Nuclear β-catenin promotes non-neural ectoderm and posterior cell fates in amphioxus embryos


  • Linda Z. Holland,

    Corresponding author
    1. Marine Biology Research Division, Scripps Institution of Oceanography, University of California San Diego, La Jolla, California
    • Marine Biology Research Division, Scripps Institution of Oceanography, University of California San Diego, La Jolla, CA 92093-0202
    Search for more papers by this author
  • Kristen A. Panfilio,

    1. Marine Biology Research Division, Scripps Institution of Oceanography, University of California San Diego, La Jolla, California
    Current affiliation:
    1. University Museum of Zoology, Downing Street, Cambridge, CB2 3EJ UK
    Search for more papers by this author
  • Roger Chastain,

    1. Marine Biology Research Division, Scripps Institution of Oceanography, University of California San Diego, La Jolla, California
    Search for more papers by this author
  • Michael Schubert,

    1. Marine Biology Research Division, Scripps Institution of Oceanography, University of California San Diego, La Jolla, California
    Current affiliation:
    1. Ecole Normale Supérieure de Lyon, 46 Allée d'Italie, 69364 Lyon Cedex 07, France
    Search for more papers by this author
  • Nicholas D. Holland

    1. Marine Biology Research Division, Scripps Institution of Oceanography, University of California San Diego, La Jolla, California
    Search for more papers by this author


In vertebrate development, Wnt/β-catenin signaling has an early role in specification of dorsal/anterior identity and a late one in posterior specification. To understand the evolution of these roles, we cloned β-catenin from the invertebrate chordate amphioxus. The exon/intron organization of β-catenin is highly conserved between amphioxus and other animals including a cnidarian, but not Drosophila. In development, amphioxus β-catenin is concentrated in all nuclei from the 16-cell stage until the onset of gastrulation when it becomes undetectable in presumptive mesendoderm. Li+, which up-regulates Wnt/β-catenin signaling, had no detectable effect on axial patterning when applied before the late blastula stage, suggesting that a role for β-catenin in specification of dorsal/anterior identity may be a vertebrate innovation. From the mid-gastrula through the neurula stage, the highest levels of nuclear β-catenin are around the blastopore. In the early neurula, β-catenin is down-regulated in the neural plate, but remains high in adjacent non-neural ectoderm. Embryos treated with Li+ at the late blastula stage are markedly posteriorized and lack a neural plate. These results suggest that in amphioxus, as in vertebrates, down-regulation of Wnt/β-catenin signaling in the neural plate is necessary for maintenance of the neuroectoderm and that a major evolutionarily conserved role of Wnt/β-catenin signaling is to specify posterior identity and pattern the anterior/posterior axis. Developmental Dynamics 233:1430–1443, 2005. © 2005 Wiley-Liss, Inc.


β-catenin functions in cell adhesion as a component of adherens junctions and is a key part of the canonical Wnt-signaling pathway. Signaling by Wnts 1, 3, and 8 results in translocation of β-catenin from the cytoplasm to the nucleus where it partners with the DNA-binding protein TCF/LEF to regulate transcription of down-stream targets (reviewed in Huelsken and Birchmeier, 2001; Pandur et al., 2002). In animals as phylogenetically distant as cnidarians, sea urchins, and vertebrates, nuclear β-catenin is localized to cells at the site of gastrulation where it functions in specification of axial polarity and/or mesendoderm (Wikramanayake et al., 2003). For example, in the sea urchin blastula, nuclear β-catenin becomes localized to vegetal cells, becoming restricted at the onset of gastrulation to a ring of cells around the blastopore (Logan et al., 1999). Similarly, after the mid-blastula transition in Xenopus, β-catenin becomes localized to nuclei in the marginal zone where the Wnt/β-catenin pathway interacts with other posteriorly expressed genes such as brachyury and caudal in specification of posterior identity (Dorsky et al., 2002; Schohl and Fagotto, 2002, 2003). Subsequently, this posterior Wnt/β-catenin center mediates anterior/posterior patterning of the nerve cord and ventral and lateral mesoderm (Kiecker and Niehrs, 2001; Schiomi et al., 2003). In addition to these late roles in axial patterning, in vertebrates, nuclear β-catenin has an early role in specification of dorsal/anterior identity and establishing the dorso/ventral axis (Moon and Kimelman, 1998; Kelly et al., 2000). This early role, in which maternal β-catenin, is concentrated in dorsal nuclei, may be Wnt-independent. However, at the gastrula and neurula stages, Wnt/β-catenin signaling is suppressed anteriorly, thus allowing proper specification of anterior tissues such as the brain and heart (Nordstrõm et al., 2002; Saneyoshi et al., 2002; Lagutin et al., 2003). Such an early role for Wnt/β-catenin in specification of dorsal/anterior structures has not been described outside the vertebrates.

The classic tool in understanding the roles of β-catenin in axial patterning during embryogenesis is Li+, which upregulates Wnt/β-catenin signaling by blocking Gsk3β, thus releasing β-catenin from a complex with Gsk3β and other proteins and allowing its translocation to the nucleus (Hedgepeth et al., 1997). Application of Li+ to embryos typically disrupts axial patterning. However, the phenotype varies depending on the organism and the concentration and time of Li+ application. Continuous exposure of cnidarian embryos to moderate concentrations (10–40 mM) results in elongated embryos with excess endoderm that lack a pharynx or tentacles (Wikramanayake et al., 2003). Similarly treated sea urchin embryos also produce excess endoderm and often exogastrulate (Cameron and Davidson, 1997). For vertebrate embryos, Li+ is typically applied as a 10–20-min pulse of a relatively high concentration (e.g., 300 mM). Such a pulse applied before the mid-blastula transition in frogs or zebrafish or at the early streak stage in the chick dorsal/anteriorizes embryos (Yamaguchi and Shinagawa, 1989; Stachel et al., 1993; Roeser et al., 1999). However, when applied after the mid-blastula transition, Li+ posteriorizes embryos (Yamaguchi and Shinagawa, 1989; Stachel et al., 1993). Manipulating expression of other components of the Wnt/β-catenin pathway has established that a shift from dorso/ventral to anterior/posterior patterning occurs at the midblastula transition in Xenopus (Kinoshita and Asashima, 1995; Schneider et al., 1996; Hamilton et al., 2001; Kiecker and Niehrs, 2001).

To gain insight into the evolution of the molecular mechanisms involved in axial patterning, we have begun an investigation of Wnt/β-catenin signaling in amphioxus, the closest living invertebrate relative of the vertebrates. Early development of the small, relatively non-yolky eggs resembles that of other invertebrates such as echinoderms with regular cleavage and gastrulation by simple invagination (Zhang et al., 1997). In contrast, later development of amphioxus embryos is more like that of vertebrates with formation of a notochord, dorsal, hollow nerve cord, paraxial muscles, and gill slits. Moreover, amphioxus embryos, like those of vertebrates, elongate from a posterior tail bud (Schubert et al., 2001). Not surprisingly, the genetic programs involved in specification and differentiation of these structures are much the same in amphioxus and vertebrates. Thus, the neuroectoderm of the early amphioxus and vertebrate gastrula is marked by the concomitant down-regulation of dll (Xdll-2 in Xenopus) and up-regulation of the early neural plate marker Sox1/2/3 (Dirksen et al., 1994; N.D. Holland et al., 1996; L.Z. Holland et al., 2000b), followed by the down-regulation of BMP2/4 (Panopoulou et al., 1998).

One advantage of amphioxus embryos is the histological simplicity of gastrulation with little movement of cells over the lips of the blastopore (Zhang et al., 1997). Thus, the blastopore, which forms around the equator of the late blastula as invagination begins, is always approximately posterior. The anterior pole is marked from the late blastula by expression of FoxQ2 (Yu et al., 2003) and is offset by about 20–30° from the animal pole (Conklin, 1932). At the mid-gastrula stage, the posterior pole coincides with the dorsal blastopore lip, shifting to the center of the blastopore as it closes. Posterior markers (e.g., Notch, brachyury and several Wnt genes) are expressed around the blastopore (P.W.H. Holland et al., 1995; L.Z. Holland et al., 2001; Zhang et al., 1997; L.Z. Holland, 2002). The first Wnt gene to be expressed is Wnt-8, which turns on at the late blastula stage throughout the mesendoderm, most strongly in a ring around the future blastopore at the ectoderm/mesendoderm boundary (Schubert et al., 2000a; Yasui et al., 2001). Most of the other amphioxus Wnts (including Wnts1 and 3, which also preferentially signal through β-catenin) become expressed around the blastopore during the gastrula and neurula stages (Holland et al., 2000a; Schubert et al., 2001). This suggests that Wnt/β-catenin signaling may be involved in anterior/posterior patterning in amphioxus as it is in sea urchins and vertebrates. To investigate the roles of Wnt/β-catenin in amphioxus development, we determined the distribution of β-catenin in amphioxus eggs and embryos and up-regulated Wnt/β-catenin signaling with Li+. Our results support a late role for Wnt/β-catenin in anterior/posterior patterning of the amphioxus gastrula and neurula. They do not suggest a role for Wnt/β-catenin signaling in axial patterning of amphioxus embryos before the late blastula stage.


Intron/Exon Organization

β-catenin proteins typically have 12 armadillo/β-catenin repeat sequences of 40–42 amino acids each, an n-terminal GSK-3 phosphorylation site and n- and c-terminal transactivation domains (Schneider et al., 2003). Structural conservation of the protein has led to the suggestion that there was a single β-catenin gene in the last common ancestor of metazoans that, like β-catenins in modern metazoans, had both cell adhesion and signalling functions (Schneider et al., 2003). Nonetheless, in spite of structural conservation of the protein, a comparison of intron/exon organization of β-catenin in human and Drosophila showed that none of the intron positions were shared between the two organisms (Nollet et al., 1996). To determine if this lack of conservation of the genomic structure is general among metazoans, we extended the comparison to include β-catenins from the cnidarian Nematostella vectensis, the sea urchin Strongylocentrotus purpuratus, and amphioxus Branchiostoma floridae. The number of introns in the coding region of β-catenin varies from 5 in Drosophila to a high of 17 in amphioxus. Human β-catenin has 13 introns in the coding region, sea urchin β-catenin has 16, while the cnidarian gene has 10. Homologies of intron positions in and near the armadillo/β-catenin repeats can easily be determined due to the high level of sequence identity among species. However, because of sequence divergence, they are less readily determined at the 3′ and 5′ ends of the coding region. As would be expected, the intron positions are highly conserved among the three deuterostome β-catenins, with 11 shared between human and amphioxus, 10 between human and sea urchin, and 12 between amphioxus and sea urchin (Fig. 1). Although only one intron is shared between Drosophila β-catenin and those of amphioxus and sea urchin, there are eight shared between β-catenins of one or more deuterostomes and that of the cnidarian. These results suggest that β-catenin has not only been highly evolutionarily conserved at the protein level but that the β-catenin gene in the common ancestor of cnidarians and bilaterians had at least 8 introns, which have been conserved in the deuterostome lineage. In contrast, Drosophila β-catenin gene has lost numerous introns.

Figure 1.

Intron positions (triangles) in β-catenins from human, Homo sapiens (H.s.), amphioxus, Branchiostoma floridae (B.f.), sea urchin, Strongylocentrotus purpuratus (S.p.), fruitfly Drosophila melanogaster (D.m.), and cnidarian Nematostella vectensis (N. v.). The number of amino acids in each protein is indicated on the right. All the β-catenins have 12 conserved armadillo/β-catenin repeats (numbered).

The Distribution of β-catenin in Early Amphioxus Embryos

The distribution of nuclear β-catenin generally reveals the cells in which the Wnt/β-catenin pathway is operating. Therefore, to investigate the roles of this pathway in amphioxus development, we determined the pattern of accumulation of nuclear β-catenin protein in eggs and embryos through the late neurula and up-regulated Wnt/β-catenin with Li+ both before and after the mid-blastula stage. In the unfertilized amphioxus egg, which is arrested at second meiotic metaphase, maternal β-catenin is distributed relatively uniformly within the cytoplasm (Fig. 2A; see also Fig. 6A). However, by the two-cell stage (1 hr post-fertilization), it becomes preferentially localized to the animal pole, which is close to the future anterior pole (Conklin, 1932; Fig. 2B; diagrammed in Fig. 6). β-catenin remains strictly cytoplasmic at the 4-cell (Fig. 2C) and 8-cell stages (Fig. 2D; see also Fig. 6A). In some embryos, there is a suggestion of some asymmetry in the distribution of cytoplasmic β-catenin at the 4-cell stage, but by the 8-cell stage, no such differences are apparent (Fig. 2D).

Figure 2.

Distribution of β-catenin protein in normal amphioxus embryos. Scale bars = 50 μm. A: Unfertilized egg with polar body (arrow) at animal pole. B: First cleavage with deeper furrow (top) marking animal pole. The polar bodies have been lost during labelling. C: A 4-cell stage. β-catenin is undetectable in the nuclei. D: An 8-cell embryo (dissociated blastomeres). E: A 16-cell stage with β-catenin in all nuclei. F: A 32-cell stage with interphase nuclei. G: A 32-cell stage with β-catenin associated with the mitotic spindle. H: Early blastula with β-catenin in nuclei and associated with cell boundaries. I: Late blastula with nuclear β-catenin (less conspicuous in mitotic cells). J: Animal pole view of early gastrula showing nuclear β-catenin in all nuclei of the animal hemisphere (less conspicuous in mitotic cells). K: Vegetal pole view of the embryo in J. The nuclei of the presumptive mesendoderm have lost β-catenin. L: Mid-gastrula in side view with blastopore at right. Nuclear β-catenin is reduced in anterior ectoderm. M: Mid-gastrula in blastopore view with highest level of nuclear β-catenin in circumblastoporal ectoderm. N: Early neurula in dorsal view. β-catenin is down-regulated in the neural plate. Highest levels of nuclear β-catenin are around the blastopore (at right) (arrowhead shows level of cross-section in P). O: Sagittal section of the neurula in N. Nuclear β-catenin is absent from the mesendoderm and neural plate, concentrated in ectodermal cells around the blastopore and elsewhere in the ectoderm largely restricted to the inner nuclear border. P: Cross-section of early neurula through level of arrowhead in N. The ectoderm bordering the neural plate (arrow) has high levels of nuclear β-catenin and is beginning to migrate over the neural plate. Q: Mid-neurula in dorsal view with neural plate mostly overgrown by non-neural ectoderm. The leading-edge cells have lost nuclear β-catenin. R: Side view of mid-neurula. Nuclear β-catenin is largely limited to the anterior and posterior ectoderm. S: Higher magnification of the anterior end of the neurula in R showing β-catenin in ectodermal nuclei. T: Sagittal section of mid-neurula. Nuclear β-catenin is most conspicuous in the anterior ectoderm (arrow) and at the ventral lip of the blastopore (arrowhead). U: Surface view of anterior end of a mid-neurula. Scattered ectodermal cells have conspicuous nuclear β-catenin. V: Late neurula with little nuclear β-catenin except in the extreme anterior and posterior ectoderm.

Figure 6.

Diagram of distribution of β-catenin in (A) normal amphioxus embryos and (B) embryos treated with Li+ at the late blastula. Arrows indicate the anterior/posterior axis. The first polar body is shown for the unfertilized egg. Both polar bodies are generally lost from embryos during antibody-labelling. Nuclei are shown as small circles. The intensity of color is proportional to the intensity of staining for β-catenin.

At the 16-cell stage, β-catenin accumulates in the nuclei of all blastomeres (Fig. 2E; see also Fig. 6A). In line with the even distribution of cytoplasmic β-catenin at the 8-cell stage, there is no apparent difference among the blastomeres in the concentration of nuclear β-catenin. Throughout the blastula stage, β-catenin remains uniformly localized to the nuclei of all the cells (Fig. 2F–I; see Fig. 6A). Scattered cells in Figure 2I that appear to have less nuclear β-catenin are, in fact, beginning to undergo mitosis and, consequently, have slightly more diffuse β-catenin in the nucleus. Even so, some β-catenin stays associated with the mitotic spindle (Fig. 2G). β-catenin also becomes concentrated at the cell boundaries at the early blastula stage as the blastomeres become more tightly adherent to one another (Fig. 2G,H). At the onset of gastrulation (4 hr after insemination), the nuclei of cells in the animal half of the embryo (presumptive ectoderm), retain nuclear β-catenin; however, it largely disappears both from the nuclei and cytoplasm of cells in the vegetal half, the presumptive mesendoderm (Fig. 2J,K; see also Fig. 6A).

To determine if nuclear β-catenin is involved in axial patterning in amphioxus embryos, we exposed embryos to seawater with Li+ partly or completely substituted for Na+ [10–486 mM Li+ (5–100% of the Na+ in normal sea water)] for varying durations during cleavage and blastula stages. The conditions were chosen to correspond to those typically used to upregulate Wnt/β-catenin signaling in both sea urchins (lower concentrations, longer duration) and Xenopus (higher concentrations, shorter duration). Continuous exposure to 10 mM Li+ from the 32–64 cell stage had no obvious effect on development. At 20 mM Li+, embryos appeared normal through the neurula stage; however, by the early larval stage (36 hr) although otherwise normal, most lacked the mouth and first gill slit. Continuous exposure to higher concentrations of Li+ inhibited hatching and resulted in highly deformed embryos as did 15–30-min pulses of 300–486 mM Li+-seawater applied at the very early blastula stage. Li+ applied for 30 min between the early and mid-blastula stages either had no effect (lower concentrations), inhibited elongation of the embryo (mid-concentrations), or resulted in exo-gastrulae (higher concentrations) (Fig. 4A). The exogastrulae resembled bells with curled edges because the endoderm did not maintain continuity at the vegetal pole and curled up around the ectoderm (Fig. 4A).

Figure 4.

Amphioxus embryos treated with Li+ (400 mM) for 30 min at the early blastula stage. Scale lines = 50 μm. Anterior is at left in B–G. A: Exogastrula resulting from treatment at 2 hr after fertilization (early blastula). The mesoderm has torn and its edges have curled up. B–G: Embryos treated with Li+ at 2.5 hr after fertilization. B: Elongated embryo 24 hr after fertilization, lacking anterior structures but with tail fin (double-headed arrow). C: Dorsal view of 24-hr embryo with curled posterior end. Somites (s) and notochord (nt) appear normal. D: A 24-hr embryo. Side view. Notochord (nt) is thickened and disorganized and the tail is twisted. The nerve cord (n) appears grossly normal. E: A 36-hr larva. Notochord (nt) is thickened and disorganized. The pharynx (p) lacks the mouth and first gill slit. Two pigment spots (arrows) in the nerve cord at this stage are unusual. The tail fin (double-headed arrow) and posterior portion of the gut (g) appear normal. F: A 36-hr larva. The anterior end is somewhat shortened. The notochord is thickened and bent. The pharynx (p) lacks mouth and first gill slit. The nerve cord has formed the first pigment spot (arrow) normally. Tail fin (double-headed arrow) is normal. G: A 36-hr larva. Both the anterior and posterior pigment spots (arrows) in the nerve cord are normal. The larva is shortened with thickened notochord (nt). The tail fin (double-headed arrow) is normal but pharyngeal structures are absent.

Embryos in which elongation was inhibited generally hatched late. Some lacked all anterior structures and had defects in epithelial fusion (Fig. 4B). Nevertheless, the tails generally formed normally (Fig. 4B, double arrows). The anterior ends of others appeared normal at the late neurula stage, but their posterior ends were curled (Fig. 4C,D). However, by 36 hr, most of these larvae had a tail fin posteriorly and heads that looked relatively normal except for deformed or missing pharyngeal structures such as gill slits and mouth (Fig. 4E–G). Nevertheless, the trunks of these larvae were generally bent and twisted with notochords that were often thicker than normal with disoriented cells (the amphioxus notochord consists mainly of muscle cells [Flood, 1975]). Posterior truncations were rare. These results show that the major effects of Li+ applied before the mid-blastula stage are either prevention of mesendoderm invagination or, if the mesendoderm does invaginate, inhibition of anterior/posterior axis elongation. Thus, down-regulation of β-catenin in the mesendoderm appears to be required for gastrulation and/or specification of the mesendoderm or the ectoderm/mesendoderm boundary.

At Gastrulation, a Wnt/β-Catenin Signaling Center Around the Blastopore Confers Posterior Identity

During the mid-late gastrula stage, nuclear β-catenin decreases in ectodermal cells near the animal pole, but remains highly concentrated in ectodermal cells around the blastopore (Fig. 2L,M). Throughout the gastrula stage, β-catenin remains undetectable in either nuclei or cytoplasm in the mesendoderm (Figs. 2L, 6A). As neurulation begins, β-catenin becomes undetectable in both the cytoplasm and nuclei of all the neural plate cells (Figs. 2N,O, 6A), suggesting that down-regulation of β-catenin may be necessary for formation of the neural plate and/or maintenance of the neuroectoderm. In the early to mid-neurula, β-catenin remains localized to nuclei of the non-neural ectoderm. In most of these cells, β-catenin becomes restricted to the inner periphery of the nuclear envelope, as shown by electron microscopy (Figs. 2O, 3). This is probably indicative of export of β-catenin from the nucleus to the cytoplasm, where it is degraded (Henderson, 2000; Wieschens and Fagotto, 2000). However, the level of nuclear β-catenin remains high in two regions: around the blastopore, suggesting that there is a posterior Wnt/β-catenin signaling center, and in non-neural ectoderm immediately adjacent the neural plate (Figs. 2N–Q, 6A). Before the neural plate rounds up, the non-neural ectoderm detaches from the periphery of the neural plate and migrates over the open neural plate, where it fuses in the dorsal midline. This sharp cut-off of nuclear β-catenin may be involved in specification of the neuroectoderm/non-neuroectoderm boundary. At the mid-neurula stage, the only cells with nuclear β-catenin are at the ventral, posterior side of the nearly-closed blastopore, in the ventral, anterior end of the embryo (Fig. 2S,T) and scattered in the ectoderm (Fig. 2U), suggesting that Wnt/β-catenin signaling is involved in specifying the identity of some ectodermal cells. By the late neurula, nuclear β-catenin is concentrated only in the ectoderm at the extreme anterior and posterior ends of the embryo (Fig. 2V).

Figure 3.

Electron micrograph of an unstained section through the ectoderm of an amphioxus neurula at the same stage as in Figure 2 immunostained for β-catenin. The dense deposits are Ni+- enhanced DAB and show β-catenin concentrated at the inner periphery of the nuclei. Magnification = 10,000×.

To determine whether localization of β-catenin to nuclei around the blastopore promotes posterior development, we transferred embryos to seawater with 85–100% of the Na+ substituted by Li+ (400–484 mM Li+) during the 30 min just prior to the onset of gastrulation. Embryos were returned to normal seawater before they began to gastrulate. At the highest concentrations of Li+, development arrests at the gastrula stage. At the lower concentrations, gastrulation initially appears normal (Figs. 5A, 6B). However, the blastopore does not close and the neural plate does not form. Instead, the gastrula flattens and elongates on a line drawn between the posterior pole and the opposite lip of the blastopore. Then, whereas normal embryos elongate with the lips of the blastopore at the posterior pole, in Li+-treated embryos, only one side of the ectoderm elongates (compare Figs. 2L and 5A). Embryos do not elongate from the lip of the blastopore, but slightly lateral to it (Figs. 5C–I, 6B). The elongating side is probably ventral/posterior, although we cannot rule out the possibility that it is dorsal/posterior, since the polar bodies are not retained during labelling and since these embryos never express neuroectodermal markers (see below). The opposite side of the Li+-treated embryos does not elongate substantially. Sometimes the archenteron disappears as the mesendoderm apparently pulls out of the archenteron (Fig. 5G). At other times, a small archenteron persists (Figs. 5H–J, 6B). In some embryos, the remains of the blastocoel contain a loose assortment of cells and extracellular material (Fig. 5I,L). In others, there is some differentiation of mesodermal tissues such as somites and a partial notochord (Fig. 5M,O).

Figure 5.

Amphioxus embryos treated with Li+ for 30 min at the late-blastula stage (3.5 hr after fertilization). Scale lines = 50 μm. A–L show distribution of β-catenin. A: Five-hr gastrula. Nuclear β-catenin concentrated in posterior blastoporal lip. Inset: Higher magnification of a section of the ectoderm in A showing strong nuclear label. Arrowhead shows border between ectoderm and mesendoderm. Scale = 5 μm. B: Blastopore view of gastrula at same stage as in A. β-catenin concentrated in ectodermal nuclei on one side of embryo with posterior identity. C: A 5.5-hr gastrula. Blastopore-view of a slightly later stage than in A. Nuclear β-catenin concentrated at the posterior and lateral sides of the blastopore. D: Six-hr gastrula. Blastopore view of an embryo at stage similar to that in C. E: Side view of embryo in D. All ectodermal cells have nuclear β-catenin. F: Longitudinal section of embryo in E. Nuclear β-catenin most prominent in ectoderm around the blastopore. G: Seven-hr embryo. Blastopore view. β-catenin is concentrated in the ectodermal nuclei. H: Ten-hr embryo. All ectodermal nuclei have β-catenin, most concentrated in the posterior extension. I: Longitudinal section of embryo in H. The endoderm lacks nuclear β-catenin. A loose aggregate of cells occurs between the endoderm and ectoderm. J: twelve-hr embryo. Blastopore at right. Posterior extension has a high concentration of β-catenin. K: Fifteen-hr embryo, which is 400 μm long. The rectangle indicates section in L. L: Section indicated by the rectangle in K. β-catenin concentrated at the inner edges of nuclei. Inside the embryo is an amorphous aggregate of extracellular material including some cells. M: Fifteen-hr embryo. In situ hybridization for brachyury. This embryo has structures resembling somites. Brachyury heavily expressed in the posterior extension and more weakly in the somites and midline of longitudinal axis. N: Twelve-hr embryo. in situ hybridization for Wnt5. Wnt5 strongly expressed at posterior tip. O: Twelve-hr embryo. In situ hybridization for Wnt1. Wnt1 strongly expressed at posterior tip. This embryo has some internal structure resembling somites. P: Twelve-hr embryo hybridized for Wnt11. Wnt11 expressed at posterior tip.

Not surprisingly, Li+ alters the distribution of β-catenin. At the gastrula stage, the level of cytoplasmic β-catenin remains relatively high, and both the mesendoderm and ectoderm, especially on the posterior/ventral side of blastopore, retain nuclear β-catenin (Figs. 5A,B, 6B). Ectodermal cells at the animal pole initially have less nuclear β-catenin than those around the blastopore (Fig. 5F), but as development proceeds, the concentration of β-catenin increases in all the ectodermal nuclei (Fig. 5H,I). Nevertheless, in spite of previous Li+ treatment, β-catenin eventually became undetectable in both the cytoplasm and nuclei of the mesendoderm (Figs. 5F,G,I, 6B). By 15 hr of development, when the embryos would normally be at the mid-neurula stage, the Li+-treated embryos are over 400 μm long compared to about 150 μm for normal embryos (compare Figs. 5K to 2Q). This suggests a role for Wnt/β-catenin signaling in tissue elongation. It is notable that in cells near the center of these embryos, β-catenin was localized to the inner border of the nuclear envelope (Fig. 5L). However, at the extremities of the embryo, particularly in the one that is growing out, it is distributed throughout the nucleus.

The posterior identity of the embryonic outgrowth is shown by hybridization of Li+-treated embryos with probes for posteriorly expressed genes. Brachyury (AmBra2), Wnt5, Wnt1, and Wnt11, which are normally expressed around the lips of the blastopore (P.W.H. Holland et al., 1995; L.Z. Holland et al., 2000a; Schubert et al., 2000b, 2001), are all expressed at the tip of the projection (Fig. 5M–P). In an embryo with some differentiation of mesodermal tissues, Ambra2 is also expressed along the midline on one side of the embryo in cells that probably have notochord identity. However, in embryos lacking mesodermal structures, the normal anterior domains of expression for AmphiWnt11 are missing (Fig. 5O). In addition, these embryos fail to express an anterior marker (FoxQ2) (Yu et al., 2003), showing that the anterior end of the Li+-treated embryo lacks anterior identity. Moreover, localization of β-catenin to the nuclei of all the ectodermal cells (Fig. 5H,I) and the lack of expression of the early neural plate marker Sox1/2/3 show that all of the ectoderm is specified as non-neural. These results demonstrate that treatment of embryos with a high concentration of Li+ for 30 min before gastrulation strongly posteriorizes amphioxus embryos. Anterior identity is lost and the dorsal ectoderm does not become specified as neural plate. We conclude that Wnt/β-catenin signaling confers posterior identity and that down-regulation of β-catenin is essential for specification and/or maintenance of anterior and neuroectodermal identities.


Evolution of Intron/Exon Organization in β-Catenin

The lack of any shared intron positions between Drosophila and human β-catenins led to the suggestion that either the gene in the common ancestor completely lacked introns or that there was a combination of intron insertion and loss during evolution that erased the original intron/exon organization (Nollet et al., 1996). Our analysis suggests that neither scenario is accurate. The high degree of conservation of intron positions between β-catenins of the cnidarian Nematostella and the deuterostomes argues for an ancestral bilaterian β-catenin gene with at least eight introns. In the deuterostome lineage, only a few introns have been gained. One, that shared among β-catenins of Drosophila and the two invertebrate deuterostomes, may have evolved de novo in the ancestral bilaterian, although a loss in Nematostella cannot be ruled out. Similarly, four introns shared between the invertebrate deuterostomes that are not present in Nematostella or Drosophila may have been gained after the cnidarian/deuterostome split, while the two intron positions common only to amphioxus and humans may have arisen de novo in their common ancestor. Loss appears to be equally uncommon in deuterostomes. Only one intron has evidently been lost from sea urchin and amphioxus β-catenin and only three in the human. In contrast, in the Drosophila lineage there has been a large loss of introns. None of the eight introns shared between the cnidarian and the deuterostome β-catenins is present in Drosophila. The only conserved intron position in Drosophila β-catenin is one that is shared with sea urchin and amphioxus. This massive loss of introns has been observed in a number of other genes not only in Drosophila but in other ecdysozoans as well (Roy and Gilbert, 2005). Our results suggest that this loss occurred after the divergence of protostomes from the cnidarian and deuterostome lineages. However, where in the protostome lineage this loss of introns occurred is not certain, since genome sequences of basal ecdysozoans such as onychophorans and of lophotrochozoans such as annelids and molluscs are not yet available.

The Distribution of β-Catenin During Early Development Is Not Conserved Among Deuterostomes

The distribution of β-catenin after fertilization and during cleavage stages differs considerably amongst the deuterostomes, suggesting that the roles of this gene in early development are not evolutionarily conserved. In amphioxus, cytoplasmic β-catenin becomes concentrated at the animal pole at the first cleavage, suggesting a possible role in maintaining animal/vegetal polarity during early cleavage, but by the 8-cell stage, distribution is apparently uniform. The pattern of cytoplasmic β-catenin in a second species of amphioxus, Branchiostoma belcheri, is the same as in B. floridae through the two-cell stage (Yasui et al., 2002). However, these authors described a persistent asymmetry of cytoplasmic β-catenin until the mid-gastrula, the latest stage examined. Although this could reflect a species difference, it may be due to differences in affinity of the antibodies used in the two studies for the phosphorylated versus the non-phosphorylated forms of β-catenin or to the lack of affinity purification of the antiserum used for B. belcheri, which on Western blots appears to recognize some additional proteins (Yasui et al., 2002; Oda et al., 2003). Indeed, in the B. belcheri study compared to ours, there was a much higher level of cytoplasmic label, which tended to mask the nuclear label (Yasui et al., 2002). In sea urchins, the distribution of cytoplasmic β-catenin is also uniform at the 8-cell stage, although a few embryos appear to have higher levels of cytoplasmic β-catenin in vegetal blastomeres at the 16-cell stage (Logan et al., 1999). In contrast to both sea urchins and amphioxus, in Xenopus, cytoplasmic β-catenin, together with other components of the canonical Wnt-signaling pathway and microtubules, is localized to the future dorsal side of the embryo by the cortical rotation.

In spite of differences in cytoplasmic distribution of maternal β-catenin, nuclear localization of β-catenin occurs in all three of the above deuterostomes after the 4th cleavage (16-cell stage), although in tunicates, it occurs one cleavage later (Imai et al., 2000). Not surprisingly, nuclear β-catenin is most prominent in the cells with the highest concentration of cytoplasmic β-catenin at the 8-cell stage. Thus, in sea urchins, β-catenin becomes preferentially localized to vegetal nuclei (micromeres and macromeres), where it functions in specification of the vegetal plate and formation of mesendoderm (Wikramanayake et al., 1998; Logan et al., 1999; reviewed in Brandhorst and Klein, 2002). Nuclear β-catenin induces secondary mesenchyme by signaling through Notch, which in turn becomes concentrated at the boundary between secondary mesenchyme and presumptive endoderm (Sherwood and McClay, 1997; Sweet et al., 1999; McClay et al., 2000). In contrast, in amphioxus, there is uniform nuclear localization of β-catenin in all cells from the 16-cell stage through the late blastula/early gastrula, which correlates with the evident absence of either a dorsalizing or vegetalizing effect of Li+ when applied before the mid-blastula stage. While a negative result may be due to incorrect timing or concentration of Li+, this result together with the lack of preferential localization of β-catenin to dorsal or vegetal nuclei suggests that in amphioxus, neither dorso/ventral polarity nor mesendodermal identity is established by signaling via β-catenin. In fact, the sharp reduction of both cytoplasmic and nuclear β-catenin in presumptive mesendoderm suggests that, in contrast with sea urchins, absence of β-catenin signaling may be necessary for mesendoderm specification in amphioxus. The down-regulation of β-catenin in presumptive mesendoderm at the late blastula stage correlates with up-regulation of the Wnt/β-catenin inhibitor Dickkopf1/2/4 in the same cells (our unpublished data), suggesting that, as in vertebrates (González-Sancho et al., 2005), induction of Dickkopf down-regulates the Wnt/β-catenin pathway. In ascidian tunicates, as in sea urchins, β-catenin accumulates in nuclei of vegetal blastomeres, specifying them as endoderm (Imai et al., 2000; Satou et al., 2001), although it does not seem to function in patterning either the dorso/ventral or anterior/posterior axis. Instead, endodermal cells with nuclear β-catenin induce the notochord, which can be respecified as endoderm by application of Li+ (Yoshida et al., 1998; Imai et al., 2001). In Xenopus, unlike other deuterostomes, β-catenin becomes preferentially incorporated at the 16-cell stage into dorsal nuclei where it functions in specification of dorsal identity and activation of organizer genes (Schneider et al., 1996; Larabell et al., 1997; Miller et al., 1999; Kelly et al., 2000; Schohl and Fagotto, 2002; Xanthos et al., 2002). Blockage of signaling via β-catenin in early embryos prevents formation of dorsal/anterior structures (Heasman et al., 2000), while upregulation of β-catenin signaling with Li+ has the opposite effect. This role of nuclear β-catenin in dorsal/anterior specification in Xenopus does not appear to be comparable to the role of β-catenin signaling in oral/aboral axis patterning in sea urchins, which is evidently mediated by localized β-catenin in the vegetal plate (Wikramanayake and Klein, 1997; Brandhorst and Klein, 2002).

A Posterior Wnt/β-Catenin Signaling Center Around the Blastopore Is Conserved Among Deuterostomes and Acts in Patterning the Anterior/Posterior Axis

Although early roles of nuclear β-catenin in deuterostomes appear to be quite divergent, later roles in patterning along the anterior/posterior (animal/vegetal) axis are evolutionarily conserved except in the tunicates, in which development is considerably modified in association with determinate cleavage. In all the other deuterostomes studied including amphioxus, a Wnt/β-catenin signaling center develops around the blastopore. It is characterized by elevated levels of nuclear β-catenin as well as Notch and brachyury and functions in specification of posterior identity. At the onset of gastrulation in amphioxus, cells at the mesendoderm/ectoderm boundary have high levels of nuclear β-catenin and express Brachyury and Wnt8 (Zhang et al., 1997; Holland et al., 2000a; Schubert et al., 2001; Yasui et al., 2001). By the mid-gastrula, Wnt1 and then Notch, Wnt3 and Wnt5 also turn on around the blastopore (P.W.H. Holland et al., 1995; L.Z. Holland et al., 2001; Schubert et al., 2001). Similarly, in the late sea urchin blastula, primary mesenchyme and veg2 cells lose nuclear β-catenin leaving a high concentration in a ring of endoderm cells at the endoderm/ectoderm boundary (Logan et al., 1999). Wnt-8 and Wnt-1 are both expressed in vegetal cells: Wnt-8 in the vegetal plate and Wnt-1 in cells adjacent the archenteron (Wikramanayake et al., 1998; Ferkowicz and Raff, 2001). Notch, which controls the localization of nuclear β-catenin in this region, and brachyury are also concentrated in cells at the endoderm/ectoderm boundary (Sherwood and McClay, 1997, 1999, 2001; Gross and McClay, 2001; reviewed in Brandhorst and Klein 2002).

Posterior Wnt/β-catenin signaling is also present throughout the vertebrates. Wnt/β-catenin signaling is concentrated at the posterior end of the zebrafish gastrula, and in the mouse and chick, in the primitive streak and node (Dorsky et al., 2002; Mohamed et al., 2004; Schmidt et al., 2004). In the late Xenopus blastula, there is a ring of high levels of nuclear β-catenin around the marginal zone (the future mesoderm) that later localizes to the dorsal and ventral lips of the blastopore (Schohl and Fagotto, 2002). The patterns of XWnt-8, brachyury (Xbra) and XWnt-11 are similar to those of nuclear β-catenin (Smith and Harland, 1991; Christian and Moon, 1993; Lemaire and Gurdon, 1994). During the neurula stage, expression of XWnt-8 ceases, while XWnt-3a and caudal also turn on around the blastopore (Epstein et al., 1997; McGrew et al., 1997; Beck and Slack, 1998, 1999; Ikeya and Takada, 2001; Lickert and Kemler, 2002). Brachyury, directly regulated by Wnt/β-catenin, stays turned on posteriorly during the neurula and tailbud stages (Gont et al., 1993; Yamaguchi et al., 1999; Arnold et al., 2000; Vonica and Gumbiner, 2002). Components of Notch-signaling (XNotch and X-Delta-1) are also expressed in posterior tissues from gastrulation onwards (Beck and Slack, 1998).

Manipulation of Wnt/β-catenin signaling shows that in amphioxus, sea urchins, and vertebrates, Wnt/β-catenin signaling functions in conferring posterior identity to tissues around the blastopore and in patterning the anterior/posterior axis (reviewed in Huelsken and Birchmeier, 2001; Kiecker and Niehrs, 2001; Brandhorst and Klein, 2002). For amphioxus, we have shown that a pulse of Li+ at the late blastula stage slows the loss of nuclear β-catenin from the mesendoderm and increases its concentration in posterior tissues, which elongate greatly and express the posterior markers Wnt1, Wnt5, Wnt11, brachyury. The failure of β-catenin to remain expressed in the mesendoderm after Li+ treatment is probably due to the concomitant upregulation of Dickkopf 1/2/4 in these cells (our unpublished data), which, after removal of the Li+, down-regulates β-catenin signaling. Enhanced degradation of β-catenin by Dickkopf-1 on withdrawal of Li+ has similarly been observed in human mesenchymal stem cells in culture (Gregory et al., 2005). Li+-treated amphioxus embryos also lose anterior identity as shown by the lack of expression of FoxQ, an exclusive marker of the anteriormost ectoderm in normal embryos (Yu et al., 2003). In addition, these embryos lose neural plate identity as shown by lack of expression of the neural plate marker Sox1/2/3 and continued expression of β-catenin in the dorsal ectoderm. The extreme elongation of amphioxus embryos treated with Li+ at the late blastula is reminiscent of Xenopus animal caps treated with Li+ or overexpressing XWnt8, β-catenin, or XWnt-11, which preferentially signals through the Wnt/JNK pathway (Tada and Smith, 2000; Kühl et al., 2001; Tada et al., 2002), and could be due to the induction of convergent extension movements. Conversely, Wnts preferentially signaling through the Wnt/Ca++ pathway inhibit convergent extension (Kühl et al., 2001; Choi and Han, 2002). In amphioxus, Wnt(s) signaling though all three pathways—the Wnt/β-catenin (Wnt-1), Wnt/Ca++ (Wnt-5), and Wnt/JNK (Wnt-11)—are expressed around the blastopore of normal embryos and at the elongating tip of the Li+-treated ones. Thus, it is likely that convergent-extension is involved in elongation of both Li+-treated and normal embryos.

In sea urchins, nuclear Wnt/β-catenin signaling from vegetal cells acts together with Notch and brachyury to pattern the animal/vegetal axis and control the position of the ectoderm/endoderm boundary (Emily-Fenouil, et al., 1998; Wikramanayake et al., 1998; Huang et al., 2000; Vonica et al., 2000; Howard et al., 2001). Up-regulating Wnt/β-catenin signaling with Li+ or injection of dominant/negative forms of GSK3-β vegetalizes sea urchin embryos, reducing the number of animal cells, increasing the proportion of endoderm, and shifting expression patterns of several genes including Brachyury toward the animal pole (Ghiglione et al., 1993; Cameron and Davidson, 1997; Emily-Fenouil et al., 1998; Gross and McClay, 2001). Blocking Wnt/β-catenin signaling has the opposite effect, inhibiting formation of endoderm (Logan et al., 1999; Huang et al., 2000) and blocking brachyury expression (Gross and McClay, 2001).

In vertebrates, Wnt/β-catening signaling is also essential for proper anterior/posterior patterning. Mouse embryos deficient in β-catenin express neither the anterior markers Hex and Otx nor the posterior marker brachyury (Huelsken et al., 2000). In Xenopus, the effects of Li+-application at the late blastula (Fredieu et al., 1997) are remarkably like those of comparable experiments in amphioxus. Forebrain and midbrain markers are not expressed, the archenteron does not form, and blastopore closure is delayed or incomplete (Fredieu et al., 1997). Experimental evidence shows that the Wnt/β-catenin and Notch pathways together with brachyury and caudal constitute a posterior signaling center that specifies and maintains posterior identity (Beck and Slack, 1998, 1999; reviewed in Gamse and Sive, 2000; Kiecker and Niehrs, 2001). After the mid-blastula transition, up-regulating β-catenin signaling (i.e., injection of XWnt-8, Li+, β-catenin overexpression) inhibits dorsal-anterior development, resulting in embryos with a shortened dorsal axis, no heads or notochords, and enlarged somites (Yamaguchi and Shinagawa, 1989; Cui et al., 1995; Kinoshita and Asashima, 1995; Fredieu et al., 1997; Kao and Elinson, 1998; Domingos et al., 2001; Hamilton et al., 2001). Gene markers of ventral and lateral mesoderm are concomitantly up-regulated (Hamilton et al., 2001). In addition, over-expression of XWnt-3a together with active Notch induces ectopic tails in animal cap explants grafted onto the neural plate (Beck and Slack, 1998, 1999), while over-expression of Disheveled, a component of both the Notch and Wnt-signaling pathways, posteriorizes neural tissue and activates posterior markers such as brachyury (Itoh and Sokol, 1997). Conversely, blocking Wnt/β-catenin or Notch signaling (e.g., injection of a dominant-negative XWnt-8, antisense Wnt-8 morpholino oligonucleotide) expands anterior neural fates and/or causes posterior defects (Takada et al., 1994; McGrew et al., 1997; Beck and Slack, 2002; Erter et al., 2001).


Our results suggest four main roles for Wnt/β-catenin signaling in early development of amphioxus: first, in formation and/or maintenance of the ectoderm/mesendoderm boundary and in specification of posterior identity; second, in formation of the neuroectoderm/non-neuroectoderm boundary; third, in tissue elongation; and fourth, in specification of cell identity in the ectoderm. In amphioxus, we found no evidence that nuclear β-catenin participates in establishing the dorso/ventral axis. Such a role for localized β-catenin in dorsal nuclei in vertebrates may represent a co-option of the pathway as a result of the evolution of yolky eggs, and precocious formation of ventral mesoderm (Kozmik et al., 2001). In amphioxus, the only consistent effects of upregulation of Wnt/β-catenin signaling before the mid-blastula stage are to inhibit development of pharyngeal structures and to impair elongation of the embryo. Inhibition of elongation suggests that Wnt/β-catenin signaling may oppose convergent/extension in amphioxus as it does in other organisms. What is evolutionarily conserved in most deuterostomes is a role for Wnt/β-catenin signaling in cells around the blastopore in establishing the ectoderm/mesendoderm boundary and giving the cells posterior identity. In amphioxus and vertebrates, this signaling center is involved in patterning along the anterior/posterior axis and in forming part of the tail organizer.


Embryo Culture and Li+ Administration

Adult amphioxus, Branchiostoma floridae, were collected during summer in Old Tampa Bay, Florida. Eggs and sperm were obtained by electrical stimulation, and embryos and larvae were raised in laboratory cultures (Holland and Holland, 1993b). The salinity of Old Tampa Bay fluctuates slowly according to the amount of local rainfall, and the animals adapt. Consequently, the concentration of Li+ required to obtain a particular morphology depends on the tonicity of the seawater to which the animals are adapted. We adjusted the tonicity of Li+ seawater, to that of the seawater at the collection site by mixing full-strength Li+ seawater (484 mM LiCl, 29 mM MgSO4, 27 mM MgCl2, 10 mM KCl, 2.4 mM NaHCO3, 10 mM CaCl2) with distilled water. To obtain a particular concentration of Li+, Li+-seawater was mixed with seawater from the collection site. This procedure avoids effects due to altered tonicity.

Cloning of β-Catenin cDNA and Genomic Analysis

A 1,340-bp partial cDNA clone including about 60% of the coding region of amphioxus β-catenin was obtained by PCR with primers 5′-GACCGCAGCTGGCGTCTGGCG-3′and 5′-GCAGGTTTACAATATGATAAGAC-3′ and cloned into the pCR 2.1 vector (Invitrogen, Inc., Carlsbad, CA). This clone was used to design nested primers that were paired with a vector-specific primer to amplify the entire cDNA as overlapping 3′ and 5′ clones from a gastrula through neurula cDNA library in pBluescript sk. Gene-specific primer sequences are as follows: Forward primers 5′-TGCGGATCCCAGGCCCTTGGTCAGCAC-CTGTCCCAC-3′and 5′-TGCGGATCCCCCCGTCTGGTCCAGAACTGCCTCTGG-3′. Reverse primers 5′-TAGGAATTCCGCGCCCGTTGCAG-GTCAGGTTCGACAG-3′ and 5′-TAGGAATTCCgGCAGCGCACGTCACAATGTTGATGTC-3′. The complete amphioxus β-catenin sequence is deposited in GenBank as accession number DQ013259. Intron/exon splice sites of amphioxus and cnidarian β-catenins were determined by comparing available cDNA sequences to sequences in the trace archives of GenBank. Because the cDNA sequence for the cnidarian Nematostella vectensis in GenBank (accession no. AF08421) is incomplete, the 5′ end of the β-catenin of N. vectensis and the intro/exon splice sites were determined from a blast search of the N. vectensis sequences in the trace archives of GenBank with the nucleotide sequence of amphioxus β-catenin. Nucleotide conservation between the β-catenin sequences of B. floridae and N. vectensis is surprisingly high (75–80% identity). For the intron/exon organization of β-catenin from the sea urchin Strongylocentrotus purpuratus, the β-catenin cDNA sequence from another sea urchin Lytechinus pictus (accession no. U34814) was blasted to the genome sequences from S. purpuratus available at

Antibody Labelling and In Situ Hybridization

For antibody labelling of embryos, an affinity-purified, polyclonal antibody generated against the amino-terminal 173 amino acids of β-catenin of the sea urchin Lytechinus variegatus was used (Miller and McClay, 1997). The amino acid sequences of amphioxus and L. variegatus β-catenin are 93% identical over the c-terminal half of this region, and 30% identical over the amino-terminal half. This antibody is known to recognize β-catenin in tunicates (Imai et al., 2000), even though the comparable percentages of identities with L. variegatus β-catenin are only 60 and 17%. The antibody was used at a dilution of 1:500. Embryos for antibody labelling and in situ hybridization were fixed in 4% paraformaldehyde in 0.1 M MOPS, 0.5 M NaCl, 1 mM EGTA, 2 mM MgSO4, pH 7.4 (L.Z. Holland et al., 1996), and stored in either 100% methanol or 70% ethanol. There was no difference in antibody-labelling between these storage methods. The protocol for antibody labelling is as previously published (Holland and Holland, 1993a). Staining was with nickel-enhanced diamino-benzidine (DAB). For transmission electron microscopy, antibody-labelled embryos in Spurr's resin were fine-sectioned and observed in an electron microscope without staining with either uranyl acetate or lead citrate. The nickel-enhanced DAB product is electron-dense.

Methods for in situ hybridization are in L. Z. Holland et al. (1996). Riboprobes were synthesized from full-length clones of AmBra1 (X91903), AmphiWnt1 (AF06194), AmphiWnt5 (AF361014), AmphiWnt3 (AF3610132), AmphiWnt8 (AF190470), AmphiFoxQ (AY163864), and AmphiSox1/2/3 (AF271787).


We thank D. R. McClay, Duke University, Chapel Hill, NC, for his kind gift of the β-catenin antibody. John Lawrence, University of South Florida, Tampa, FL, generously provided laboratory space during the amphioxus breeding season.