Gastrulation proceeds by the coordination of signaling processes and cell movements that produce the three germ layers and establish the anterior–posterior axis. Distinct genetic pathways control axis patterning and these axis-patterning pathways also control gastrulation cell movements (Kodjabachian et al., 1999; Solnica-Krezel and Cooper, 2002). In zebrafish, early morphogenetic movements include epiboly, the process of cell spreading over the yolk cell; involution, the inward movement of cells under the external surface to produce the mesodermal and endodermal cell layers; and convergence and extension of cells toward the midline and animal pole, respectively (Kane and Adams, 2002; Solnica-Krezel and Cooper, 2002). Cell adhesion systems, including cadherin adhesion, have been implicated in early morphogenetic cell movements in the zebrafish embryo (Kane and Adams, 2002; Solnica-Krezel and Cooper, 2002).
Cadherin cell adhesion molecules represent one of several classes of cell adhesion molecules that play critical roles in many developmental processes (Yagi and Takeichi, 2000; Perez-Moreno et al., 2003). This family of transmembrane proteins has extracellular domains that bind homophilically with cadherin molecules on opposing cells, and a single transmembrane domain is followed by the cytoplasmic domain. Peripheral membrane proteins, including the catenins, bind cytoplasmic domain sequences that link cadherin adhesion molecules with the actin cytoskeleton and intracellular signaling molecules (Perez-Moreno et al., 2003).
E-cadherin was the first known cadherin family member; it was identified by its ability to regulate compaction of the mouse morula (Kemler et al., 1977; Takeichi, 1977, 1988). The spatiotemporal expression pattern of E-cadherin in mouse, chick, and zebrafish implicates this molecule in many developmental processes, including early embryogenesis, epithelial tissue formation, and neurogenesis (Takeichi, 1988; Yagi and Takeichi, 2000; Babb et al., 2001). In zebrafish, E-cadherin (cdh1) is expressed maternally in all blastomeres. At the onset of gastrulation, cdh1 message accumulates in the region of the shield, which is the dorsal organizing center equivalent to Spemann's organizer. As the embryonic axis forms during gastrulation, cdh1 continues to be expressed in the anterior chordamesoderm, which is another critical signaling center that regulates anterior–posterior patterning of the neural axis. Later, cdh1 expression is activated in newly forming tissues, including central nervous system placode structures, nephric ducts, the gut, and discrete regions of the brain (Babb et al., 2001).
Maternally provided message allows E-cadherin knockout embryos to develop until cavitation, when the trophectoderm and inner cell mass form (Larue et al., 1994; Riethmacher et al., 1995). At this point, without zygotic contribution, E-cadherin levels fall, and the trophectoderm disintegrates. As a consequence, no information exists about the function of E-cadherin during early morphogenesis of the vertebrate embryo proper.
By using morpholino knockdown experiments, we examined cdh1 function during early embryogenesis in zebrafish. Morpholino oligonucleotides (MOs) are chemically modified analogs of antisense oligonucleotides that block expression of target genes and have been extremely useful for studying zebrafish development (Ekker, 2000; Ekker and Larson, 2001). Injection of translation blocking morpholino oligonucleotides inhibited both maternal and zygotic Cdh1 expression, causing irregular blastomere cleavage and inhibited epiboly and gastrulation cell movements. Most embryos injected with cdh1 MO arrested within the first day of development. Significantly, cdh1 MO-injected embryos showed defects in prechordal plate survival and head organizer activity, indicating that E-cadherin is not only required for morphogenetic cell movement but is also required for tissue formation and axis patterning in the early embryo.
Cdh1 Expression in Early Zebrafish Embryos
We previously described the expression of cdh1 message (Babb et al., 2001). Here, we describe the expression of Cdh1 in early zebrafish embryos by using an antiserum raised against a Cdh1 peptide sequence. This antiserum specifically recognized two polypeptide species in immunoblot analysis, one at 120 kDa and another at 140 kDa. Both polypeptides could be competed by incubating excess immunizing peptide with the antiserum, but an irrelevant peptide did not prevent antibody binding (Fig. 2B). In younger embryos (less than 13 hours postfertilization [hpf]), the 140-kDa polypeptide was more abundant than the 120-kDa polypeptide. At later times, the 120-kDa polypeptide was more abundant than the 140-kDa polypeptide, and by 29 hpf, the 140-kDa polypeptide was nearly undetectable (Fig. 2B). The 140-kDa polypeptide likely represents an intracellular biosynthetic intermediate whose proteolytic processing or glycosylation has not completely matured. The presence of the 140-kDa species correlates with the presence of intracellular vesicular structures that label with this antibody by immunofluorescence. This intracellular form of Cdh1 appears to be held in vesicles, ready to assemble new junctions in the rapidly developing embryo. All immunohistochemical labeling using Cdh1 antibody could also be competed using the immunizing peptide (Fig. 1O, inset). In addition, Cdh1 antibody reactivity in immunoblot and immunohistochemical experiments was reduced in cdh1 MO-injected embryos (see below). Together, these data demonstrate that the Cdh1 antibody is monospecific. Cdh1 expression was analyzed by using this antibody in the zebrafish early embryo.
In initial cleavage stage embryos, Cdh1 was concentrated in cleavage furrows. Two-cell stage embryos were labeled using Cdh1 antibody and horseradish peroxidase-conjugated secondary antibodies and diaminobenzidine (DAB)-peroxidase precipitate detection. Cdh1 was detected in the cleavage furrow between the two blastomeres and within the forming cleavage furrow for the four-cell stage (Fig. 1A,B). Cdh1 was also detected within the cytoplasm of blastomeres and within the yolk cell (Fig. 1B). Cdh1 was maintained in cleavage furrows in eight-cell stage and later embryos (Fig. 1C,D). All blastomeres express Cdh1 during early epiboly stages (Fig. 1E,F). Cdh1 was concentrated in junctions between blastomeres, within cytoplasmic vesicles and within the yolk cell (Fig. 1F,H). Cdh1 expression in the yolk cell was prominent near the yolk syncytial layer and near the most vegetal yolk syncytial layer nuclei that form foci for the yolk cell microtubule array (Strahle and Jesuthasan, 1993; Solnica-Krezel and Driever, 1994; Jesuthasan and Stahle, 1997; Jesuthasan, 1998; Fig. 1F,G). The accumulation of Cdh1 at cell–cell contacts was strongest in the enveloping layer on the surface of the embryo, but deep cells also show cell contact labeling by using Cdh1 antibody (Fig. 1H; see video 1 in Supplemental Material, which is available at http://www.mrw.interscience.wiley.com/suppmat/1058-8388/suppmat/).
During gastrulation, Cdh1 accumulates in the anterior mesendoderm. Lateral views of 75% epiboly stage embryos showed strong Cdh1 staining in the forming embryonic axis (Fig. 1I,J). Two-photon sections through the anterior axis region showed that Cdh1 protein was accumulating in anterior mesendodermal cells (Fig. 1K,L; see video 2 in Supplemental Material).
At segmentation stage, Cdh1 expression was refined. It was reduced in most cells and was induced in several structures. Ectoderm cells express high levels of Cdh1 throughout embryogenesis (Fig. 1N and data not shown). Cdh1 was strongly expressed in the notochord and the medial floor plate of the developing neural tube of the 15.5 hpf segmentation stage embryo (Fig. 1N). Expression of Cdh1 protein closely correlated with our previous description of cdh1 message expression during early embryogenesis (Babb et al., 2001).
E-Cadherin Morpholino Knockdown Has Immediate Effects on Cleavage Stage Embryos
To directly determine the functional implications of cdh1 expression in early embryos, we suppressed Cdh1 translation by using MO injection experiments (Nasevicius and Ekker, 2000). Two MOs were designed to hybridize near the translation initiation start site (see Experimental Procedures section). Control experiments were performed to determine the specificity of the MO injection experiments. No differences in development were detected, and there was no statistical difference in the number of embryos that survived when we compared uninjected and negative control embryos injected with either buffer alone, rhodamine–dextran or negative control MO (data not shown). Two independent MOs specific for cdh1 produced the same MO-induced phenotype (described in detail below). Cdh1 expression levels were suppressed by cdh1 MO injection (Figs. 1O,P, 2A). The extent of phenotypic changes induced by cdh1 MO injection (described in detail below) correlated with the reduction of Cdh1 expression by immunoblot (Fig. 2A). Significantly, phenotypic changes induced by cdh1 MO injection could be rescued by coinjecting synthetic message encoding Xenopus E-cadherin, which shows no significant sequence similarity with either zebrafish cdh1 MO (Figs. 3, 6, 9; Table 1). Together, these experiments strongly support the conclusion that effects observed using cdh1 MO injections are specifically due to reduced Cdh1 maternal and zygotic function.
Table 1. cdh1 MO Injection Inhibits Both Epiboly and Gastrulation
Body axis length (μm)
aP values determined by Student's t-test. MO, morpholino oligonucleotide.
mRNA + cdh1 MO
Body axis length
Control vs. mRNA
P = 0.57
P = 0.98
Control vs. cdh1 MO
P = 5.04 × 10−7
P = 1.33 × 10−9
Most embryos injected with cdh1 MO did not survive the first 24 hours of embryogenesis (Fig. 3). Coinjection of synthetic message encoding Xenopus E-cadherin partially rescued the effect of cdh1 MO, significantly increasing the proportion of embryos that survived. Embryos injected with synthetic mRNA encoding Xenopus E-cadherin alone displayed no statistical difference in survival compared with control MO-injected embryos (Fig. 3), and no obvious morphological defects were observed in embryos injected with Xenopus E-cadherin mRNA (Figs. 3, 6, 9; Table 1; and data not shown). Specifically, there was no detectable difference between control and Xenopus E-cadherin mRNA injected embryos in our analysis of early blastomere adhesion, epiboly, gastrulation, presence of hatching gland, survival, and general morphology.
In normal embryos and in control MO-injected control embryos, blastomere cleavage through the 32-cell stage occurs at stereotypical, regular orientations and produces daughter cells of approximately equal size (Fig. 4, top panel; see video 3 in Supplemental Material). In embryos injected with cdh1 MO at the one-cell stage, blastomere cleavage planes formed at irregular angles, and these cleavages produced blastomeres of unequal size (Fig. 4, middle and bottom panels; see video 3 in Supplemental Material). Cdh1 was localized at cleavage furrows in early cleavage stage embryos (Fig. 1A–D), suggesting that Cdh1 within junctional complexes of early blastomeres regulates cleavage orientation and partitioning of cytoplasm.
To examine the consequence of cdh1 MO injection on junctional complexes in cleavage stage embryos, Cdh1 immunofluorescence was performed. Optical sections through embryos were collected by using two-photon microscopy, and projection images were generated from three-dimensional renderings of these image volumes. In eight-cell embryos, Cdh1 was detected in cleavage furrows, in vesicular structures within the cytoplasm of blastomeres, and in the yolk cell cytoplasm (Fig. 1O). In eight-cell embryos that were injected with cdh1 MO at the one-cell stage, Cdh1 staining was reduced in cleavage furrows, in the blastomere cytoplasm, and in the yolk cell cytoplasm (Fig. 1P). Reduction of Cdh1 expression in these eight-cell embryos was not complete, indicating that maternal Cdh1 expression was activated before injection of cdh1 MO into one-cell embryos. However, the reduced Cdh1 expression had dramatic consequences on junctional complexes. One junction in the cdh1 MO-injected eight-cell embryo in Figure 1P was discontinuous (arrow), and another junction in this embryo was curled in on itself (arrow). This radically defective junction morphology induced by cdh1 MO injection is associated with asymmetric blastomere cleavage.
Normally, in later cleavage stage embryos, blastomeres become tightly adherent on the top of the yolk cell, in a process that has been termed compaction (Fig. 1, top panel: 1.75 hpf and 2.1 hpf; Jesuthasan, 1998). Blastomeres of embryos injected with cdh1 MO did not form a compact cell mass on top of the yolk cell (Fig. 1, middle and bottom panels: 1.75 hpf and 2.1 hpf). In some cdh1 MO-injected embryos, blastomeres were shed from the embryo (see video 3 in Supplemental Material). In summary, inhibiting Cdh1 expression interferes with normal cleavage symmetry and compaction of blastomeres in cleavage stage embryos.
Epiboly Rate Was Reduced in E-Cadherin Morpholino Knockdown Embryos
Despite severe effects of cdh1 MO injection on the cleavage stages, embryos recover from these early defects and initiate epiboly. Epiboly rate for control embryos was compared with that of cdh1 MO-injected embryos, and the extent of epiboly was reduced in cdh1 MO-injected embryos, compared with control MO-injected embryos (Fig. 5; see video 3 in Supplemental Material). We measured the extent of epiboly in embryos at 8 hpf by in situ hybridization using a probe for ntl, which labels the germ band, a ring of cells at the vegetal border of the embryo. The average percent epiboly was reduced from 77% in controls to 62% in cdh1 MO-injected 8 hpf embryos (Fig. 6A,B; Table 1). Therefore, inhibiting Cdh1 function reduces epiboly cell movements. Microinjecting Xenopus E-cadherin mRNA alone did not significantly affect epiboly rate, but this RNA did partially rescue the effect of cdh1 MO treatment on epiboly (average percent epiboly of 8 hpf embryos that were injected with both cdh1 MO and Xenopus E-cadherin mRNA was 68%; see Fig. 6A–D; Table 1).
Zebrafish genes that control early morphogenetic movements were identified in various mutagenesis screens (Hammerschmidt et al., 1996; Kane et al., 1996; Solnica-Krezel et al., 1996). Mutations that affect epiboly are characterized by defects of cell movements to cover the yolk cell, which cause the yolk cell to lyse. These defects in epiboly can also result in a bifurcated embryonic axis when axis formation initiates, despite the delay in spreading of cells over the vegetal pole, producing an embryonic axis that bends and extends around the delayed germ band rather than extending along a normal midline axis (Kane et al., 1996; Kane and Adams, 2002). Many cdh1 MO-injected embryos showed this bifurcated body axis phenotype (Fig. 7). In cdh1 MO-injected embryos that showed severe epiboly delay, blastomeres frequently failed to envelop the yolk cell, and consequently, the subsequent forces produced by gastrulation cell movements caused the yolk cell to lyse (see video 4 in Supplemental Material).
Gastrulation Defects Due to E-Cadherin Morpholino Knockdown Leads to Defects in Prechordal Plate Survival
Involution cell movements initiate when epiboly is approximately 50% complete, at 6 hpf (Fig. 5, top panel). In cdh1 MO-injected embryos at 6 hpf, epiboly was less than 50% complete but gastrulation was still initiated (Fig. 5, middle and bottom panels). Therefore, inhibiting Cdh1 expression uncouples epiboly cell movements and gastrulation initiation. Previous studies have also shown that epiboly and gastrulation can be uncoupled, for example, by microtubule depolymerization (Strahle and Jesuthasan, 1993; Solnica-Krezel and Driever, 1994).
Gastrulation progress was measured to determine whether the rate of gastrulation cell movements was delayed. In situ hybridization of ntl in 8 hpf control and MO-injected embryos was performed to label axial mesendoderm. The length of the newly forming body axis was measured (Fig. 6; Table 1). The average length of the embryonic axis at 8 hpf in control embryos was 210 μm, while the average length of the embryonic axis was reduced to 117 μm in cdh1 MO-injected embryos (Fig. 6A,B; Table 1). The shape of the axial mesendoderm and presumptive prechordal plate labeled by ntl and gsc in situ hybridization, respectively, was changed from a columnar shape in control embryos to a wider, shorter, wedge shape in cdh1 MO-injected embryos (Figs. 6A,B,E,F). This ntl and gsc staining pattern was reminiscent of that seen in gastrulation mutants (Solnica-Krezel and Cooper, 2002), suggesting convergence and extension defects. The shape of the ntl and gsc expression domains and body axis length were partially rescued by comicroinjecting Xenopus E-cadherin synthetic mRNA (Fig. 6A–H).
Convergence and extension cell movement defects result in abnormal somite morphology and shorter body length (Solnica-Krezel and Cooper, 2002). Overall morphology of 13 hpf embryos was observed by differential interference contrast (DIC) microscopy (Fig. 8A–D). Compared with controls, cdh1 MO-injected embryos showed a reduced body length. Additionally, cells were frequently shed from cdh1 MO-injected embryos, indicating cell adhesion defects in these embryos (Fig. 8B). By using DIC microscopy, the dorsal surface of 13 hpf embryos was examined, focusing on anterior somite morphology. Somites in control embryos are cuboidal following the neural keel (Fig. 8C). In cdh1 MO-injected embryos, somites were expanded in the medial–lateral axis and compressed in the anterior–posterior axis (Fig. 8D), a characteristic of embryos displaying convergence and extension defects (Solnica-Krezel and Cooper, 2002). In addition, convergence and extension of paraxial mesoderm was examined by using papc in situ hybridization. Somitic and presomitic paraxial mesoderm failed to converge on the midline, was expanded in the medial–lateral axis, and compressed in the anterior–posterior axis in cdh1 MO-injected embryos, as compared with control embryos (Fig. 8E,F). In summary, defects in cdh1 MO-injected embryos suggest the existence of a convergence and extension cell movement defect.
Gastrulation not only involves the orchestration of cell movements that produce three germ layers, but also orchestration of signaling processes that control anterior–posterior patterning and initiate tissue formation and organogenesis (Pourquie, 2001; Rallu et al., 2002). Morphological analysis of pharyngula stage embryos (24 hpf) showed that the hatching gland, a derivative of the prechordal plate, was reduced or absent in cdh1 MO-injected embryos (Fig 9D, arrow). A component of the prechordal plate, the head organizer induces forebrain formation and patterns the anterior–posterior axis in a gradient from rostral to caudal. Defects in head organizer signaling severely impair forebrain patterning (Rallu et al., 2002). In cdh1 MO-injected embryos, forebrain structures were severely reduced; midbrain and hindbrain defects were less severe than those in the forebrain (Fig. 9A,B,D,E). Forebrain defects were examined by using acetylated tubulin labeling of axon tracts and two-photon microscopy. In control embryos, anterior axon tracts were well developed (Fig. 10A,B), but cdh1 MO-injected embryos showed significant reduction in anterior commissure, postoptic commissure, and supraoptic tract formation (Fig. 10C–F). In the most severe cases, the tract of the postoptic commissure was also greatly reduced (Fig. 10E,F). Significantly, coinjection of cdh1 MO and Xenopus E-cadherin mRNA restored hatching gland and forebrain development (Fig. 9G,H), possibly by rescue of prechordal plate survival and head organizer signaling. Posterior structures did not show patterning defects (Fig. 9C,F); although, the ectoderm showed adhesion defects, and ventroposterior structures were disorganized, which may be a consequence of gastrulation cell movement defects (Fig. 9C,F). In summary, prechordal plate derived structures failed to form when Cdh1 expression was blocked by MO injection. This morphology is consistent with reduced head organizer activity. Alternatively, defects could be a direct consequence of Cdh1 loss in the forebrain.
Goosecoid is a homeodomain-containing transcription factor that is expressed in the presumptive head organizer and prechordal plate and regulates body axis formation (Ferreiro et al., 1998; Yao and Kessler, 2001). To examine whether Cdh1 knockdown prevents head organizer and prechordal plate formation during gastrulation initiation, goosecoid (gsc) expression was examined by in situ hybridization in 7 hpf embryos (Fig. 6E–H), a time point shortly after the induction of gsc expression. Both control and cdh1 MO-injected embryos expressed gsc in the presumptive prechordal plate (Fig. 6E,F). The shape of the gsc expression domain was altered, presumably due to defects in convergence and extension, but there was a similar labeling intensity and number of cells within the expression domain. These data indicate that induction of the head organizer and presumptive prechordal plate domain was not affected by Cdh1 knockdown. Therefore, failure in development of prechordal plate derivatives occurs at some later point, perhaps by failure to maintain the integrity of the head organizer.
The disposition of the hatching gland was followed during development by using hgg1 in situ hybridization. In control embryos, hgg1 labeling increased from 10 hpf to 15 hpf, and the expression domain was confined to the ventral region of the anterior mesoderm (Fig. 6 I,J, M,N,Q,R). In cdh1 MO-injected embryos, hgg1 labeling was detected with similar intensity as control embryos at 10 hpf (Fig. 6K,L), but fewer hgg1-positive cells were detected at later stages (Fig. 6O,P,S,T). In addition, hgg1-positive cells were expanded in their distribution, being found in dorsal regions of the head mesoderm. At 15 hpf, cells within the area of the developing hatching gland and prechordal plate of cdh1 MO-injected embryos were undergoing extensive apoptosis, as detected using acridine orange staining (Fig. 6W,X). Control embryos displayed very limited apoptosis at this stage (Fig. 6U,V). Together, these data indicate that Cdh1 expression promotes prechordal plate survival, which may maintain subsequent signaling from the head organizer, a prechordal plate derivative.
In this study, zebrafish Cdh1 was found to be required for discrete developmental processes in the early embryo. Initially, maternal Cdh1 regulates cleavage orientation, cytoplasmic partitioning in blastomeres and blastomere compaction. Subsequently, embryos injected with cdh1 MO were able to initiate epiboly but exhibited reduced rates of epiboly and gastrulation morphogenetic cell movements. Most cdh1 MO-injected embryos did not survive the first day of development. However, MO inhibition of gene expression produces a hypomorphic series of phenotypes due to variable reduction in protein expression, and consequently, some less affected cdh1 MO-injected embryos survive past the first day. Embryos injected with cdh1 MO that survived beyond 24 hpf displayed phenotypes that indicated a failure in prechordal plate derivative survival and consequently, a reduction in head organizer activity. Importantly, all observed cdh1 morphant phenotypes were rescued by coinjecting synthetic Xenopus E-cadherin mRNA.
Gene disruption experiments have revealed important developmental roles for E-cadherin (Larue et al., 1994; Riethmacher et al., 1995). E-cadherin plays an essential role in early mouse development; E-cadherin is maternally expressed, initially distributed in the trophectoderm and the inner cell mass. Later, epithelial tissues, like ectoderm, gut, and nephric tubules, express E-cadherin (Takeichi, 1988). Also, E-cadherin expression is induced in discrete regions of the brain (Shimamura and Takeichi, 1992; Stemmler et al., 2003). This expression pattern is analogous to that of zebrafish cdh1 message. Mouse knockout experiments showed that E-cadherin is required for trophectoderm maintenance (Larue et al., 1994; Riethmacher et al., 1995). Development of embryos lacking E-cadherin fails due to the disintegration of this extraembryonic epithelial tissue. Thus, these experiments did not reveal a function for the protein during development of the embryo proper.
The shotgun (shg) gene encodes Drosophila E-cadherin (DE-cadherin; Tepass et al., 1996; Uemura et al., 1996). The shg mutant phenotype shows a requirement for DE-cadherin during dynamic rearrangement of epithelial tissues in the early embryo. Our findings using MO injection to inhibit Cdh1 expression in zebrafish are consistent with the shg mutant phenotype. Epiboly and gastrulation are the first morphogenetic movements of the early zebrafish embryo, and both were sensitive to reduced Cdh1 expression.
Zebrafish epiboly mutants share three phenotypic features: reduced rate of epiboly cell movements; increased frequency of yolk cell lysis; and bifurcated body axis (Kane et al., 1996; Kane and Adams, 2002). All three defects were produced by cdh1 MO injection. Epiboly cell movements are driven by microfilament- and microtubule-based motility mechanisms within the yolk cell that drive yolk syncytial layer movements and blastomere cell movements that cover the yolk cell (Kane and Adams, 2002). Depolymerizing microtubules in epiboly stage embryos reduces epiboly rate, and in these embryos, epiboly and gastrulation initiation are uncoupled (Strahle and Jesuthasan, 1993; Solnica-Krezel and Driever, 1994; Jesuthasan and Stahle, 1997; Jesuthasan, 1998). Blocking Cdh1 expression also uncoupled epiboly and gastrulation initiation. Cdh1 was localized to membrane-microtubule association regions at the yolk syncytial layer, indicating that Cdh1 may link membrane adhesion sites and the microtubule cytoskeleton. Although the connection between the cadherin–catenin complex and the microtubule cytoskeleton is far less well understood than that of the actin cytoskeleton, evidence exists that cadherins control cell architecture through microtubules (Perez-Moreno et al., 2003), raising the possibility that inhibiting Cdh1 expression may uncouple the yolk cell microtubule cytoskeleton membrane contacts within the yolk syncytial layer.
Mutations in genes that regulate gastrulation cell movements are categorized into three groups: genes that regulate dorsal mesoderm formation and central nervous system patterning; genes that control ventral signaling pathways that regulate ventroposterior structure formation; and genes that control convergence and extension cell movements (Hammerschmidt et al., 1996; Solnica-Krezel et al., 1996; Solnica-Krezel and Cooper, 2002). Mutations can affect processes that overlap more than one of these categories. Reduced Cdh1 expression affected caudal central nervous system patterning, and their morphology was consistent with convergence and extension cell movement defects during gastrulation, including abnormal shape of ntl and gsc expression domains and in the abnormal somite morphology of the gastrulating embryos. Gastrulation cell movements occur by the coordinated migration of sheets of blastomeres toward the midline, involution of superficial cells, intercalation, and elongation of mesendodermal cells during extension of the forming body axis (Solnica-Krezel and Cooper, 2002). Our findings indicate that gastrulating cells may use E-cadherin–mediated cell–cell contact as a migration substrate and for cell movements that drive convergence and extension. Additional experiments will be required to test our hypothesis that specific cellular behaviors (migration and radial intercalation) are affected by loss of Cdh1-mediated adhesion.
Paraxial protocadherin (papc) was identified as a target for noncanonical Wnt signaling that regulates convergence and extension cell movements during gastrulation in Xenopus and zebrafish (Yamamoto et al., 1998; Kim et al., 2000; Heisenberg and Tada, 2002; Solnica-Krezel and Cooper, 2002). These effects were shown by papc overexpression and dominant negative expression. Knockout of the gene in mouse (mpapc) showed no defects (Yamamoto et al., 2000). De Robertis and colleagues speculate that there may be redundant protocadherin family members that compensate for papc. Knockdown of Cdh1 expression affected the shape of the papc expression domain but did not affect the level of papc expression in the gastrulating zebrafish embryos (see Fig. 8E,F). Any Cdh1 effects on convergence and extension would be independent of papc, and Cdh1 may be another target for noncanonical Wnt signaling during vertebrate gastrulation.
Other cadherin family members are expressed in the early zebrafish embryo. N-Cadherin (cadherin-2) is first expressed at the shield stage (Bitzur et al., 1994), and N-cadherin was shown to functionally substitute for E-cadherin (Malicki et al., 2003). VN-cadherin (cadherin-11) is first expressed at the 60% epiboly stage (Franklin and Sargent, 1996). These cadherins may be partially redundant during epiboly and gastrulation, permitting these processes to proceed in cdh1 knockdown embryos, albeit abnormally.
Previous studies in Xenopus showed that cadherins regulate cell–cell interactions and cell migration during gastrulation (Brieher and Gumbiner, 1994; Heasman et al., 1994). Our studies confirm and extend these previous experiments. Disrupting cadherin function during early Xenopus development caused defects in epiboly and gastrulation (Brieher and Gumbiner, 1994; Heasman et al., 1994; Lee and Gumbiner, 1995). In Xenopus, E-cadherin is not a maternal cadherin (Choi and Gumbiner, 1989), and gastrulation cell movements are controlled by other maternal cadherins (Brieher and Gumbiner, 1994; Heasman et al., 1994). However, E-cadherin is maternally expressed in mammals and chicken (Takeichi, 1988). We propose that the conserved E-cadherin expression pattern reflects the conserved function for this protein between zebrafish and mammals during gastrulation. In support of this assertion, E-cadherin in Drosophila embryos is maternally provided and controls early morphogenetic cell movements (Tepass et al., 1996; Uemura et al., 1996), indicating that E-cadherin function during dynamic cell rearrangements and movements in the early embryo is an ancestral function.
For Cdh1 to function during morphogenetic cell movements, cadherin adhesion must be regulated and coordinated. Cadherin-mediated cell migration requires both strong adhesion and disengagement. In cells expressing cadherin/α-catenin fusion proteins that cannot disengage cadherin adhesion, coordinated cell migration within a monolayer is defective (Nagafuchi et al., 1994). Several signaling pathways act to coordinate gastrulation, and many of these activities should have E-cadherin as a regulatory target. For example, DE-cadherin and myosin VI interact during border cell migration in the egg chamber, which is required for Drosophila oogenesis (Geisbrecht and Montell, 2002). Thus, cadherin function during development can be controlled by regulating gene expression and by signaling processes that regulate adhesive activity of the cadherin protein.
Various regulatory pathways may impinge on cdh1 to regulate the expression of this cell adhesion molecule and its activity during gastrulation. For example, a transcription factor, Snail, was shown to regulate E-cadherin levels in various vertebrates and in Drosophila (Oda et al., 1998; Batlle et al., 2000; Cano et al., 2000). In mouse, Snail binds E-box consensus sequences in the promoter of its gene and, thereby, represses transcription (Batlle et al., 2000). Snail overexpression suppresses E-cadherin gene expression and induces an epithelial-to-mesenchymal transition (Cano et al., 2000). However, Snail represses transcription of other key regulators of the epithelial phenotype (Ikenouchi et al., 2003), and restoring E-cadherin expression is not sufficient to rescue the effects of Snail overexpression (Cano et al., 2000). Cdh1 expression and its adhesive activity are likely to be controlled by several developmental signaling pathways to regulate cell adhesion and migration during gastrulation cell movements.
In addition to cell movement defects, a subset of zebrafish gastrulation mutants showed reduced or absent head organizer activity. In the cdh1 morphant, derivatives of the prechordal plate were reduced or absent, and reduced forebrain development indicates that subsequent signaling activity from the head organizer was impaired. Our goosecoid expression analysis showed that the prechordal plate was induced normally. In addition, prechordal plate derivatives labeled by hgg1 in situ hybridization did not remain within their ventroanterior location; instead, these cells migrated to more dorsoposterior regions in the head mesoderm. Failure to respect compartmental boundaries indicates that Cdh1-mediated adhesion is responsible for maintaining tissue integrity.
In the cdh1 morphant, prechordal plate derivatives and cells located in the ventral-most region of the neural tube (floor plate) died by apoptosis around 15 hpf. Consequently, inductive signals for anterior brain development may have been compromised. Apoptosis of prechordal plate derivatives may be caused by adhesion defects in cdh1 MO-injected embryos. Anoikis is cell death due to defects in substrate adhesion (Frisch and Francis, 1994), and the cadherin binding protein β-catenin was found to regulate anoikis (Orford et al., 1999). Alternatively, survival signal receptors may be impaired due to adherens junction disruption. Various receptors are concentrated in the adherens junction (Perez-Moreno et al., 2003), which may be mislocalized as a consequence of down-regulating Cdh1 expression. In cdh1 MO-injected embryos, cells of the prechordal plate may not receive normal survival signals, leading to cell death and the subsequent loss of head organizer activity.
In conclusion, these studies show that cdh1 regulates critical morphogenetic cell movements in the early vertebrate embryo. In addition to the cell migration defects, the presence of the Cdh1 adhesion molecule is necessary to maintain integrity and survival of prechordal plate derivatives. Failure to maintain these structures may have resulted in neural axis patterning defects due to reduced head organizer activity. Analysis of cdh1 function during organogenesis stages will require additional experimentation.
Zebrafish (Danio rerio) were raised and kept under standard laboratory conditions (Westerfield, 2000) in accordance with Indiana University Policy on Animal Care and Use. For some experiments, 0.2 mM phenylthiourea was added to prevent melanization. Ages of the embryos are given as hours postfertilization (hpf).
Morpholino Oligonucleotide and mRNA Injection
Translation blocking antisense MOs: mphEcad-m54-m79 5′-TAA ATC GCA GCT CTT CCT TCC AAC G-3′ (which hybridize to bases −54 to −79 from the start codon); mphEcad-st-25 5′-ATC CCA CAG TTG TTA CAC AAG CCA T-3′ (which hybridizes to the start codon to +25); and standard control 5′-CCT CTT ACC TCA GTT ACA ATT TAT A-3′ purchased from Gene Tools (Philomath, OR) were used as described (Nasevicius and Ekker, 2000). MOs were designed according to Gene Tools targeting guidelines. MO sequences were then compared with databases by using BLAST, and no significant similarities were found to any sequences other than zebrafish cdh1. Notably, these cdh1 MOs do not show significant sequence similarity with Xenopus E-cadherin sequences. MOs were injected into one- to eight-cell stage embryos at 4.2 μg/μl (0.5 mM) in Daneau buffer [58 mM NaCl, 0.7 mM KCl, 0.4 mM MgSO4, 0.6 mM Ca(NO3)2, 5.0 mM HEPES pH 7.6]. The total volume injected was 5–10 nl (21–42 ng per embryo).
Synthetic capped Xenopus E-cadherin mRNA was transcribed from Xenopus E-cadherin cDNA in the pSP64T vector (Kintner, 1992; Heasman et al., 1994) by using the SP6 mMessage mMachine (Ambion, Austin, TX). mRNA was microinjected into one-cell stage embryos at 0.5 μg/μl in Daneau buffer. The total volume injected was 5–10 nl (2.5–5 ng per embryo).
Cdh1 Polyclonal Antibody
A synthetic peptide (N-DKDLPPFAGPFKVEPQGDTSKN-C) derived from zebrafish Cdh1 extracellular domain (Babb et al., 2001) was synthesized, conjugated to KLH, and used to immunize New Zealand White rabbits (Covance, Richmond, CA).
For immunoblotting, zebrafish embryos were solubilized in sodium dodecyl sulfate (SDS) buffer (10 mM Tris-HCl, pH 7.5, 2 mM ethylenedi- aminetetraacetic acid, 1% SDS), separated by using SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose and blocked in 3% ovalbumin in TBST (10 mM Tris pH 7.5, 0.1 M NaCl, 0.1% Tween 20). Membranes incubated with antiserum (1:15,000 dilution) and anti-rabbit horseradish peroxidase (HRP) -conjugated secondary antibody (1:10,000 dilution; Amersham, Arlington Heights, IL) were processed for chemiluminescence (ECL kit, Amersham) and exposed to film (Kodak Bio-Max ML, Eastman Kodak, Rochester, NY).
For whole-mount immunolabeling, embryos (6 hpf and younger) were fixed in 20% dimethyl sulfoxide (DMSO)/80% methanol at −20°C overnight. Older embryos were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) overnight at 4°C. Before immunolabeling, embryos were blocked for 1 hr in PBS with 0.5% Triton X-100, 5% horse serum, 0.2% bovine serum albumin, 1% DMSO. All rinses were performed using PBS with 0.5% Triton X-100.
For HRP labeling, zebrafish Cdh1 antiserum was used at 1:2,500 dilution, followed by biotinylated anti-rabbit antibody (BA-1000, Vector Laboratories, Burlingame, CA) at 1:250 dilution. Endogenous peroxidase activity was quenched by using 0.3% H2O2. Embryos were then incubated in HRP–avidin-D (Vector Laboratories) at 1:1,000 dilution, and color was developed by using the DAB substrate kit for HRP (Vector Laboratories). Embryos were mounted in glycerol and imaged by using a Leica MZ12 dissecting microscope (Leica Microsystems, Inc., Deerfield, IL) equipped with a SPOT RT camera (Diagnostic Instruments, Sterling Heights, MI).
For immunofluorescence, embryos were incubated in Cdh1 anti-peptide antibody at 1:2,500 dilution or anti-acetylated tubulin antibody (T-6793, Sigma, St. Louis, MO) at 1:5,000 dilution, followed by either Texas Red-conjugated anti-rabbit or anti-mouse secondary antibodies made in goat (Jackson Immunoresearch Labs, West Grove, PA) at 1:100 dilution and Texas Red-conjugated anti-goat tertiary antibody made in donkey (Jackson Immunoresearch Labs) at 1:100 dilution. Embryos were then mounted in PBS for viewing. Image volumes were acquired by using a Bio-Rad MRC1024 laser scanning confocal system (Bio-Rad, Hercules, CA) equipped with a Titanium-Sapphire laser for 2-photon illumination mounted on a Nikon inverted microscope. These two-photon image volumes were rendered by using Voxx, a voxel based three-dimensional near real time rendering program, developed at the Indiana Center for Biological Microscopy (Clendenon et al., 2002) (http://www.nephrology.iupui.edu/imaging/voxx/index.htm).
In Situ Hybridization
Embryos were fixed in 4% paraformaldehyde in PBS overnight at 4°C and whole-mount in situ hybridization of zebrafish embryos was performed as described (Liu et al., 1999). Digoxigenin-labeled riboprobes for ntl (Schulte-Merker et al., 1992), gsc (Thisse et al., 1994), papc (Yamamoto et al., 1998), and hgg1 (Thisse et al., 1994) were synthesized from cDNA as run-off transcripts from linearized templates by using the Genius System DIG RNA Labeling Kit (Roche, Indianapolis, IN). Embryos were mounted in glycerol and imaged by using a Leica MZ12 dissecting microscope equipped with a SPOT RT camera (Diagnostic Instruments, Sterling Heights, MI).
Acridine Orange Staining
To detect apoptosis, live embryos were removed from their chorions and incubated for 2 min into Embryo Medium (Westerfield, 2000) containing 5 μg/ml of acridine orange (A-3568, Molecular Probes, Eugene, OR). After several rinses in Embryo Medium, epifluorescence images were collected by using a Nikon Diaphot microscope (Nikon, Inc., Melville, NY) equipped with DIC optics and a SPOT RT camera (Diagnostic Instruments).
We thank Drs. C. Nusslein-Volhard, S. Amacher, C. Kintner, J. Heasman, J. Postlethwait, and M. Tada for probes and plasmids. Technical assistance by Sharon Bledsoe is gratefully acknowledged. J.A.M. was supported by National Institute of Health, and S.G.B. was supported by a postdoctoral fellowship from Prevent Blindness America/Fight For Sight.