Cadherins constitute a large family of calcium-dependent cell–cell adhesion molecules that have been shown to play important roles during the development and tissue organization of multicellular organisms (for reviews see Gumbiner, 1996; Huber et al., 1996a; Takeichi, 1995; Vleminckx and Kemler, 1999). Several subfamilies of cadherins have been defined, including classical cadherins, desmosomal cadherins, and protocadherins (Kemler, 1992; Suzuki, 1996a). Classical cadherins are transmembrane glycoproteins, characterized by their highly conserved cytoplasmic domains and the presence of five homologous repeats in the extracellular region of the molecule. The cytoplasmic domain links classical cadherins to the actin cytoskeleton via cytosolic proteins known as catenins. Desmosomal cadherins possess an extracellular region that is similar to the classical cadherins, but differ in their cytoplasmic domains, which links them instead to the intermediate filament network (for review see Garrod, 1993). Protocadherins are a relatively recently discovered and structurally divergent group of cadherins (Sano et al., 1993; Wu and Maniatis, 1999). They are believed to be the common precursors of the cadherins because their extracellular cadherin repeats display the highest intramolecular homology among all the cadherins. In addition, their gene clusters were found on different chromosomal locations than those of other cadherins, and also, they are found in a wide range of organisms, occurring in invertebrates as well as vertebrates (Suzuki, 1996b). So far, however, little is known about the physiological roles of protocadherins in vivo.
With the isolation of rat LI-cadherin (Liver-Intestine cadherin), a new cadherin was identified that is structurally different from any of the known cadherins (Berndorff et al., 1994). Its extracellular cadherin repeats clearly bear homology to those of classical cadherins. However, instead of five, LI-cadherin contains seven cadherin repeats. The cytoplasmic domain of LI-cadherin consists of only 20 amino acids and displays no similarity to the cytoplasmic domains of the other cadherin subfamilies. Together with Ksp-cadherin, another cadherin possessing a similar structure (Thomson et al., 1995), LI-cadherin represents a distinct group within the cadherin superfamily known as 7D-cadherins (7 Domain cadherins; Kreft et al., 1997). The stable adhesion mediated by most cadherins is largely dependent upon their ability to associate with the cytoskeleton. However, LI-cadherin is still capable of mediating cell–cell adhesion, even when its cytoplasmic and transmembrane domains are deleted and replaced by a glycosyl phosphatidylinositol anchor (Kreft et al., 1997). To date, no intracellular binding partners have been identified for LI-cadherin.
Many classical cadherins show a wide tissue distribution pattern in the adult organism. In contrast, LI- as well as Ksp-cadherin are very restricted in their expression. In rats, LI-cadherin is expressed exclusively in a subset of hepatocytes and in intestinal epithelial cells (Berndorff et al., 1994), and Ksp-cadherin has been only found in kidney epithelial cells in rabbits (Thomson et al., 1995). In the intestine, LI-cadherin is co-expressed with the classical cadherin, E-cadherin. Both cadherins exhibit a basolateral cell surface distribution, but only E-cadherin has been found to accumulate in adherens junctions. Whereas E-cadherin has been shown to be instrumental in the maintenance of epithelial integrity and cell polarity (Behrens et al., 1985; Hermiston and Gordon 1995a; McNeill et al. 1990), nothing is known about the role of LI-cadherin in epithelial tissue organization.
Classical cadherins also play important roles during vertebrate development. The inactivation of E-cadherin in the mouse embryo through gene knockout, demonstrated that this cadherin is essential for the maintenance of the early mouse blastocyst trophectoderm (Larue, 1994; Riethmacher et al., 1995). Later in development, differential expression of several cadherin subtypes is thought to control important morphogenetic events such as cell condensation, cell sorting, and tissue stratification during embryogenesis (Gumbiner, 1996; Takeichi, 1995). The creation of knock-outs for N- and VE-cadherin demonstrated the vital roles of these cadherins in morphogenetic processes like heart and vasculature development (Radice et al., 1997; Vittet et al., 1997). Moreover, cadherins play important roles in the control of tissue differentiation, as shown for E- and N-cadherin in the differentiation of embryoid bodies (Larue et al., 1996), and for P-cadherin in mammary gland development (Radice et al., 1997).
So far, the role of LI-cadherin during embryonic development has not been investigated. To obtain an initial insight into the developmental function of this cadherin, we characterized the temporal and spatial expression pattern of LI-cadherin during mouse embryogenesis. Our results show that LI-cadherin is expressed rather late in development (embryonic day 12.5). Except for transient expression in the urogenital sinus and common bile duct at day 13.5 p.c., LI-cadherin was found exclusively in the intestinal epithelial layer, where it is co-expressed with E-cadherin. Interestingly, LI-cadherin expression coincides with the morphological transformation of the multilayered epithelium into a single-layered columnar epithelium, a developmental stage during which major tissue remodeling, growth, and differentiation occurs in the embryonic intestine.
Isolation and Characterization of Mouse LI-Cadherin cDNA
To gain insights into the extent of conservation of LI-cadherin between species and to obtain species-specific probes for the investigation of LI-cadherin in mouse, we first isolated the cDNA clone for mouse LI-cadherin. A full-length rat LI-cadherin cDNA (Berndorff et al., 1994) was used to screen a 17-day p.c. mouse embryo cDNA library. A single clone was isolated that contained a 3.5-kb cDNA insert with an open reading frame of 2.6 kb (Fig. 1; GenBank accession number: BankIt 287270 AF 177669). The predicted mouse LI-cadherin protein showed a 91% amino acid sequence identity with its rat homologue (data not shown) and also appears to share all of its structural properties. The extracellular region of mouse LI-cadherin contains seven cadherin-like domains, EC1 to EC7. Four cystein residues that are present in the domain most proximal to the transmembrane domain of all classical cadherins are preserved in EC7, but two additional cysteins were found in EC4 that are not present in classical cadherins. In addition, mouse LI-cadherin lacks sequence motifs responsible for calcium-binding between the second and third extracellular domain, whereas these motifs are present in all other domains. Furthermore, a short cytoplasmic domain of 20 amino acids was identified that does not resemble those of any other cadherins, but is homologous to a cytoplasmic domain found in the rat LI-cadherin.
Expression of LI-Cadherin in Adult Mouse Tissues
The expression pattern of LI-cadherin was determined in various tissues of adult mice. Northern blot analysis revealed a transcript of 3.6 kb in small intestine and colon as predicted from the length of the LI-cadherin cDNA. This transcript was not detected in heart, brain, lung, liver, skeletal muscle, spleen, kidney, testis, and stomach (Fig. 2A). Increased exposure time of the Northern blot by a factor of 30 revealed a weak signal indicating RNA of the same molecular weight in spleen (data not shown). The weak intensity of the signal indicates that the abundance of LI-cadherin mRNA in this tissue is several orders of magnitude lower than in small intestine and colon. It was surprising that LI-cadherin was not detected in mouse liver since LI-cadherin was isolated originally from the rat hepatoma cell line HepG2 and has also been found in the livers of adult rats (Berndorff et al., 1994). The lack of LI-cadherin in mouse liver was confirmed by Western blot and RT-PCR (Fig. 3). A protein band of the expected molecular weight of 120 kDa was detected in tissue lysates from intestine but not from liver, although the gel was loaded with significantly more total protein from liver than from intestine as determined by Coomassie Blue staining (Fig. 3A). Concordantly, RT-PCR analysis did not identify LI-cadherin transcript in liver, but did confirm the presence of the mRNA in the intestine (Fig. 3B). In comparison, E-cadherin was detected in both tissues by Western blot and RT-PCR analyses.
LI-cadherin specific immunostaining of a cross-section of adult small intestine shows that LI-cadherin is expressed in all epithelial cells of the crypt and along the villi (Fig. 4). LI-cadherin occurs at regions of cell-cell contact and is absent from the apical surface of the epithelial cells. Mesenchyme and muscle cell layers did not show any LI-cadherin expression.
Temporal Expression of LI-Cadherin During Mouse Embryogenesis
The temporal expression of LI-cadherin during embryonic development was analyzed by Northern blot and RT-PCR. Northern blots performed on poly A+ RNA from embryos at different developmental stages show a single mRNA band of a molecular weight of 3.6 kb for embryonic day 17 (Fig. 5A). A much weaker signal of the same size was detected at day 15, with no signal being observed at day 11 and day 7 p.c. To define more precisely the onset of LI-cadherin mRNA transcription prior to day 15, we performed RT-PCR with samples of total RNA derived from mouse embryos at days 10.5, 11.5, and 12.5 p.c. A clear signal the size of the predicted PCR-poduct was detected at day 12.5 p.c., a much less intense signal at day 11.5d p.c., and no signal at all at day 10.5 p.c. (Fig. 5B). Given the high sensitivity of RT-PCR, we conclude that LI-cadherin expression is turned on at day 11.5. In comparison, the E-cadherin RT-PCR product was detected in embryos of all three stages.
Expression Pattern of LI-Cadherin During the Development of the Intestine
Our results obtained from Northern blots and RT-PCR suggest that LI-cadherin transcription begins at 11.5d p.c. Immunostainings for LI-cadherin and E-cadherin were performed with tissue sections of fetal intestine from embryos at day 11.5 p.c. until day 16.5 p.c. LI-cadherin protein was not detected at embryonic day 11.5 (Fig. 6A), but first strong expression of LI-cadherin had appeared by day 12.5 p.c. (Fig. 6B) and persisted through all following stages. This confirmed our results obtained with Northern blots and RT-PCR.
At all developmental stages examined, LI-cadherin expression in the intestine was confined to the epithelial cell layer (Figs. 6–8). LI-cadherin was distributed on whole cell surfaces of the deeper layer of the epithelium, but was restricted to the basolateral cell surfaces in the outer epithelial layer facing the lumen of the intestine. Figure 6B depicts a cross-section of the midgut within the umbilical hernia at day 12.5, showing positive staining for LI-cadherin in the multilayered epithelium of the intestine surrounded by mesenchyme. At day 13.5 p.c., strong staining for LI-cadherin was detected in the epithelium of all regions of the intestine proximal to distal, including the duodenum, proximal midgut, and hindgut (Fig. 7A). At this stage, additional staining was detected in the epithelia of the urogenital sinus (Fig. 7A) and the common bile duct at the point where it enters the duodenum (Fig. 7B).
During fetal development, the mammalian intestine undergoes considerable changes in morphology and cell differentiation (Madara et al., 1981; Simon-Assmann and Kedinger, 1993). During this transition, the undifferentiated stratified epithelium develops into a single-layered columnar epithelium and villi are formed. These morphological changes occur in the mouse embryo between days 14.5 and 16.5 p.c. as depicted in Figure 8. While the intestinal epithelium is still a cylinder consisting of multiple cell layers at day 14.5 (Fig. 8A–A″), the formation of villi becomes clearly visible by day 15.5 p.c. (Fig. 8B–B″). Sections of day-15.5 embryos also show that the epithelium at the tips of villi have developed into a single cell layer. By day 16.5, villi are well defined and the whole epithelium consists of a single cell layer (Fig. 8C,C′). LI-cadherin was consistently expressed in the epithelial cells through the whole process of morphogenetic transformation of the epithelium (Fig. 8A–C).
Parallel experiments performed with antibodies against E-cadherin revealed that this cadherin is already present in the intestine at day 11.5 p.c. (Fig. 6A). At all stages, the expression pattern was similar to LI-cadherin and confined to the epithelial cell layer. In addition to the intestine, E-cadherin was detected in the epithelia of a variety of other tissues, including liver, pancreatic primordium (see Fig. 7A″,B′), lung, epidermis, and kidney (not shown). Occasionally, a punctate staining pattern was detected at the apico-lateral cell–cell contacts presumably indicating the adherens junctions (arrows in Fig. 7C′). No such staining pattern was observed for LI-cadherin (Fig. 7C).
Immunostaining for β-catenin, a protein known to bind to the intracellular domain of classical cadherins, overlapped with that of E-cadherin in the epithelium. However, additional β-catenin was detected in the mesenchyme and presumptive muscle layer surrounding the epithelium (Fig. 8A″,B″), regions where no E-cadherin was detected. This suggests that a different type of cadherin may be present in this tissue.
Differential Expression of LI-Cadherin in Mouse Adult Tissues
Previous studies of adult rat tissues have shown that LI-cadherin is expressed in both liver and intestine (Berndorff et al., 1994). Surprisingly, our studies revealed that, in mice, LI-cadherin is abundant in colon and small intestine, but not in liver. Comparison of promoter and enhancer regions of rat and mouse LI-cadherin genes may reveal whether different control elements or DNA-binding regulatory proteins are responsible for the species-specific expression pattern of LI-cadherin. At this point, the functional role of LI-cadherin in rat liver remains to be explored as well as why a closely related species seems not to require the adhesion molecule in this tissue. In addition to the strong signals observed in the intestine, the Northern blot analysis also revealed a weak signal in spleen. We conclude that there must be small amounts of LI-cadherin in this organ. Support for this conclusion was found in a recent publication by Ohnishi and colleagues (2000) who identified a cadherin that proved to be LI-cadherin after we compared sequences (not shown). The authors showed the presence of the cadherin in B-cells which reside in the marginal zone of the spleen. Hence, it is very likely that our signal in the Northern blot analysis originates from mRNA of B-cells but not from spleen tissue itself.
Immunohistochemistry of the bladder tested negative for LI-cadherin (data not shown), although its precursor tissue, the urogenital sinus, tested positive for LI-cadherin in the embryo (see Fig. 7A). In the same analysis placental tissue also tested negative for LI-cadherin (data not shown).
Expression of LI-cadherin During Mouse Intestinal Development
To gain an initial insight into the role of LI-cadherin during development, we analyzed its expression in mouse embryos. We found that LI-cadherin gene expression begins rather late in development at embryonic day 12.5. With the exception of a transient expression in the epithelia of the common bile duct and the urogenital sinus, the expression is restricted to the basolateral cell surfaces of the epithelial layer of the intestine. During the development of the fetal intestine, many genes are differentially expressed along the proximal to distal axis (for review see Hermiston et al., 1994). Using mouse embryos of day 12.5 and older, we prepared numerous crossections covering all regions of the intestine in every examined stage and never observed the absence of LI-cadherin in the epithelium beginning with day 12.5 p.c. However, indirect immunolabeling as performed in this study, is only semiquantitative and we cannot exclude the possibility that LI-cadherin may be present in different amounts in epithelial cells from different regions of the intestine. It remains to be determined whether LI-cadherin plays roles in the development of proximal-distal polarity of the intestine or whether it underlies regulative control mechanisms that determine differential gene expression along the proximal-distal axis. Since the expression of LI-cadherin does not correlate with the separation or fusion of tissues, it seems not to participate in morphogenesis via a differential adhesion mechanism, as has been described for classical cadherins involved in early development (Takeichi, 1987).
The present study reveals that LI-cadherin occupies the same cell surfaces in the intestinal epithelium as E-cadherin. E-cadherin, a classical cadherin, is known to play an instrumental role in the maintenance of polarity in epithelial cells and the formation of the junctions that keep epithelial cell layers intact and performing their physiological functions while constituting a physical barrier. The importance of this cadherin in the maintenance of a functional intestinal epithelium in the adult organism was demonstrated by Hermiston and colleagues (1995a, 1996). Other research has shown that R-cadherin, another classical cadherin, is also expressed in fetal intestinal epithelial cells (Sjödin et al., 1995). It remains to be explored why different types of cadherins occur simultaneously in the intestinal epithelium.
Although LI- as well as E- and R-cadherin are characterized as adhesion molecules, there might be a difference in the regulation of adhesion mediated by the two cadherin subfamilies, i.e., 7D- and classical cadherins. A major molecular difference between LI- and E-cadherin lies in the composition of their cytoplasmic domains. E-cadherin is known to interact with intracellular proteins that link it to the actin cytoskeleton. This cadherin loses most of its adhesive strength when its intracellular domain is truncated, consequently preventing it from connecting to the actin cytoskeleton (Nagafuchi and Takeichi, 1988; Ozawa et al., 1990). Results from numerous studies suggest that E-cadherin may be a target for regulation by signal transduction pathways modulated by certain growth factors, changing the interaction of the cadherin with the cytoskeleton (Behrens et al., 1993; Hamaguchi et al., 1993; Matsuyoshi et al., 1992; Sastry and Horwitz, 1996; Shibamoto et al., 1994; for reviews see Daniel and Reynolds, 1997). It has been shown that these same growth factors can impact tissue morphogenesis (Brieher and Gumbiner, 1994; Montesano et al., 1991; for review see Gumbiner, 1996). To date, LI-cadherin appears to mediate adhesion independent of its cytoplasmic domain and also seems to lack intracellular binding partners (Berndorff et al., 1994; Kreft et al., 1997). For this reason, LI-cadherin is expected not to be regulated by the same mechanisms as E-cadherin (Kreft et al., 1997). One role LI-cadherin may play is the maintenance of close cell–cell contacts between cells when E-cadherin transiently loses its full adhesive properties. LI-cadherin presumably maintains its adhesive function as it is not dependent on interaction with intracellular components and is not likely to be regulated by the same growth factors (Kreft et al., 1997). Since it is not bound to the submembraneous cytoskeleton, LI-cadherin may exhibit a flexible and persistent adhesion between cells, allowing for dynamic cell behaviour during remodeling and expansion of a tissue layer without the loss of close, cell–cell contact. It is remarkable that LI-cadherin expression begins before the fetal intestinal epithelium undergoes a transition from an undifferentiated, multilayered epithelium to a columnar monolayered epithelium while villi are formed. Whether LI-cadherin has an influential role on this morphogenetic process needs to be explored in future studies.
cDNA Cloning and Sequencing
A 17-day Embryo 5′-Stretch Plus lambda-gt 10 cDNA library (Clontech) was screened with the full-length cDNA of rat LI-cadherin (Berndorff et al., 1994) according to the manufacturer's instructions. Probes were digoxigenin-labeled using the DIG Nucleic Acid Detection Kit (Boehringer Mannheim, FRG). The cDNA inserts of positive clones were excised with Not I and ligated into Not I restricted pSK+ (Stratagene). The inserts were sequenced using the ABI-Sequencer 377 (Perkin Elmer).
RNA Isolation and Northern Blot Hybridization
Total RNA from mouse adult tissues was isolated by using TRIZOL reagent (Gibco BRL, Grand Island, NY) according to the manufacturer's protocol. Approximately 15 μg of total RNA from each tissue was separated on a 1.2% agarose gel containing formaldehyde, transferred to a nylon membrane (Hybond-N; Amersham Pharmacia, Freiburg, Germany), and UV cross-linked. Standardization of RNA levels was achieved by comparing the intensity of the ribosomal RNA bands in the ethidium bromide staining of the gel. A digoxigenin-labeled antisense RNA transcript (DIG RNA Labeling Mix, Roche Diagnostics, Mannheim, Germany) consisting of 900 bp of the 5′ end of mouse LI-cadherin cDNA was generated and used as a probe at a concentration of 5 ng/ml. Hybridization was performed at 68°C overnight in DIG Easy Hyb hybridization buffer (Roche Diagnostics, Mannheim, Germany). After hybridization, the membrane was washed at a final stringency of 0.1 × SSC/ 0.1%SDS at 68°C. Detection of the digoxigenin-labeled nucleic acids was carried out by enzyme immunoassay with luminescence substrate according to the manufacturer's protocol (DIG Nucleic Acid Detction, Roche Diagnostics, Mannheim, Germany). RNA transfer to the membrane was controlled by hybridization with a 340-bp mouse GAPDH-antisense cRNA using the same protocol.
Hybridization of Northern blots containing mouse embryonic poly A RNA.
LI-cadherin cDNA containing the entire coding region for LI-cadherin (except for EC 1) was labeled with 32P-phosphate using the RadPrime Labeling System (GIBCO-BRL). Embryonic mouse tissue Northern blots (Clontech) were prehybridized with ExpressHyb-solution (Clontech) and hybridized in the same solution with radioactive labeled DNA probe added at 42°C overnight. The filter was washed with 0.1 × SSC and 0.05% SDS at 50°C.
Preparation of total RNA and subsequent first-strand cDNA synthesis were conducted using the Micro Scale RNA Separator Kit (Clontech) and the Advantage RT-for-PCR Kit (Clontech) according to the manufacturer's instructions. RNA was prepared from liver and intestinal tissues of adult mice, whole mouse embryos at days 10.5 and 11.5 p.c., and decapitated embryos of day 12.5 p.c. (mouse strain NMRI, Charles River, FRG). cDNA synthesis was performed using 1 μg total RNA isolated from each tissue, and the same volumes of resulting cDNA solutions (20 μl) were used for PCR. PCR was conducted using primers flanking a predicted 657-bp product for E-cadherin (nt 537–563 and nt 1,193–1,167; see Nagafuchi et al., 1987) and a predicted 555-bp product for mouse LI-cadherin (primers nt 1844–1867 and nt 2398–2379, see Fig. 1). PCR was performed for 30 cycles as follows: denaturing, 94°C, 45 sec; annealing, 60°C, 45 sec; extension, 72°C, 2 min; final extension 72°C, 7 min. PCR products were analyzed by electrophoresis in 2% agarose gels and ethidium bromide staining.
cDNA coding for the EC1 domain of mouse LI-cadherin was amplified by PCR and cloned into plasmid pDS56 (Bujard et al., 1987). The resultant plasmid pDS56-EC1 encoded a His6-tagged form of the EC1 domain. Using pDS56-EC1, the His6-tagged EC1 domain was overexpressed in Eschicheria coli DH10B as described by Le Grice (1990). The EC1-peptide was purified from bacterial lysates by affinity chromatography using the Ni-NTA Spin Kit (Qiagen, Hilden, FRG), and used for the induction of antibodies in rabbits. Rabbits were immunized with 50 μg of protein, followed by boosts after 3 weeks, and then every 6 weeks following, each time with 50 μg of protein. The specificity of the resulting polyclonal antibody for LI-cadherin was confirmed by comparison with a cross-reacting polyclonal antibody specific for rat LI-cadherin (Berndorff et al., 1994) on Western blots, and by immunofluorescent labeling of tissue sections of adult mouse intestine (data not shown). Rabbit polyclonal antibody against the extracellular domain of mouse E-cadherin was a kind gift of Dr. R. Kemler, Max-Planck-Instiute for Immunobiology (Freiburg, Germany). Rabbit polyclonal antibody against β-catenin was purchased from Sigma (Product No. C-2206). Secondary antibodies were purchased from Dianova (goat anti rabbit-cy-2 conjugated IgG), Sigma (goat-anti-rabbit-TRITC-conjugated IgG), and Dako (swine-anti-rabbit-horseraddish peroxidase-conjugated IgG).
SDS-PAGE and Immunoblotting
Small pieces of liver and intestine of adult mice (strain NMRI, Charles River, FRG) were excised and immediately homogenized for 30–60 sec in 2× SDS-electrophoresis buffer (Laemmli, 1970) using an Ultra Turrax. The lysate was centrifuged for 30 min at 100,000g to sediment insoluble material. Proteins in the supernatant were separated by 7.5% SDS-PAGE. Gels were stained with Coomassie Blue or protein was electrophoretically transferred onto nitrocellulose membranes. Membranes were blocked for 30 min in blot washing buffer (TBST: 15 mM Tris/HCl, pH7.5, 120 mM NaCl, 0.1% Tween 20) containing 5% nonfat dry milk, followed by incubation for 60 min with primary antibody diluted in TBST containing 5% nonfat dry milk. Membranes were washed twice in TBST for 15 min and incubated for 60 min with horseradish peroxidase-conjugated secondary antibody diluted in TBST containing 5% nonfat dry milk. After another wash with TBST, bound antibody was detected by the ECL detection system, according to the manufacturer's instructions (Amersham Corp., Arlington Heights, IL).
Tissue Preparation and Immunofluorescence Staining
Mouse embryos and intestines of adult mice (strain NMRI, Charles River, FRG) were isolated, briefly washed in PBS, and fixed in Bouin's fixative (71% picric acid, 5% glacial acetic acid, 9% formaldehyde) for 30–60 min at room temperature. In the case of embryos of 14.5, 15.5, and 16.5 days p.c., the intestine was dissected from the embryos and fixed separately. The specimens were infiltrated with 30% sucrose in PBS at 4°C, then embedded in Tissue-Tek (Sakura) and stored at −80°C. Cryosections, 5–8-μm-thick, were prepared using a Microm cryostat, fixed for 30 min on ice with 4% para-formaldehyde in PBS, and then washed in PBS for 10 min at room temperature. Sections were immersed for 10 min in fixative blocking buffer (25 mM glycine, 25 mM NH4Cl, 25 mM lysine in PBS). After another wash in PBS (10 min), sections were treated with blocking buffer (0.2% BSA, 1% goat serum in PBS), and then exposed to primary antibody for 60 min. The sections were washed three times for 5 min with PBS and exposed to secondary antibody for 60 min. After three 5-min washes in PBS, the slides were briefly immersed in H2O and sections mounted in Elvanol (1.5% Mowiol [Hoechst], 33% glycerol, 0.1% paraphenylene diamine in PBS pH 8.2). Slides were viewed and photographed using a Zeiss Axiophot epifluorescence microscope (Carl Zeiss, Germany). Identification of embryonic tissues was done using the Atlas of Mouse Development (Kaufman, 1992).
The authors thank Bertolt Kreft, Helmut Wurst, Otmar Huber, and Claudia Fieger for critically reading the manuscript; Claire Grainger for proofreading the manuscript; Barbara Kosel for her excellent technical assistance; and Rolf Kemler for E-cadherin-specific antibody.