SEARCH

SEARCH BY CITATION

Keywords:

  • cell adhesion;
  • placenta development;
  • labyrinth;
  • mouse embryo;
  • gene replacement

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

The specific roles of classical cadherins at key morphogenetic events during development are still not fully understood. As part of a project to study cadherin function during early mammalian development, we generated mice carrying an HA-epitope tagged Cdh1 (E-cadherin) cDNA knocked into the Cdh1 locus, similar to the previously described mouse mutants in which we forced Cdh2 (N-cadherin) expression in the Cdh1 expression domain. As expected and in contrast to Cdh1Cdh2/Cdh2 and Cdh1−/−, our Cdh1HA/HA mutant embryos form proper trophectoderm, implant and undergo both gastrulation and neurulation. However, Cdh1HA/HA mice display an unexpected phenotype at embryonic day 10.5. Cdh1HA/HA embryos are smaller, paler and suffer from an insufficient nutrient supply. We detected a reduced expression of Cdh1HA specifically in the extraembryonic ectoderm and in the labyrinth layer, whereas expression in the embryo proper was normal. With this approach, we show for the first time that Cdh1 is essential for the correct formation of the placenta. Placentas without Cdh1 expression are impaired and incapable of establishing a proper connection between the embryonic and the maternal blood vessels for efficient nutrient and oxygen transport. Developmental Dynamics 239:2330–2344, 2010. © 2010 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Early embryonic development and cell differentiation fundamentally rely on cell–cell adhesion that is mainly mediated by classical cadherins. In the embryo, cell adhesion between the same cell types and the discrimination of cells from different germ layers are established by the cell-type specific expression of Cdh1 (E-cadherin, cadherin 1) and Cdh2 (N-cadherin, cadherin 2) (Stemmler, 2008; Nishimura and Takeichi, 2009), whereas Cdh3 (P-cadherin, cadherin 3) is found in the decidua and the placenta (Nose and Takeichi, 1986). In mammals, Cdh1 is already required during the initial preimplantation developmental steps in the processes of morula compaction and blastocyst formation. Absence of Cdh1 leads to a failure in trophectoderm development, and embryos die within their zona pellucida, unable to hatch (Larue et al., 1994; Riethmacher et al., 1995; De Vries et al., 2004). Later in development and adulthood, Cdh1 expression is mainly confined to epithelia with very important roles in tissue integrity, homeostasis and organ function. For example, a lack of Cdh1 in the lactating mammary gland induces precocious apoptosis, and mice without Cdh1 in the epidermis die due to an impaired inside-out barrier function of the skin (Boussadia et al., 2002; Tunggal et al., 2005). Very strikingly, the expression of Cdh1 and Cdh2 is mutually exclusive and switching in expression correlates with important morphogenetic events (Takeichi, 1988; Gumbiner, 2005). During gastrulation, when cells from the epiblast delaminate and undergo epithelial–mesenchymal transition (EMT) to form the newly emerging mesoderm, Cdh1 is down-regulated and Cdh2 expression initiated. A similar switch in cadherin expression is observed during neurectoderm formation (Hatta and Takeichi, 1986). Importantly, aberrant changes in cadherin expression and activation of the EMT program are observed during tumorigenesis. Cdh1 down-regulation and de novo expression of Cdh2 coincides with the onset of invasion, formation of metastases, and poor prognosis (Li et al., 2001; Thiery and Sleeman, 2006; Nieto, 2009; Vandewalle et al., 2009).

Most of our knowledge about cadherin function has been gathered from cell culture studies or from animal models either by overexpression of cadherin genes or by loss-of-function using knock-out or knock-down approaches. Although these studies helped to get a detailed view of cadherin structure and function in the cell, a comprehensive understanding of the unique properties of individual cadherin superfamily members in embryogenesis is still lacking. Cdh1 provides a polarized, epithelial cellular phenotype, whereas Cdh2 induces cell migration and depolarization during EMT. How these very similar cadherins provide such different cellular features is unknown. A gene replacement approach whereby one cadherin gene replaces another is an elegant in vivo tool to gain further insight into the unique functions of cadherins during development. Recently, we forced the expression of Cdh2 in the Cdh1 expression domain using a gene replacement approach. In Cdh1Cdh2/+ mice, Cdh2 was co-expressed with Cdh1 in a correct spatio and temporal manner in addition to its endogenous expression (Stemmler et al., 2005; Kan et al., 2007). Homozygous mutant mice (Cdh1Cdh2/Cdh2) were unable to form a functional blastocyst, indicating a specific requirement for Cdh1 during trophectoderm formation that Cdh2 cannot supply (Kan et al., 2007).

Only little is known about the role of cadherins during placenta formation. The development of the placenta is dependent on various pathways and the interaction of maternal and embryonic tissues. In mice, several mutations are described that affect formation of the placenta and lead to improper function in providing nutrients and oxygen to the embryo (Rossant and Cross, 2001; Watson and Cross, 2005). Cdh1 is found in specific layers of the placenta, but Cdh3 is predominantly expressed in the decidua and the placenta (Nose and Takeichi, 1986). However, in mice Cdh3 gene ablation results in fertile females and a normal placenta formation during pregnancy (Radice et al., 1997). In humans pre-eclampsia and the HELLP syndrome are severe life-threatening disorders that occur in 5–8% maternities during or after pregnancy. The exact causes of these disorders are unknown, but it is suggested that a shallow implantation might lead to hypoxia. Consequently, maternal endothelium is damaged and placental growth restricted, followed by an immune response (Haram et al., 2009; Kanasaki and Kalluri, 2009). Pre-eclampsia is usually treated by delivery, but in severe cases abortion is inevitable. In these disorders as well as in gestational trophoblastic diseases, like hyadatiform mole, Cdh1 expression is frequently altered in comparison to normal pregnancies. Both an increase in Cdh1 expression and an aberrant down-regulation have been documented in different studies (Zhou et al., 1997; Floridon et al., 2000; Xue et al., 2003; Brown et al., 2005; Batistatou et al., 2007; Blechschmidt et al., 2007). However, the role of cadherin-mediated adhesion for placenta formation and tissue homeostasis and in the described disorders remains elusive.

In this study, we generated a knock-in allele by placing the coding region for an epitope-tagged version of Cdh1 into the Cdh1 locus. We demonstrate that the gene replacement strategy does not affect Cdh1 expression in either the developing embryo or adult mice. By analyzing the unexpected lethality of homozygous mutant embryos at embryonic day (E) 10.5, we show an as yet unknown function of Cdh1 in formation of a proper placenta and in establishing the connection between the maternal and the embryonic blood vessels.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Generation of Mice With a cDNA Coding for Cdh1-HA Knocked Into the Cdh1 Locus

Cdh1 was replaced by a cDNA that codes for an HA-tagged version of Cdh1 (Fig. 1A). Embryonic stem (ES) cells were transfected with the targeting vector to insert the cDNA into the ATG codon, and homologous recombination was verified by Southern blot analysis (Fig. 1B). Transient expression of Cre-recombinase in ES cells lead to efficient removal of the neomycin resistance cassette (Fig. 1B; Supp. Fig. S1A, which is available online). Two independent clones were used to establish germ line transmitting mice by ES-cell injection into blastocyst. We analyzed the protein expression of the targeted allele in Cdh1Cdh1-HA/+ (Cdh1HA/+) heterozygous mice at E14.5. The total amount of Cdh1 was compared with the fraction derived from the Cdh1HA allele. Anti-Cdh1 and anti-HA immunoblots displayed similar spatial protein synthesis from wild-type (wt) and knock-in alleles (Fig. 1C). The expression of both alleles was further analyzed on E14.5 embryonic sections by immunohistochemistry. HA-tagged protein was correctly localized to the basolateral membranes, similar to endogenous Cdh1 (Fig. 1D; Supp. Fig. S1B). The analysis shows that this targeting strategy results in proper gene expression from the knock-in allele.

thumbnail image

Figure 1. Generation of Cdh1HA/+ mice. A: A schematic representation of the Cdh1 locus, the targeting construct, the generated allele after successful homologous recombination and the deletion of the selection cassette. Exons are indicated in black, Cdh1 cDNA in yellow, sequence coding for HA-tag in red, SV40 polyA in gray, neomycin resistance (neoR) by open boxes and loxP sites by black triangles. The positions of restriction enzyme sites are given for BamHI (B), EagI (E), HindIII (H), KpnI (K), SacI (S), SpeI (Sp), and XbaI (X), and Southern blot probes are labeled using small letters. B: Representative images of the Southern blot analysis using the indicated restriction digests and probes indicate successful homologous recombination in embryonic stem (ES) cells and deletion of the selection cassette in the resulting mice. C: Expression of the knock-in allele in various organs of embryonic day (E) 14.5 embryos analyzed by immunoblotting. Expression is detected by anti-HA antibody and shows the correlated expression at comparable levels between wild-type (wt) and transgene-derived Cdh1 in gut (g), kidney (k), liver (li), lung (lu), skin (sk), and stomach (st). Consistent with the endogenous Cdh1 expression, no protein is found in the brain (b) or the heart (h). D: Immunohistochemistry with indicated antibodies on sections of E14.5 embryos showing protein expression of the knock-in allele is consistent with endogenous Cdh1 expression and the absence of anti-HA staining in the wt embryos. Scale bar = 100 μm.

Download figure to PowerPoint

Expression of the Cdh1HA Knock-in Allele Confers Proper Embryogenesis and Tissue Integrity

We first checked the expression and the function of the HA-tagged protein in a couple of control experiments. Because the gene replacement of Cdh1 with Cdh2 resulted in a trophectoderm defect at E3.5 (Kan et al., 2007), we focused on the preimplantation development of embryos from heterozygous Cdh1HA/+ intercrosses. In vivo and in vitro morulae of Cdh1HA/HA embryos compact and form proper blastocysts similar to their littermates as shown by time-lapse recording (Supp. Fig. S2A–C′,G; Supp. Movie M1). Subsequent immunofluorescent labeling for total Cdh1 and HA-tagged Cdh1 expressed from the knock-in allele showed that the protein stained with comparable intensities in the homozygous mutants as in either their heterozygous or their wt littermates (Supp. Fig. S2D–F″). This indicates that the Cdh1HA knock-in allele provides sufficient protein to allow blastocyst formation. Moreover, the detected expression levels are comparable between homozygous mutants and both their heterozygous and wt littermates.

Expression levels of knock-in and wt alleles were compared in more detail in ES cells, generated from blastocyst outgrowth (Supp. Fig. S3A). RNA and protein were isolated to compare the expression levels among wt, heterozygous and homozygous ES cells. Expression from the homozygous Cdh1HA/HA ES cells was reduced to 34% of the wt transcript levels. Protein was reduced to 30–50% and to 20–30% of Cdh1HA/+ and Cdh1+/+ levels, respectively (Supp. Fig. S3B–E). Because Cdh1 is known to interact with and to crosstalk to several signaling cascades (Pece et al., 1999; Pece and Gutkind, 2000; Larue and Bellacosa, 2005; Stemmler, 2008), we analyzed key components of Wnt/β-cat-, MAPK-, and Akt pathways in ES cells. We found a reduction in p-Erk1/2 levels in Cdh1HA/HA lysates, whereas the total Erk1/2 levels are unaffected (Supp. Fig. S3F). The reduction in signaling-competent dephospho-β-cat detected in the Cdh1HA/HA ES cells likely represents the reduced cadherin-bound pool due to reduced presence of Cdh1-HA protein, because Topflash reporter assays did not reveal significant changes in Wnt/β-cat signaling (Supp. Fig. S3F,G). Similar to experiments on preimplantation embryos, Cdh1HA/HA ES cells showed proper basolateral membrane localization of Cdh1 (Supp. Fig. S4A–C″). In addition, comparative analysis of pluripotency by marker gene expression and of differentiation potential in teratomas gave similar results for both ES cell lines (Supp. Fig. S4D–K).

We then examined the molecular interaction of Cdh1 derived from the knock-in allele (Cdh1-HA) by immunoprecipitation to rule out a potential interference of the HA-tag on protein function. The interaction to β-catenin and association to proteins that mediate the connection to the actin cytoskeleton, like Eplin, were similar between the wt Cdh1 and Cdh1-HA (Supp. Fig. S5). Thus, the generated tagged protein shared proper functioning ability with its wt counterpart. Because the expression was reduced to 20–30%, we analyzed whether Cdh1HA/HA ES cells integrate properly into the intracellular matrix (ICM) or are excluded from specific germ-layers in blastocyst injections. Chimeric mice analyzed at E11.5 show compareable contribution of ES cells with all germ layers, indicating equal adhesiveness of homozygous mutant ES cells and wt ICM cells because they mix properly with cells of the host embryo (Supp. Fig. S6A–D′). In adult chimeras ES-cell descendants were identified in epithelia and show comparable anti-Cdh1 staining (Supp. Fig. S6E–J).

Finally, we analyzed whether knock-in derived Cdh1-HA is able to establish proper tissue integrity and homeostasis in Cdh1-depleted epithelia during late embryogenesis and adulthood. Gene ablation of a conditional Cdh1 allele in combination with Cdh1HA was carried out in the intestinal epithelium in Vil::Cre;Cdh1HA;R26RΔ/+ mice. The promoter that drives Cre expression is activated from E10.5 until adulthood in the epithelium of the gut (el Marjou et al., 2004) and mice were analyzed at the age of 19 weeks. In contrast to Vil::Cre;Cdh1Δ/Δ mice that die during embryogenesis (not shown), our mutants did not show a detectable phenotype during adulthood. Genotyping of isolated intestinal epithelial cells at the age of 19 weeks confirmed efficient recombination (Fig. 2A) by detection of only a weak polymerase chain reaction (PCR) product from a minor fraction of unrecombined cells. Consistently, recombination of the R26Rflox allele was found in all samples of the small and large intestine (Fig. 2B–D′). By immunohistochemistry of the proximal and distal small intestine and of the colon, we detected Cdh1 protein in Vil::Cre; Cdh1HA;R26RΔ/+ and Vil::Cre; Cdh1HA/+;R26RΔ/+ mice without any differences (Fig. 2E–H′; Supp. Fig. S7). Cdh1 protein expressed exclusively from the knock-in allele was confined to the basolateral cell membrane. Taken together, these control experiments demonstrate that the Cdh1HA allele shares proper function with the wt allele, although expression in ES cells is reduced to 30–50% and to 20–30% compared with heterozygous and wild-type ES cells, respectively.

thumbnail image

Figure 2. Mice that express Cdh1 in the intestine only from the knock-in allele are normal. Analysis of Vil::Cre;Cdh1HA;R26RΔ/+ mice. A: Genotyping of isolated gut epithelia cells of the proximal (prox), intermediate (mid) intestine, colon (col), and tail of the same animal, compared with control tail DNA from Cdh1+/+ (+/+), Cdh1flox/+ (flox/+), and Cdx1:: Cre;Cdh1Δ/+ (Δ/+) mice (Hierholzer and Kemler, 2010). The detected alleles by the presence of corresponding polymerase chain reaction (PCR) products are given on the right. B–D′: β-Gal staining of longitudinally opened whole-mount (B) and sectioned distal intestine (C) and colon (D,D′) of Vil::Cre;Cdh1HA;R26RΔ/+ mice. PCR and β-gal staining shows that >90% of alleles in the epithelium are recombined. E–J: Anti-Cdh1 immunohistochemistry of proximal (E,E′,H,H′), distal intestine (F,I) and colon (G,J) sections showing normal gut morphology of villi and crypt regions and unaltered Cdh1 expression levels in Vil::Cre;Cdh1HA;R26RΔ/+ mice (E–G) and Vil::Cre;Cdh1HA/+;R26RΔ/+ controls (H–J). Boxes in D, E, H indicate the position of blow-ups in D′, E′, H′, respectively. Scale bars = 25 μm in D′,E′,H′; 100 μm in C–J; 500 μm in B.

Download figure to PowerPoint

Gastrulation and Neurulation of Cdh1HA/HA Embryos Are Normal but They Die at Approximately E10.5

Subsequently, we analyzed postimplantation development of Cdh1HA/HA mice. Up to E9.5 homozygous embryos were found in a Mendelian ratio with normal morphology. The heart, the somites, regionalization of the brain, the limb bud primordia, the optic eminence and the otic pit were all clearly identified in both the Cdh1HA/HA and the Cdh1HA/+ embryos (Fig. 3A,C), indicating that gastrulation and neurulation were normal. However, Cdh1HA/HA embryos isolated between E10.5 and E11.5 displayed an unexpected phenotype and died shortly afterward. They appeared pale, with reduced blood in both the yolk sac and the embryonic tissues, and were also arrested in growth, resembling E9.5 embryos (Fig. 3B,D). Consistent with the data shown above, immunohistochemistry staining of E9.5 embryos revealed that both the Cdh1-HA and the Cdh1 proteins were present in comparable amounts in all the Cdh1 expressing cells of homozygous mutant and control littermates (Fig. 3E–E″,F–F″). The epithelial marker Cytokeratin 8 was detected in embryos of both genotypes with comparable intensities, indicating proper epithelial specification (Fig. 3E′″,F′″). Because Cdh1 is not expressed in cells of the cardiovascular system, we reasoned that a heart and/or vessel disorder is very unlikely to cause the phenotype.

thumbnail image

Figure 3. A phenotype of Cdh1HA/HA embryos is detected at embryonic day (E) 10.5. A–D: Embryos were isolated at E9.5 (A,C) and E10.5 (B,D), and homozygous (A,B) and heterozygous mutants (C,D) are shown. Cdh1HA/HA embryos develop normally until E9.5 but are both paler and slightly smaller than their heterozygous littermates at E10.5. They start to die at around E11.5. E–F′″: Immunohistochemistry staining of sections at E9.5 of homozygous mutant (E,E′,E″,E′″) and heterozygous control embryos (F,F′,F″,F′″) with anti-Cdh1 (E,F,E′,F′), anti-HA (E″,F″) and TROMA1 against cytokeratin 8 (CK8) (E′″,F′″). Comparative analysis shows similar expression levels of Cdh1 and the epithelial marker CK8 of homozygous and heterozygous knock-in embryos. Boxes in E and F indicate the position of blow-ups in E′–F′″. Scale bars = 500 μm in A–F, 25 μm in E′–F′″.

Download figure to PowerPoint

Cdh1HA/HA Embryos Show a Placental Defect Accompanied by a Reduction in Cdh1 Expression

In search of possible explanations for the embryonic lethality seen in the mutants, we detected a specific difference in extraembryonic tissues (Fig. 4). Cdh1 and β-catenin staining was already reduced at E6.5 in the extraembryonic ectoderm (ExE) of the Cdh1HA/HA embryos, compared with heterozygous controls (Fig. 4A,D and C,F, respectively). Detection of the protein derived from only the knock-in allele by anti-HA gave similar results to anti-Cdh1 staining, although the protein in heterozygous embryos is barely detectable in the ExE due to gene dosage effects (Fig. 4B,E). The ExE together with the ectoplacental cone and the allantois contributes to the formation of the embryonic part of the placenta. At E10.5 when the phenotype became apparent, the overall size of the embryonic part of the placenta was much smaller in the Cdh1HA/HA mutant embryos than in their heterozygous littermates (Fig. 4G,H). The four different layers of the embryonic part of the placenta, the trophoblast giant cells, the spongiotrophoblast, the labyrinth, and the chorionic plate were clearly identifiable in placentas of homozygous mutant and control embryos (Fig. 4G′,H′). However, the size of the labyrinth layer was greatly reduced in the Cdh1HA/HA placenta, whereas all of the other layers were of similar size as in the Cdh1HA/+ placenta. In addition, we found less intermingling of the maternal and the embryonic blood vessels as indicated by the restriction of the embryonic nucleated blood cells to superficial layers that did not deeply penetrate into the labyrinth layer. The labyrinth layer is derived from the ExE and continuously expresses Cdh1 in the control placenta (Fig. 4L). In contrast, the smaller labyrinth layer in the homozygous mutant was devoid of any Cdh1 staining (Fig. 4J). The loss of Cdh1 was specific to the labyrinth layer and was not observed in the yolk sac (Fig. 4I,J) or in the embryo proper (Fig. 3). Of interest, no reduction in Cdh1 transcript levels in the labyrinth of homozygous compared to heterozygous mutants was detected. This indicates that mRNA expression from the knock-in allele was normal in the homozygous mutant placenta but labyrinth-specific reduction in protein level likely occurred on a posttranscriptional level (Fig. 4M,N). In summary, the analysis shows that the restricted loss of Cdh1 expression in the ExE of homozygous mutant embryos leads to a reduction in size of the labyrinth layer of the placenta. Likely, the disabled placenta does not provide sufficient nutrients and oxygen support for further development and leads to embryonic lethality around E11.

thumbnail image

Figure 4. Cdh1 protein expression is specifically reduced in the extraembryonic ectoderm and placental tissue of Cdh1HA/HA embryos. A–F: Transverse sections of the extraembryonic part of Cdh1HA/HA (A–C) and Cdh1HA/+ embryos (D–F) at embryonic day (E) 6.5 stained by immunohistochemistry with anti-Cdh1 (A,D), anti-HA (B,E) and anti-β-catenin (C,F) antibodies. Boxes in A–E mark positions of the insets given in the same image. The plane of sections for the upper and the lower panel are given in the cartoons in C and F, respectively. Specifically, reduced expression from the knock-in allele is observed in the extraembryonic ectoderm (ExE) of homozygous mutants in comparison to the visceral endoderm (VE), leading to a reduced membrane staining of β-cat in the extraembryonic ectoderm. G–H′: Hematoxylin/eosin staining of E10.5 placentas of Cdh1HA/HA (G,G′) and Cdh1HA/+ embryos (H,H′), with a detailed view of the trophoblast giant cell layer (TGC), spongiotrophoblast layer (Sp), labyrinth layer (Lb), and the chorionic plate (CP) shows a reduction in the size of the labyrinth in the homozygous mutant placenta. I–L: Cdh1 protein expression levels are unchanged in the embryo and the yolk sac of homozygous mutants or controls (I,K). Expression is lost in the labyrinth layer of Cdh1HA/HA E10.5 placentas (J), in contrast to Cdh1HA/+ (L). M–N′: Cdh1 RNA in situ hybridization of homozygous knock-in (M,M′) and control (N,N′) placentas indicates that similar amounts of Cdh1 transcripts are found in both genotypes. Note that the placental TGC-CP axis is from top to bottom (G–L) and from left to right (M–N′). Boxes in A,B,D,E,G,H,M,N indicate the position of blow-ups in A′,B′,D′,E′,G′,H′,M′,N′, respectively. Scale bars = 100 μm in A–F,I–L,M′,N′, 25 μm in A′,B′,D′,E′, 500 μm in G,H, 200 μm in G′,H′,M,N.

Download figure to PowerPoint

Reduction of Marker Gene Expression Specific for the Labyrinth and Spongiotrophoblast Layer in Cdh1HA/HA Embryos

The different layers and cell types in the developing placenta can be identified by specific marker gene expression. To obtain a comprehensive view of the observed defects in Cdh1HA/HA embryos we performed RNA in situ hybridization to analyze specific genes (Fig. 5; Supp. Fig. S8). In agreement with the observation of a reduction in size of the labyrinth layer, cells with Dlx3 expression are largely reduced in Cdh1HA/HA embryos (Fig. 5E–F′). Similarly, Gcm1 and Syna that specifically label cells associated with fetal blood spaces and maternal blood sinuses, respectively, are detected in fewer cells as in the heterozygous control. (Fig. 5K–N′). The development of the spongiotrophoblast layer is likely affected as well, because Mash2-, Tpbpa-, and Vegfr1-positive cells are decreased in number in the homozygous mutant placenta (Fig. 5C–D′,I–J′; Supp. Fig. S8E–F′). A slight increase in cells expressing trophoblast-specific markers (Hand1, Plf1, Pl1) are detected in Cdh1HA/HA mutants, indicating either normal development and proliferation of this layer or a compensatory effect for the lack of a proper labyrinth and spongiotrophoblast layer (Fig. 5A–B′,G–H′; Supp. Fig. S6C–D′). The marker gene analysis corroborated the findings of a size-reduced labyrinth layer and revealed that reduced Cdh1 expression in the ExE affects normal formation of other layers of the placenta as well.

thumbnail image

Figure 5. Marker gene analysis of different layers of the placenta at embryonic day (E) 10.5 reveals impaired development of the labyrinth layer. A–M: Sagittal sections at the medial plane of the placenta of Cdh1HA/HA and Cdh1HA/+ embryos are stained by in situ hybridization for Hand1 (A–B′), Mash2 (C–D′), Dlx3 (E–F′), Plf1 (G–H′), Tpbpa (I–J′), Gcm1 (K–L′), and Syna (M–N′). These markers specifically label cells of the layers as indicated on the left side of the panels. Reduced staining intensities for Mash2, Tpbpa, and Dlx3 in the mutants indicate a reduction in the Lb that affects also Sp differentiation. Boxes in A–N indicate the position of blow-ups in A′, –N′, respectively. Scale bars = 500 μm in A–N, 50 μm in A′–N′.

Download figure to PowerPoint

Malformation of the Placenta in Homozygous Mutants Is Accompanied by Increased Apoptosis and Decreased Proliferation

We further analyzed the mutant placenta for cell death and proliferation (Fig. 6). Sections of embryos were labeled by TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling) staining at E10.5. Homozygous mutants were still alive, as indicated by the beating of their hearts. Whereas no difference was observed in the number of apoptotic cells in the embryo proper of homozygous mutants when compared with the controls the amount of TUNEL-positive nuclei in the placenta of Cdh1HA/HA was increased (Fig. 6A,C–C″,M). Here, apoptosis was found in the labyrinth layer and the chorionic plate (Fig. 6C,M). At E11.5, when embryos were dying, almost all cells of the homozygous mutant embryos and of their placentas were positive in the TUNEL assay, indicating embryonic death and resorption (not shown). To address the differences in proliferation, pregnant females of heterozygous intercrosses were injected intraperitoneally with bromodeoxyuridine (BrdU) 2 hr before isolation of the embryos at E10.5. Staining for PCNA did not reveal any major differences between both genotypes (Fig. 6E–H). Almost all nuclei of cells in the placentas of Cdh1HA/HA and Cdh1HA/+ mice contained PCNA protein, indicating that they were in the cell cycle. The short pulse of BrdU gave a more detailed view of cells in S-phase. In homozygous mutants, many BrdU-negative cells in the labyrinth (open arrow) and the trophoblast giant cell/spongiotrophoblast layer (open arrowhead) were identified (Fig. 6I,K,N), whereas in the heterozygous placentas, nearly all cells were BrdU-positive (Fig. 6J,L,N). An increase in the size of the nuclei and an improper placental connection indicated by a large gap between embryonic and maternal tissues was detected in homozygous mutants and highlights the incomplete intermingling of blood vessels and branching morphogenesis (Fig. 6I,K). Our results show that in addition to increased apoptosis, the cell cycle time was increased in Cdh1HA/HA mutant placentas, resulting in a reduction in cell numbers.

thumbnail image

Figure 6. Analysis of cell death and proliferation in Cdh1HA/HA mutant placentas at embryonic day (E) 10.5. A–D″: TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling) staining (green) and DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride; blue) in Cdh1HA/HA (A,C–C″) and Cdh1HA/+ embryos (B,D–D″) show an increase in cell death in homozygous mutant embryos and a slight increase in the corresponding placental tissue, while almost no apoptotic cells are found in control embryos and placentas. E–H: PCNA immunohistochemistry on sections of Cdh1HA/HA (E,G) and Cdh1HA/+ placentas (F,H). I–L: Pregnant females were injected peritoneally with bromodeoxyuridine (BrdU), and E10.5 embryos were isolated together with their placentas 2 hr later. Sections of Cdh1HA/HA (I,K) and Cdh1HA/+ placentas (J,L) were stained with an anti-BrdU antibody and counterstained with Carazzi's hematoxylin to visualize nuclei. Homozygous mutant placentas show more BrdU-negative cells, both in the labyrinth layer (open arrows) and in the trophoblast giant cell layer (open arrowheads). BrdU-positive nuclei of the Cdh1HA/HA labyrinth and trophoblast giant cell layer are indicated by filled arrows and arrowheads, respectively. Note that nuclei in the trophoblast giant cell layer (TGC) layer of Cdh1HA/HA mutants are enlarged in comparison to the control nuclei. M,N: Quantification of apoptotic (M) and, BrdU-positive cells (N) of the labyrinth layer with n = 3 in both experiments. Standard error of the mean (SEM) is given. Paired student's t-test indicate significant changes compared with control with P = 0.04 (M) and P = 0.006 (N). Boxes in A,B,E,F,I,J indicate the position of blow-ups in C–C″,D–D″,G,H,K,L, respectively. Scale bars = 250 μm in A,B,E,F,I,J, 100 μm in C–D″,G,H,K,L.

Download figure to PowerPoint

Blastocyst Complementation Assay Rescues the Lethality of Cdh1HA/HA Embryos

To confirm that the phenotype in homozygous mutant embryos is due to a dysfunction of the placenta, we tested whether the defect could be rescued by providing extraembryonic wild-type cells to allow proper nutritional support for efficient embryonic development. We addressed this issue in a blastocyst complementation experiment. In a developing embryo, tetraploid cells (4n) can only contribute to the trophectoderm and extraembryonic tissues. This effect can be exploited to generate postimplantation conceptuses that are entirely generated by diploid (2n) ES cells (Zambrowicz et al., 1997; Hadjantonakis et al., 1998; Tanaka et al., 2009). Because embryos derived from 100% ES cells usually do not develop until term, we focused on embryonic stages. Homozygous Cdh1HA/HA ES cells were injected into 4n Rosa26 blastocysts and isolated at different time-points after reimplantation (Fig. 7A–C). Embryos at E10.5 showed normal development and no growth retardation (Fig. 7A). β-Galactosidase staining confirmed that there was almost no contribution of cells from the host embryo, except to the yolk sac (Fig. 7A, arrow) and the incorporation of few 4n cells, e.g., trapped in the otic pit. Similar results were obtained at E13.0; a normal embryo developed from only ES-cell descendents (Fig. 7B). Embryos of control Cdh1HA/+ ES cells were obtained with the same frequency at E11.5 (Fig. 7C) and E13.5 (not shown), demonstrating a similar developmental and differentiation capacity of ES cells of both genotypes. The rescue of development provides strong support that the embryonic lethality of Cdh1HA/HA embryos is attributed to improper placental function, due to the specific and unexpected reduction of Cdh1 expression in the ExE.

thumbnail image

Figure 7. The placenta defect of Cdh1HA/HA embryos is rescued in tetraploid complementation experiments. A–C: Embryos derived from the injection of 2n Cdh1HA/HA (A,B) or Cdh1HA/+ embryonic stem (ES) cells (C) into 4n Rosa26 blastocysts. The contribution of tetraploid host cells is analyzed by β-gal staining and was found in the yolk sac (arrow, dashed line) and the otic vesicle (A). Cdh1HA/HA ES cells give rise to living embryos of embryonic day (E) 13.0 if placental tissue is provided by wt cells. Estimated developmental stages are given. Scale bars = 1 mm in B,C, 500 μm in A.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

As part of the analysis of Cdh1 and Cdh2 function, we generated a Cdh1 knock-in allele that carries a Cdh1 cDNA insertion. Homozygous mutant embryos properly implant and undergo key morphogenetic events like gastrulation and neurulation. The sudden embryonic death indicates improper placental function due to a vital role of Cdh1 in the ExE that forms the embryonic part of the placenta. Our results show that reduced Cdh1 expression in the ExE led to decrease in the size of the placenta. Homozygous mutant placentas show reduced cell division rates and increased apoptosis, resulting in a smaller labyrinth layer and improper intermingling of embryonic and maternal blood vessels as well as reduction in labyrinth- and spongiotrophoblast-specific marker gene expression. Based on our findings, especially the observation of reduced Mash2 and Dlx3 staining, the progenitor cells of the trophoblast giant cell/spongiotrophoblast and labyrinth layer are largely reduced. This leads to limited branching morphogenesis and abnormal labyrinth and spongiotrophoblast differentiation in the homozygous mutants. Because trophoblast giant cells are apparently not reduced in number, but originate from the same progenitor pool, it is suggested that Cdh1 is required to maintain the progenitor pool and to control cell differentiation. The reduction of Cdh1 expression in the homozygous knock-in ectoplacental cone might lead to precocious differentiation into parietal trophoblast giant cells. In summary, Cdh1 is required for proper formation and function of the placenta in establishing sufficient nutrients and oxygen transport to the embryo.

During early postimplantation development, nutrient transport is achieved by diffusion through the decidual tissue up to E9–E10 (Rossant and Cross, 2001; Watson and Cross, 2005; Cross et al., 2006). Placenta development starts at E8.5, when the allantois fuses to the chorion (chorioallantoic attachment). Shortly afterward, folds are formed in the chorion and fetal blood vessels originating from the allantois grow into the spaces to form the chorionic plate. The chorionic trophoblast differentiates into three cell layers, labyrinth, spongiotrophoblast and trophoblast giant cells, and starts to proliferate. Until birth, the labyrinth layer is further branched and villi are formed to provide a large surface to maximize contact between maternal and embryonic blood vessels (Rossant and Cross, 2001; Watson and Cross, 2005; Cross et al., 2006). From E9.5 onward, the major nutrient transport is provided by the function of the placenta. Placental defects that manifest in inappropriate vascularization, patterning, dilation, or perfusion result in impaired development or embryonic lethality (Pardi et al., 2002). Several gene mutations display placenta phenotypes that are comparable to our Cdh1HA/HA mutant mice. These include targeted deletions of genes encoding for αv-integrin, Birc6/Bruce, Connexin31, 45, c-myc and fibroblast growth factor receptor-2 (FGFR2; Bader et al., 1998; Xu et al., 1998; Kruger et al., 2000; Plum et al., 2001; Hitz et al., 2005; Dubois et al., 2008). For example, the inactivation of Fgfr2 leads to a failure in chorioallantoic fusion in 30% of homozygous mutant embryos. The remaining Fgfr2 mutant embryos show a reduction in the labyrinth layer that is characterized by a massive decrease in proliferation of the trophoblast cells and inefficient vascularization (Xu et al., 1998). In Birc6/Bruce homozygous mutant embryos, the defect is restricted to the spongiotrophoblast layer which is lost between E11.5 and E14.5 (Hitz et al., 2005). Defects in the formation and function of the labyrinth layer have been described for targeted mutations for example in Mash2, p38α and K8/K19 genes (Guillemot et al., 1994; Adams et al., 2000; Tamai et al., 2000). Additionally, Mash2−/− and K8−/−;K19−/− embryos show absence of the spongiotrophoblast layer. The best correlation with the Cdh1HA/HA phenotype is seen in mice carrying a homozygous mutation for p38α kinase. These mutant embryos die around E10.5 and show a reduction in size of the labyrinth, including inefficient intermingling between embryonic and maternal blood vessels, whereas other layers are normal (Adams et al., 2000). Whether the similar phenotypes of the described targeted mutations and of Cdh1HA/HA mice are attributable to the impairment of the same pathways and processes remains elusive. It is noteworthy that a connection between Cdh1-mediated cell-adhesion and p38α MAP kinase signaling was observed in embryos as well as in tumor cells. The lack of p38α activity during gastrulation leads to accumulation of mesodermal descendents at the primitive streak region because Cdh1 protein is not properly degraded (Zohn et al., 2006). In malignant melanoma, the loss of Cdh1 expression enhances NFκB activity, presumably through a p38α MAP kinase mechanism. The re-expression of Cdh1 in these cells blocks p38α MAP kinase and NFκB activity (Kuphal et al., 2004). This may indicate that Cdh1 and p38α MAP kinase are functionally linked in specific contexts by means of the reciprocal correlation between Cdh1 expression and the activity of p38α MAP kinase.

Alternatively, the phenotype in our Cdh1HA/HA mice may be related to receptor tyrosine kinases (RTKs), because homozygous mutations in Egfr (epithelial growth factor receptor, EGFR) and Met (c-Met) show similar placental defects (Bladt et al., 1995; Schmidt et al., 1995; Sibilia and Wagner, 1995). EGFR interaction with Cdh1 leads to modifications in receptor activity and results in the induction of MAPK activity in keratinocytes and mammary epithelial cells (Hoschuetzky et al., 1994; Pece and Gutkind, 2000; Fedor-Chaiken et al., 2003). The fact that we observed reduced activated Erk1/2 (p-Erk1/2) in Cdh1HA/HA ES cells supports this idea (Supp. Fig. S3F). Accordingly, in our mutant placenta, EGFR-mediated signaling might be reduced due to the lack of Cdh1. It is tempting to speculate that impaired molecular interactions with any of the described proteins are responsible for the phenotype in Cdh1HA/HA mutants. On the other hand, we cannot exclude that the failure of proper placenta formation is due to improper tissue integrity. A reduction in adhesion might lead to inefficient proliferation and cell survival in the trophoblast stem cell niche. The amount of cells in S-phase were decreased in our mutants and an increase in apoptotic cells was detected simultaneously. Cdh1 has been shown previously to act as a survival factor in epithelial cells of the mammary gland (Boussadia et al., 2002), which might also be true for the labyrinth layer. The question of how the lack of Cdh1 induces placental defects on a molecular level needs to be addressed by future experiments.

Consistent with the results from other mutations causing placental dysfunction, the defect in labyrinth layer formation was rescued by the blastocyst complementation assay. Cdh1HA/HA embryos develop beyond E10.5 if the ExE is derived from tetraploid cells of wt origin. Similarly, phenotypes of Mash2−/− and p38α−/− embryos are rescued by the same approach, confirming that the phenotype in all mutants is partially or completely attributable to the inappropriate formation of the labyrinth layer (Guillemot et al., 1994; Adams et al., 2000).

Placenta development is dependent on a functional trophoblast and various pathways (for review see Watson and Cross, 2005). Cdh3 is the member of the cadherin superfamily predominantly expressed during development in the decidual tissue and in the placenta. The lack of Cdh3 does not cause placental defects or reduced litter sizes (Radice et al., 1997). Although the expression of Cdh1 has been found in extraembryonic tissues and specific layers of the developing placenta as well, the function of Cdh1 for placenta morphogenesis has not been addressed (Nose and Takeichi, 1986; Brown et al., 2005; Batistatou et al., 2007). In humans, a trophoblast giant cell layer is found in the placental bed, and an invasive extravillous cytotrophoblast layer mediates attachment to the uterine wall (Rossant and Cross, 2001). The cytotrophoblast is covered by a multinucleated cell-layer, the syncytiotrophoblast, that forms the outside of the chorionic villi, which is comparable to the labyrinth in rodents. During normal pregnancies, Cdh1 can be found in the anchoring placental villi, the cytotrophoblasts and the extravillus trophoblasts (Coutifaris et al., 1991; Getsios et al., 2000). Subsequently, the protein is lost during syncytiotrophoblast differentiation (Shih Ie et al., 2002). Altered Cdh1 expression is likely the cause for certain complications during pregnancy, such as pre-eclampsia or some spontaneous abortions, and can be detected in gestational trophoblastic syndromes (Zhou et al., 1997; Floridon et al., 2000; Xue et al., 2003; Batistatou et al., 2007). In pre-eclampsia, an incomplete invasion of the extravillus trophoblast into the uterine endometrium was identified, which correlates with aberrant Cdh1 expression (Zhou et al., 1997; Brown et al., 2005). The analyses indicate that Cdh1 needs to be tightly controlled for proper placenta function and to avoid diseases. They support the finding that insufficient Cdh1 synthesis as detected in our mice leads to impaired pregnancy.

Why the expression of Cdh1 is specifically reduced in the ExE and its descendents is puzzling, particularly because we observe equal levels of Cdh1 mRNA expression between homozygous mutants and control embryos, but no protein is generated from the knock-in allele. One possibility is that expression between embryonic and extraembryonic structures is regulated by fundamentally different mechanisms. In our knock-in allele we used an SV40 polyadenylation site that confers proper mRNA stability in the embryo, but it might generate an mRNA that is differentially regulated in the ExE and inefficiently translated. Interestingly, we observed a similarly reduced expression in the ExE of Cdh1lacZ reporter mice that carry a targeted insertion of a lacZ gene in the Cdh1 locus (Stemmler et al., 2005, and data not shown). Thus, the reduction in Cdh1 expression in the ExE could result from the lack of the endogenous 3′-untranslated region (UTR). An important function of the 1.8-kb-long 3′- UTR of Cdh1 is suggested, because it contains regions that are conserved between different species. This could suggest that the endogenous 3′-UTR contains sequences required for interactions of protein, microRNAs or other regulatory small RNAs important for proper Cdh1 translation in the placenta that are not provided by the artificial exogenous polyA sequences. However, other possibilities have to be considered. A general reduction in gene expression of the knock-in allele between two- and five-fold on protein and mRNA level was detected. This is presumably due to the architecture of the generated knock-in transcript that lacks mRNA splicing. Intron-less transcripts share features with RNA from a retroviral origin that can affect the translation efficiency and nuclear export (Le Hir et al., 2003). Although knock-in gene expression is sufficient to allow development of the embryo proper, we cannot exclude that transcript levels are dropping below a threshold concentration that is essential for correct placenta formation. The contribution of gene-dosage effects to the phenotype is seen in mutants that express Cdh1 from only one knock-in allele (Supp. Fig. S9). If Cdh1 gene dosage is reduced to approx. 50% in Cdh1+/− compared with Cdh1+/+ embryos no histological alteration in placenta development or changes in Cdh1 protein level detected by immunohistochemistry is found (Supp. Fig. S9A–D′) and Cdh1+/− mice are viable (Larue et al., 1994). If the level of gene expression of Cdh1HA/HA mice is further decreased in Cdh1HA/− mutants, embryos are only found in a subset of deciduas and severely affected at E9.5 (Supp. Fig. S9E–G′). The size of the placenta is even smaller than in Cdh1HA/HA embryos. Although further reduction of Cdh1 expression in Cdh1HA/− mutants leads to an additional phenotype in the embryo proper, the level of Cdh1 expression levels may contribute to the phenotype of the placenta.

With the generation of a Cdh1 knock-in allele in which a tagged version of Cdh1 cDNA is under the control of the Cdh1 locus, we generated a hypomorphic allele that for the first time demonstrates the importance of Cdh1 function during placenta development. Previous results about the role of Cdh1 in tissue integrity and homeostasis are extended, and we show that the labyrinth layer formation is dependent on the presence of Cdh1.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Generation of Cdh1HA/HA Knock-in Embryos

A gene-targeting strategy was used as described (Stemmler et al., 2005; Kan et al., 2007). Briefly, a genomic fragment of the Cdh1 locus between −1.5 kb and +11 kb was used to insert a cDNA coding for a C-terminally HA-tagged version of Cdh1 combined with a pEGFP-N3-derived SV40 polyadenylation signal (Invitrogen) into the ATG codon. At the same site, a TK::neoR cassette flanked by loxP sequences was inserted in reverse order downstream of the Cdh1 cDNA for positive selection. Upstream of the genomic Cdh1 fragment, a HSV::tk cassette was included for negative selection. Cloning and sequence details are available upon request. Thirty micrograms of SwaI-linearized targeting vector was electroporated into E14.1 ES cells, with subsequent ES cell selection using 250 μg/ml G418 (Sigma) and 2 μM Ganciclovir (Cymeven) for 10 days. Surviving ES cells were analyzed by Southern blot for homologous recombination as described (Stemmler et al., 2005; Kan et al., 2007). Two independent clones were injected into blastocysts and the resulting chimeric mice were crossed to CMV::Cre deleter mice (Schwenk et al., 1995) for removal of the neoR cassette and backcrossed to C57BL/6 mice for 7 to 10 generations. Animal husbandry and all experiments were performed according to the German Animal Welfare guidelines.

Mouse Breeding and Genotyping

Heterozygous Cdh1HA/+ mice were crossed to C57BL/6 mice and inter se to obtain homozygous mutant embryos. Embryonic stages were determined by the appearance of vaginal plugs at E0.5. For conditional gene ablation, Cdh1HA/+ mice were crossed to Vil::Cre and subsequently to mice carrying homozygous conditional Cdh1 knock-out (Cdh1flox) and Gt(ROSA)26Sortm1Sor/J (Rosa26 reporter; R26Rflox) alleles (Soriano, 1999; Boussadia et al., 2002; el Marjou et al., 2004). Tetraploid blastocyst complementation experiments were done using 4n Gt(ROSA)26Sor/J (Rosa26) embryos as described (Zambrowicz et al., 1997; Hadjantonakis et al., 1998; Tanaka et al., 2009). Genotyping was performed by PCR using tail biopsies, yolk sacs or individual embryos. To analyze Cre-mediated recombination efficiency in the gut, DNA of epithelial cells from adult mice was isolated. Longitudinally opened pieces of gut samples were incubated in 30 mM ethylenediaminetetraacetic acid (EDTA)/Hank's balanced salt solution to detach the epithelia and cells were separated from the lamina propria by centrifugation. The following primers were used: Vil::Cre (Cre_as: GCC AGG TAT CTC TGA CCA GA and Vil-Cre_s: CAA GCC TGG CTC GAC GGC C), CMV::Cre (Cre_s: CAA GTT GAA TAA CCG GAA ATG and Cre_as), Cdh1flox (wt/flox/floxdel; pE10.2: CTT ATA CCG CTC GAG AGC CGG A, pI10as.3: TGA CAC ATG CCT TTA CTT TAG T and In5_s: TGT TCC AAG CCT GCT TTC TT), or (wt/flox; pE10.2 and pE11as.2: GTG TCC CTC CAA ATC CGA TA), Cdh1HA (HA; MMEC_s: CCA TCT CAA GCT CGC GGA TAA C and MMEC_as: CGG CTT CAG AAC CAC TGC C), Cdh1HA (wt; Ecad5′UTR_s: CCAAGAACTTCTGCTAGAC and E-cad_as: TACGTCCGCGCTACTTCA), Cdh1HA (wt/mut; Ecad5′UTR_s, E-cad_as and Ecad_s: AAG CTG GCG GAC ATG TAC).

Time-Lapse Microscopy

Preimplantation embryos were recovered in M2 Medium (Sigma) by flushing both the oviduct and the uterus. Time-lapse recording was performed as described (Hiiragi and Solter, 2004). Embryos were isolated at E2.5, mounted in KSOM (Gynemed), and recorded for 24 hr in 15-min intervals. DNA of individual embryos was isolated for later genotyping.

ES Cell and Teratoma Generation

Blastocysts were isolated from heterozygous intercrosses and plated in 96-well plates on primary embryonic fibroblasts in DMEM supplemented with 15% fetal calf serum (PAN) and 500 U/ml LIF (Chemicon). After attachment and the outgrowth of trophectoderm cells, the ICM was dissociated, transferred to fresh feeder-cell containing wells and expanded once the ES cell colonies became visible. ES cells were genotyped and clones with the correct karyotype were analyzed further. A total of 1 × 107 cells were injected subcutaneously into 6- to 8-week-old nude mice in a single-cell suspension and teratomas of 0.5–1 cm2 were isolated 3–4 weeks afterward. Small pieces were then either used for genotyping or embedded into paraffin for histological analysis.

Antibodies

The following antibodies were used: gp84 (Vestweber and Kemler, 1984; 1:200), anti-Cdh1 (BD, 1:3,000), HA.11 (Covance, 1:2,000), TROMA-1 (Kemler et al., 1981; 1:50), β-catenin (BD, 1:3,000), dephospho-β-catenin (Millipore, 1:2,000), anti-Eplin (Abe and Takeichi, 2008; 1:200), anti-p120 (Zymed, 1:2,000), anti-Erk1/2 (Cell Signaling, 1:2,000), anti-p-Erk1/2 (Cell Signaling, 1:2,000), anti-Akt (Cell Signaling, 1:2,000), anti-p-Akt (Cell Signaling, 1:2,000), anti-Gapdh (Calbiochem, 1:20,000), anti-PCNA (DAKO, 1:1,000), anti-BrdU (Serotec, 1:1,000), anti-GATA4 (Santa Cruz), anti-Oct4 (Santa Cruz, 1:200), alexaFluor488-, alexaFluor594, and alexaFluor680-conjugated secondary antibodies (Molecular Probes, 1:200–1:5,000), IRDye800 (Rockland Immunochemicals, 1:5,000), and horseradish peroxidase-conjugated antibodies (Dianova, 1:10,000).

Immunofluorescence Labeling and Confocal Microscopy

Whole-mount immunofluorescence labeling on preimplantation embryos and on ES cells plated on feeder cells on cover slips was carried out as described previously (Kan et al., 2007). Embryos were mounted in micro drops on glass-bottom dishes, ES cells were mounted on slides mounted with Antifadant Solution (Citifluor), and both were analyzed with a Leica TCS SP2 UV laser scan head attached to a Leica DM IRE2 inverted microscope using Leica Confocal and IMARIS imaging software (Bitplane).

Immunohistochemistry, BrdU labeling, X-Gal and TUNEL Staining

Postimplantation embryos were dissected with or without their placenta, and adult tissue samples were isolated and rinsed in phosphate buffered saline (PBS). Small pieces of intestine and colon were flushed with PBS and partially opened longitudinally. Fixation in 4% paraformaldehyde (PFA) was done between 2 hr and overnight, followed by dehydration and paraffin-embedding. Specimens were sectioned at 7 μm and processed further for immunohistochemistry, in situ hybridization or hematoxylin/eosin staining.

Antigen retrieval and immunostaining were done using Tris/EDTA buffer, EnVision Plus solution (DAKO) and diaminobenzidine (DAB) substrate (Sigma), as described in detail previously (Hierholzer and Kemler, 2010).

For BrdU labeling of both the placenta and the embryos, pregnant females were subjected to a 2-hr pulse of BrdU incorporation by intraperitoneal injection of 100 μg BrdU/PBS (Sigma) per gram of body weight. After embryo dissection, embedding and sectioning, the sections were dehydrated and boiled for 2 min in 10 mM Citric acid, pH 6.0, for antigen retrieval. After incubation for 20 min in 3% H2O2/TBS, slides were treated with 2M HCl for 1 hr at 37°C and subsequently with 0.1 M Boric acid, pH 8.5, for 10 min. Antibody incubation and immunostaining was done according to the immunohistochemistry procedure, and slides were counterstained for 2 min with Carazzi's hematoxylin to visualize nuclei.

X-Gal staining was carried out as described (Stemmler et al., 2005; Kan et al., 2007). Embryos or tissue samples were fixed in PBS/1% formaldehyde/0.2% glutaraldehyde/2 mM MgCl2/5 mM EGTA/0.02% NP-40 for 15–45 min, washed three times in PBS/0.02% NP-40, and incubated overnight in PBS/2 mM MgCl2/5 mM K3Fe(CN)6/5 mM K4Fe(CN)6/0.01% sodium desoxycholate/0.02% NP-40/1 mg/ml X-gal. After post-fixation with 4% PFA, some specimens were embedded in paraffin, sectioned at 7 μm and counterstained with nuclear fast red (Sigma).

TUNEL Staining

Embryos were isolated at E10.5 with the yolk sac and the placenta, fixed, dehydrated, and paraffin-embedded. After sectioning, TUNEL staining was performed using In Situ Cell Death Detection Kit (Roche) according to the manufacturer's instructions.

In Situ Hybridization

Embryos were fixed in 4% PFA overnight, dehydrated through a graded series of methanol, stored overnight in 100% methanol and then embedded into paraffin. Subsequently, specimen were sectioned at 7 μm, dewaxed, and rehydrated. After digestion of RNA-bound proteins with 10 μg/ml proteinase K/0.1 M Tris-HCl pH 7.5 for 10 min at 37°C, slides were treated with 0.1 M triethanolamine dropwise supplemented with acetic anhydride to a final concentration of 0.25% for 10 min at room temperature. Slides were then incubated with a digoxigenin-labeled riboprobe for Cdh1 (Stemmler et al., 2003) and placenta-specific marker genes (Hitz et al., 2005; Natale et al., 2006; Simmons et al., 2008) at 70°C overnight. Bound probe was detected with alkaline phosphatase-coupled anti-digoxigenin antibody (Roche) and BM purple substrate (Roche).

Immunoblotting and Immunoprecipitation

Immunoblotting was carried out as described previously (Hecht and Stemmler, 2003; Kan et al., 2007). Organs were isolated in PBS and lysed in lysis buffer (20 mM Tris-HCl pH 7.6, 150 mM NaCl, 2 mM MgCl2, 2 mM EDTA, 0.5% Triton X-100, 0.5% Igepal, 10% Glycerol, 1 mM PMSF, complete protease inhibitor cocktail [Roche]) using a Dounce homogenizer. Insoluble cellular material was removed by centrifugation for 10 min at 20,000 g. Twenty micrograms of total protein was separated on an 8% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gel. For immunoprecipitation, the cells were resuspended in lysis buffer. Five hundred micrograms of total protein were incubated in a 500 μl volume supplemented with 20 μl equilibrated Protein G-Agarose beads (Roche) that had been saturated with 500 ng of the appropriate antibody. After the removal of unbound protein by repeated washing in TBS/Tween, pelleted beads were incubated in 25 μl of 2× SDS-LB (100 mM Tris-HCl pH 6.8, 4% SDS, 10% β-mercaptoethanol, 20% glycerol, 0.2% bromphenol blue) for 5 min at 100°C, and 10 μl was separated on an 8–10% SDS-PAGE gel. Quantitative protein detection was carried out using a Li-Cor Odyssey Infrared Imaging System according to the manufacturer's instructions.

RNA Isolation and qPCR

Total cell RNA was isolated by Trizol (Invitrogen), and 2 μg of DNaseI-treated RNA was reverse-transcribed by a SuperscripIII kit (Invitrogen). Relative quantification was performed with oligodT-generated cDNA corresponding to 1 ng of initial RNA per qPCR reaction. The Absolute QPCR ROX Mix (ABgene) and the Universal Probe Library (Roche) with an ABI 7300 system (Applied Biosciences) were used according to the manufacturer's instructions. The following primers were used for Cdh1+ and Cdh1HA (5′EcadUPL_s: AGT GTT TGC TCG GCG TCT and 5′EcadUPL_as: GCA AAG CCA TGA GGA GAC C); for Cdh1HA (3′EcadUPL_s: CAC CCC CTT ACG ACT CTC TG and HAUPL_as: GAC GTC ATA AGG ATA TCC AGC A) and for Actb (bActUPL_s: AAG GCC AAC CGT GAA AAG AT and bActUPL_as: GTG GTA CGA CCA GAG GCA TAC).

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

The polyclonal rabbit anti-mouse Eplin antibody was a kind gift of Dr. Masatoshi Takeichi, placenta-specific in situ probes were kindly provided by David R. Natale and James C. Cross, and we thank Dr. Sylvie Robine for providing the Vil::Cre mice. We are grateful to Drs. Rolf Kemler, Verdon Taylor and David R. Natale for helpful discussions and critically reading the manuscript, to Benoît Kanzler for support in generating transgenic mice and tetraploid morula aggregation and to Janine Seyfferth for excellent technical assistance. I.B. is a PhD student of the Faculty of Biology, University of Freiburg. This work was supported by the Max-Planck Society.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
DVDY_22375_sm_SuppFig1.tif5801KSupporting Information Figure 1. Detailed analysis of Cdh1HA/+ gene targeting. A: Representative images of the Southern blot analysis using the indicated restriction digests and probes indicate successful homologous recombination in embryonic stem (ES) cells without aberrant locus modifications and proper deletion of the selection cassette in resulting mice. B: Immunohistochemistry with indicated antibodies on sections of embryonic day (E) 14.5 embryos showing protein expression of the knock-in allele that is consistent with endogenous Cdh1 expression and the absence of anti-HA staining in wild-type (wt) embryos; n.d., not done. Scale bar = 100.
DVDY_22375_sm_SuppFig2.tif1043KSupporting Information Figure 2. Homozygous Cdh1HA/HA embryos form normal blastocysts. A–C: Compacted morulae of heterozygous Cdh1HA/+ intercrosses were isolated at embryonic day (E) 2.5 and observed in KSOM medium by time-lapse recording for blastocyst formation every 15 min over 24 hr. Representative images at the beginning (A–C) and at the end of the experiment (A′,B′,C′) are shown. D–F″: Immunofluorescent staining of embryos isolated at E3.5 with anti-Cdh1 (D,E,F), and anti-HA (D′,E′,F′) antibodies are imaged using the same settings and are shown together with merged channels (D″,E″;F″). Intensities of anti-Cdh1 staining show comparable intensities of homozygous mutants and control littermates. The HA-tagged portion of Cdh1 shows higher intensity staining in the homozygous mutants, compared with the heterozygous mutants. Both endogenous and knock-in allele expression overlap at the basolateral membrane in trophectoderm and ICM. G: Summary of in vitro blastocyst formation of cultured embryos between E2.5 and E3.5. The number of embryos that formed a blastocyst was found in a normal Mendelian ratio, indicating that homozygous embryos form trophectoderm as efficiently as their heterozygous and wt littermates. Similar results were obtained with in vivo-formed blastocysts that had been directly flushed from the uterus. Scale bar = 10.
DVDY_22375_sm_SuppFig3.tif1267KSupporting Information Figure 3. Molecular analysis of Cdh1HA/HA and Cdh1HA/+ embryonic stem (ES) cells established from blastocyst outgrowth. A: Polymerase chain reaction (PCR) genotyping of ES cells obtained using primers Ecad5′UTR_s, E-cad_as and Ecad_s. B,C: qPCR of Cdh1 transcripts using primers detecting both transcripts derived from wild-type (wt) and knock-in origin (B) and detecting only Cdh1HA transcripts (C). In homozygous ES cells transcripts were found at 34% of wild-type levels. A 26% reduction was found in Cdh1HA/+ ES cells based on accumulated expression from one wt and one knock-in allele. Data were normalized to Actb levels, values from at least two experiments are given and error bars represent standard error of the mean (SEM). D: Immunoblot scanned with Li-Cor Odyssey system simultaneously incubated with anti-Cdh1 (intracellular), gp84 (Cdh1 extracellular), and anti-Gapdh antibodies, an anti-HA blot and a typical quantitative analysis for single channels of all scans. For each cell line 5, 10 and 30 μg of total protein were loaded. E: Results from quantification of three different immunoblots indicating a three- to five-fold reduction of protein levels in Cdh1HA/HA vs. Cdh1+/+ and a two- to three-fold reduction in Cdh1HA/HA vs. Cdh1HA/+ samples. These changes are similar for the different anti-Cdh1 antibodies, indicating that very likely the affinity of the antibody to the Cdh1-HA protein is not altered. In a corresponding comparison of the anti-HA immunoblot, the amount of protein from the knock-in allele is two-fold more abundant in the homozygous mutant than in heterozygous ES cells. F,G: Analysis of proteins involved in signaling cascades that are regulated by Cdh1, like Wnt-, Mapk-, and Akt-signaling. E: The amount of β-cat, dephospho-β-cat (dp-β-cat), and p-Erk1/2 signals are reduced in the Cdh1HA/HA ES cells, indicating that signaling that activates Erk1/2 by phosphorylation is reduced. F: Cdh1HA/HA, Cdh1HA/+, Cdh1+/+ ES cells were transfected with the Topflash reporter to monitor Wnt activity, alone or in combination with stabilized β-cat (βS33A) and Lef1 or a constitutively active form of Lef1 (LefΔN-VP16) (Hecht et al., 2000) that was used as positive control. No difference in reporter gene activity of the three cell types is detected, indicating that the reduction in signaling-competent, unphosphorylated β-cat is not reflecting a reduction in Wnt/β-cat signaling but in cadherin-bound β-cat. Three independent experiments were done, error bars represent SEM and activity of Topflash alone in Cdh1HA/HA ES cells is set to 1.
DVDY_22375_sm_SuppFig4.tif3754KSupporting Information Figure 4. Cdh1HA/HA embryonic stem (ES) cells are phenotypically normal and differentiate into all three germ-layers. A–E′: Immunofluorescence staining of Cdh1HA/HA (A–A″,D,D′), Cdh1HA/+ (B–B″,E,E′) and Cdh1+/+ (C–C″) ES cells for Cdh1 and HA together with merged staining and for Oct4 and GATA4 to verify pluripotency. Nuclei are stained by DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). Cadherin staining is found at comparable intensities in both the homozygous mutant and the controls, whereas expression of the knock-in gene, detected by anti-HA antibody, is higher in the homozygous mutant ES cells due to gene dosage effects. Oct4 is present and GATA4 is absent in all ES cells, indicating that cells are undifferentiated and pluripotent. Only the nucleus of a feeder cell stains Oct4-negative (D) and GATA4-positive (D′). F–K: Histological (F,I) and immunohistochemical analysis (G,H,J,K) of consecutive sections of teratomas analyzed three weeks after subcutaneous injection of Cdh1HA/HA (F–H), Cdh1HA/+ (I–K) ES cells into nude mice. Histological analysis of several sections show similarities in differentiation potential from all ES-cell lines. Derivatives of all three germ layers are found, like epithelia both from ectodermal and from endodermal origin that express Cdh1 from both wt and knock-in allele. Scale bars = 10 μm in A–C″, 20 μm in D–E′, 100 μm in F–K.
DVDY_22375_sm_SuppFig5.tif1930KSupporting Information Figure 5. The binding of Cdh1 to β-catenin and assembly of the cadherin-catenin complex is not affected by the C-terminal HA-tag. A–D: A total of 500 μg protein from embryonic stem (ES) cell lysates of Cdh1HA/HA (HA/HA), Cdh1HA/+ (HA/+), and Cdh1+/+ (+/+) cells were incubated with protein-A agarose beads alone (beads), immunoprecipitated (IP) with either anti-Cdh1 (IP: Cdh1) or anti-β-cat (IP: β-cat) antibodies. and immunoblotted (IB) with anti-β-cat (A), anti Cdh1(B), anti-HA (C), anti-Eplin (D), and anti-p120 antibodies as indicated on the right side of the figure together with 1/20 of input (lysate). Densitometric measurements of the corresponding bands are given as relative values normalized to IgG in A and B (+/+ is set to 1.00). The interaction of Cdh1-HA protein with β-cat is found with no detectable differences when comparing IP and co-IP of lysates from homozygous mutants and heterozygous or wild-type (wt) ES cells (A,B). β-Catenin co-immunoprecipitation efficiently for all genotypes confirming a similar binding affinity for β-catenin to Cdh1 in wt, heterozygous or homozygous mutant cells. Similar findings were obtained with IP and co-IP of anti-HA immunoblots (C). Additionally, the presence of Eplin, a protein suggested to link α-catenin to the cadherin-catenin complex at the actin cytoskeleton (Abe and Takeichi, 2008), was analyzed in the immunoprecipitated protein complex. Similar levels of Eplin in all immunoprecipitations also show that binding of indirectly linked proteins are not changed in Cdh1HA/HA cells (D). Although p120 levels were similar in lysates of all genotypes, a light reduction in p120 protein in Cdh1HA/HA was co-precipitated with anti-Cdh1 consistent with the reduced Cdh1-HA levels in those cells (E). Filled arrowheads indicate the corresponding full-length protein, and the α and β isoforms of Eplin in D, open arrowheads mark the IP immunoglobulins and the asterisk in C mark p120 from a previous antibody incubation (E). Gapdh detection was used as loading control (F).
DVDY_22375_sm_SuppFig6.tif3758KSupporting Information Figure 6. Cdh1HA/HA embryonic stem (ES) cells mix well with the ICM and show equal contribution to all three germ layers in chimeric animals. A–J: β-Galactosidase (β-gal) staining (A–D′) and immunohistochemistry (E–J) of chimeric embryos or tissues of 35-week-old adult mice derived from Cdh1HA/HA (A–B′,E–J′) and Cdh1HA/+ (C–D″) ES cells injected into Rosa26 wt blastocysts. Embryos are stained as whole-mounts (A,C) and sections are shown (B,B′,D,D′), demonstrating β-gal-negative ES cell-derived cells (filled arrowheads) contributing to β-gal-positive epithelia of the host embryo (open arrowheads). Cells of the homozygous mutant ES cells are still present in adult Cdh1-expressing cells detected by anti-HA immunohistochemistry (H–J), without a detectable reduction in expression, based on anti-Cdh1 staining (E–G) in skin of the ear (E,H), the small intestine (F,I) and the pancreas (G,J). Arrows label the positions of the same structures positive for anti-HA and anti-Cdh1 staining. Boxes in B,D indicate the position of blow-ups in B′,D′, respectively. Scale bars = 1 mm in A,C; 250 μm in B,D; 25 μm in B′,D′; 100 μm in E–J.
DVDY_22375_sm_SuppFig7.tif9133KSupporting Information Figure 7. Mice that express Cdh1 in the intestine only from the knock-in allele share similar expression levels with their heterozygous littermates and do not show any change in phenotype. Immunohistochemistry of Cdh1 staining of sections from different parts of the gut in 19-week-old mice. A–F″: Proximal (A–A″,D–D″) and distal (B–B″,E–E″) parts of the intestine and the colon (C–C″,F–F″) of Vil::Cre;Cdh1HA/Δ;R26RΔ/+ mutants (A–C″) and Vil::Cre;Cdh1HA/+;R26RΔ/+ control littermates (D–F″) are shown at different magnifications. No differences in protein localization or in staining intensity between homozygous and heterozygous mutants are detected. Scale bars = 500 μm in A–F, 100 μm in A′–F′), 25 μm in A″–F″).
DVDY_22375_sm_SuppFig8.tif8333KSupporting Information Figure 8. Additional marker gene analysis of different layers of the placenta at embryonic day (E) 10.5. A–B″: Sections of Cdh1HA/HA and Cdh1HA/+ embryos of Fig. 5 stained with Hand1 (A,B), Mash2 (A′,B′), and Dlx3 riboprobes (A″,B″). Dashed lines schematically separate the different layers of the placenta. C–H′: Sections at the medial plane of the placenta of Cdh1HA/HA and Cdh1HA/+ embryos stained for Pl1 (C–D′), Vegfr1 (E–F′), and Vegfr2 (G–H′). These markers specifically label cells of the layers as indicated on the left side of the panels. Increased staining in the Cdh1HA/HA placenta is observed for the trophoblast giant cell layer (TGC) -specific gene Pl1. A reduction in the number of Vegfr1- and Vegfr2-positive cells but not in staining intensities is detected in the Cdh1HA/HA placenta, confirming impaired Lb formation in the homozygous mutants. Boxes in C–H indicate the position of blow-ups in C′, –H′, respectively. Scale bars = 250 μm in A–B″, 500 μm in C–H, 50 μm in C′–H′).
DVDY_22375_sm_SuppFig9.tif6901KSupporting Information Figure 9. Analysis of Cdh1+/+, Cdh1+/−, Cdh1HA/−, and Cdh1HA/+ phenotypes at embryonic day (E) 9.5. A–H′: Embryos of Cdh1+/+ (A,A′,C,C′), Cdh1+/− (B,B′,D,D′), Cdh1HA/− (E,E′,G,G′), and Cdh1HA/+ (F,F′,H,H′) analyzed histologically by hematoxylin/eosin (A–B′,E–F′) and by Cdh1 immunohistochemistry staining (C–D′,G–H′). No differences are detected in placenta size, layer formation and Cdh1 protein staining in Cdh1+/− embryos compared with Cdh1+/+ or Cdh1HA/+ embryos. The phenotype of Cdh1HA/− is more severe than of Cdh1HA/HA embryos. Most Cdh1HA/− implantation sites contain only clumps of cells of the embryo proper and the remaining sites contain retarded malformed embryos with severely impaired placentas with a smaller Lb than of Cdh1HA/HA (E–E′). The Lb is devoid of Cdh1 staining (G,G′), only the yolk sac is Cdh1-positive as seen in G′. Boxes in A–H indicate the position of blow-ups in A′–H′, respectively. Scale bars = 500 μm in A–H; 50 μm in A′–H′.
DVDY_22375_sm_SuppMovie.mov8920KSupporting Information Movie. Movie M1. A 24-hr time-lapse analysis of blastocyst formation of Cdh1HA/HA mutant embryos. Embryos are flushed at embryonic day (E) 2.5, the morula stage, and mounted in KSOM for recording. Genotypes of embryos are indicated by a series of still images at the beginning and at the end of the movie: Cdh1HA/HA (ki/ki), Cdh1HA/+ (ki/+) and Cdh1+/+ (+/+). All embryos form blastocysts in vitro. After being recorded, embryos were subjected to genotyping.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.