Deficiency in crumbs homolog 2 (Crb2) affects gastrulation and results in embryonic lethality in mice


  • Zhijie Xiao,

    1. Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Jaakko Patrakka,

    1. Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Masatoshi Nukui,

    1. Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Lijun Chi,

    1. Department of Medical Biochemistry and Molecular Biology, Oulu Center for Cell Matrix Research, Biocenter Oulu, Laboratory of Developmental Biology, University of Oulu, Oulu, Finland
    Search for more papers by this author
  • Dadi Niu,

    1. Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Christer Betsholtz,

    1. Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Timo Pikkarainen,

    1. Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, Stockholm, Sweden
    Search for more papers by this author
  • Seppo Vainio,

    Corresponding author
    1. Department of Medical Biochemistry and Molecular Biology, Oulu Center for Cell Matrix Research, Biocenter Oulu, Laboratory of Developmental Biology, University of Oulu, Oulu, Finland
    • Department of Biochemistry and Molecular Biology, University of Oulu, Oulu, Finland
    Search for more papers by this author
  • Karl Tryggvason

    Corresponding author
    1. Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, Stockholm, Sweden
    • Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Karolinska Institutet, 171 77 Stockholm, Sweden
    Search for more papers by this author


This article is corrected by:

  1. Errata: Erratum: Deficiency in crumbs homolog 2 (Crb2) affects gastrulation and results in embryonic lethality in mice Volume 241, Issue 2, 431, Article first published online: 16 December 2011


The Crumbs family of transmembrane proteins has an important role in the differentiation of the apical membrane domain in various cell types, regulating such processes as epithelial cell polarization. The mammalian Crumbs protein family is composed of three members. Here, we inactivated the mouse Crb2 gene with gene-targeting techniques and found that the protein is crucial for early embryonic development with severe abnormalities appearing in Crb2-deficient embryos at late-gastrulation. Our findings indicate that the primary defect in the mutant embryos is disturbed polarity of the epiblast cells at the primitive streak, which affects epithelial to mesenchymal transition (EMT) during gastrulation, resulting in impaired mesoderm and endoderm formation, and embryonic lethality by embryonic day 12.5. These findings therefore indicate a novel role for the Crumbs family of proteins. Developmental Dynamics 240:2646–2656, 2011. © 2011 Wiley Periodicals, Inc.


Crumbs homolog 2 (Crb2) is one of the three mammalian homologues of the Drosophila Crumbs (Crb) protein that regulates epithelialization and differentiation of the apical membrane during blastoderm stages (Tepass et al., 1990). Impaired Crb function disturbs organization of epithelial cells that are derived from the embryonic ectoderm and leads to epithelial cell apoptosis (Tepass et al., 1990), while Crb overexpression generates expansion of the apical surface (Wodarz et al., 1995). Crb is involved in the control of cell–cell adhesion and epithelial cell polarity, and it functions by interacting with the stardust and bazooka protein complexes (Tepass and Knust, 1993; Tanentzapf and Tepass, 2003).

Of the three mammalian Crb proteins, Crb1 and Crb3 have been shown to regulate apical–basal polarity in complex with other proteins, such as Pals1/Patj and Par3/Par6/aPKC (orthologs of Drosophila stardust/DPatj and Bazooka/Par6/aPKC, respectively; Roh et al., 2002, 2003; van de Pavert et al., 2004; Margolis and Borg, 2005; Shin et al., 2006; Shin and Margolis, 2006). Human Crb3 has also been shown to be involved in formation of intercellular tight junctions in vitro (Lemmers et al., 2004; Fogg et al., 2005). Mutations in the CRB1 gene underlie human retinitis pigmentosa (RP), and some cases of Leber congenital amaurosis (LCA), a progressive degenerative retinopathy leading to visual impairment in childhood (den Hollander et al., 1999, 2001, 2002). It localizes to the subapical region above the adherens junction between photoreceptor and Muller glial cells, where it is required for the adherence between these cells. Consistent with these findings, mice deficient in Crb1 show defects in retina adherens junctions after 3–9 months of age (van de Pavert et al., 2004). Crb3 knockout has not yet been reported. In humans, Crb2 expression has been detected by reverse transcriptase-polymerase chain reaction (RT-PCR) in retina, adult brain, and kidney, but its function is poorly understood (van den Hurk et al., 2005).

Our large-scale mouse glomerulus gene profiling project identified Crb2 as a highly podocyte specific gene in adult mouse kidney (Takemoto et al., 2006). In zebrafish, Crb2b knockdown led to glomerular foot process effacement and loss of the slit diaphragm structure (Ebarasi et al., 2009). Here, we inactivated the mouse Crb2 gene through homologous recombination in embryonic stem (ES) cells, and found, somewhat surprisingly, that Crb2 is crucial for the early steps of embryogenesis, the protein being essential at late-gastrulation and onward. In Crb2 deficiency, the primary defect appears to be disturbed epiblast cell polarity, which affects the EMT at the primitive streak, leading to defects in the mesoderm and endoderm formation. As a result, several organ primordia, including neuroepithelium, gut, and heart, fail to form properly. Thus, these findings indicate a novel role for the Crb family of proteins.


Crb2 Is Expressed During Gastrulation and Early Organogenesis Stages in the Mouse

The murine Crb2 gene contains 13 exons, localizes to chromosome 2, and encodes a 1,282-amino acid transmembrane protein. Based on the SMART database, the Crb2 protein has a conserved 38-amino acid intracellular domain and an extracellular domain with 15 endothelial growth factor (EGF) -like and 3 laminin G domain-like repeats. To address the potential developmental roles of Crb2, we studied its expression pattern during early mouse embryogenesis using the whole-mount in situ hybridization technique. Expression was analyzed on embryonic day (E) 7–E8.5 embryos, representing those from the middle-streak stage to the ones containing 12 somites.

The in situ hybridization analysis indicated that Crb2 is widely expressed throughout the epiblast and the lateral plate mesoderm at E7 (Fig. 1A). At E7.5–E7.75, Crb2 continues to be expressed at these regions, and expression is also found on the apical side of the embryonic endoderm, and extra-embryonic amnion and allantois (Fig. 1B). At E8–E8.5, Crb2 expression is detected also in the heart tube, foregut, and the apical side of the somite epithelium (Fig. 1D–F). Stronger Crb2 expression is detected on the apical sides of the splitting lateral plate mesoderm, and the apical side of the neural ectoderm at trunk region. Crb2 is not expressed in the notochord plate or the extra-embryonic endoderm (Fig. 1F). In summary, the Crb2 gene is expressed in early embryonic cells and typically in embryonic regions undergoing dramatic rearrangement, such as the developing neuroepithelium which proceeds with neural tube closure, the anterior splitting lateral plate mesoderm that wraps the pericardial cavity, and the differentiating somite epithelium. These developmental events involve processes of EMT or mesenchymal to epithelial (MET) transition, indicating that Crb2 may have important roles in these processes during the early embryonic development.

Figure 1.

Crb2 expression during embryonic development. Analysis of Crb2 expression during embryonic day (E) 7–E8.5 by whole-mount in situ hybridization. A: At E7, the in situ hybridization signal is detected throughout the ectoderm and the lateral plate mesoderm, as well as the extra-embryonic mesoderm (the latter not shown). ec, ectoderm; e-en, embryonic endoderm; la-m, lateral plate mesoderm; ps, primitive streak. B: During E7.5–E7.75 Crb2 is ubiquitously expressed except that it is not detected in extra-embryonic endoderm (ex-en). In the mesoderm, Crb2 signal is stronger at the splitting lateral plate mesoderm (la-m). am, amnion; ex-m, extra-embryonic mesoderm; me, mesoderm. C: Hybridization of E7.75 embryos with anti-sense (as) and sense (s) probes demonstrates the specificity of the signal obtained with the anti-sense probe. D–F: Panels show Crb2 expression during E8–E8.5. Note that in E the somatic Crb2 expression is at the apical side of the somite epithelium (arrows). fg, foregut; ht, heart tube; so, somite; n, notochord plate. Scale bar = 100 μm.

Crb2 Knockout Impairs Gastrulation and Is Embryonic Lethal

By gene targeting and homologous recombination in ES cells, all coding exons of the Crb2 gene were replaced with an enhanced green fluorescent protein (eGFP) lox-Ub1- EM7- Neo- lox cassette (Fig. 2A). Crb2+/− mice were mated to generate +/+, +/−, and −/− offspring. The genotype analyses (Fig. 2B) revealed, however, that no viable −/− mice or embryos were obtained after E12.5. Hence, Crb2 deficiency is embryonic lethal, indicating a critical early developmental role for this protein. To determine the time of intrauterine lethality more closely, we genotyped embryos each day from E7 to E10.5. When summing up this data, embryos with all respective genotypes were identified at the expected Mendelian frequencies (Fig. 2C). The lack of the Crb2 transcript in Crb2-null embryos was confirmed by RT-PCR (Fig. 2D).

Figure 2.

Complete deficiency in Crb2 results in early embryonic lethality. A: Gene-targeting strategy. This is a schematic presentation and not to scale. B: Genotyping of embryos by polymerase chain reaction (PCR) confirms that embryos homozygous for the null mutation (−/−) do not have the wild-type allele, whereas heterozygous embryos (+/−) contain both the wild-type and the mutated allele. See the Experimental Procedures section for the primers used. C: Summary of the genotyping results for embryos and 8- to 10-day-old mice. D: Analysis of Crb2 expression by reverse transcriptase (RT)-PCR shows that the Crb2 transcript is lost in the null mutant. Total RNA was extracted from E8 littermate embryos. G3PDH (glyceraldehyde-3-phosphatedehydrogenase) was used as a template loading control.

Embryos heterozygous for the Crb2-null allele were viable and fertile and did not reveal any notable phenotypes. In contrast, the Crb2−/− embryos enlarged in size until around E10.5 (Fig. 3a), but no embryonic structures were distinguishable at this stage, and by E12.5, these mutants were resorbed completely. Closer inspection revealed that the Crb2-deficient embryos appeared morphologically normal until E7.5, a stage when gastrulation has normally progressed for approximately 1 day. However, by E7.75, the Crb2-deficient embryos had already acquired several severe developmental anomalies as judged by histological inspection of the knockout embryos and their wild-type littermate controls. Indeed, the head fold, heart tube, foregut invagination, and somite development were all disturbed in the absence of Crb2 function (Fig. 3b, B′,E′). Furthermore, the primitive streak of the mutant embryos was clearly wider than that in normal controls (Figs. 3b,I′,N′, c), whereas the mesoderm germ layer of these embryos was more compact and reduced in thickness Fig. 3b,J′,O′). Moreover, the anterior end of the Crb2-deficient primitive streak had not shifted proximally as normally occurs after E7.5 (Fig. 3b,C′,F′, arrowheads). Hence, the data indicate that Crb2 deficiency leads to severe gastrulation defects, resulting in impaired somitogenesis and organogenesis and embryonic death by E12.5.

Figure 3.

Crb2−/− embryos display malformations from the embryonic day (E) 7.75 late-gastrulation stage onward. a: Comparison of littermate controls (Con., +/+ or +/−) and Crb2−/− embryos. At E7.5, the control and Crb2−/− embryos do not show any obvious differences. At E7.75, the Crb2−/− embryos clearly exhibit abnormalities, as manifested, for example, by the retarded primitive streak (ps) and the flat head fold (hf). The null embryos exhibit holistic defects at E8 but keep expanding until E10.5. For each stage, photographs of the embryos were taken under the same magnification. al, allantois; exe-ys, extra-embryonic yolk sac; pla, placenta. b: Hematoxylin–eosin staining of series of sections from E7.75 littermate control and Crb2−/− embryos. A, representative photographs of E7.75 control and Crb2−/− embryos. B–G and H–Q show, respectively, sagittal and transversal sections corresponding to the positions indicated in A. B′, C′, E′, F′, J′, I′, O′, and N′ show blown-up views of the boxed area in frames B, C, E, F, J, I, O, and N. Comparison of the frames H and M show the absence of somites (so) in the Crb2−/− embryos, whereas the frames L and Q reveal that blood cells (bl) do not form in the Crb2−/− extra-embryonic mesoderm at E7.75. fg, foregut diverticulum; ht, heart tube; me, mesoderm. c: Measurement of the width of the primitive streak at E7.75–E8 shows that it is clearly wider in the Crb2−/− embryos than in their wild-type controls (P < 0.0001 by t-test, n = 10).

Lack of Crb2 Function Causes Defects in Development of the Extra-embryonic Structures and Failure in Early Organogenesis

Besides the early embryonic defects, absence of Crb2 function leads to severe malformations of the extra-embryonic structures, namely amnion, allantois, and yolk sac development. Notably, amniotic cavity of the mutant embryos was small (data not shown), and the allantois of the Crb2-deficient embryos did not elongate and did not fuse properly to the chorion, but instead fused to the caudal yolk sac region (Fig. 3b,B,E). Blood island formation in the yolk sac became apparent first at around E8 (data not shown), and the subsequent yolk sac vasculogenesis was also impaired, while this process is well on its way in the normal embryos at E7.5 and E9.5 (Fig. 3b,L,Q, and data not shown).

Cer1 is an anterior visceral endoderm marker before E7.75 (Biben et al., 1998). Its expression was almost identical in control and mutant embryos at E7.5 (Fig. 4, Cer1), which demonstrated that the anterior–posterior body axis formation during early gastrulation process was normal (Perea-Gomez et al., 2001). Subsequently, Cer1 serves, however, as a marker for somite patterning, because its expression is confined to the nascent and anterior half of newly formed somites (Biben et al., 1998). Indeed, Cer1 gene expression was severely reduced with no clear patterned expression in the E9 Crb2−/− embryos (Fig. 4, Cer1). Further studies on somitogenesis revealed that the Crb2-deficient embryos developed somitic mesodermal defects at E7.75 as judged by changes in Mox1 marker gene expression. The Mox1-positive presomitic mesoderm did not generate well-segmented and -patterned somites. Somitogenesis process was disturbed altogether, and only a few pre-somite pairs were observed at E9 when the control embryos had already developed around 20 pairs of somites (Fig. 4, Mox1). Thus Crb2 deficiency has a drastic effect on somitogenesis.

Figure 4.

Crb2−/− embryos fail in early organogenesis. Analysis of the expression of different marker genes by whole-mount in situ hybridization in control and Crb2−/− embryos. For each marker, photographs of the embryos were taken under the same magnification. All the embryos are orientated so that the head is to the left. al, allantois; ave, anterior visceral endoderm; exe-ys, extra-embryonic yolk sac; fg, foregut diverticulum; hg, hindgut diverticulum.

In addition to the somites, loss of Crb2 function leads to a severe deficiency in the cardiovascular system development. This conclusion is based on the finding that the heart tube is completely missing in the mutant embryos (Fig. 3b,B′,E′). We noted only some clusters of scattered blood forming islands in the Crb2-deficient embryos by E10.5 (Fig. 3a, E10.5), indicating that these embryos have some capacity to develop certain primitive blood cells but are not sufficient to develop a well-structured heart primordium.

The Crb2-deficient embryos also lack the foregut and hindgut diverticulum as shown by the entirely exposed Shh-positive notochord at the ventral surface of the mutant embryo. In contrast, both ends of the notochord grew dorsally along with gut invagination in the control embryos (Fig. 4, Shh, black arrows). The Crb2-deficient embryonic endoderm showed an expanded phenotype instead of being restricted only to the inner lining of the gut (Fig. 4, Shh, red bars, red arrows show corresponding Shh signal in controls and mutants). Note that the Shh-labeled notochord was discontinuous since E7.75, suggesting a defective notochord formation as well (Fig. 4, Shh, arrow heads). Concurrently, the Crb2 null embryo did not start the “turning” process, which is the inversion of germ layers occurring during E8–E9 of the mouse embryonic development.

Crb2 Is Required for Epithelial–Mesenchymal Transition (EMT) in the Primitive Streak

The marker studies revealed that a variety of cell lineages for different embryonic structures had the capacity to develop in the absence of Crb2, but the cells failed to develop functional structures, which suggests a gastrulation failure. Gastrulation is characterized by a process where the epiblast cells in the primitive streak lose their polarized morphology and undergo EMT to generate the motile mesodermal cells and the definitive endoderm germ layer critical for subsequent morphogenesis (Hashimoto et al., 1987; Tam et al., 1993). In this process, expression of E-cadherin, a cell–cell adhesion molecule important for epithelial morphology, becomes down-regulated in cells undergoing EMT (Thiery, 2002; Baum et al., 2008). Along with this, the basement membrane located between the germ layers is degraded locally, as indicated by the loss of perlecan expression (Tam et al., 1993). In addition, the differentiated mesodermal cells acquire a specific morphology and are surrounded by a typical extracellular matrix containing for example fibronectin, which controls cell migration between the germ layers during gastrulation (Boucaut et al., 1990; Klinowska et al., 1994).

To address a possible role of Crb2 in the control of EMT, we analyzed expression of the above mentioned proteins. We found that at E7.75, when no E-cadherin was detected in the mesoderm of wild-type embryos, its expression remained in that of the mutant ones (Fig. 5A,B, stars). Furthermore, the perlecan-negative region in the primitive streak of the Crb2-deficient embryos was much wider than in their wild-type controls, indicating a larger posterior area undergoing EMT in the absence of Crb2 (Fig. 5C,D). After E8, no E-cadherin expression is detected in the mutant mesoderm either, whereas staining for fibronectin indicated dramatically different staining patterns between the wild-type and mutant embryos (Fig. 5E,F). Indeed, there was very little fibronectin present in the mesoderm of the mutant embryos, whereas an abundant and widespread signal was detected in that of the wild-type embryos. These findings reflect severe defects in mesoderm formation, because mesoderm itself is a source of fibronectin (Suzuki et al., 1995). Thus, we conclude that Crb2 is critical for the EMT program, and the defects in the process severely affect subsequent organogenesis.

Figure 5.

Crb2 deficiency leads to defects in the epithelial to mesenchymal transition (EMT) at the primitive streak during late-gastrulation. Photographs show immunofluorescence staining of transversal sections in the posterior region of control and Crb2−/− embryos at embryonic day (E) 7.75–E8.25. A–D: double staining for E-cadherin (green) and perlecan (red). E,F: single staining for fibronectin. Cell nuclei are visualized by DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). Panels A, C, and E represent control embryos, whereas B, D, and F show the stainings in the Crb2-deficient ones. Each pair of the embryos is photographed under the same magnification. am, amnion; bm, basement membrane; ec, ectoderm; me, mesoderm; ps, primitive streak.

Expression of Bmp4, a Mesodermal Inducer, Is Down-regulated in Crb2 Deficiency

Bmp4, Wnt3a, and Fgf8 are involved in mesoderm induction (Takada et al., 1994; Crossley and Martin, 1995; Ciruna and Rossant, 2001). Thus, changes in expression of these factors may lie behind the defective gastrulation due to Crb2 deficiency. Of interest, the Bmp4 expression was severely down-regulated in the Crb2-deficient embryos as analyzed at E7–E7.75, while the amnion and allantois expression of Bmp4 was retained (Fig. 6, Bmp4). At this stage, Brachyury and Fgf8 expression in the primitive streak remain unchanged in the Crb2−/− embryos (Fig. 6, Brachyury, Fgf8). Therefore, mesodermal defects seen in the mutant embryos may be partly explained by decreased Bmp4 expression.

Figure 6.

Bmp4 expression is down-regulated in the Crb2−/− embryos. Analysis of the expression of Bmp4, Brachyury, and Fgf8 by whole-mount in situ hybridization in control and Crb2−/− embryos. For each marker, photographs of the embryos were taken under the same magnification. All the embryos are orientated so that the head is to the left. exe-ys, extra-embryonic yolk sac; ps, primitive streak.

Crb2 Deficiency Leads to Neuroepithelium Malformation

We found that the neuroepithelium of the Crb2-deficient embryos is poorly developed, and the neural tube closure did not occur. Furthermore, we found that perlecan staining in the basement membrane beneath the malformed neuroepithelium was discontinuous at E8.25 (arrowheads in Figs. 7B,D, 8D). By E9.5, the Crb2-deficient neuroepithelium became partially disassembled, lost the sheet-like structure and mixed with the mesodermal cells (Fig. 7E,F). Knowing that epithelial cell adhesion to extracellular matrix is important for cell survival, we were interested in studying whether the defects in the basement membrane integrity lead to increased apoptosis. However, the TUNEL assay did not reveal differences in the rate of apoptosis in the disorganized neuroepithelium between the knockout and wild-type embryos (Fig. 7G,H). Since the Crb proteins and their polarity complex partners have been shown to be involved in the establishment of tight junctions and maintenance of the apical–basal polarity (Grawe et al., 1996; Tepass et al., 2001; Roh and Margolis, 2003; Lemmers et al., 2004; van de Pavert et al., 2004), we further investigated if the disruption of Crb2−/− neuroepithelium is associated with alterations in the expression of tight junction proteins. We analyzed ZO-1, occludin, and claudin-5 expression during E8–E8.25, the time points when the Crb2−/− neuroepithelium has started to develop obvious malformations and when Crb2 function could become critical. However, no Crb2 deficiency-related changes in the expression of these proteins or their location were observed (Fig. 8A–D, and data not shown). Moreover, the localization of the tight junctions in Crb2-deficient ectoderm was comparable with controls by transmission electron microscopy (TEM; Fig. 8E,F). Thus, in the early stages of the mouse development, Crb2 is dispensable for the formation and maintenance of neuroepithelial tight junctions. We also stained the E8 neuroepithelium for ezrin, a plasma membrane-cytoskeleton linker protein, which localizes to the apical domain in epithelial cells. However, similar staining was seen in the Crb2-deficient neuroepithelium as in their littermate controls. Nevertheless, defects in basement membrane integrity, illustrated by the discontinuous perlecan staining and leading to disorganized neuroepithelium, clearly indicate that loss of Crb2 affects neuroepithelial cell polarity.

Figure 7.

Deletion of Crb2 results in abnormal neuroepithelium. A,B: Photographs show immunofluorescence staining of transversal sections for E-cadherin (green) and perlecan (red) in the head-fold region of the embryonic day (E) 8.5 control and knockout embryos. Nuclei are visualized by DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). C,D: Double immunofluorescence staining for Pan-cadherin (green) and perlecan (red) in the E8.5 control and Crb2−/− embryos. The photographs show transversal sections of the trunk region. E,F: Double immunofluorescence staining for Pan-cadherin (green) and fibronectin (red). Photographs show transversal sections at the head level of E9.5 control and Crb2−/− embryos. G,H: Adjacent sections to E and F were stained with the TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling) assay to detect apoptotic cells (green). Nuclei are visualized by DAPI. The photographs were taken under the same magnification. am, amnion; bm, basement membrane; me, mesoderm; ne, neuroepithelium; se, surface ectoderm.

Figure 8.

Deletion of Crb2 does not alter the localization of ezrin, a protein of the apical cytoskeleton, or that of a tight junction protein ZO-1. A–D: Photographs of the immunofluorescence stainings of transversal sections in the head fold region of embryonic day (E) 8–E8.25 control and Crb2−/− embryos. A and B show double immunofluorescence staining for ezrin (green) and perlecan (red), whereas the sections in C and D are stained for ZO-1 (green) and perlecan (red). Nuclei are visualized by DAPI (4′,6-diamidine-2-phenylidole-dihydrochloride). E,F: TEM picture of tight junctions (arrows). Photographs show sagittal sections of control and Crb2−/− neuroepithelium. Each pair of the photographs was taken under the same magnification. bm, basement membrane; me, mesoderm; ne, neuroepithelium.


The findings of our study show that Crb2, one of the three mammalian Crb proteins, is crucial for the early mouse embryonic development. In embryos lacking a functional Crb2 gene, severe developmental abnormalities appear during late-gastrulation, resulting in a failure of organogenesis and embryonic death by E12.5. Several findings suggest that the primary defect is a disturbed apicobasal polarity of the epiblast, which leads to impaired EMT at gastrulation. These findings, therefore, indicate a novel role for the Crb protein family.

Consistent with a developmental role for Crb2, expression analysis by in situ hybridization at E7–E8.5 indicated that the Crb2 gene is widely expressed during the early mouse development. Therefore, its embryonic expression pattern appears to differ from that of Crb1, which has an important role in retina. Indeed, Crb1 is expressed later than Crb2, being found exclusively in the eye after E11.5 and in the brain after E10.5 (den Hollander et al., 2002). The early embryonic expression pattern of mouse Crb3 is currently not clear, but based on EST profiles, it is expressed in embryos. Both mouse and human Crb3 are also widely expressed in adult tissues and cultured epithelia (Makarova et al., 2003). Crb3, whose extracellular part is completely different from that of Crb2, is quite extensively studied for its function in controlling epithelial cell polarization and tight junction formation by means of its interaction with the polarity complexes Pals1/Patj and Par3/Par6/aPKC (Roh et al., 2003; Lemmers et al., 2004; Shin et al., 2006; Assemat et al., 2008).

We also tried to detect the Crb2 protein by antibodies generated in our lab, but they failed to detect the protein in immunohistochemical stainings on embryonic sections. Furthermore, although the knockout cassette introduced the eGFP gene into the Crb2 locus, no expression of the eGFP protein was detected in the Crb2-deficient embryos. This indicates that some important elements(s) needed for the Crb2 gene expression were lost during the gene targeting process in which the entire genomic region between exons 1 and 13 was deleted. However, regarding the embryonic expression of the Crb2 protein, we want to point out a previous study by Lee et al. (2007), who found that pan-Crb antibodies give an apical staining in the E7.5 epiblast as well as in the E8.5 neuroepithelium. With our Crb2 mRNA expression data and that concerning the other Crb family members (see above), this positive signal has to at least partially represent Crb2. It is also of interest to highlight the observation that in some cells the Crb2 mRNA seems to localize apically (Fig. 1E,F). Remarkably, similar observations have been made for the Crb transcript in Drosophila epithelial cells (Li et al., 2008). These findings very likely reflect the phenomenon of coupling mRNA targeting to localized translation.

In embryos deficient in Crb2, the earliest abnormalities appear at E7.75, which include delayed E-cadherin down-regulation in the nascent mesoderm and widening of the primitive streak, as indicated by the absence of perlecan, an ubiquitous component of basement membranes. Similarly, at a somewhat later stage, perlecan staining indicated defects in the integrity of the neuroepithelial basement membrane, which very likely contributes to the subsequent neuroepithelium disorganization. With the knowledge that basement membranes are crucial for maintaining epithelial cell polarity (Li et al., 2003), the observed phenotypic abnormalities suggest that, in the absence of Crb2, epiblast cell polarity is disturbed, and this severely affects gastrulation EMT. Some mesoderm and endoderm is formed in the mutant embryos, but it is in fact evident already at E7.75 that there are severe abnormalities in the formation of the head fold, heart tube, foregut invagination, and somite development. Moreover, staining for fibronectin at E8.25 indicated dramatically reduced levels of this extracellular matrix protein, demonstrating that lack of Crb2 indeed severely affects mesoderm formation. Not only is the amount of mesoderm reduced, but there may also be qualitative defects. In this regard, it is noteworthy that fibronectin-deficient embryos develop similar phenotypes as the Crb2-deficient ones with defects in the embryonic mesoderm, neural tube, and vasculature, as well as with posteriorly truncated body axis (George et al., 1993). Analysis of many other marker genes, namely Cer1, Mox1, Fgf8, Lim1, and Foxf1, demonstrated rather normal expression initially (Figs. 4, 6, and data not shown) but down-regulation at later stages, which is consistent with the fact that the mutant embryos develop normally until the late-gastrulation stage. Based on the observation by Lee et al. (2007) that Crb proteins are not expressed in the nascent mesoderm, it seems that we can exclude the possibility that Crb2 ablation directly affects the migration of the nascent mesoderm cells away from the primitive streak. Regarding the study by Lee et al., it is also of high interest to point out that, similarly to Crb2 ablation, ablation of Epb4.1l5, a FERM domain-containing protein shown to associate with the Crb complex, was found to disrupt gastrulation with mutant cells often found trapped in the primitive streak at an intermediate stage of the EMT, thereby resulting in embryonic death before mid-gestation. In the Drosophila null mutant of Yurt, an Epb4.1l5 ortholog, the photoreceptor apical domain is expanded, indicating that Yurt negatively regulates the activity of Crb (Laprise et al., 2006). However, considering the fact that inactivation of Epb4.1l5 does not affect Crumbs expression (Lee et al., 2007) and the similar phenotypes of the Crb2- and Epb4.1l5-deficient mice, it is more likely that in mammalian cells Epb4.1l5 is a positive effector of Crb2, possibly connecting the Crb proteins to the actin cytoskeleton (Lee et al., 2007). It is also of interest to note that in both mutant embryos gastrulation seems to develop initially normally. We do not have an explanation for why the early gastrulation is not affected in the embryos, but we speculate that lack of the proteins may also affect the recruitment of the epiblast cells to the primitive streak, and this process is affected as gastrulation progresses and cells more lateral to the primitive streak are recruited.

Thus, we propose that the abnormal EMT in the primitive streak results in the defective mesoderm and endoderm formation, leading to a complete failure in embryonic development. However, it is possible that Crb2 deficiency affects embryonic development also by means of other mechanisms than EMT. In this regard, we observed that Bmp4 expression was down-regulated in the mutant embryos before the presence of obvious malformations. Bmps are well known to be involved in neurulation and the neural tube closure process (Vieira et al., 2010; Stern, 2005). Moreover, Bmp4 controls early mesoderm patterning and presomitic mesoderm formation, as shown in studies with embryos with targeted inactivation of the Bmp4 gene (Winnier et al., 1995).

Finally, in addition to EMT, Crb2 may also contribute to the MET process, such as differentiation of the presomitic mesoderm to epithelialized somites. Whether the mesodermal Crb2 expression directly plays a role in de novo somite epithelium formation is not clear due to the multiple earlier defects in the Crb2-null embryos, but the condensed Crb2 expression in the splitting lateral mesoderm, which forms splanchnopleure and somatopleure, cannot be ignored. This question could in the future be addressed with the help of an inducible gene-targeting technique.


Generation of Crb2-Deficient Mice

Crb2-deficient mice were generated by gene targeting using the Velocigene technology (Regeneron Pharmaceuticals, Inc.; Valenzuela et al., 2003). An eGFP-containing cassette (eGFP lox-Ub1- EM7- Neo- lox cassette) was used to replace the entire coding region (exons 1–13), with fusions precisely at the translation initiation and termination sites. This resulted in a deletion of approximately 20 kilobases of the Crb2 genomic sequence. Correctly targeted ES cells (derived from the 129S6SvEv/C57BL6 mouse strain) were identified using the loss-of-negative-allele assay (Valenzuela et al., 2003), and the proper integration of the knockout cassette was verified by sequencing of the integration sites (both the 5′ and 3′ junctions). The mice used in this study were backcrossed seven generations onto the C57BL/6 background. Crb2+/− mice were mated with each other to obtain wild-type, heterozygous, and null mutant mice. The Crb2 mouse strain has now been backcrossed for over nine generations, and no phenotypic changes have been observed during this process.


Genomic DNA was isolated from toes of 8- to 10-day-old mice or from 7- to 14.5-day embryos by a proteinase K digestion, chloroform extraction, and ethanol precipitation. In case of the embryos, a part of the extra-embryonic yolk sac was excised, and used for genotyping. Genotyping was performed by PCR using the HotStarTaq DNA polymerase (QIAGEN). For the detection of the wild-type allele, the primer pair “WTL” (5′-TGGCTGGGTAGGACCTA GTG-3′) and “WTR” (5′-GACTCCGG CAGACTCTATGG-3′) generating a 974-bp band was used (see Fig. 2 for the location of the primers). The null allele was detected with the WTL primer and a primer specific for the insertion of the eGFP-containing cassette; “KOR”: 5′-GAACTTCAGGGTC AGCTTGC-3′). This primer pair generates a 639-bp band. The PCR procedure used was: 1 cycle of 95°C/15 min; 30 cycles of 95°C/1 min, 55°C/1 min, 72°C/1 min; 1 cycle of 72°C/10 min. The PCR products were analyzed on 1% agarose gels.


RT-PCR on total RNA from E8 littermate control and null mutant embryos was performed to confirm the absence of Crb2 expression in the Crb2−/− embryos. A total of 1 μg of the RNA was reverse transcribed using the Superscript III Reverse Transcriptase (Invitrogen). Aliquots of the generated cDNAs were used for PCR with the conditions described above for mouse genotyping but with a primer pair from the 3′ untranslated region (UTR) region of the Crb2 cDNA (forward primer: 5′-CCTGGATGCTAAGCGAG AAG-3′; reverse primer: 5′-ACTACA GCCCTCTCCCCAGT-3′). This primer pair generates a 488-bp product. As a template loading control, a similar assay was performed with glyceraldehyde 3-phosphate dehydrogenase-specific primers.

Histology Analysis

Embryos were dissected, and fixed in 4% paraformaldehyde/PBS for several hours at room temperature or overnight at 4°C, then washed with PBS, followed by dehydration in a methanol series, and embedding in JB-4 (Polysciences, Inc.). They were then sectioned at 5–10 μm, and stained by hematoxylin and eosin with a standard procedure. Sections were photographed with QImaging Exi digital camera (QImaging.).

Whole-Mount In Situ Hybridization

Whole-mount in situ hybridization experiments were performed following protocols simplified from standard procedures (Wilkinson, 1992). Because wild-type and heterozygous embryos developed normally until adulthood without difference, they both were used as controls for the null mutant embryos. The Crb2 sense and antisense probes were made based on the same 3′-UTR 488-bp fragment as described above. The probe for Mox1 was provided by Michael Ray (Vanderbilt University). Probes for Cer1, Wnt3a, Fgf8, BMP4, Lim1, and Shh were obtained as gifts from Eddy DeRobertis, Andy McMahon, Seppo Vainio, Gail Martin and Brigid Hogan. The probe for Brachyury was from Jaime Rivera (University of Massachusetts Medical School). The probe for Foxf1 was from Leif Lundh (Gothenburg University). Single-stranded RNA probes were labeled with digoxigenin-UTP according to the manufacturer's instructions (Roche). Following the whole-mount in situ labeling, the results were documented by photographing (Leica DFC 340), post-fixed, embedded, and sectioned as in the histology studies for the exact localization of the transcripts.

Immunohistochemistry and TEM

For immunofluorescence staining, embryos were fixed in 4% paraformaldehyde/PBS for several hours at room temperature or overnight at 4°C, followed by several washes with PBS and incubation in 10% sucrose/PBS for 30 min at room temperature. Thereafter, the embryos were snap-frozen in OCT (Sakaru) on dry ice. Cryosections of 8 μm were fixed with cold acetone (−20°C), followed by blocking in 5% normal goat serum. Primary antibodies were incubated overnight at 4°C or for 1 hr at room temperature. After several washes with PBS, sections were incubated with Alexa Fluor 488- or 546-conjugated secondary antibodies. For double-labeling, the antibody-incubation steps were performed sequentially. Some sections were triple-stained with DAPI (1:2,000). The sections were photographed (QImaging) under fluorescence microscopy (Leica), and analyzed by the Volocity software (PerkinElmer, Inc.). The following primary antibodies were used: mouse monoclonal anti-E-cadherin (BD Biosciences) and anti-Pan-cadherin (Sigma), rat monoclonal anti-perlecan (clone HK-102, Seikagaku Corporation), polyclonal rabbit antibodies against fibronectin (Sigma), ezrin (Abcam), ZO-1, occludin, and claudin-5 (all from Zymed Laboratories). Secondary antibodies were purchased from Invitrogen Molecular Probes. For ZO-1 and occludin staining, sections were treated with 10 μg/ml of proteinase K for 2.5 min at room temperature in between the acetone fixation and blocking. The E-cadherin antibody was kindly provided by Bjorn Obrink. TEM analysis was done as previously described (Patrakka et al., 2007; Ebarasi et al., 2009).


Sections were prepared in the same way as those used for the immunofluorescence stainings. Apoptotic cells were detected using the In Situ Cell Death Detection Kit, Fluorescein (Roche). Nuclei were visualized using DAPI. The results were analyzed and documented as above.


The t-test was used to determine the statistical difference in the width of the primitive streak between wild-type and Crb2-null embryos. The analysis was performed using the GraphPad Prism 5 software (GraphPad Software Inc., La Jolla, CA). P < 0.05 was considered significant.


We thank Lwaki Ebarasi for stimulating discussions and support; Mechael Shen for Cryptic, Schöler for Oct4 in situ probe; Jane McGlade and Nancy F. Silva-Gagliardi for the EPB41L5 antibody, and troubleshooting that results were not included in this study. K.T. and C.B. were funded by the Knut and Alice Wallenberg Foundation, the Swedish Research Council, the Swedish Foundation for Strategic Research, and the Novo Nordisk Foundation. K.T. was funded by the Söderberg's Foundation and S.V. was funded by the Academy of Finland.