Dystroglycan is involved in skin morphogenesis downstream of the Notch signaling pathway

Expression of X-dg during epidermal ectoderm differentiation

We previously described the expression pattern of X-dg transcripts in Xenopus laevis during development (Moreau et al., 2003). Using an X-dg antisense RNA probe, we detected the transcripts in a variety of tissues and cell types. In this paper, we focus the analysis on X-dg expression in the epidermal ectoderm during the course of its differentiation.

We examined the expression pattern using whole-mount, in situ hybridization and subsequent sections (Figure 1). In epidermal ectoderm, the X-dg signal appeared at stage 14/15 in the posterior region of the embryo (Figure 1A) and progressed to the anterior region up to stage 19 (Figure 1B). The expression was scattered from stage 15 to stage 19, then it spread out, as observed at stage 24 (Figure 1C). The signal was still observed at stage 29/30 (Figure 1D).

Dg is not expressed by CCPs

During neurulation (stages14–19), a subset of the epidermal sensorial layer gives rise to CCPs, which express α-tubulin and differentiate by intercalating radially into the outer layer (see Introduction). Since α-tubulin and X-dg are simultaneously expressed in the sensorial layer, we were interested by their relative pattern of expression in this layer. To specify the spatiotemporal expression of X-dg versus α-tubulin, we used double whole mount in situ hybridization and subsequent histological sections using X-dg and α-tubulin cRNA probes.

At stage 15, we observed expression of the two genes without any overlapping of the two expression domains (Figure 1E). From late stage 15 and onward, the CCPs began to intercalate radially into the superficial layer, while the X-dg expression became stronger in the sensorial layer (Figure 1, F–H). At stage 19, all the CCPs reached the superficial layer, and the X-dg–expressing cells were now adjacent in the inner layer of the ectoderm (Figure 1G). The inner cells continued to express X-dg and progressively flattened and spread out underneath the outer layer, while the multiciliated cells reached their fully differentiated state (stage 24, Figure 1H). It appeared that, in the sensorial layer, cells that expressed α-tubulin did not express X-dg. The two cell types were distributed in an evenly spaced pattern, suggesting a lateral inhibition mechanism of regulation.

Dg depletion disrupts epidermal ectoderm differentiation

We knocked down X-dg by using two dg antisense morpholinos (DgMo). Their specificity and efficiency were demonstrated previously (Bello et al., 2008; Hidalgo et al., 2009). Injections with 16 ng of morpholinos in each cell were performed in both left blastomeres at the four-cell stage. We previously described this amount leading to a significant reduction of Dg protein, which did not induce too drastic morphological abnormalities in the embryos (Bello et al., 2008). The uninjected half of each embryo represents the internal control of the experiment. The toxicity of morpholinos and specificity of phenotypes observed were controlled in the same conditions by injection of a mismatch dg morpholino oligonucleotide (CMo).

The surfaces of the embryos were studied by scanning electron microscopy (SEM) during differentiation of epidermal ectoderm. Observations were performed on stage 13 embryos, as this stage corresponds to the beginning of epidermal morphological differentiation (Billett and Gould, 1971), and on stage 28 embryos, as the epidermis is fully differentiated in this stage. We focused our observations to the lateral trunk epidermis, since ciliated cell distribution in differentiated epidermis varies with body region (Nokhbatolfoghahai et al., 2005). There was no difference between the epidermis of CMo-injected and uninjected embryos, so we selected photographs from CMo-injected embryos for all the figures. At stage 13, the epidermal surface of the CMo- or DgMo-injected sides of embryos were composed of polygonal cells with prominent and lineal intercellular limits. The size of these cells was nearly identical (Figure 2, A and D). The outer ectodermal surface is devoid as in regular development of multiciliated cells at this stage.

At stage 28, the epidermal surface of the CMo-injected sides of embryos were composed of a heterogeneous mosaic of regular polygonal cells. The ciliated cells (identified by large clusters of cilia emerging from their apical surfaces) were interspaced regularly between nonciliated cells. The nonciliated cells were mainly secreting cells. Distinct straight lineal and prominent boundaries were evident between the cells (Figure 2B). Ciliated cells made contact with four or five of the large secreting cells. This arrangement represents a specific pattern similar to the corolla of flowers; the multiciliated cell occupies the central area of the corolla (Figure 2C). The epidermal surfaces of the DgMo-injected sides of embryos were composed of cells arranged in irregular polygonal patterns. Their boundaries appeared flattened. The surfaces of large secreting cells were greater than in controls (Figure 2E), suggesting cell proliferation was reduced. This reduction of proliferation was confirmed by using an anti-phosphohistone H3 antibody as a mitosis marker. At stage 24, the proliferation was reduced by 35% in epidermis of DgMo-injected sides of the embryos (n = 20; the positive cells were counted on the same surface area of the lateral trunk epidermis [Supplemental Figure S1]). The number of ciliated cells was reduced, and they were irregularly scattered and often positioned side by side, a pattern that was never observed in controls. The typical corolla of flowers pattern disappeared (Figure 2F). To quantify the decreasing number of ciliated cells, indirect immunofluorescence using an antibody to acetylated α-tubulin was performed (Figure 2G). At stage 28, the quantification (in the same surface area of the lateral trunk epidermis) on both DgMo-injected and noninjected sides of the embryos (n = 10) showed a 30% reduction of differentiated ciliated cells in the injected side (Figure 2H). To analyze the organization of cells in the two ectodermal layers, ultrathin sections were performed. The control showed two layers of cuboidal cells at stage 13; the outer-layer cells were tightly cohesive, unlike the inner-layer cells (Figure 2I). At stage 24, in the inner layer, cells were flat and closely associated (Figure 2J). At stage 13, the epidermal layers of the DgMo-injected sides were nearly similar to the control sides (Figure 2K). At stage 24, the cells of the superficial layers were greater, and numerous cells were interspaced between the outer and inner layers. The inner cells as the outer cells were greater; in addition, they were loosely associated, extended numerous filopodia, and did not flatten (Figure 2L). Among the cells interspaced between the outer and inner layers, we identified (by double in situ hybridization with X-dg and α-tubulin) ciliated cell precursors that were not able to reach the outer layer. We never observed X-dg–expressing cells among the interspaced cells (Figure 2, M and N), suggesting that, without Dg, there are failures in the radial intercalation processes.

The previous results demonstrate that the structure of the epidermis is affected by the loss of Dg, essentially at the level of cell proliferation, distribution of multiciliated cells, and cell adhesion.

Dg is critical for the intercalation of the precursors of ciliated cells

Dg is expressed only in the sensorial layer, but the reduction of Dg affects the structure and organization of the two layers of the epidermis, and especially the number of ciliated cells of the outer layer. This suggests that Dg acts in a non-cell-autonomous manner. To analyze Dg function, we grafted the outer-layer ectoderm from a control embryo onto the inner layer of an injected host embryo at the beginning of gastrulation. The explant corresponds to the prospective region of epidermal ectoderm, and the graft was done in the same region, namely the ventral ectoderm (Figure 3A). The donor cells were labeled by the injection of RNA encoding membrane-localized forms of green fluorescent protein (GFP; GFP-ras mRNA) in the four animal blastomeres at the eight-cell stage. The host embryo was injected in the same conditions with DgMo or CMo. At stage 28, when the larval skin was fully differentiated, the epidermis in the graft region was excised. In this tissue, the superficial cells coming from the donor expressed GFP at cell–cell contacts. The apical junctions were marked with antibodies to ZO-1. The cells of the sensorial layer coming from the host by radial intercalation were GFP-negative. The ciliated cells were identified by indirect immunofluorescence using acetylated α-tubulin antibody. The surface layer cells were sliced from apical to basal surface using confocal microscopy (Figure 3). The number of acetylated α-tubulin-positive cells/total cell number in the same surface area was identical in control and DgMo embryos (n = 8), suggesting Dg does not regulate the Notch lateral inhibition process. In hosts injected with CMo, the ciliated cells were GFP-negative, regularly interspaced, and in contact with four to five large GFP-positive cells. The GFP cells were polygonal; their basal and apical surfaces were identical in size (Figure 3, B and B′). Small, GFP-negative cells corresponding to nonciliated cells were also interspaced in this layer. The apical membranes were continuously labeled with ZO-1 antibodies. The apical and basal contacts between cells were narrow. We never observed the juxtaposition of two ciliated cells. The ciliated cells extended from the apical to the basal region of the surface layer, with a basal surface larger than the apical one (Figure 3, B and B′). These results suggest that all host CCPs have achieved their radial intercalation and are fully differentiated at stage 28 (Figure 3D). In host embryos injected with DgMo, the apical domain of all cells was larger than in controls, and the tight contacts were not affected. In contrast, the basal surface of the cells was irregular, with fewer intercellular contacts, suggesting defects in cell–cell adhesion. Both the ciliated cells and the small GFP-negative cells were irregularly interspaced. Among the acetylated α-tubulin-positive cells, some extended from the apical to basal domain of the tissue as in the control, but others were trapped below the surface (Figure 3, C, C′, and E), suggesting that their radial intercalation was incomplete (Figure 3, E and F). These results show that Dg is required for CCP intercalation and maintenance of lateral cell–cell contacts in the epidermal outer layer.

Given the defects observed in cell adhesion, and the essential role of cell adhesion and ECM molecules in intercalation, we then investigated the expression of E-cadherin, fibronectin, and laminin in embryos with reduced Dg expression.

Dg depletion inhibits laminin-1 deposition and disorganizes fibronectin meshwork in ECM

To investigate the effect of Dg depletion to ECM assembly, embryos were injected unilaterally at the four-cell stage with CMo or DgMo and GFP-ras mRNA to delineate cell borders. At stage 19, the lateral trunk ectoderm was peeled off and then labeled with anti-laminin or anti-fibronectin antibodies. The inner side of the ectoderm was observed by confocal microscopy (Figure 4A). In control explants, the inner ectodermal cells were polygonal; their GFP tagged membranes appeared cohesive. Fibronectin was assembled into fibrillar structures and formed an extensive meshwork across the cell surface (Figure 4, B and C). Laminin appeared assembled into short fibrils across the cell surface (Figure 4D). In the control the membrane staining appears tight and relatively linear, while in DgMo it is wider with apparent gaps, suggesting defects in cell adhesion. Consequently the intercellular space was enlarged (Figure 4E). The fibronectin fibrils localized on cell surface, but not across the intercellular space (Figure 4F). In contrast to fibronectin, the ECM was completely devoid of laminin (Figure 4G).

Dg depletion reduces E-cadherin expression at the cell–cell contacts, but β₁ integrin is not affected

As the intercellular adhesions are affected when Dg is knockdown, we wondered whether the expression of E-cadherin and β₁ integrin might be affected in Dg-Mo embryos. Indirect immunofluorescence, performed at stage 19, revealed a weaker expression of E-cadherin at the level of cell–cell contacts in the two epidermal layers of Dg-Mo embryos when compared with controls (unpublished data). These differences were amplified at stage 24, where the immunoreactivity for E-cadherin was almost undetectable at the cell membranes (Figure 4, H, I, and L). In contrast, β₁-integrin distribution was never affected (Figure 4, J, K, and L). Four independent Western blot analyses of Mo-injected tissues from stage-24 embryos using anti-Dg, anti–E-cadherin, and anti–β₁-integrin antibodies were performed. No significant variation of E-cadherin and β₁ integrin was detected, but Dg was absent in the Mo-injected epidermis (Figure 4M). To test whether the reduction in E-cadherin expression at the cell membranes was due to perturbation of E-cadherin transcription, we quantified the gene expression level of E-cadherin in Mo-injected embryos. Surprisingly quantitative real-time PCR (qRT-PCR) analysis normalized to ornithine decarboxylase, showed a significant up-regulation of E-cadherin expression (Figure 4N).

These results show that Dg controls laminin deposition and E-cadherin localization to the cell membrane, but it does not perturb the expression of E-cadherin mRNA and protein. Further studies will address whether E-cadherin mislocalization is a direct consequence of the loss of Dg and/or an indirect consequence of the absence of laminin.

Dg is a key element of epidermal differentiation

To test whether the abnormal morphology of inner cells in DgMo-injected embryos is associated with a failure of their differentiation, we examined DgMo-injected embryos using a molecular marker, the transcription factor P63, which is expressed during differentiation of the sensorial layer (Lu et al., 2001; Barton et al., 2009). The embryos were unilaterally injected with either the CMo or DgMo. The injected cells were followed using fluorescein isothiocyanate (FITC)-dextran. In CMo embryos observed at stage 24 by indirect immunofluorescence, the sensorial layer of cells was flattened and expressed P63 (Figure 5A). In DgMo epidermis, cells in the deep layer did not express P63, suggesting that they failed to differentiate (Figure 5B). On the other hand, qRT-PCR experiments showed a decrease of X-p63 expression in DgMo embryos after the neurula stage (Figure 5C). All together, these results suggest that Dg is a key element of epidermal differentiation as it contributes to maintain the expression of the known marker of cell differentiation in the deep layer. They also suggest that the regulation of X-p63 transcription might be downstream of Dg–ligand interactions.

These results show depletion of Dg affects the interactions between cells and laminin involved in epidermal differentiation. To attempt to confirm and validate these results, overexpression of a mutated Dg missing the intracellular β-C-terminus tail (Dg-ΔCyto) was performed. SEM analyses of embryos expressing Dg-ΔCyto showed phenotypes partially similar to those of DgMo embryos. As in DgMo, ciliated cells were irregularly scattered and their number at the embryonic surface was reduced (Figure S2, A and C). Similar to DgMo, in situ hybridization for α-tubulin transcripts showed that some of the positive cells were not able to reach the outer layer in the injected side of the embryos, compared with the noninjected one, suggesting that their intercalation was affected (Figure S2B). However, analysis of P63 distribution revealed that the protein was expressed in the nuclei (Figure S2D). Concerning the distribution and organization of laminin (Figure S2, E–F), we observed that linear deposition of laminin under the deep layer of epidermal cells was disrupted (compare Figure 7, J and K, later in the article with Figure S2E). Laminin was irregularly deposited at basal poles of cells in the deep layer (Figure S2E, white arrows). Its organization in the matrix was affected, since the fibrils were shortened and, in particular, the network of laminin fibrils appeared discontinuous (Figure S2F).

Notch signaling pathway regulates X-dg transcription in the epidermal ectoderm

Notch is a well-known regulator of lateral inhibition, a contact-mediated signaling whereby a cell that adopts a particular fate inhibits its immediate neighbors from acquiring the same fate. In the epidermal ectoderm, lateral inhibition occurs during gastrulation (Deblandre et al., 1999), before the expression of Dg (Figure 1E). So, inhibiting Notch signaling by expressing a DNA-binding mutant of Su(H) (Su(H)^DBM) leads to a dramatic increase in the density of the CCPs, while activating Notch signaling by expressing the intracellular domain of Notch (Notch^ICD) results in a complete loss of α-tubulin-expressing cells (Deblandre et al., 1999). Since we never observed overlapping expression of α-tubulin and X-dg in cells within the inner layer at stage 15, we hypothesized that the Notch pathway might regulate X-dg expression. At the four-cell stage, we injected 1) mRNA encoding Su(H)^DBM into both left blastomeres or 2) Notch^ICD mRNA in one ventral blastomere. In su(H)^DBM-injected embryos observed at stage 19, the α-tubulin-expressing cells in the injected side were adjacent and formed several layers in thickness, while the number of X-dg–expressing cells in the inner layer of the ectoderm was considerably reduced, compared with the control side (Figure 6, A and C). At stage 24, in the injected side, a large fraction of CCPs remained trapped below the outer layer of the epidermis, the sensorial layer cells did not flatten, and only a few of them expressed X-dg; in the more severe cases, there was a total absence of X-dg expression (Figure 6D). These results were confirmed by qRT-PCR, which also showed a decrease in the expression of X-dg (stages 15–24) following Notch inhibition (Figure 6G). Similar to what was observed in DgMo-treated embryos, indirect immunofluorescence showed an absence of P63 protein expression following Notch inhibition (Figure 6J) and qRT-PCR revealed a decreased expression of X-p63 mRNAs (Figure 6L). As expected, inhibiting the Notch pathway increased the number of CCPs, but this inhibition also dramatically reduced the number of X-dg–expressing cells in the sensorial layer of epidermal ectoderm. Consequently, p63 was down-regulated. In contrast, the activation of the Notch pathway by notch^ICD corresponded to a total loss of the α-tubulin-positive cells in the inner layer of the epidermis at stage 19, while the number of X-dg–expressing cells increased. These cells were adjacent and organized in multiple layers (Figure 6E). At stage 24, in the injected side, the X-dg expression increased. X-dg–expressing cells were organized in a single layer (Figure 6F) or in several layers of thickness (unpublished data) and had not flattened, probably because the amount of space available became restricted. All the nuclei of the inner cells expressed P63 (Figure 6K). qRT-PCR showed an increase of X-dg and X-p63 mRNAs for each stage from stage 15 to stage 24 (Figure 6, H and M, respectively).

These results demonstrate that inhibition or activation of the Notch pathway prevents or promotes transcription of X-dg, respectively, and that Dg is downstream of Notch.

The rescues of phenotypes reveal an epistatic relationship among Notch, Dg, and P63

The results discussed in the preceding section strongly suggest an epistatic relationship among Notch, Dg, and P63. To confirm this possibility, we attempted to rescue the su(H)^DBM phenotypes by overexpressing X-dg. We coinjected su(H)^DBM and the full-length X-Dg (FL-dg; Moreau et al., 2003) FL-dg mRNA in the two left cells of four-cell embryos. The quantification of CCPs using acetylated α-tubulin antibodies showed a twofold increase in the outer layer of the injected side relative to the control (Figure 7, A–E). At stage 24, the shape of cells, the organization of the inner cell layer, and the nuclear expression of P63 were all rescued (see Figure 7, F and H). We also observed that laminin deposition (Figure 7, I–K) and organization (Figure 7L) were restored in the ECM. Thus the concomitant blocking of Notch and overexpression of X-dg rescued the intercalation processes with a number of differentiated ciliated cells substantially increased. It rescued the structure of the deep layer, the deposition and organization of laminin, and the expression of P63. To further investigate the cross-talk between Notch and Dg, we tested whether the DgMo phenotypes might be reversed by activation of the Notch pathway. Embryos were coinjected with DgMo and Notch^ICD mRNA at the four-cell stage in the left ventral blastomere. We found a loss of the α-tubulin–positive cells in the epidermis, as expected, since the Notch pathway, which is upstream of Dg expression, is activated (Figure 7, M and M′). On the other hand, we showed nuclear expression of P63 protein in the inner epidermal cells of the injected side (Figure 7N). We also showed laminin deposition and its fibrillar organization at the basal pole of inner cells (Figure 7, O and P). This demonstrated that laminin deposition and organization, as well as differentiation of inner cells, were rescued. This rescue may be related to the presence of Dg protein. When Notch^ICD was overexpressed, the number of X-dg–expressing cells was greater, followed by a titration of DgMo leading to the traduction of Dg protein (Figure S3).

The above results, obtained by inhibition/activation of the Notch pathway, and these rescued phenotypes clearly show that Dg is involved in skin morphogenesis downstream of the Notch signaling pathway. They also show an epistatic relationship among Notch, Dg, and P63 in control of skin morphogenesis.

Embryos, morpholinos, RNA, and injections

Xenopus laevis embryos were obtained using standard methods and staged according to Nieuwkoop and Faber (1994). The Dg morpholino antisense oligonucleotide (DgMo) and the mismatch dg control morpholino (CMo) were previously described (Bello et al., 2008; Hidalgo et al., 2009). Depending on the experiment, morpholino oligos were injected either in both left blastomeres at the four-cell stage (16 ng per blastomere), or in the four animal blastomeres at the eight-cell stage (8 ng per blastomere). For Western blot analysis, embryos were injected in both blastomeres at the two-cell stage (32 ng of morpholinos). The templates for notch^ICD and su(H)^DBM, full-length dg (FL-dg), and mutated dg missing the β-C-terminus tail (dg-ΔCyto) mRNA synthesis have been described (Chitnis et al., 1995; Wettstein et al., 1997; Bello et al., 2008; Hidalgo et al., 2009). mRNA injections were delivered at the four-cell stage into the animal region of the two left blastomeres for su(H)^DBM (1.5 ng per blastomere), FL-dg (200 pg per blastomere), and dg-ΔCyto (1.3 ng per blastomere) or left ventral blastomere for notch^ICD (1 ng). Membrane-tagged GFP mRNA (GFP-ras mRNA) was injected as cell membrane tracer and FITC-dextran as a cell lineage tracer.

SEM and thin sections

SEM was carried out as previously described (Steinman, 1968; Billett and Gould, 1971). A Cambridge S-260 scanning electron microscope (Cambridge, UK) was used for observations. Thin sections (0.5 μm) were stained with a toluidine blue solution. Light imaging was done with a Nikon Eclipse 80i (Nikon Instrument, France).

Hybridization

Whole-mount, in situ hybridization was carried out as described elsewhere (Harland, 1991). The embryos were fixed in MEMFA (0.5 M MOPS, pH 7.4, 100 mM EGTA, 1 mM MgSO4, 3.7% formaldehyde). The antisense or control sense-strand RNA probes from X-dg (Moreau et al., 2003) and α-tubulin (Deblandre et al., 1999) were generated from linearized plasmids using either digoxigenin (DIG) or fluorescein RNA-labeling mix (Roche, Basel, Switzerland). For double whole-mount, in situ hybridization, the embryos were hybridized with both probes at the same time under standard conditions. After detection of the first probe with the DIG alkaline phosphatase (ALP) antibody (1:2000; Roche) and BCIP/NBT (Sigma-Aldrich, St. Louis, MO), embryos were washed in MeOH 100%; rehydrated in MeOH/MAB (100 mM maleic acid, 150 mM NaCl); and blocked in MAB, 2% Boehringer Mannheim Blocking reagent (Roche), and 20% heat-inactivated sheep serum. Following incubation with fluorescein ALP antibody (1:5000; Roche), the ALP reaction was performed with Fast Red (Sigma).

Transplants

Surface cell transplants were performed on early-gastrula-stage embryos in Danilchik's medium as described by Drysdale and Elinson (1992). An outer layer of ectoderm from a membrane-tagged GFP mRNA-injected embryo was grafted on the inner layer of a noninjected, CMo- or DgMo-injected embryo. After healing, embryos were transferred to 0.1× MMR medium (Euromedex, Souffelweyersheim, France) with antibiotics. Embryos were allowed to develop until at least stage 28 and fixed in MEMFA. The fluorescent transplant was peeled out from the embryo and treated for immunochemistry.

Immunochemistry

Embryos fixed in MEMFA were incubated with the mouse monoclonal 6-11B-1 antibodies that recognize acetylated α-tubulin (1:100; Sigma) and with the anti–mouse ALP-conjugated antibody (1:40,000; Jackson Immunoresearch, West Grove, PA). BCIP/NBT was used for the color reaction. Positive cells were counted in both the injected and the uninjected sides using ImageJ software (http://rsb.info.nih.gov/ij/download.html).

Epidermis fragments were peeled on embryos (stage 19 or 24), fixed with MEMFA or 4% PFA on ice (for cadherin and integrin detection), and dehydrated in ethanol. For frozen sections, embryos were embedded in cold-water fish gelatin/sucrose as previously described (Bello et al., 2008). The sections were performed at 14-μm thickness in a cryostat (CM3050 S, Leica). Peeled fragments or sections were blocked in 10% goat serum. The following primary antibodies were used: rabbit anti-GFP (1:500; Abcam, Cambridge, MA), mouse anti–acetylated tubulin 6-11B-1 (1:500; Sigma), mouse anti–ZO-1 (1:50; Invitrogen, Carlsbad, CA), rabbit anti–laminin-1 (1:25; Sigma), mouse anti–fibronectin 4H2 (1:100; D. DeSimone, University of Virginia), mouse anti–β₁ integrin (1:500; D. Alfandari, University of Massachusetts), anti–E-cadherin 5D3 (1:100; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA), anti-P63 (1:50; Abcam); mouse anti–β-Dg (1:50; A. Menarini Diagnostics, Florence, Italy). After undergoing washing, sections or fragments were incubated with the appropriate secondary antibody: anti–rabbit FITC-conjugated (1:40; Jackson Immunoresearch), anti–rabbit tetramethylrhodamine B isothiocyanate. (1:100; Jackson Immunoresearch), anti–mouse CY3-conjugated (1:100; Sigma), anti–mouse immunoglobin G1 (IgG1) Alexa Fluor 633 (1:1000; Molecular Probes, Invitrogen), anti–mouse IgG2b Alexa Fluor 568 (1:1000; Molecular Probes, Invitrogen). Nuclear staining was carried out with Hoechst H33258 (1:1000; Sigma).

qRT-PCR

Total RNAs were isolated from the epidermis of 20 embryos (stage 15, 19, and 24) using the RNeasy mini kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. All primers were designed using Primer Express Software. PCR reactions were carried out using SYBR green (Applied Biosystems, Bedford, MA) on a 7700 Sequence Detector cycler (Applied Biosystems). All experiments were repeated at least three times on separate injections and the qRT-PCR was also performed in triplicate. The results were analyzed using the 2^−ΔΔCt method (Livak and Schmittgen, 2001). The relative expression of X-dg is shown normalized to the expression of the housekeeping gene ODC. The qRT-PCR X-dg, X-p63, and E-cadherin primers are as follows:

• X-dg sense: TCCCCACAGAATTTCTCGCT

• Antisense: GTAACCAAGAGGGCAATGTCTCC

• p63 sense: AGCAAAGTCAGCCACCTGGA

• Antisense: GTTTTGGCAATCTGACAGTAGAGC

• E-cadherin sense: CGTGTGGCTTTGCGGACT

• Antisense: CTGCATTTATGGAAGGCACAAC

Western blots

Morpholinos were injected in both cells at the two-cell stage. At stage 24, embryos were homogenized in lysis buffer (1% Triton X100, 150 mM NaCl, pH 7.5, 10 mM Tris, 1 mM EDTA, 1 mM ethyleneglycoltetraacetic acid [EGTA], 0.5% NP-40, 0.2 mM phenylmethyl sulfonyl fluoride [PMSF]) supplemented with protease inhibitors. Samples were separated by 10% SDS–PAGE and transferred to nitrocellulose membrane (Hybond, GE Healthcare, Waukesha, WI). The membrane was incubated with the following primary antibodies: mouse anti–β Dg 43DAG1/8D5 (1:25; Abcam), mouse anti–E-Cadherin 5D3 (1:500; Developmental Studies Hybridoma Bank), mouse anti–β₁ integrin (1:500; D. Alfandari). It was then incubated with anti–mouse antibody coupled to peroxidase (1:10,000; Jackson Immunoresearch). Identification was made by chemiluminescence (SuperSignal West Pico Chemiluminescent substrate kit, Pierce, Thermo Scientific, Lafayette, CO). The membrane was then dehybrided and incubated with mouse anti–α-tubulin antibody DM1A (1:10,000; Sigma) and with anti–mouse antibody conjugated with peroxidase (1:10,000; Jackson Immunoresearch). Identification was made by chemiluminescence as previously. For quantification, blots were exposed on x-ray films for various time points, and the films were scanned and analyzed using ImageJ software.

Proliferation assays

Proliferation assays were performed in whole embryos (n = 20), fixed in MEMFA at stage 24, and then stained with the rabbit anti–human phosphohistone H3 antibody (1:100; Ser 10, mitosis marker, Euromedex, Souffelweyersheim, France) and the anti–rabbit ALP-conjugated (1:5000; Jackson Immunoresearch). Positive cells were counted in both the injected and the uninjected sides using ImageJ software.

Membrane fluorescence intensity quantification

After acquisition, images were imported into ImageJ software for analysis. Mean pixel intensity of a fixed area of the cell membrane was quantified and corrected by the subtraction of the mean pixel intensity of an area of the same size from the cytoplasm. Twenty measurements were taken from each group, and all experiments were repeated at least three times on separate injections. Results are expressed as the ratio of the average of the mean intensity ± SE.