Nectin-2 and N-cadherin interact through extracellular domains and induce apical accumulation of F-actin in apical constriction of Xenopus neural tube morphogenesis

Neural tube formation is an essential and characteristic morphogenetic event of vertebrate development, involving the dynamic rearrangement of cells, including mediolateral intercalation and apical constriction (Davidson and Keller, 1999; Colas and Schoenwolf, 2001). Apical constriction is seen in neural tube formation in all vertebrates except for teleosts. It is a morphological change in which the epithelial cells undergo acute contraction in their apical surface, accompanied by cell elongation in the apicobasal axis, thus taking on a wedge-like shape. This cell-shape change is followed by the invagination of the neuroepithelium, which forms a neural groove along the dorsal midline flanked by neural folds: bulges of neural ectoderm. Eventually, the folds fuse at the midline, completing the formation of the neural tube (Colas and Schoenwolf, 2001).

Apical constriction is observed not only in vertebrates, but also in invertebrates in circumstances such as invagination of the mesoderm and dorsal closure in Drosophila melanogaster (Lecuit and Lenne, 2007). To the extent that the molecular mechanisms have been elucidated, similar molecular pathways apparently govern this process in both phyla. Small GTPases, including Rap1 for vertebrates (Haigo et al., 2003), and RhoGEF2, Rap1 and Rho1 for invertebrates (Barrett et al., 1997; Häcker, 1998; Sawyer et al., 2009), play important roles. As a common downstream target of these pathways the apically localized actomyosin tensile system is activated via Rho kinases and generates a force to narrow the cell surface, ensuring apical constriction (Wei et al., 2001; Dawes-Hoang et al., 2005; Nishimura and Takeichi, 2008). Recent studies have shown that the Shroom family of actin-binding molecules is involved in the apical constriction of both cultured cells and Xenopus embryos and may facilitate the apical accumulation of actin filaments (F-actin) (Haigo et al., 2003; Hildebrand, 2005; Fairbank et al., 2006; Lee et al., 2007; Nishimura and Takeichi, 2008). Although the apical actin bundle is required for apical constriction (Burnside, 1971; Karfunkel, 1972; Colas and Schoenwolf, 2001; Corrigall et al., 2007; Lecuit and Lenne, 2007; Lee and Harland, 2007; Kinoshita et al., 2008), the detailed mechanism of apical accumulation of F-actin is just beginning to be elucidated. Here we show that nectin-2, a transmembrane cell adhesion molecule, is required for the apical constriction of neuroepithelial cells during Xenopus neural tube formation, by facilitating the apical accumulation of F-actin.

Nectin is a member of the immunoglobulin (Ig) superfamily, with four isoforms in mammals, and contains three extracellular Ig-like domains, a single-pass transmembrane region and four conserved amino acids of a binding motif for afadin, an F-actin binding protein that connects nectin to the F-actin in the intracellular region in its C-terminus (Takai and Nakanishi, 2003; Takai et al., 2008). Nectin is known to mediate cell adhesion and is implicated in signal transduction with platelet-derived growth factor (PDGF) receptor and integrins (Takai and Nakanishi, 2003; Takai et al., 2008). Immunohistochemical studies of MDCK cells and the mouse small intestine showed that nectin was preferentially localized to adherens junctions (AJs), which the F-actin bundle underlies intracellularly (Takeichi, 1988; Dejana et al., 1995; Takahashi et al., 1999; Satoh-Horikawa et al., 2000). Mouse nectin is also expressed in embryonic epithelial cells, including the neuroepithelium and developing cranial nerve ganglia (Okabe et al., 2004). These expression patterns suggest that nectin might play roles not only in the maintenance of adult tissues but also in developmental processes, such as formation of the neural tissue. However, even though the nectins are expressed from the early embryonic stages, no obvious defects in embryogenesis have been seen in knockout mice for the individual nectin genes (nectin-1, -2 or -3) (Bouchard et al., 2000; Inagaki et al., 2005). This lack of phenotype could be due to redundant functioning among the nectin genes.

In Drosophila, Echinoid (Ed), which shares some similar properties with vertebrate nectin, has been implicated in the embryonic morphogenesis (Wei et al., 2005; Laplante and Nilson, 2006; Lin et al., 2007). Studies suggested that Ed controls apical constriction by assembling actin cables at the interface of juxtaposed cells expressing and lacking Ed. Therefore, structural similarities between nectin and Ed cause us to speculate that vertebrate nectin may function in embryonic morphogenesis.

In this study, we found that Xenopus nectin-2 induced apical constriction in cooperation with N-cadherin during neurulation, and that the interacting point of these two molecules lay in the extracellular domains. We first found that nectin-2 was strongly expressed in the neuroepithelium throughout neurulation. Depletion of nectin-2 caused a closure defect of the neural tube accompanied by impaired apical constriction. Conversely, nectin-2 overexpression in non-neural ectoderm induced ectopic apical constriction with apical F-actin accumulation. However, this effect of nectin-2 did not require the known molecular linkage between nectin-2 and F-actin, via afadin. As N-cadherin has been shown by us previously to be required for neural tube folding (Nandadasa et al., 2009), we tested whether the action of nectin is also associated with N-cadherin. We found that N-cadherin and nectin-2 bind through their extracellular domains and that they cooperatively promote apical constriction. Furthermore, we found that the role of nectin in F-actin accumulation may be mediated by the intracellular domain of N-cadherin through β-catenin. Our findings provide the first mechanistic evidence of the function of nectin in vertebrates and the regulatory mechanism by which the neuroepithelial cells undergo apical constriction.

We searched Xenopus nectin in the NCBI BLAST server (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and JGI X. tropicalis genome database v4.1 (http://genome.jgi-psf.org/Xentr4/Xentr4.home.html) against amino acid sequences of mouse nectins and identified UniGene cluster numbers for putative Xenopus nectin genes: Xenopus laevis nectin-1 (Xl.49696), nectin-2 (Xl.27064), nectin-3 (Xl.59402) and nectin-4 (Xl.73379); Xenopus tropicalis nectin-1 (Str.73359), nectin-2 (Str.47530), nectin-3 (Str.44628) and nectin-4 (Str.5750). Electronic northern was performed by calculating the ratio of expressed sequence tag (EST) expression levels of each gene from the oocyte to the tailbud stages, using the EST Profile in the UniGene database.

Xenopus embryos were obtained by standard methods, fertilized in vitro, dejellied in 3% cysteine solution (pH 7.8) and cultured at 13°C. Microinjection was performed in 0.1× Steinberg’s solution containing 3% Ficoll PM400 (Amersham), and the injected embryos were cultured until they reached the desired stage. Embryos were injected at the four-cell stage into the dorsal-left-side blastomere in knockdown experiments or into both of the ventral blastomeres in overexpression experiments. As needed, membrane-targeted green fluorescent protein (memGFP) or red fluorescent protein (memRFP) was co-injected as a tracer. Embryos were staged according to Nieuwkoop and Faber (Nieuwkoop and Faber, 1994). The aberrant surface phenotype was visually judged and defined by a rough surface and ectopic pigmentation.

Whole-mount in situ hybridization (WISH) was performed as described previously (Harland, 1991; Goda et al., 2009). β-Galactosidase staining was performed in advance of the WISH process as described previously (Chung et al., 2005).

RT-PCR analysis for nectin-2 expression was performed as described previously (Chung et al., 2005) with the following primers: forward, 5′-TAGGATCCGTGAAAGCTGATCCAAAAAAC-3′; reverse, 5′-GGAATTCTTACGTCTCTTTAGAGTCGTAG-3′.

We obtained Xenopus laevis nectin-2, afadin and N-cadherin cDNA clones from an XDB3 cDNA database (http://xenopus.nibb.ac.jp/): clone numbers XL170j24 (for nectin-2α; GenBank accession number GU290315), XL219e23 (nectin-2β; GU290316), XL293e12ex (afadin) and XL289n05ex (N-cadherin). nectin-2 rescue construct was generated by changing nucleotides at the morpholino target site of nectin-2α as follows, indicated in small letters: (–16) 5′-AAtAttATatcGGTAAATG-3′ (+3). The nectin-2 deletion constructs were generated by removing the sequences of nectin-2α that correspond to the following amino acids: nec2-ΔC4 (A467-V470), nec2-ΔIC (R353-V470), nec2-ΔEC (L37-L313). For afadin cDNA, as the XL239e12ex sequence was partially lost, we referred to the sequence of Xenopus tropicalis afadin in the genome database (JGI Protein ID: 153941) to reconstruct Xenopus laevis afadin (GenBank accession number GU290317). The coding sequence of N-cadherin was subcloned from XL289n05ex using PCR with the following primers: forward 5′-CACCATGTGCCGGAAAGAGCCCTT-3′ and reverse 5′-AGTTCAGTCGTCGCTCCCTCCGTACAT-3′. Xenopus laevis E-cadherin was obtained from a stage 10 cDNA library, prepared as described previously (Chung et al., 2005), using PCR with the following primers: forward 5′-TAGATATCACCATGGGGTTGAAGAGGCCCT-3′ and reverse 5′-GCGATATCTTAATCCTCATCACCTCCATAC-3′. Xenopus laevis C-cadherin construct was as previously described (Lee and Gumbiner, 1995). For the extracellular domain constructs, the following amino acid sequences were used: nec2-ex-GST/-FLAG (M1-N325), Ncad-ex-FLAG (M1-T722), Ecad-ex-FLAG (M1-G696), Ccad-ex-FLAG (M1-D703). A β-catenin-binding domain deleted N-cadherin (Ncad-Δβ) was generated by deleting the sequence that corresponds to the amino acids G839-D906. Rescue constructs of N-cadherin and Ncad-Δβ were generated by changing nucleotides at morpholino target site of FLAG-tagged N-cadherin (Ncad-FLAG) or Ncad-Δβ (Ncad-Δβ-FLAG), respectively, as follows indicated in small letters: (–4) 5′-CACCATGTGtaGaAAgGAaCCaTTt-3′ (+21).

Antisense morpholino oligonucleotides (MOs) were purchased from Gene Tools, LLC (OR), and consisted of the following sequences: nec2-MO, 5′-CACCATTTACCCCGATCCTCTTATC-3′; Ncad-MO, 5′-GAAGGGCTCTTTCCGGCACATGGTG-3′ (Nandadasa et al., 2009); control MO, 5′-CCTCTTACCTCAGTTACAATTTATA-3′. mRNAs were synthesized using mMESSAGE mMACHINE SP6 (Ambion) and purified on a NICK column (GE Healthcare).

For western blot, embryos or explants at desired stages were lysed in lysis buffer [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 50 mM NaF, 5 mM EDTA, 0.5% NP-40] with protease inhibitors. The supernatant of the lysate was sampled and denatured by the same volume of 2× SDS sample buffer (0.5 M Tris-HCl pH 6.8, 10% SDS, 50% glycerine, 2-mercaptoethanol). After boiling for 5 minutes, the samples were processed in SDS-PAGE, blotted onto the PVDF membrane (Bio-Rad), labeled with the primary and secondary antibodies and detected using an ECL Plus kit (GE Healthcare). For GST pull-down assays, 293T cells were transiently transfected with the indicated constructs by Lipofectamine (Invitrogen). Forty hours after transfection, the cells transfected with GST were lysed in the lysis buffer containing 0.17% NP-40 with protease inhibitors and incubated with Glutathione Sepharose 4B (Pharmacia). For other constructs, conditioned media were used in the pull-down process. The precipitates were then washed with the lysis buffer and subjected to western blot. For co-immunoprecipitation of endogenous β-catenin, Ncad-FLAG, Ncad-Δβ-FLAG or FLAG-tagged nectin-2 (nec2-FLAG) was injected into the animal pole of four-cell-stage embryos at 2 ng for Ncad-FLAG and Ncad-Δβ-FLAG and 1 ng for nec2-FLAG. The embryos were cultured until they reached stage 9, and their animal caps were dissected and lysed in the lysis buffer, followed by centrifugation at 15,000 rpm (17,400 g) for 30 minutes. The supernatant was subjected to co-immunoprecipitation reaction with Protein A Sepharose beads (GE Healthcare) using mouse anti-FLAG antibody (Sigma). The beads were washed with the lysis buffer, and subsequently analyzed by western blot.

For immunohistochemistry, embryos were fixed in MEMFA (0.1 M MOPS, pH 7.4, 2 mM EGTA, 1 mM MgSO₄, 3.7% formalin) for 1 hour at room temperature (RT) and then placed in PBST (PBS with 0.1% Tween 20). As needed, they were sliced using a Microslicer (DTK-3000W; DSK, Kyoto, Japan). After incubation in blocking solution [10% fetal bovine serum (FBS) in PBST] for 1 hour at RT, the samples were incubated with primary antibodies overnight at 4°C. After washing three times with PBST, they were incubated with secondary antibodies overnight at 4°C and then washed extensively with PBST before imaging. For F-actin staining, samples were prepared as above and an aliquot of 4 U/ml Alexa Fluor 546 phalloidin (Molecular Probes) was added to the secondary antibodies. For the N-cadherin staining, embryos were fixed in Dent’s fixative (Dent et al., 1989) and, after being blocked in 10% FBS, were incubated with primary and secondary antibodies, which were diluted in Can Get Signal immunostain solution A (Toyobo, Osaka, Japan).

For western blot, the following antibodies were used at the indicated dilutions. The primary antibodies were rabbit anti-GFP (1:2000; Molecular Probes), mouse anti-GST (1:1000; Santa Cruz), rabbit anti-FLAG (1:1000; Sigma), rabbit anti-N-cadherin (1:100; Abcam) and mouse anti-β-catenin (1:1000; Sigma). The secondary antibodies were sheep anti-mouse IgG HRP-conjugated (1:10,000; GE Healthcare) and donkey anti-rabbit IgG HRP-conjugated (1:10,000; GE Healthcare). Anti-nectin-2 antibody was produced from the Xenopus nectin-2 cDNA fragment corresponding to amino acids V128-T216, which was fused to pGEX-6P-1 (GE Healthcare). This protein was purified, cleaved from the GST by a protease and used to immunize mice. The antiserum was used for western blot after its specificity was checked (see Fig. S5A in the supplementary material).

For immunohistochemistry, the following antibodies were used at the indicated dilutions. The primary antibodies were rabbit anti-GFP (1:200; MBL, Nagoya, Japan), mouse anti-α-tubulin (1:500; Sigma), rabbit anti-FLAG (1:200; Sigma), mouse anti-FLAG (1:200; Sigma) and mouse anti-N-cadherin (1:200; Abnova). The secondary antibodies were Alexa Fluor 488 goat anti-rabbit IgG (1:500; Molecular Probes) and Cy5-conjugated goat anti-mouse IgG (1:500; Jackson ImmunoResearch).

Images of fluorescently labeled samples were obtained using LSM 510 META (Zeiss) with a Plan-Neofluar 20×/0.5 numerical aperture objective, a C-Apochromat 40×/1.2 numerical aperture water-immersion objective, an argon laser (488 nm) and helium-neon lasers (543 nm and 633 nm). Images were obtained as 512×512 or 1024×1024 pixels with z-series, processed with LSM 510 software (Zeiss) for the maximum projection of z-series and with Photoshop 7.0 (Adobe Systems) for tilting and trimming, and analyzed by using ImageJ v1.42. All statistical analyses were performed with the paired Student’s t-test.

We found putative orthologs for all four mouse nectins in both Xenopus laevis and Xenopus tropicalis (see Fig. S1A in the supplementary material; see Materials and methods). Their estimated expression levels, based on the EST database (i.e. electronic northern), and RT-PCR analysis showed that nectin-2 is the predominant form expressed in early Xenopus embryos (see Fig. S1B,C in the supplementary material). Therefore, we focused on nectin-2 in subsequent experiments.

We found that Xenopus has two paralogs of nectin-2: nectin-2α and nectin-2β (62.6% identical). Compared with mouse nectin-2, they showed good conservation of the extracellular region (41.5% identical to Xenopus nectin-2α; 41.9% to Xenopus nectin-2β), and complete conservation of the C-terminal afadin-binding motif (see Fig. S1D in the supplementary material). In addition, nectin-2 tagged with venus (vns), a variant of yellow fluorescent protein (Nagai et al., 2002), (nec2-vns) was localized to the plasma membrane (see Fig. S1E in the supplementary material), as reported for mouse nectin-2 (Takahashi et al., 1999).

We next performed WISH to see the spatial expression of nectin-2. At the onset of gastrulation, nectin-2α was uniformly expressed in the ectodermal layer (data not shown). From the beginning of neurulation, nectin-2α expression became more prominent in the superficial layer of the neural plate, along the entire anteroposterior axis, as well as continuing to be weakly expressed in non-neural ectoderm (Fig. 1A,B). This pattern was maintained throughout neurulation (Fig. 1C-F). In the neural plate, nectin-2α was expressed as two lines along the anteroposterior axis, which we thought probably corresponded to the hinge points at which apical constriction is most prominent (Davidson and Keller, 1999; Wallingford and Harland, 2002; Lee et al., 2007). The staining specificity was confirmed by the lack of staining with a sense probe for nectin-2α (Fig. 1G,H). The expression pattern of nectin-2β was almost identical to that of nectin-2α (data not shown). These spatiotemporally regulated expression patterns of nectin-2 during neurulation imply its involvement in neural development.

To investigate the roles of nectin-2 in early development, we next performed a knockdown experiment by using an MO against nectin-2 (nec2-MO), which effectively depleted the two variants of nectin-2 (Fig. 2A,B). Although the nec2-MO-injected embryos did not show any abnormalities until the early neurula stage, at the beginning of neural tube formation they failed to form complete neural fold and neural ridge (Fig. 2F), which is one of the hallmarks of neurulation and appears as a pair of pigmented lines along the dorsal side (Fig. 2D), on the MO-injected side. The pigmented neural fold lines were completely absent in severely affected embryos. Transverse sections through the trunk region clearly showed the normally raised morphology of neural ridges in an uninjected embryo (Fig. 2E) and the much flatter morphology in a nec2-MO-injected morphant embryo (Fig. 2G). This effect of nec2-MO was partially rescued in a dose-dependent manner (Fig. 2H-J) by co-injection with a rescue construct (nec2-res; Fig. 2A,C), confirming that the neural tube defect may be due to the lack of nectin-2. We also confirmed, by WISH, that the nec2-MO injection did not affect the expression patterns of neural differentiation marker genes (see Fig. S2 in the supplementary material). These results showed that nectin-2 is required for neural fold formation and thus for proper neural tube closure, without affecting the differentiation of the neural tissue.

To identify how the neural folding process was affected by the nec2-MO, we next investigated the morphant phenotype at the cellular level by sectioning the MO-injected embryos. On the uninjected side, the apical surface of the superficial cells was constricted, showing the characteristic wedge-like cell morphology (Fig. 3A,A′, arrow). By contrast, the nec2-MO-injected cells did not show apical constriction (Fig. 3A,A′, asterisk). Quantifications of the apical-surface length and cell height showed that apical constriction and cell elongation were suppressed on the nec2-MO-injected side, and these effects of nec2-MO were partially rescued by nec2-res (Fig. 3B,C). Together, these results indicate that nectin-2 is required for the apical constriction of neuroepithelium during neurulation.

We further examined the F-actin localization in nectin-2 knockdown embryos. Control embryos showed the accumulation of apical F-actin in neuroepithelium (Fig. 3D,E,H,I). In nec2-MO-injected embryos, however, this accumulation was severely impaired in the MO-containing cells (Fig. 3F,G,J,K). Thus, nectin-2 is required cell-autonomously for F-actin to accumulate properly at the apical side of the Xenopus neuroepithelium.

To further assess the function of nectin-2, we carried out an ectopic overexpression experiment by injecting 100 pg of nectin-2 mRNA into the region that gives rise to non-neural ectoderm, which does not normally undergo apical constriction (control embryos; Fig. 4A,B,E,F). In contrast to the control, nectin-2-injected embryos exhibited abnormal pigmentation and a rough epithelial surface (Fig. 4C,D). In addition, F-actin staining revealed acute shrinkage of the apical cell surface accompanied by the ectopic accumulation of F-actin in these cells (Fig. 4G,H; Table 1). This effect became more pronounced as the amount of injected nectin-2 mRNA was increased (Fig. 4M). When sectioned, some cells in the nectin-2-injected embryo showed ectopic wedge shape with constricted apical surfaces, increased cell height and, importantly, apically accumulated F-actin (Fig. 4K,L) in contrast to the cuboidal shape of normal ventral epithelial cells (Fig. 4I,J). The surface length and apicobasal height of these ectopically constricted cells were similar to those of neuroepithelial cells (Fig. 4N,O; compare with Fig. 3B,C). Thus, these data all support the idea that nectin-2 could induce ectopic apical constriction, which is an apparently similar constriction movement to that seen in the neural plate (Fig. 4P), accompanied by F-actin accumulation at the apical side of the cells. Finally, we confirmed that nectin-2 was preferentially localized to the apical cell membrane when ectopically expressed in Xenopus embryos (Fig. 4Q,R), as previously reported for cultured cells and mouse tissues (Takai and Nakanishi, 2003; Okabe et al., 2004), indicating that nectin-2 is expressed at, and functions in, the very place that apical constriction occurs.

Interestingly, the ectopic apical constriction was not observed in all the daughters of the injected cells; rather, the constricted cells were scattered throughout the ectoderm. One possibility for this is that the distribution of overexpressed nectin-2 might be uneven in the tissue, and the cells with higher concentration of nectin-2 could undergo constriction.

Nectin is connected to F-actin via afadin (Takai and Nakanishi, 2003; Takai et al., 2008). Therefore, we hypothesized that nectin-2 might induce apical constriction through its intracellular interaction with F-actin, by promoting the accumulation of F-actin at the apical region. To test this hypothesis, we prepared three deletion constructs of nectin-2 (Fig. 5A): nec2-ΔC4 (which lacks the four afadin-binding amino acids of the C-terminal region), nec2-ΔIC (which lacks the entire intracellular region) and nec2-ΔEC (which lacks the three extracellular Ig-like domains). As expected from the previous study (Takahashi et al., 1999), full-length nectin-2 and nec2-ΔEC bound to afadin, whereas nec2-ΔC4 did not (see Fig. S3 in the supplementary material), supporting the evidence of physical linkages from nectin-2 to F-actin through afadin.

We then used these deletion constructs in overexpression studies. Unexpectedly, 100 pg of nec2-ΔC4 and -ΔIC mRNA induced the aberrant epithelial surface morphology in the non-neural ectoderm, with ectopic apical constriction and characteristic accumulation of F-actin (Fig. 5B-F,I-L; Table 1). These effects were comparable to that of full-length nectin-2 (100 pg). By contrast, the overexpression of 100 pg of nec2-ΔEC mRNA showed less impact on the surface morphology (Fig. 5B,G,H), even though there was a weak effect compared with uninjected control, and its cell shape and F-actin distribution were almost indistinguishable from those of the control embryos (Fig. 5M,N; Table 1). Transverse sections showed ectopic wedge-like cell morphology in embryos overexpressed with nec2-ΔC4 and -ΔIC (Fig. 5O-R) but not nec2-ΔEC (Fig. 5S,T). These results indicate that the extracellular domain of nectin-2, rather than the intracellular afadin-binding motif, is required for it to induce apical constriction and F-actin accumulation. We further examined whether the extracellular domain is sufficient for the apical constriction in neuroepithelium by rescuing the defective neural fold phenotype. Co-injections of nec2-ΔC4 or -ΔIC with nec2-MO partially rescued the morphant phenotype, whereas the effect of nec2-ΔEC was not significant (Fig. 5U), suggesting that the extracellular domain of nectin-2 may be sufficient for neural fold formation through the apical constriction.

To rule out effects from inappropriate expression of nectin-2 constructs, we examined their expression levels and locations in the non-neural ectoderm. We confirmed by immunohistochemistry that all of the constructs were appropriately localized to the plasma membrane and showed similar expression levels (see Fig. S4A in the supplementary material). Further examination by western blot revealed that the protein levels of nec2-vns and nec2-ΔEC-vns were slightly lower than those of the other two constructs (see Fig. S4B in the supplementary material). As nec2-ΔEC-vns was expressed at about the same level as nec2-vns, which efficiently induced ectopic apical constriction, the results indicate that nec2-ΔEC lacks the ability to induce ectopic constriction compared with the full-length protein.

The findings that apical constriction is independent of the known nectin-F-actin linkage but requires the extracellular domain of nectin-2 indicated that some sort of extracellular machinery was required to effect this morphological change. Published studies indicate that the extracellular domain of homodimerized nectin-2 binds a nectin dimer on neighboring cells to promote cell adhesion (Takai and Nakanishi, 2003; Takai et al., 2008). Therefore, we next examined the binding between extracellular domains of nectin-2 by a pull-down assay, using a C-terminally truncated form of nectin-2 that consisted of only the extracellular (ex) domain (nec2-ex). Unexpectedly, we could not detect a homophilic interaction between the nectin-2 extracellular domains (Fig. 6A), suggesting that nectin-2 is unlikely to promote apical constriction by itself.

We then sought other molecules that might interact with nectin-2 extracellularly and associate with F-actin intracellularly. The most probable candidate was cadherin, because it is one of the most commonly expressed transmembrane proteins localized to AJs, and importantly, interacts with F-actin (Wheelock and Johnson, 2003; Gumbiner, 2005). The cadherin family is composed of many subfamilies, which are expressed in specific tissues (Halbleib and Nelson, 2006). During Xenopus neurulation, N-cadherin is preferentially expressed in the neural tube from the early neurula stage (Detrick et al., 1990; Fujimori et al., 1990), and has been shown previously to be required for neurulation movements (Nandadasa et al., 2009). Therefore, we tested whether nec2-ex could interact with the extracellular domain of N-cadherin (Ncad-ex). In contrast to the lack of a detectable homophilic interaction in nectin-2, Ncad-ex clearly co-precipitated with nec2-ex (Fig. 6A), indicating that these molecules may interact through direct and heterophilic binding in the extracellular space.

We next examined whether this nectin-N-cadherin interaction could induce apical constriction, by overexpressing either nectin-2 or N-cadherin or both. At a low dosage of mRNA (50 pg), neither nectin-2 nor N-cadherin alone caused obvious changes in the non-neural ectoderm (Fig. 6B-E; Table 1). However, their co-expression, at the same dose, effectively induced marked ectopic apical constriction and F-actin accumulation (Fig. 6F-H; Table 1). Moreover, to assess whether nectin-2 and N-cadherin associate either in cis or trans interaction, we separately injected them into adjacent blastomeres of four-cell-stage embryos and found that cells at the border between nectin-2-expressing and N-cadherin-expressing areas constricted in a significantly lower frequency compared with the co-injection of them (Fig. 6I; Table 1). These results indicate that nectin-2 may function cooperatively with N-cadherin to effect apical constriction in the cis interaction.

As nectin-2 (100 pg) alone induced ectopic apical constriction in the non-neural ectoderm, which expresses E-cadherin and C-cadherin but not N-cadherin (Choi and Gumbiner, 1989; Ginsberg et al., 1991; Levi et al., 1991), we examined, by pull-down assays, whether E- or C-cadherin had a similar interaction with nectin-2. However, we could not detect an interaction between the extracellular domain of nectin-2 with that of E- or C-cadherin (Ecad-ex or Ccad-ex; Fig. 6A), which highlighted the specificity of the interaction between nectin-2 and N-cadherin. Although we could not prove a biochemical interaction between nectin-2 and E- or C-cadherin, in overexpression experiments, they enhanced the aberrant surface morphology (Fig. 6H) and the ectopic apical constriction, albeit at lower frequencies than induced by the nectin-2 (50 pg)-N-cadherin pair (Fig. 6J-M; Table 1). Therefore, although the interaction between nectin-2 and N-cadherin was stronger and more effective for inducing apical constriction than the interaction between nectin-2 and E- or C-cadherin, the fact that some apical constriction occurred suggests that E- and/or C-cadherin interact with nectin-2 at a level sufficient to induce ectopic apical constriction, which may account for the phenotype induced by the overexpression of nectin-2 mRNA (100 pg) alone in non-neural ectoderm. We also found that the expression level of endogenous nectin-2 protein was lower in the ventral side than the dorsal side of the neurula embryos (see Fig. S5B in the supplementary material), suggesting that the ventral expression level of nectin-2 may not be enough to induce apical constriction with E- or C-cadherin.

The cooperative effect of nectin-2 and N-cadherin on ectopic apical constriction implies that N-cadherin has an in vivo role in neuroepithelial apical constriction. N-cadherin has been implicated in neural tube morphogenesis (Nandadasa et al., 2009), and we found that, by using N-cadherin MO (Ncad-MO), the N-cadherin-depleted cells failed to undergo apical constriction, and their apical accumulation of F-actin was inhibited (Fig. 7A,B), resembling the nectin-2 knockdown phenotype (Fig. 3F,J). This similarity supports our hypothesis that they play a cooperative role in neural tube closure. In addition, N-cadherin protein is apically located in neural plate cells normally and in the epidermis when ectopically expressed (Nandadasa et al., 2009). These results led us to hypothesize that nectin-2 is required for the function of N-cadherin in the neural ectoderm, and for its localization to the apical cytoplasm.

We first tested this hypothesis by assaying the synergy between the two proteins. When a low dose of either nec2-MO or Ncad-MO was injected into the dorsal side of the embryos, the neural fold formation was weakly affected (Fig. 7C). When combined, however, these MOs more severely disrupted the neural fold formation (Fig. 7C), indicating that the molecular interaction between nectin-2 and N-cadherin may play a crucial role in neuroepithelial folding. We next assessed the interdependence of these proteins for their subcellular localization. As reported (Nandadasa et al., 2009), N-cadherin was strictly localized to the apical portion of the neuroepithelial cells in control embryos (Fig. 7D,E). However, this localization was abrogated by the depletion of nectin-2 (Fig. 7F,G), without affecting the level of the N-cadherin protein (see Fig. S6 in the supplementary material). Nectin-2 was also apically located in the Xenopus neuroepithelium (Fig. 7H,I). However, its localization was not affected by the knockdown of N-cadherin (Fig. 7J,K). These results suggest that N-cadherin requires nectin-2 for its apical localization in the neuroepithelial cells, probably through its extracellular domain.

If N-cadherin is responsible for the apical accumulation of F-actin, associated proteins must mediate the linkage between them in the cytoplasm (Gumbiner, 2005). Although which molecules mediate this linkage is a matter of controversy (Drees et al., 2005; Yamada et al., 2005), one of the key molecules that may mediate it is β-catenin (Gates and Peifer, 2005). As cadherins possess a well-conserved β-catenin-binding region at the C-terminus (Ozawa et al., 1990; Pokutta and Weis, 2007), we designed a mutant N-cadherin construct lacking the putative β-catenin-binding region (Ncad-Δβ), which did not bind to the endogenous Xenopus β-catenin (see Fig. S7A in the supplementary material).

To assess the requirement for β-catenin to mediate the linkage between F-actin and N-cadherin during neural tube closure, we injected Ncad-Δβ mRNA into the dorsal side of four-cell-stage embryos. At the mid-neurula stage, Ncad-Δβ-injected embryos showed significantly abnormal neural fold formation (Fig. 7L), and phalloidin staining revealed a partial loss of apically located F-actin (Fig. 7P,Q), compared with control embryos (Fig. 7N,O). This may result from a competition between endogenous and mutant N-cadherin for the extracellular binding to nectin-2. Furthermore, we performed a rescue experiment for the N-cadherin morphants using rescue constructs of full-length N-cadherin (Ncad-res) and Ncad-Δβ (Ncad-res-Δβ; see Fig. S7B,C in the supplementary material). The defective neural fold phenotype of Ncad-MO-injected embryos was rescued by Ncad-res but not by Ncad-res-Δβ (Fig. 7M). Together, these results indicate that nectin-2 facilitates the apical localization of N-cadherin, followed by the accumulation of F-actin probably through the β-catenin-mediated linkage, which leads to apical constriction and proper neural tube closure.

In early vertebrate development, neural tube closure is one of the essential, and most dynamic, morphological processes. In this study, we found that nectin-2 plays a crucial role in this process by regulating apical constriction in collaboration with N-cadherin.

Among the four types of nectin, only nectin-2 mRNA was present at significant levels during early Xenopus embryogenesis, although whether other nectins are contributed maternally as proteins remains to be clarified. As Xenopus tropicalis also has these four nectins, like other vertebrates including humans, the four nectin genes are unlikely to reflect the tetraploidization of the Xenopus laevis genome, but seem to have evolved as independent genes. Interestingly, a recent study showed that zebrafish nectin-1 is expressed in the anterior neural keel of the early neurula and in the eye and brain of the larva (Helvik et al., 2009), implying a conserved function for nectins in the regulation of developmental morphogenesis.

In this study, we identified a novel function of nectin-2 in vertebrates as a component of the neural apical constriction machinery. Our results indicate that apical constriction seems to be induced by the extracellular domain of nectin-2 but not by its intracellular afadin-binding motif linked to F-actin. This suggests that the F-actin linkage through afadin may not be strong enough to cause F-actin accumulation, although afadin might still serve as a mediator for cell adhesion and for the maintenance of tissue integrity. Our finding also highlights some functional differences from the previous findings in Drosophila, which include the necessity of the intracellular domain of Ed to regulate cell morphology (Lin et al., 2007). This difference implies an evolutionally diverged function of these structurally related adhesion molecules for the same cell-morphological change.

Overexpression experiments show that it is unlikely that the ectopic apical constriction caused by overexpressed nectin-2, nec2-ΔC4 or -ΔIC was owing to the formation of cis dimers with endogenous nectin-2, and exerted through their intact afadin-F-actin linkage, because the extracellular domain of nectin-2 did not show detectable homophilic binding. In addition, the transmembrane region may not be responsible for this phenotype, because nec2-ΔEC, which contains this region, did not induce ectopic apical constriction. Thus, the effect caused by overexpression of nectin-2, nec2-ΔC4 or -ΔIC may not result from cis interaction with endogenous nectin-2.

Relationships between nectin family molecules and cadherins have been reported as a co-localization pattern (Takahashi et al., 1999; Miyahara et al., 2000; Okabe et al., 2004) and as an indirect connection mediated by associated intracellular proteins (Tachibana et al., 2000; Pokutta et al., 2002). However, no direct interaction between nectins and cadherins has been shown previously. Moreover, we found that N-cadherin was dependent on nectin-2 for its apical accumulation in the neuroepithelium, which may be required for apical constriction. These findings serve as the first evidence for the direct interaction and functional relationship between these adhesion molecules. Although we cannot rule out the possibility that nectin-2 may also function for cell adhesion in neuroepithelium independently of cadherins (Takahashi et al., 1999), the clearer interaction of nectin-2 and N-cadherin than that of nectin-2 homodimer suggests a functional importance of this heterophilic interaction in vivo.

We also found that the interaction between nectin-2 and cadherin is subtype specific: i.e. nectin-2 interacted with N-cadherin more strongly than with E- and/or C-cadherin. In addition, N-cadherin is predominantly expressed in the neuroepithelium from the beginning of neurulation (Detrick et al., 1990; Fujimori et al., 1990), where nectin-2 is also expressed at a high level. Thus, the strong, specific interaction and spatiotemporally concurrent expression of nectin-2 and N-cadherin appear to be crucial for apical constriction and folding of the neural tube. Recently, Nandadasa et al. (Nandadasa et al., 2009) showed that N-cadherin and E-cadherin are specifically required for the assembly of cortical F-actin in the neural plate and in the non-neural ectoderm, respectively, suggesting that N- and E-cadherin have different functional specificities in the different tissues. Thus, the identification of nectin-2 as a functional partner of N-cadherin may also provide a clue to understanding the molecular basis of the divergent functions of certain cadherin family proteins in vivo. For the future, it is intriguing to narrow down interacting domains between nectin-2 and N-cadherin.

Taking our results together with those of previous studies, we propose the following model for the apical constriction of the neuroepithelium during Xenopus neurulation. At the start of neurulation, the expression of nectin-2 is upregulated in the superficial layer of the neuroepithelium, where N-cadherin is also expressed. The nectin-2 protein preferentially localizes to the AJs by an unknown mechanism and binds to N-cadherin through its extracellular domain, facilitating the apical accumulation of N-cadherin. The accumulated N-cadherin promotes the translocation of F-actin, which is indirectly associated with the cytoplasmic domain of N-cadherin, to the apical cytoplasm. Finally, the apical F-actin bundle contracts via the force of activated myosin, which may be phosphorylated by Shroom and ROCK activity (Lee et al., 2007; Nishimura and Takeichi, 2008). In this model, the nectin-2-N-cadherin system serves as pre-constriction machinery to confer cells with the competence to undergo apical constriction. To understand this process further, it will be important to analyze the relationship between the nectin-cadherin machinery and the other regulatory systems such as the Shroom-ROCK pathway in apical constriction.

We thank Dr B. M. Gumbiner for the C-cadherin construct, Drs R. Mayor, Y. Mimori-Kiyosue and S. Koshida for helpful discussions and comments, and members of the Division of Morphogenesis, NIBB, for technical assistance, discussions, and comments. This work was supported by grants from the Ministry of Education, Science, Sports, and Culture of Japan to N.U. and NIH (RO1HD044764-06) to S.N. and C.W. Deposited in PMC for release after 12 months.