G-protein-coupled signals control cortical actin assembly by controlling cadherin expression in the early Xenopus embryo

The Xenopus blastula is a hollow ball of cells (blastomeres) with no skeletal tissue and little extracellular matrix. At this early stage, shape and rigidity of the whole embryo are provided by the cortical actin skeleton of each component blastomere. In the absence of cortical actin, the embryo collapses under its own weight and the movements of gastrulation do not occur(Kofron et al., 2002; Tao et al., 2005). The mechanism by which the cortical actin skeleton is generated and maintained is therefore an important problem in early development.

We have shown previously that the pattern and density of cortical actin changes through the cell cycle. During interphase, each blastomere possesses a dense cortical actin network, which is replaced during mitosis by a much sparser network of actin filament bundles(Lloyd et al., 2005). Dissociation of the blastomeres, so that cell contact and intercellular signals are prevented, causes all blastomeres to adopt the sparse configuration of actin, which is reversible by reassociation(Lloyd et al., 2005). Previous expression screening experiments identified two proteins that are required for assembly of cortical actin. Receptors for the signaling phospholipid lysophosphatidic acid (LPA) and an orphan G-protein-coupled receptor (GPCR)named Xflop, were each found, by gain- and loss-of-function experiments, to be both necessary and sufficient for assembly of the dense cortical actin network(Lloyd et al., 2005; Tao et al., 2005). These experiments indicated the novel fact that intercellular signaling plays a major role in generating the appropriate density and pattern of cortical actin assembly in the early embryo.

It therefore becomes important to determine the sites of actin assembly at the cell surface and the way the G-protein-coupled receptors control the amount and pattern of actin assembly, so as to generate the appropriate shape and mechanical rigidity of the whole embryo.

Recent elegant experiments using epithelial cells in culture have shown that during initial contacts between epithelial cells the cytoplasmic domains of transmembrane cadherins are sites of actin assembly(Ehrlich et al., 2002; Jamora and Fuchs, 2002; Kovacs et al., 2002a; Kovacs et al., 2002b; Vaezi et al., 2002). E-cadherin engagement in cultured cells causes activation of Rac1 and Cdc42,both of which are involved in actin polymerization and organization(Betson et al., 2002; Kim et al., 2000; Kovacs et al., 2002a; Nakagawa et al., 2001; Noren et al., 2001). Actin nucleation proteins such as members of the Arp2/3 complex and formin 1 are associated with nascent cadherin-mediated adhesive contacts(Kobielak et al., 2004; Kovacs et al., 2002b; Verma et al., 2004), as are nucleation-promoting factors such as Ena/Vasp and cortactin(Helwani et al., 2004; Scott et al., 2006; Vasioukhin et al., 2000). However, the detailed mechanism by which cortical actin assembles at the cadherin complex is still unknown (Adams and Nelson, 1998; Drees et al., 2005; Perez-Moreno et al., 2003; Yamada et al.,2005).

Could cadherins be the sites of assembly of cortical actin in the Xenopus blastula? It is known that the early blastomeres are held together by calcium-dependent adhesion(Nomura et al., 1988; Turner et al., 1992). The calcium-dependent adhesion protein C-cadherin (also known as EP-cadherin)(Choi et al., 1990; Ginsberg et al., 1991) is expressed during the egg-to-blastula stages and has been shown, by antisense-mediated mRNA depletion, to be absolutely required for cell adhesion at this stage (Heasman et al.,1994).

In this paper, we show that C-cadherin is essential for the assembly of the dense cortical network seen in interphase blastomeres, but not for the sparse network seen during cell divisions. First, we show that C-cadherin proteins are organized as punctae on the surfaces of the blastomeres and that the actin filament network is associated with these punctae. Second, we show by both gain- and loss-of function experiments that the level of C-cadherin expression is both necessary and sufficient for the dense cortical actin network. Increasing the level of C-cadherin expression leads to increased cortical actin density, whereas decreasing C-cadherin expression, by depletion of its mRNA, reduces it. We also show that the juxtamembrane domain of C-cadherin,which binds p120 catenin, is essential for its control of actin assembly. Consistent with this, we show that p120 catenin is required for dense actin network assembly and that a mutant form of p120 catenin that lacks the binding site for C-cadherin fails to increase actin assembly. These data show that the cadherin complex is both necessary and sufficient for assembly of the normal pattern and density of cortical actin in the Xenopus blastula, and that this activity requires p120.

If cortical actin assembly depends on the expression level of C-cadherin,then LPA signaling, and signaling through the G-protein-coupled receptor Xflop, could act either at the level of actin nucleation, or at the level of C-cadherin presentation on the cell membrane (or both). We show here that both signaling pathways control the level of C-cadherin expression on the cell surface. These data show for the first time how C-cadherin levels on the cell surface are controlled in the Xenopus embryo and the consequences of this control for actin assembly.

All frogs (Xenopus laevis) were obtained from Nasco (Fort Atkinson, WI). Full-grown oocytes were obtained by manual defolliculation and cultured in oocyte culture medium (OCM). To prepare 800 ml OCM: 480 ml Leibpvitz L-15 with glutamine (Sigma Aldrich), 320 ml water, 4 ml Pen-Strep Solution (MP Biochemicals, LLC), 320 mg BSA and adjust pH to 7.6-7.8 using 5 M NaOH. Antisense oligonucleotides or synthetic mRNAs (doses as indicated in text) were injected into defolliculated oocytes. After oligonucleotide injection, oocytes were cultured for 2-3 days before maturation. After mRNA injection, oocytes were matured overnight, starting on the day of the injections. Progesterone (1-2 μM) was used to mature oocytes. Matured oocytes were fertilized by the host transfer technique(Heasman et al., 1991).

To generate a p120 catenin mRNA to which the MO could not bind, we made a cDNA construct with three third-base alterations using the following primer pair for PCR cloning: forward,5′-CGGAATTCTATGGACGAACCTGAGTCTGAAAGTCCG-3′(underlined are the EcoRI site for subcloning, the ATG start codon,and the three substitutions at the third base of each codon, respectively);reverse, 5′-TACGCAGGCAACCTGTAGTG-3′. Xflop R112A mutant was generated by PCR-based mutagenesis, using the following PCR primers(underlined nucleotides are designed to substitute arginine with alanine):

Xflop5′-BglII,5′-AGCAGATCTATGGCGTGTAATCAGAGCTGTGAATAC-3′;

R112A r,5′-CACTGTAGCCACAAAAGCATCCATGGCTATACAGCT-3′;

R112A f, 5′-TGTATAGCCATGGATGCTTTTGTGGCTACAGTGTTC-3′;and

Xflop3′-EcoRI,5′-GTTGAATTCTATCCTGTCCTTTTTGATGACCTCCTTC-3′.

The primer pair of Xflop5′BglII and R112A r was used to generate fragment 1 using wild-type Xflop cDNA as template, whereas the primer pair of R112A f and Xflop3′EcoRI was used to generate fragment 2 using wild-type Xflop cDNA as template. The primer pair of Xflop5′BglII and Xflop3′-EcoRI was then used to generate the full-length Xflop R112A mutant using the purified fragments 1 and 2 as template. The full-length Xflop R112A was inserted into BamHI and EcoRI-cut pCS107 vector. All PCR products were verified by sequencing.

A Xenopus tropicalis C-cadherin construct was isolated from an arrayed cDNA library constructed in vector pCS107 (a gift from Aaron Zorn,University of Cincinnati, OH), linearized with Asp718, then transcribed with SP6 RNA polymerase. Xenopus C-cadherin mutants[C-cad(G-A)-HA (referred to below as C-cad G-A), C-cad ΔCBD](Paulson et al., 2000) were synthesized with SP6 RNA polymerase from a pCS2+ vector linearized with NotI. Xenopus p120 iso1, p120 iso1 MO-resistant and the Arm1 deletion mutant (p120 ΔArm1) were constructed in pCS2-MT vector with a Myc tag at the N-terminus of p120. For RNA synthesis, we linearized the vector with NotI and transcribed from the SP6 promoter. pCS107 Xflop and Xflop R112A and LPA2 were each linearized with Asp718 and transcribed with SP6 RNA polymerase. Message Machine (Ambion) Kits were used for all in vitro transcriptions.

The antisense oligodeoxynucleotides (designated AS below) or morpholino oligonucleotides (designated MO below) used were (asterisks indicate phosphorothioate-modified residues):

C-cad AS61 (Heasman et al.,1994),5′-C^*C^*T^*CTCCAGCTCCCT^*A^*C^*G-3′;

P120 catenin AS1, 5′-TGCATCCCTCCATCCTGT-3′ (no modification);

P120 catenin MO1 (Fang et al.,2004), 5′-ACTCTGGCTCATCCATATAGAAAGG-3′;

Xflop oligonucleotides (Tao et al.,2005) 1s,5′-A^*A^*G^*GGAACACTGTAG^*C^*C^*A-3′and 5s,5′-G^*T^*T^*GTACGTTTTGGC^*T^*G^*G-3′;

LPA1 MO (Lloyd et al.,2005), 5′-TTCACTTCAGATGTCAGTCATGCTG-3′; and

LPA2 MO (Lloyd et al.,2005), 5′-ACCTCCAATGTTACAGCGCAGCCTC-3′.

For F-actin single staining, caps were excised at the late blastula stage(St9) (Nieuwkoop and Faber 1967) and fixed with 3.7% formaldehyde, 0.25%glutaraldehyde in PBS, 0.1% Tween 20 (FG fix) for 10 minutes. Caps were then washed three times for 10 minutes each and stained with 5 U/ml Alexa 488-conjugated phalloidin (Molecular Probes, Oregon) for 3 hours at room temperature or overnight at 4°C. For anti-C-cadherin/F-actin double staining, the phalloidin-stained caps were blocked with 10% normal goat serum at room temperature for 1 hour and then incubated with 5 ng/ml anti-C-cadherin monoclonal antibody (6B6, Developmental Study Hybridoma Bank, Iowa) overnight at 4°C, followed by extensive washing with PBS, 0.3% Triton X-100. A Cy5-conjugated goat anti-mouse IgG (Jackson Laboratory, 1:300) was then added,followed by 2 hours incubation at room temperature. In every experiment, 3-5 caps were also incubated with secondary antibody only as negative controls. For anti-C-cadherin single staining, 2% TCA in water was used to fix animal caps for 30 minutes at room temperature. After anti-C-cadherin staining, caps were dehydrated in a methanol series and cleared in Murray's Clear [from Murray and Kirschner, as cited by Dent and Klymkowsky(Dent and Klymkowsky, 1989)],before examination by confocal microscopy.

A Zeiss LSM 510 confocal microscope was used. A Fluar UV/20×NA0.75 objective was used for lower magnification imaging, and a C-Apochromat 63×/NA1.3 water lens or a Plan-Neofluar40×/NA1.3 oil lens was used with 1.5-2× digital zoom for higher magnification imaging. All images were gained with the size of 512×512 pixels (8 bit for experiments represented in Figs 6, 7 and 8, 12 bit in experiments represented in Figs 2, 3, 4 and 5, 9). The intensity of phalloidin staining of each pixel measured by LSM510 software was used to quantify the levels of F-actin. The level of F-actin for each imaged area was the average(Mi) of intensity measurements from all 512×512 pixels. The level of F-actin for each experimental treatment group was the average of 5-10 animal caps (Mt=∑Mi/i, where i=number of imaged areas). The linear range of pixel intensity measurement was from 0 to 255 for 8-bit images or from 0 to 4095 for 12-bit images. In all experiments, the confocal settings were optimized to allow more than 90% of pixels to have intensity measurement within the linear range, which might introduce variations among control groups from experiment to experiment. Each histogram shown in Figs 2, 3, 4, 5, 6, 7, 8 and 9 represents the mean pixel intensity (Mt) ±s.d. from 5-10 animal caps. The two-tailed t-test was used to generate P values.

To detect the total protein levels of p120 or C-cadherin, five embryos were homogenized with 250 μl of ice-cold PBS, 1% Triton X-100 containing 1 mM PMSF and a 1:100 dilution of protease inhibitor cocktail (PIC, Sigma P8340),and cleared by centrifugation at 750 g for 10 minutes at 4°C. The supernatants were then transferred into precooled Eppendorf tubes. An equal volume of 4× sample buffer was added to the cleared supernatants.

To detect the membrane levels of C-cadherin or its binding proteins,including p120, plakoglobin and β-catenin, five embryo samples were homogenized in volumes of 50 μl per embryo in ice-cold membrane preparation buffer (250 mM sucrose, 10 mM HEPES-NaOH, 2 mM MgCl₂, 1 mM EGTA,0.5 mM EDTA, 1 mM PMSF, 1:100 dilution of PIC). Lysates were cleared by centrifugation at 700 g for 10 minutes at 4°C. Supernatants were transferred into precooled clean tubes and centrifuged at 21,000 g for 30 minutes at 4°C. Pellets were resuspended with 30 μl of 2× sample buffer.

Two to three embryo equivalents of membrane protein, or one embryo equivalent of total protein, were boiled for 5 minutes and separated by 6% SDS PAGE for 2 hours at 80 volts. Gels were blotted onto nitrocellulose membranes using a semi-dry apparatus (BioRad). All membranes were subsequently blocked(1 hour at room temperature) with 5% non-fat dry milk in PBS, 0.1% Tween 20 and incubated in primary antibodies overnight at 4°C. Antibody conditions were as follows, all diluted in blocking buffer: primary antibody anti-C-cadherin (Developmental Study Hybridoma Bank, 6B6) 1:500 with secondary antibody GaM-HRP (Jackson ImmunoResearch) 1:2500; primary antibody anti-humanγ-catenin (BD Transduction laboratory, C26220) 1:500 with secondary antibody GaM-HRP 1:2500; primary antibody anti-β-catenin (Sigma, C2206)1:2000 with GaR-HRP (Jackson ImmunoResearch) 1:2500; primary antibody anti-p120 (rabbit polyclonal) (Fang et al., 2004) 1: 5000 with GaRb-HRP (Jackson ImmunoResearch) 1:2500. Membranes incubated with HRP-conjugated secondary antibodies were detected with ECL developing solution (Amersham) and exposed to X-ray films (Hyperfilm,Amersham) at variable times to obtain unsaturated bands.

C-cadherin is expressed on the surface membranes of blastomeres during the cleavage and blastula stages (Choi et al.,1990; Levi et al.,1991). Cadherin clusters have been reported to be present as punctae in the membranes of epithelial cells during their initial contact,before becoming resolved into adherens junctions(Adams et al., 1998; Adams et al., 1996; Vasioukhin et al., 2000; Yonemura et al., 1995). To analyze the pattern of expression of C-cadherin in Xenopus blastulae,we followed the method described in Kofron et al.(Kofron et al., 2002) (see also Fig. 1A). Dissected animal caps from late blastulae (Stage 9) were fixed for 30 minutes in 2%trichloroacetic acid and stained as whole-mounts with a monoclonal antibody raised against C-cadherin (6B6) (Choi et al., 1990). Animal caps were cleared in Murray's Clear and examined in a series of optical slices, starting at the blastocoelic surface. When individual 1-1.5 μm optical slices were assembled into stacks of 30-70, C-cadherin appeared to extend in a continuous line around each adjacent cell surface, as previously seen in conventional histological sections(Heasman et al., 1994). However, when individual 1.2 μm optical slices were examined, C-cadherin could be seen as a series of punctae around adjacent plasma membranes(Fig. 1B). Individual punctae were measured by pixel counting and had areas of 0.7±0.3μm² spaced by distances of 1.9±0.99 μm. This can be seen particularly clearly in grazing sections through cell surfaces (arrows in Fig. 1C). When animal caps were dissociated in calcium- and magnesium-free buffered saline, C-cadherin was lost from the cell surfaces (Fig. 1D) (see also Hausen and Riebesell, 2002). When cells were returned to calcium- and magnesium-containing saline and allowed to contact each other, punctae of C-cadherin rapidly assembled at the sites of membrane apposition(Fig. 1E).

Since we have shown previously that dissociated blastomeres lose the dense cortical actin network after cell dissociation and reassemble it when they are allowed to reassociate (Lloyd et al.,2005), this suggested the hypothesis that the dense actin network might assemble on punctae of C-cadherin. To test this hypothesis, we fixed animal caps from late blastulae in formaldehyde-glutaraldehyde fixative (see Materials and methods), stained them with Alexa 488-coupled phalloidin to reveal the cortical actin network, and then stained them as whole-mounts with anti-C-cadherin antibodies. Fig. 1F,G show cortical actin filament bundles associated with C-cadherin-containing punctae (arrows), interspersed with regions of cadherin-containing membrane not associated with actin.

These data suggested that C-cadherin-containing complexes might be required for cortical actin assembly at the membranes in Xenopus blastulae,and that the pattern and density of cortical actin in the embryo could be controlled by the expression of C-cadherin itself, in addition to other possible levels of control. To test this, we first depleted the maternal mRNA encoding C-cadherin, by injecting an antisense deoxyoligonucleotide (oligo)complementary to part of this mRNA, into cultured full-grown oocytes as described by Heasman et al. (Heasman et al., 1994). The oocytes were matured in vitro using 1 μM progesterone and fertilized by the host transfer method(Heasman et al., 1991). Fig. 2A,B (left panels) show C-cadherin protein levels on the cell surfaces in control and C-cadherin mRNA-depleted embryos at the late blastula stage. There is a dramatic reduction in the level of staining, consistent with previous findings (Heasman et al.,1994). The density of the cortical actin network was correspondingly decreased in the C-cadherin-depleted embryos (right panels of Fig. 2A,B). The normal, dense actin network characteristic of interphase blastomeres was replaced by a sparser network similar to that seen in dividing or dissociated cells. Second, C-cadherin mRNA (250-1000 pg per oocyte) was injected into oocytes before maturation. These were matured in vitro and fertilized as above and the affects on cortical actin assembly examined at late blastula stage. Fig. 2C shows that the level of C-cadherin staining (left panel) and the density of cortical actin (right panel) were both increased in C-cadherin-overexpressing animal caps compared with the controls (Fig. 2A). Pixel intensities from 5-10 caps were measured using the Zeiss LSM 510 software. Statistically significant quantitative changes in actin staining confirmed the results seen in the individual images(Fig. 2D). These data show that the dense actin network is cadherin-dependent, whereas the sparse actin network seen in dividing or dissociated cells is cadherin-independent.

The function of transmembrane cadherins is regulated by their cytoplasmic domains, which bind a number of cytoplasmic proteins(Pokutta and Weis, 2002). Two major functional domains of the cytoplasmic tail are the juxtamembrane region(JMR) and the distal β-catenin/plakoglobin (γ-catenin)-binding domain (CBD). We used mutant forms of C-cadherin to test the roles of each of these domains in actin assembly. First, we injected mRNA encoding a mutant Xenopus C-cadherin in which the residues GGG (aa 731-733) in the JMR were replaced with AAA. This mutant form is unable to bind p120 catenin(Thoreson et al., 2000). Unlike wild-type C-cadherin, overexpression of this mutant form did not increase the density of the cortical actin(Fig. 3A, middle lower panel);in fact, it consistently decreased it (Fig. 3B). A mutant form lacking the distal catenin-binding domain(C-cad ΔCBD, deletion of aa 839-896) did not significantly alter the density of cortical actin (Fig. 3A, right lower panel, Fig. 3B). Immunostaining shows that both mRNAs were expressed(Fig. 3A, upper panels). These data suggest that both the JMR and CBD domains are required for C-cadherin to assemble the dense actin network.

In this experiment, the endogenous cadherin was still present and might have affected trafficking or presentation of the mutant cadherins. To control for this, we injected the two mutant mRNAs into embryos whose endogenous C-cadherin had been depleted by antisense oligo injection into the oocytes. Fig. 4A shows that C-cadherin mRNA depletion dramatically reduced the level of cadherin protein on the cell surface. Both mutant forms of cadherin were efficiently translated, as shown by immunocytochemical staining with anti-C-cadherin antibody. C-cad G-A was expressed on the cell surface and rescued cell adhesion (Fig. 4A). However, it did not rescue the cortical actin network. The only sites of actin polymerization were in dense membrane projections at cell boundaries (arrowed in Fig. 4B). However, over the rest of the cell cortex, there was no dense cortical actin network in these cells. C-cad ΔCBD was not expressed at the cell surface, but instead accumulated in large cytoplasmic vesicles(Fig. 4A,B). This explained the fact that it did not rescue cell adhesion and precluded an assessment of the function of this mutant in actin assembly. Actin-rich cell-surface projections were also seen in these cells, but there was no rescue of the cortical actin network over the rest of the cell surfaces. The overall levels of polymerized actin were quantitated by pixel intensity, as shown in Fig. 4C. Neither of the mutant forms of cadherin rescued the loss of polymerized cortical actin caused by C-cadherin depletion. These data show that the p120-binding site of C-cadherin is required for C-cadherin to assemble a dense cortical actin network.

The requirement for the JMR of C-cadherin in cortical actin assembly suggested that p120 catenin might be required for this function. P120 catenin is expressed both maternally and zygotically in Xenopus. To test its role in cortical actin assembly at the blastula stage, a morpholino oligo(MO), previously shown to block the translation of p120 catenin mRNA(Fang et al., 2004), was injected into cultured Xenopus oocytes (10-40 ng per oocyte), which were subsequently matured in vitro and fertilized. The amount of p120 catenin protein was assessed by western blotting and found to be reduced in a dose-dependent manner by the mid-blastula stage(Fig. 5B). The density of the cortical actin network was also significantly reduced by p120 catenin depletion (Fig. 5A,C). This experiment was repeated with an antisense deoxynucleotide oligo, complementary to a different region of the target mRNA (AS1, 7-9 ng per oocyte), which had the same effect (data not shown). The effects of both the MO and the antisense oligo were rescued by reintroducing p120 catenin mRNA (shown for the antisense oligo experiment in Fig. 5C), showing that in both cases, the effects were specific for p120 catenin depletion. Thus, p120 catenin is required for cortical actin assembly at the blastula stage.

Next, synthetic p120 catenin mRNA in doses of 250 to 1000 pg was injected into oocytes and the cortical actin examined at the late blastula stage. Fig. 5A,D show that the density of the cortical actin network was increased by the overexpression of p120 catenin. We conclude that p120 catenin is both necessary and sufficient to generate a dense cortical actin network.

p120 catenin is thought to regulate the steady-state levels of cadherins on the cell surface (Chen et al.,2003; Davis et al.,2003; Elia et al.,2006; Ireton et al.,2002; Perez-Moreno et al.,2006; Xiao et al.,2003). We therefore asked whether p120 catenin controls cortical actin assembly by controlling the level of C-cadherin on the surface of each blastomere. Animal caps from p120 catenin-depleted blastulae were stained with 6B6 monoclonal antibody against Xenopus C-cadherin. Fig. 5E shows that depletion of p120 catenin caused a significant reduction in C-cadherin on the cell surface at the late blastula stage and, conversely, that injection of p120 catenin mRNA into the early embryo caused increased levels of C-cadherin at the cell surface (Fig. 5F). These data show that p120 catenin plays an essential role in controlling the steady levels of C-cadherin in the Xenopus embryo and, through this,controls the level of cortical actin assembly.

To confirm that interaction between C-cadherin and p120 catenin is required for their function in controlling the dense actin assembly, we generated a p120 catenin mutant that lacks the cadherin-binding domain (spanning the first Arm repeat, p120 ΔArm1). The first Arm repeat has been shown to be necessary to target p120 catenin to cadherin on the cell membrane, but not to affect p120 catenin regulation of Rho GTPases(Anastasiadis et al., 2000; Yanagisawa et al., 2004). Overexpression of this construct did not increase either the density of cortical actin or the level of C-cadherin on the cell surface in Xenopus embryos (Fig. 5G,H), showing that the dense cortical actin assembly requires interaction between p120 catenin and C-cadherin.

We have shown previously that the dense actin network assembly in the Xenopus blastula is controlled by two maternally expressed G-protein-coupled receptors, LPA1 and Xflop(Lloyd et al., 2005; Tao et al., 2005). We have shown above that the C-cadherin-p120 catenin complex is a site of control of the dense cortical actin network. We therefore explored the possibility that one mechanism of Xflop and LPA function might be through regulation of C-cadherin levels. First, mRNAs encoding LPA1 (400 pg per embryo) or Xflop(250 pg per embryo) were injected into the animal cytoplasm at the 2-cell stage, and the embryos allowed to develop to the late blastula stage before fixation and staining for C-cadherin. Fig. 6A-C show that both mRNAs caused upregulation of C-cadherin levels. To distinguish between increased total protein and altered cellular distribution, we assayed the total C-cadherin levels by western blotting. Fig. 6D shows that LPA or/and Xflop overexpression increased the total level of C-cadherin at the late blastula stage.

We next considered whether the two GPCRs could be acting as structural components of the cadherin complex, rather than signaling through G proteins. Since the ligand for the LPA receptors, LPA, causes an increase in cortical actin (Lloyd et al., 2005),signal transduction must be occurring. However, this has not been shown for Xflop, as its ligand(s) is unknown. We therefore created the point mutation R112A, which has been shown to block the activation of G proteins by the motif of aa 111-113 (DRF) in the second intracellular loop of GPCRs(Rovati et al., 2007), mRNA encoding Xflop R112A was injected into each cell of the 2-cell embryo in a wide range of concentrations, up to 1 ng. In no case did it mimic the wild-type Xflop in increasing the level of cortical actin(Fig. 7). These data confirm that signaling through the LPA and Xflop receptors controls the level of C-cadherin at the late blastula stage.

If the LPA and Xflop signaling pathways control cortical actin assembly by regulating the amount of C-cadherin on the cell surface, then overexpression of C-cadherin should rescue the depletion of these receptors. We therefore injected 500 pg of C-cadherin mRNA into oocytes depleted of either LPA1 or Xflop by injection of antisense oligos into cultured oocytes. In both cases, the dense cortical actin network was restored(Fig. 8B-E). Depletion of either Xflop or LPA reduced both the membrane(Fig. 8A,B,C) and total(Fig. 8D,E insets) levels of C-cadherin, further supporting the notion that Xflop and LPA receptors control the level of C-cadherin at the late blastula stage in Xenopus.

As shown above, signaling through both Xflop and LPA receptors controls the amount of cadherin on the cell surface, on which actin is polymerized. Actin filaments can be assembled on a number of different cell-surface proteins and so it is possible that these receptors also control actin assembly on other surface proteins. If so, overexpression of the receptors would rescue cortical actin in the absence of C-cadherin. To test this, we depleted C-cadherin using antisense oligos injected into cultured oocytes, as above, and injected mRNAs encoding either Xflop or LPA1 at the 2-cell stage. Xflop overexpression (200 pg mRNA) did not rescue the dense cortical actin network in the absence of cadherin, although it did increase the cortical actin density in control embryos (Fig. 9), suggesting that it acts only through the cadherin complex. LPA1 overexpression did increase cortical actin density after depletion of C-cadherin (not shown). This confirms a previous finding that the addition of the LPA ligand to dissociated cells in culture, which do not express cadherins, led to increased actin assembly (Lloyd et al.,2005). This suggest that LPA signaling, at least at higher than physiological levels, can control the assembly of actin by other mechanisms in addition to its role in cadherin expression.

Previous work has shown that the cortical actin skeleton, in addition to its known roles in the movements of gastrulation and in cytokinesis, also provides a skeletal framework for the embryos at the blastula stage. Abrogation of the cortical actin skeleton at the late blastula stage causes collapse of the entire embryo and failure of gastrulation movements(Kofron et al., 2002). It is therefore of great interest to understand the mechanism of cortical actin assembly in the early embryo. Previous work also showed that dissociated cells, in which cell contact and signaling are prevented, lose their dense cortical actin, and that the G-protein-coupled receptors LPA1 and LPA2, and Xflop, are required for cortical actin assembly in vivo(Lloyd et al., 2005; Tao et al., 2005). However,the mechanism of action of these receptors, and whether other aspects of cell contact are required, have not been explored. Here, we show that there is a direct relationship between the expression of C-cadherin on the cell surface and the assembly of cortical actin in the cytoplasm, and that the level of cell-surface cadherin is controlled by signaling through GPCRs. This interacting system results in the level of cell cohesion and rigidity seen at the late blastula stage.

It is already known that cadherin-containing punctae form at sites of initial contact of epithelial cells in culture(Helwani et al., 2004; Kobielak et al., 2004; Kovacs et al., 2002b; Scott et al., 2006; Vasioukhin et al., 2000; Verma et al., 2004), and that these associate with actin filaments and actin assembly components such as Arp2/3 (Kovacs et al., 2002b; Verma et al., 2004), formin 1(Kobielak et al., 2004) and Ena/Vasp (Scott et al., 2006). Here we show, in a developing system in vivo, that such punctae and their associated actin filament assemblies form the skeleton of the early embryo. The presentation of C-cadherin on the cell surface is clearly a rate-limiting step of this process, as its overexpression increases the dense cortical actin network, whereas its depletion causes loss of the dense cortical actin.

The importance of C-cadherin for dense cortical actin assembly is supported by the finding that p120 catenin was also found to be both necessary and sufficient for this process. p120 catenin is known to be important for stabilizing cadherin on the cell surface(Davis and Reynolds, 2006; Elia et al., 2006; Perez-Moreno et al., 2006) and this proved to be the case here. p120 overexpression caused increased amounts of both C-cadherin and cortical actin, whereas depletion caused the reverse. In addition, a p120 mutant lacking the cadherin-binding site failed to increase either C-cadherin expression on the surface, or cortical actin assembly, showing that the level of C-cadherin expression on the cell surface requires its interaction with p120. It is possible that p120 controls cortical actin assembly at other levels, in addition to cadherin presentation. Our data suggest that p120-mediated cortical actin assembly requires its interaction with C-cadherin. The fact that a cadherin construct lacking the p120-binding site was expressed at high levels on the cell surface, but caused loss of cortical actin filaments, supports this hypothesis. p120 catenin has been shown to bind signaling intermediates in actin assembly such as Rac1 and Cdc42(Noren et al., 2000), so it will be interesting to see whether these interactions are also required for actin assembly in Xenopus. A cadherin construct lacking theβ-catenin-binding domain and that does not bind β-catenin(Pokutta and Weis, 2002)failed to significantly increase cortical actin assembly. However, in the absence of endogenous C-cadherin, this construct was not transported to the cell membrane and did not rescue cell adhesion. The role of the CBD in cortical actin assembly could therefore not be tested with this mutant. However, our observations do suggest a potential role for the CBD in protein trafficking and membrane presentation of cadherin. The high level of membrane expression of the cadherin mutant lacking the p120-binding site might also be due to the intact CBD in this mutant.

Once it became clear that the late blastula cells somehow titrate the assembly of cortical actin against the level of C-cadherin on the surface,this immediately raised the novel possibility that intercellular signaling through GPCRs might control the density and pattern of cortical actin in the blastula at the level of cadherin expression. This indeed proved to be the case. Depletion or augmentation of the receptors LPA1 and LPA 2, and Xflop,previously shown to be required for cortical actin assembly(Lloyd et al., 2005; Tao et al., 2005), caused corresponding loss and increase of C-cadherin expression on the cell surface. Furthermore, overexpression of C-cadherin by mRNA injection was able to rescue cortical actin assembly after depletion of either LPA1 or Xflop in the blastula. These data implicate the two signaling pathways in controlling the level of expression of cadherin on the cell surface. It will be of major interest to discover whether these signaling pathways are the sites of action of cell cycle components, as the dense cortical actin network is lost during cell division.

One interesting and unexplained fact in these studies is that there appears to be two actin-containing networks in the blastula cells. Cells that are dividing in vivo, or dissociated in calcium- and magnesium-free saline, or depleted of C-cadherin, all lose their dense cortical actin network, but retain a sparse network, even in the absence of cadherin on the cell surface(Fig. 2) (see also Lloyd et al., 2005). This sparse network is clearly cadherin-independent and is independent of the dense network. We do not know if the reverse is the case, as we have not found any independent way of removing the sparse network. It is possible the LPA and Xflop signaling also control the sparse network, although experimental details are still lacking. We also do not know how the sparse network is associated with the cell cortex. There are several known sites of actin insertion into the cell membrane, in addition to cadherins. For example, α-spectrin and moesin, members of FERM (band 4.1/ezrin/radixin/moesin) proteins are maternally expressed in Xenopus oocytes(Carotenuto et al., 2000; Thorn et al., 1999), and could play a role in attaching the sparse actin network to the cell membrane. Some integrins (β1, for example) are present at the blastula stage(Gawantka et al., 1992) and could be sites of actin assembly.

Many more questions remain. We do not know the mechanism by which LPA and Xflop signaling control the level of C-cadherin on the cell surface. It is also possible that LPA and Xflop control actin assembly at the cadherin complex at levels other than via C-cadherin expression. They might also control the sparse network through mechanisms that do not involve cadherin expression. The Xenopus blastula represents a simple in vivo system with which to address these questions.

This work was supported by Cincinnati Children's Hospital Research Foundation and the National Institutes of Health (RO1-HD044764). The authors declare that they have no competing financial interests.