were blocked in 1% bovine serum albumin (BSA)/PBS for 1 hr, washed in PBS, and incubated in 0.05% BSA/PBS with the appropriate antibodies (2 hr, RT). The following antibodies were used: monoclonal anti–β-catenin (Sigma, 1:200), polyclonal anti–β-catenin (Sigma, 1:800), anti–N-cadherin (Transduction Labs, 1:200), anti-Nestin (Developmental Studies Hybridoma Bank, 1:30), anti-pH3 (Upstate, 1:200), and polyclonal anti-PKCζ (Santa Cruz, 1:400). After washing in PBS, secondary antibodies were applied (1 hr, RT; fluorochrome coupled anti-rabbit and anti-mouse, Jackson, 1:300). The sections were washed and mounted (AF Citifluor + DAPI 1 μg/ml). Pictures were taken with an Axioscop2 (Zeiss, Germany).

For whole-mount β-Gal–staining embryos were fixed (4% paraformaldehyde [PFA], 0.2% glutaraldehyde in PBS, 15 min, RT), washed (3× PBS, 10 min, RT), incubated in PBS + 0.02% NP40 (10 min, RT), and stained using X-Gal in buffer (10 mM K3Fe(CN)6,10 mM K4Fe(CN)6-3H2O, 2 mM MgCl2, 1 mg/ml X-Gal, 0.02% NP40). Finally 5-μm horizontal paraffin sections were made and counterstained with Nuclear Fast Red.

In Situ Hybridization

Plasmids containing mouse N-cadherin or β-catenin cDNA were used to synthesize digoxigenin-labeled antisense riboprobes according to the supplier's protocols (Roche, Invitrogen) and purified on spin columns (Qiagen). In situ hybridization on tissue sections was performed as previously described (Schaeren-Wiemers and Gerfin-Moser, 1993; Yamamoto and Henderson, 1999). Briefly, E10.0 mouse embryos were fixed overnight (4% PFA in PBS), cryopreserved (20% sucrose in PBS), and embedded in OCT (Miles). Cryosections (10 μm) were acetylated (0.1 M triethanolamine/HCl, pH 8.0, 0.25% acetic anhydride, 15 min at RT) and hybridized with riboprobe (400 ng/ml) overnight at 68°C. The sections were washed (1× standard saline citrate, 50% formamide, 0.1% Tween-20, 2× 45 min, 68°C) and blocked in the presence of 20% inactivated goat serum before incubation with alkaline phosphatase–conjugated anti–digoxigenin-Fab fragments (Roche, 1:3,000). After extensive washing, hybridized riboprobes were revealed by performing a nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) reaction.

Electron Microscopy

Embryos were fixed in 4% PFA+0.1% glutaraldehyde (ON). The heads were removed and thoroughly rinsed in 0.1 M phosphate buffer, pH 7.4 (PB) before postfixation (1% osmium tetroxide, 6.86% sucrose in PB, in the dark, 1 hr, RT). After several rinses with PB, tissues were stained en bloc with aqueous 1% uranyl acetate (1 hr in the dark, RT). The embryonic tissue was then dehydrated in an ascending series of ethanol (30 min, each step) followed by incubations in propylene oxide (2× 10 min) and finally embedded in Durcopan (Fluka, Neu-Ulm, Germany). The tissue was polymerized at 60°C for 2 to 3 days.

From embedded embryonic heads, serial ultra-thin horizontal sections (70 nm) were cut on a Reichert Ultracut S microtome (Leica) and collected on Formvar-coated slot copper grids. Electromicrographs were taken on a LEO 906 electron microscope (Zeiss, Germany).

Western Blot Analysis

Embryonic forebrain tissue was dissected in ice-cold PBS and immediately transferred to lysis buffer and triturated (25 mM HEPES pH 7.4, 150 mM NaCl, 0.5% NP40, 1 mM phenylmethyl sulfonyl fluoride, 10 mM NaF, 0.1 mM Na3Va4, 2 mM NaMo, 2 mM EDTA, Complete protease inhibitor [EDTA-free, Roche]). Lysates were centrifuged (17,000 – g, 20 min, 4°C) and precleared with protein G–agarose (2 hr, Roche), and supernatants were incubated with either anti–N-cadherin (Transduction Labs) or anti–β-catenin (Sigma) antibodies overnight and finally treated for 2 hr with protein G–agarose. Immunoprecipitates were washed 5 × with lysis buffer and analyzed by SDS-PAGE.

TUNEL Analysis

Whole-mount TUNEL staining was performed as described (Yamamoto and Henderson, 1999). TUNEL labeling on sections was performed on 5-μm paraffin sections with fluorochrome-coupled secondary antibodies according to the manufacturer's protocol (Roche, Germany).

Acknowledgements

We thank Frank Sager and Evelyn Wirth for their excellent technical help to this work, Sigrun Nestel for help with the EM, and Brian Pendergrass for genotyping. We thank Susan McConnell and Jean Hébert for providing the FoxG1-Cre line; P. Soriano for providing the ROSA26-R animals; Randy Cassada, Natalia Kan, and Hermann Rohrer for critical reading and constructive comments on the manuscript. The Rat-401 anti-Nestin antibody developed by Dr. S. Hockfield was obtained from the Developmental Studies Hybridoma Bank maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA 52242.

Ancillary