Nodal/activin signaling establishes oral–aboral polarity in the early sea urchin embryo



Components of the Wnt signaling pathway are involved in patterning the sea urchin primary or animal–vegetal (AV) axis, but the molecular cues that pattern the secondary embryonic axis, the aboral/oral (AO) axis, are not known. In an analysis of signaling molecules that influence patterning along the sea urchin embryonic axes, we found that members of the activin subfamily of transforming growth factor-β (TGF-β) signaling molecules influence the establishment of AO polarities in the early embryo. Injection of activin mRNAs into fertilized eggs or treatment with exogenously applied recombinant activin altered the allocation of ectodermal fates and ventralized the embryo. The phenotypes observed resemble the ventralized phenotype previously reported for NiCl2, a known disrupter of AO patterning. Sensitivity to exogenous activin occurs between fertilization and the late blastula stage, which is also the time of highest NiCl2 sensitivity. These results argue that specification of fates along the embryonic AO axis involves TGF-β signaling. To further examine TGF-β signaling in these embryos, we cloned an endogenous TGF-β from sea urchin embryos that is a member of the activin subfamily, SpNodal, and show through gain of function analysis that it recapitulates results obtained with exogenous activins and NiCl2. The expression pattern of SpNodal is consistent with a role for nodal signaling in the establishment of fates along the AO axis. Loss of function experiments using SpNodal antisense morpholinos also support a role for SpNodal in the establishment of the AO axis. Developmental Dynamics 231:727–740, 2004. © 2004 Wiley-Liss, Inc.


Recent experiments have shown that components of the Wnt (Emily-Fenouil et al., 1998; Wikramanayake et al., 1998; Logan et al., 1999; Huang et al., 2000; Vonica et al., 2000) and the bone morphogenetic protein (BMP) 2/4 (Angerer et al., 2000) signaling pathway play important roles in patterning cell fates along the sea urchin primary or animal–vegetal (AV) axis. In contrast, the molecular mechanisms underlying patterning of the second embryonic axis, the aboral/oral (AO) axis, are poorly understood. The AO axis is not morphologically apparent in the early embryo, but is discernible by the end of gastrulation as a regional specialization of the embryonic epithelium. By this stage, the oral epithelium surrounding the mouth is separated from the aboral epithelium (which covers most of the rest of the embryo) by rows of ciliated cells, the ciliary band. Lineage tracing in Strongylocentrotus purpuratus embryos has shown that the ectoderm derives almost entirely from the animal blastomeres or mesomeres of the 16-cell embryo and that the oral and aboral ectodermal territories bear a consistent relationship to the cleavage planes in this species (Cameron et al., 1989). These results combined with early asymmetries in territory specific genes suggest that the AO axis is first specified during cleavage (Davidson, 1989). However, isolated blastomeres of the four-cell stage sea urchin embryo can each develop into a normal pluteus (Hörstadius, 1973). In addition, the AO axis can be entrained by various physical and chemical treatments arguing that commitment along this axis probably does not occur until the mesenchyme blastula/early gastrula stage (Hardin et al., 1992).

Chemical treatments that alter specification along the AO axis act primarily between fertilization and the mesenchyme blastula stage. For example, the presumptive oral side of the embryo exhibits elevated cytochrome oxidase activity (reviewed in Czihak, 1971) and treatments that abolish (Child, 1948; Czihak, 1963) or alter (Pease, 1941, 1942; Coffman and Davidson, 2001) this respiratory asymmetry alters AO polarity. Similarly, NiCl2 treatment radializes and ventralizes sea urchin embryos; the oral ectodermal territory is expanded, whereas the aboral ectoderm is reduced (Hardin et al., 1992). In NiCl2-treated embryos, directed tilting of the archenteron to one side fails to occur and the primary mesenchyme cells (PMCs), which normally form two bilateral clusters, are radially arrayed.

Coffman and Davidson (2001) recently proposed that the respiratory asymmetry influences the activity of components of cell signaling pathways and/or transcription factors (Coffman and Davidson, 2001). One transcription factor that has been shown to influence AO cell fates is sea urchin goosecoid (SpGsc). SpGsc expression is restricted to the oral ectoderm, where it represses aboral ectoderm genes (Angerer et al., 2001). How SpGsc is initially activated and restricted is not well understood. In other organisms, goosecoid is a downstream target of Nodal (Toyama et al., 1995), a member of the activin/Vg1/nodal subfamily of transforming growth factor-β (TGF-β) signaling molecules (Schier, 2003).

To better understand the signaling pathways that pattern cell fates along the AO axis, we examined the effects of the ectopic expression of members of the activin subfamily of TGF-β signaling molecules on sea urchin development. We found that injection of either human or Xenopus activin βB or Xnr-2 RNAs or treatment with recombinant human activin radializes and ventralizes embryos reminiscent of the effects of NiCl2 on the establishment of AO polarities. To identify the endogenous TGF-β molecule responsible for these effects, we used degenerate polymerase chain reaction (PCR) to identify members of the activin subfamily of TGF-βs. This resulted in the identification of a sea urchin nodal homolog (SpNodal). We show that SpNodal is an endogenous molecule that patterns the AO axis during sea urchin development. SpNodal ectopic expression expands the oral ectoderm and reduces aboral ectoderm. Perturbation of SpNodal by antisense morpholino injection generates the reciprocal phenotype. In addition, the expression of SpNodal transcripts in a subset of ectodermal cells supports a role for endogenous SpNodal in patterning cell fates along the AO axis. These results are in close agreement with those reported for nodal signaling in another species of sea urchin (Duboc et al., 2004).


Microinjection of Human Activin βB Radializes and Ventralizes Embryos

When Lytechinus variegatus zygotes were injected with in vitro transcribed RNA encoding human activin βB, the embryos display a bell shape at the gastrula and pluteus stages and contain multiple short spicules radially arranged around the hindgut in the vegetal half of the embryo close to the blastopore (Fig. 1B,E). The archenteron grows straight toward the blastocoele roof instead of turning toward one side of the embryo. The phenotype observed was similar to that reported for embryos treated with NiCl2 (Hardin et al., 1992). At the gastrula stage (Fig. 1B,C), thickenings are discernible in the animal ectoderm foreshadowing the bell shape that forms at the pluteus stage in both activin-injected and NiCl2-treated embryos (Fig. 1E,F). Identical results were obtained when Xenopus activin βB and the closely related TGF-β, Xenopus nodal-related 2 (Xnr2), RNAs were injected (data not shown) and when embryos were treated with human recombinant activin. The altered shape of the embryo and the radially arranged spicules (Fig. 1H) compared with controls (Fig. 1G) indicate that activin, like NiCl2-treatment, ventralizes and radializes the embryo.

Figure 1.

Activin radializes and ventralizes embryos in a manner similar to NiCl2. A–C: Gastrula stage L. variegatus embryos injected with human activin βB RNA (B) lack bilateral symmetry evident in control embryos (A). Activin-injected embryos appear bell shaped, the archenteron grows straight upward, and embryos display a circumferential thickening of the ectoderm on the animal side of the embryo (arrowheads in B) and resemble 0.5 mM NiCl2-treated embryos of the same stage (C). D–F: As development proceeds, the bilateral symmetry of a normal pluteus stage embryo is clearly evident in controls (D), whereas activin-injected (E) and NiCl2-treated (F) embryos retain a bell-shape and appear radialized. Similar results are obtained when embryos are treated with human-recombinant activin. G,H: Embryos show the characteristic bell shape and multiple short spicules radially arrayed around the blastopore (H), in contrast to the bilaterally symmetric skeletal rods of control embryos (G). G and H are viewed by polarizing optics.

Microinjection of Activin βB Results in an Expanded Oral Ectoderm

Since NiCl2-treatment has been reported to alter the allocation of embryonic fates along the AO axis and radialize the embryo (Hardin et al., 1992), we asked whether there was a similar rearrangement of fates in activin-injected embryos. L. variegatus embryos injected with human activinβB RNA were compared with NiCl2-treated embryos by immunofluorescent staining with tissue specific antibodies. Both activin-injected and NiCl2-treated embryos show an expanded oral ectoderm and a reduced aboral ectoderm when labeled with the ectodermal marker EctoV. In normal pluteus stage embryos, EctoV stains only the facial epithelium that surrounds the mouth and a portion of the foregut, whereas much of the embryonic epithelium is not labeled (Fig. 2A). In contrast, in activin βB-injected (Fig. 2B) and NiCl2-treated (Fig. 2C) embryos, the majority of the embryonic surface is labeled with EctoV. Unlike control embryos where a large portion of the surface is unlabeled by EctoV, only a small area of epithelium that surrounds the blastopore (Fig. 2C, arrowheads) and a small area protruding from the mouth (Fig. 2B,C, asterisks) are unlabeled.

Figure 2.

Comparison of tissue specific marker expression in activin-injected and NiCl2-treated embryos. L. variegatus embryos were injected with activin βB RNA or treated with 0.5 mM NiCl2, cultured to the pluteus stage, immunolabeled with tissue specific antibodies, and compared with controls. A–C: EctoV monoclonal antibody recognizes the oral epithelium that surrounds the mouth on the ventral side of the embryo (asterisk in A), and a portion of the foregut of control embryos. Embryos injected with activin βB (B) or treated with NiCl2 (C) show most of the ectoderm labeled with EctoV, only a small area surrounding the blastopore remains unlabeled (arrowheads in C). D–F: Control pluteus stage embryos labeled with CBA demonstrate a band of staining a few cells thick between the oral and aboral ectoderm. In activin-injected (E) and NiCl2-treated (F) embryos CBA staining is displaced toward the blastopore on the opposite side of the embryo. G–I: Monoclonal antibody 6a9 recognizes primary mesenchyme cells that align along the bilaterally symmetric skeletal rod in controls (G). These same cells are clustered around the blastopore in activin-injected (H) and NiCl2-treated (I) embryos. J,M: MHC polyclonal antibody recognizes the circumferential muscle surrounding the foregut (J) and Endo1 monoclonal antibody recognizes the endoderm of the mid- and hindgut (M) of control plutei. K,L,N,O: MHC staining is seen associated with the foregut in activin-injected (K) and NiCl2-treated (L) embryos and a tripartite gut forms in all embryos examined (N,O). All embryos are positioned with the oral side or ventral surface at the top.

Ciliary Band Is Shifted Vegetally in Activin-Injected Embryos

The ciliary band delineates the boundary between the oral and aboral epithelium of the pluteus and can be recognized by immunocytochemical staining with the monoclonal antibody anti-CBA (monoclonal UH295; Wessel and McClay, 1985). In control plutei, the ciliary band cells were visible as cells outlining the facial epithelium that surrounds the mouth (Fig. 2D). In activin-injected and NiCl2-treated embryos, the ciliary band was shifted vegetally to surround the reduced aboral epithelium that encircles the blastopore (Fig. 2E,F).

Activin-Injected Embryos Are Radialized, and the Skeletogenic Mesenchyme Cells Are Misplaced

Normally, the ectoderm of the embryo provides informational cues that position the skeletogenic or PMCs and determine the shape of the larval skeleton (Armstrong et al., 1993; Armstrong and McClay, 1994). In NiCl2-treated embryos reallocation of the ectoderm results in disorganized PMCs, multiple spicule centers, and the embryo loses its bilateral symmetry (Hardin et al., 1992). NiCl2 effectively radializes the ectodermal signals that pattern the skeleton of the embryo. To determine whether a similar displacement of the PMCs is observed in activin-injected embryos, we used monoclonal antibody 6A9 (Ettensohn, 1990) to view the position of PMCs in activin-injected embryos compared with controls and NiCl2-treated embryos. In control embryos, PMCs are seen distributed along the bilaterally symmetrical skeletal rods (Fig. 2G). In contrast, the PMCs in activin-injected (Fig. 2H) and NiCl2-treated embryos (Fig. 2I) are restricted to the vegetal portion of the embryo surrounding the hindgut close to the blastopore/future anus.

Endoderm Specification and Secondary Mesenchyme Cells Are Unaffected by Activin

To determine the effects of activin injection on the regionalization of the endoderm and differentiation of other mesenchyme cells in the embryo, we labeled embryos with antibodies that define different regions of the gut. Anti-MHC labels muscle cells that surround the foregut (Wessel et al., 1990) of control embryos (Fig. 2J), whereas Endo1 is a monoclonal antibody that labels the surface of cells in the mid- and hindgut (Wessel and McClay, 1985). In activin-injected and NiCl2-treated embryos the gut, although somewhat less developed than in controls, forms a straight tube that runs through the center of the embryo and appears to be regionally specified into a foregut, midgut, and hindgut. Circumesophageal muscle cells form normally around the tip of the archenteron to define the foregut (Fig. 2K,L), whereas EndoI labels the mid- and hindgut (Fig. 2N,O) as in controls, although distinct constrictions between these regions are not always apparent.

Dose Response of Embryos to Recombinant Activin

To determine whether the dose response and the timing of activin's effect is consistent with results reported for NiCl2-treatment, we examined the phenotype of embryos treated with exogenously applied human recombinant activin-A. In these experiments, activin-A was added to embryo cultures at different concentrations. As with NiCl2-treatment (Hardin et al., 1992), a graded effect was observed on embryonic AO axis patterning when different concentrations of activin-A were added to embryos in culture (Figs. 3A–F, 4). As the concentration of activin-A was increased, embryos showed an increasingly severe phenotype. At the highest concentrations used (100 nM), the embryos did not gastrulate and appeared to be arrested in their development (Fig. 3F). Between 25 and 50 nM activin-A, the embryos resembled those seen with microinjection of 2 pg of activin βB RNA or treatment with 0.5 mM NiCl2. Embryos displayed a characteristic bell shape and were radialized and ventralized. As the concentration of recombinant activin-A was reduced, the phenotypes were intermediate and resembled those reported for intermediate concentrations of NiCl2 (Hardin et al., 1992), showing an expansion of the oral hood and broadening of the angle between the larval arms (Fig. 3B).

Figure 3.

Embryos show a dose response to recombinant activin. When increasing concentrations of activin-A are added to L. variegatus embryos in culture (A), a graded response (arrowheads) was observed at the pluteus stage with intermediate phenotypes (B,C), resembling those seen at intermediate concentrations of NiCl2. At concentrations between 25 and 100 nM, most embryos form the characteristic radialized and ventralized bell shape (D–E). At higher concentration, embryos do not gastrulate (F).

Figure 4.

Dose response and timing of the effect of recombinant activin. Recombinant activin-A was added to L. variegatus embryo cultures at concentrations between 0 and 100 nM at fertilization (Fert) or at the hatching blastula (Blast) stages. As concentrations were increased, embryos became progressively more radialized at both time points examined. Values are shown as the percentage of embryos with the corresponding morphology (A, most radialized and ventralized; B, less radialized and ventralized; C, least radialized and ventralized; D, normal plutei). The highest percentage at each concentration is shown in bold type.

Timing of the Effect of Activin

Because it has been shown that NiCl2 can influence the allocation of AO fates at any time between fertilization and the late blastula stage (Hardin et al., 1992), we asked if activin treatment was similarly effective at these times. We treated embryos with 50 nM recombinant human activin-A at fertilization and just after hatching. We found that embryos displayed a ventralized and radialized phenotype when recombinant activin-A was added to the culture at any time between fertilization and the late blastula stage (Fig. 4). These effects suggest that the sensitivity to activin occurs with a similar time course to that of NiCl2.

Effect of Activin and NiCl2 Treatment on S. purpuratus Embryos

To determine whether activin had similar effects on axial patterning in other species of sea urchins, we tested different concentrations of human recombinant activin on S. purpuratus embryos. We found that S. purpuratus embryos were less sensitive to activin-A (Fig. 5A–C), requiring higher concentrations of activin than were needed to radialize and ventralize L. variegatus embryos. Because of this reduced sensitivity to activin we examined the response of S. purpuratus embryos to NiCl2 (Fig. 5D–I) and found S. purpuratus embryos to be more sensitive to NiCl2 than L. variegatus embryos. L. pictus embryos were also examined and found to closely resembled L. variegatus in their response to both activin and NiCl2 (data not shown).

Figure 5.

Treatment of S. purpuratus embryos with recombinant activin-A and NiCl2. A–C: When S. purpuratus embryos are treated with increasing concentrations of activin-A, they exhibit the radialized and ventralized phenotype at higher concentrations than those required to produce the same phenotype in L. variegatus embryos (compare with Fig. 3). D–I: In contrast, lower concentrations of NiCl2 are required to obtain the bell-shaped phenotype in S. purpuratus embryos, which is first evident at 1/100th the concentration of NiCl2 that will radialize and ventralize L. variegatus embryos (Hardin et al., 1992). S. purpuratus embryos fail to gastrulate (I) at lower concentrations than those that radialize and ventralize L. variegatus embryos.

Cloning of Sea Urchin Nodal

To clone endogenous sea urchin TGF-β family members that might be involved in patterning the AO axis, we used degenerate PCR. By using this strategy, we identified the TGF-β family member, nodal. We initially isolated a 231-bp DNA fragment from S. purpuratus genomic DNA. The S. purpuratus PCR product was used to probe both 7 and 14 H S. purpuratus cDNA macroarray filter sets. All the clones identified from this screen encoded overlapping SpNodal cDNAs (Fig. 6A). NCBI Blast comparisons of the PCR product sequence (Fig. 6B) revealed that the SpNodal deduced amino acid sequence is more similar to vertebrate nodals and P. lividus nodal than to the other TGF-β proteins that have been identified in the sea urchin (Fig. 6C).

Figure 6.

SpNodal cDNAs and sequence comparisons. A: Alignment of the cDNAs obtained from screens and 5′ rapid amplification of cDNA ends (medium gray, 5′-untranslated region [UTR]; light gray, coding region; dark gray, 3′-UTR). B: Comparison of the deduced amino acid sequence of the SpNodal polymerase chain reaction product to the same region in vertebrate and Paracentrotus lividus nodal proteins (Sp, S. purpuratus; Pl, P. lividus; Mm, Mus musculus; Xl, Xenopus laevis; BF, Branchiostoma floridia; Hs, Homo sapiens; Hr, Halocynthia roretzi; Dr, Danio rerio). C: Sequence comparison of SpNodal and other non-nodal transforming growth factor-βs identified in the sea urchin. (Only amino acids showing identity with those in SpNodal are boxed in gray).

The deduced amino acid sequence of the longest cDNA revealed the following putative domains based on similarity to nodals from other species: a multibasic cleavage site (amino acids 306-309) and 7 cysteines predicted to form a C-terminal “cysteine knot” typical of TGF-βs (Fig. 7). Further comparisons with nodal from other species revealed that the longest macroarray clone was missing the most N-terminal amino acids that constitute the signal sequence characteristic of TGF-βs (SpNodal1; Fig. 6B). 5′-rapid amplification of cDNA ends (RACE) was used to obtain additional 5′ sequence. This sequence (SpNodal4; Fig. 6B) included 285 bp of 5′-untranslated region (UTR) and 63 bp encoding 21 amino acids that included the missing signal sequence. The 5′-RACE sequence was confirmed with a partial SpNodal 5′ sequence (SpNodal5; Fig. 6B) obtained from an S. purpuratus “unigene screen” (Poustka et al., 1999, 2003). A genomic Southern using SpNodal cDNA as probe and genomic DNA from three individuals indicate a single nodal-related gene is present in the S. purpuratus genome (data not shown). Comparison of the full-length deduced amino acid sequence of SpNodal with that of P. lividus nodal revealed 69% identity and 84% similarity overall. Comparison of the two pro domains revealed greater similarity (89% identity and 94% similarity), whereas the mature domains were less similar (67% identity and 80% similarity).

Figure 7.

SpNodal cDNA and deduced amino acid sequence. The deduced sequence of the combined 5′ rapid amplification of cDNA ends (RACE) cDNA and the longest cDNA from the macroarray screen reveals a putative SpNodal protein of 450 amino acids. The underlined sequence was removed to create ΔSpNodal for antisense morpholino rescue experiments, because it removes the morpholino recognition sequence, shown shaded in light gray. A putative proprotein convertase recognition site, RARR, is shaded in dark gray, the sequence in gray is the predicted processed ligand sequence. The seven cysteines that are predicted to form the characteristic cysteine knot of transforming growth factor-βs are outlined. The double arrow denotes the sequence to which primers were designed for 5′-RACE. The single arrows designate the forward and reverse sequences recognized by the degenerate primers used to clone the initial polymerase chain reaction product.

SpNodal Temporal and Spatial Expression

To determine when in the sea urchin embryo SpNodal is expressed, we performed nonquantitative reverse transcriptase (RT) -PCR analysis on RNA isolated from different stages of sea urchin development and found SpNodal in all stages examined (Fig. 8A). These data suggested that SpNodal transcript is present maternally and is expressed throughout early development. Further analysis using quantitative PCR supported these findings and showed that the levels of SpNodal transcript expression increase from the unfertilized egg through the mesenchyme blastula stage (Fig. 8B,C). Whereas low levels of SpNodal transcript were detected in the unfertilized egg (∼2,851 copies/reaction), levels increase dramatically during the blastula stages (∼285,037 copies/reaction).

Figure 8.

Expression analysis of SpNodal by polymerase chain reaction (PCR). A: Reverse transcriptase-PCR analysis of SpNodal RNA levels at different stages of development reveals transcripts are present at low levels in the unfertilized egg and throughout early stages of development. SpMito PCR was used as a loading control. B: The mean cT (cycle threshold) values for SpNodal at different stages of development obtained by quantitative PCR (QPCR) analysis are shown for comparison. The cT values represent the mean of triplicate QPCR reactions of 0.5 μg of cDNA standardized against duplicate reactions of S. purpuratus mitochondrial rRNA (SpMito) controls. C: QPCR reveals that SpNodal expression gradually increases from the egg and early cleavage stages to peak at the 60-cell and mesenchyme blastula stages (the lowest mean cT values). Delta-Delta cT values, the differences between the values from each stage and the controls, were calculated to determine the mean relative difference in SpNodal expression in the egg and embryonic stages and used to estimate the number of copies of SpNodal transcripts/reaction.

To determine where in the sea urchin SpNodal is expressed, we performed whole mount in situ hybridization on embryos from different stages using sense and antisense digoxigenin-labeled SpNodal probes. SpNodal expression was observed in all blastomeres at early stages of development (Fig. 9A,B). Although SpNodal transcript levels are low in these early stages, consistent labeling was observed compared with sense controls (Fig. 9D,F). Localized SpNodal transcripts were first evident in the early blastula stage (Fig. 9C). This localization is restricted at the mesenchyme blastula stage to one side of the embryo (Fig. 9G–L), and no staining was observed using sense probes at these stages (Fig. 9F). At the early gastrula stage, transcripts were detected in the epithelium closest to the tip of the developing endodermal tube, which is predictive of the future oral side (Fig. 9M,N).

Figure 9.

In situ hybridization analysis of SpNodal transcript expression in the S. purpuratus egg and early embryo. A,B: In situ hybridization analysis using SpNodal antisense probe reveals uniform weak transcript expression in the egg (A) and early cleavage stages (B). D,E: No staining was observed in similar stages labeled with SpNodal sense probes. C,G–I: During blastula stages, SpNodal is enriched in a subset of blastomeres. J–L: This asymmetric localization is further restricted at the mesenchyme blastula stage to the ectoderm on one side of the embryo. Expression is seen in ectodermal cells that occupy a subdomain consistent with expression on the presumptive oral side based on the expression pattern in gastrula stage embryos. F: No labeling was observed with sense probe at any stage examined. M,N: Gastrula stage embryos show SpNodal expression on the side closest to the advancing archenteron (M, lateral view; N, vegetal view). In B and E, embryos are viewed from the vegetal side. C,G–K: Blastula stage embryos are positioned with the presumptive oral side on the left. The embryos in L and N, are the same as those shown in K and M, respectively, viewed from the direction shown by the arrows.

SpNodal Perturbation Affects AO Patterning

To determine the effects of perturbing SpNodal function, we first injected in vitro transcribed capped SpNodal RNA into both S. purpuratus and L. variegatus zygotes. Embryos microinjected with SpNodal RNA (∼3–4 pg/embryo) displayed the characteristic activin/Ni+2 phenotype in both species. Embryos are ventralized and radialized and demonstrate a bell-shaped phenotype consistent with a role for SpNodal in patterning the AO axis. Embryos also show reduced pigment cell numbers (Fig. 10B), suggesting an expanded oral and a reduced aboral ectoderm. They also demonstrate multiple spicules clustered around the blastopore instead of the bilaterally symmetrical skeleton seen in control embryos (compare Fig. 10A′,B′).

Figure 10.

Perturbation of SpNodal expression supports a role in aboral/oral (AO) patterning. A,B: Microinjection of SpNodal RNA recapitulates the activin and Ni+2 phenotype, embryos display the characteristic bell shape at the pluteus stage and show reduced pigment cell numbers (B) compared with controls (A). A′,B′: These embryos also show multiple, short, radially arrayed spicules toward the vegetal side of the embryo (B′) in contrast to the bilaterally symmetric rods of controls (A′). C: Microinjection of SpNodal morpholino generates a somewhat radialized phenotype, because the gut does not bend as in controls. Also, these embryos also have an increased pigment cell number, indicating an expanded aboral ectoderm. C′: SpNodal antisense morpholino-injected embryos also have a thickened band of epithelium at the oral side (arrow in C) and elongated elaborate spicules (C′) distinct from those seen in controls (A′) and SpNodal-injected embryos (B′). Irrelevant morpholino injections did not alter development. D–I′: The specificity of the SpNodal morpholino is demonstrated in rescue experiments. E–F′: Embryos injected with ΔSpNodal RNA alone display the radialized and ventralized phenotype (E,E′) as in B, whereas embryos injected with SpNodal antisense morpholino display the reciprocal phenotype (F,F′). The SpNodal antisense morpholino phenotype is rescued by simultaneous injection of ΔSpNodal RNA (G,G′) and display a morphology that closely resembles that of glycerol-injected controls (D,D′). Although these embryos develop with a normal shape and internal structures, they generally lack the degree of pigmentation (I,I′) seen in controls (H,H′). The focal plane is at the surface of the embryo in H and I and in the middle of the embryo in H′ and I′. J: Analysis of gene expression in SpNodal RNA and antisense morpholino-injected embryos by reverse transcriptase-polymerase chain reaction (RT-PCR) supports altered AO patterning. BMP2/4 transcripts are up-regulated by SpNodal RNA injection and down-regulated by SpNodal morpholino antisense injections when compared with control glycerol-injected embryos. In contrast, Spec1 and Tbx, both aboral ectoderm markers, are down-regulated by SpNodal RNA injections and up-regulated by injection of SpNodal antisense morpholino injections. SpMito RT-PCR showed equivalent loadings of cDNA in these samples.

In reciprocal experiments, we examined the consequences of the loss of SpNodal function by microinjecting SpNodal morpholino antisense oligonucleotides into S. purpuratus zygotes. Embryos injected with an SpNodal antisense morpholino (∼2 pl of 1.6–2.7 μM/embryo) recognizing nucleotides 233-257 (Fig. 7) display a thickened band of ciliated cells on their oral side and increased pigmentation, suggesting an expanded aboral ectoderm (Fig. 10C). The phenotype obtained from morpholino antisense injection is distinct from that obtained from SpNodal or activin ectopic expression and NiCl2-treatment. Most notable was the increased pigment cell number and a thickened band of cilia on their oral side. In addition, the embryos show a unique skeletal phenotype with elongated spicules shaped much like those of control embryos but displaced in the blastocoele. The morpholino-injected embryos never appear bell-shaped and, because the skeletal spicules often run perpendicular to the long axis of the embryo, frequently assume a wedge shape (Fig. 10C,C′,F,F′). When a standard control morpholino was injected at comparable concentrations no morphological effect was observed (data not shown).

To demonstrate the specificity of the SpNodal antisense morpholino, we deleted the sequence from the 5′-UTR of SpNodal (underlined sequence in Fig. 7) that binds the morpholino (ΔSpNodal). When embryos are injected with the ΔSpNodal RNA, they produce the characteristic bell-shaped phenotype also observed with full-length SpNodal (Fig. 10E,E′). However, coinjection of the morpholino and ΔSpNodal RNA rescues most features of the characteristic knockdown phenotype of the morpholino (Fig. 10G,G′) and more closely resembles control embryos of the same stage (Fig. 10D,D′). While the rescued embryos are normally shaped, they are not as pigmented (Fig. 10I,I′) as their sibling control embryos (Fig. 10H,H′). Greater than 60% of the coinjected embryos displayed a normal shape at the pluteus stage, <30% showed a phenotype that either more closely resembled the morpholino or the ΔSpNodal RNA injection phenotype, whereas the remaining 10% were indeterminate.

To further determine how SpNodal perturbations affect AO patterning, we examined the expression of AO markers by RT-PCR. Spec1 is normally restricted to the aboral ectoderm of unperturbed embryos (Hardin et al., 1985). RT-PCR analysis of Spec1 at the hatched blastula stage indicates SpNodal RNA-injection leads to reduced Spec1 expression and SpNodal morpholino injection leads to increased Spec1 expression compared with controls (Fig. 10J). In contrast, SpBMP2/4, which is expressed primarily in the presumptive oral ectoderm (Angerer et al., 2000) is increased by SpNodal RNA injection and reduced by SpNodal morpholino injection (Fig. 10J). The T-box transcription factor Tbx2/3 is normally expressed in the aboral ectoderm and is affected by perturbations that alter AO specification (Gross et al., 2003). When we examined the expression of Tbx in SpNodal RNA- and morpholino-injected embryos, we found expression is down-regulated and up-regulated, respectively, at the hatched blastula stage compared with glycerol-injected controls (Fig. 10J). These results are consistent with a reduced aboral ectoderm in SpNodal RNA-injected embryos and increased aboral ectoderm in SpNodal morpholino-injected embryos.


The molecular mechanisms that underlie specification of the secondary embryonic or AO axis in the sea urchin embryo are poorly understood. In this report, we demonstrate that members of the activin/nodal subfamily of TGF-β signaling ligands pattern cell fates along the sea urchin AO axis. In addition, we have identified sea urchin nodal (SpNodal) as the endogenous signaling molecule that recapitulates the effects of exogenous activin and mimics the effects of NiCl2-treatment, one of the few treatments that has been shown to affect specification of fates along this axis (Hardin et al., 1992).

Nodal is a member of the activin/Vg1/nodal subfamily of TGF-β signaling molecules (Schier, 2003). Other members of the TGF-β family that have been identified in sea urchins include BMP2/4 (Angerer et al., 2000), BMP5/7 (Ponce et al., 1999), and univin (Stenzel et al., 1994). Sea urchin BMP2/4 has been shown to regulate the position of the ectoderm/endoderm boundary and to promote aboral ectoderm differentiation (Angerer et al., 2000). BMP2/4 signaling acts to repress aboral genes in the oral ectoderm territory; however, BMP2/4 overexpression prevents the expression of only a subset of aboral genes and does not uniformly oralize the embryo (Gross et al., 2003). Univin (Stenzel et al., 1994) has been shown recently to play a role in skeletal rod growth (Zito et al., 2003), whereas BMP5/7 (Ponce et al., 1999) has yet to be assigned a function in this embryo.

Our results are similar to those reported for nodal function in another species of sea urchin, Paracentrotus lividus (Duboc et al., 2004), with some important differences. Our data indicate that SpNodal is present maternally and, in SpNodal knockdown experiments, aboral ectoderm genes are expressed. In contrast, Duboc et al. report that the nodal morpholino-injected P. lividus embryos do not produce either oral or aboral ectoderm. These differences might be explained by species-specific differences in the timing of expression of nodal and/or the inherent differences in the program that establishes AO polarity in the two species examined. Sea urchin embryos display a species-specific relationship between the position of the initial cleavage planes and the orientation of the AO axis. Although the orientation of the early cleavage planes bears a consistent relationship to the AO axis in S. purpuratus and L. pictus embryos, this finding is not the case for all species (Cameron et al., 1989; Cameron and Davidson, 1991; Henry et al., 1992; Henry, 1998). The different sensitivity to activin and NiCl2-treatment in L. variegatus and S. purpuratus embryos that we observed may also reflect such an underlying difference in the timing and execution of AO axis specification. Nevertheless, SpNodal RNA injection resulted in oralization and ventralization at similar concentrations in both S. purpuratus and L. variegatus arguing that nodal signaling is involved in AO axis specification in these species as well as P. lividus.

SpNodal transcripts are expressed during the time that AO patterning occurs and are preferentially expressed in the prospective oral ectoderm on one side of the embryo. Microinjection of SpNodal RNA leads to a radialized phenotype with a reduction in pigment cell numbers in the epithelium consistent with expanded oral and reduced aboral ectoderm. SpNodal RNA injection also leads to up-regulation of BMP2/4, which normally is progressively restricted to the oral ectoderm where it leads to the repression of a subset of aboral genes (Angerer et al., 2000; Gross et al., 2003). SpNodal RNA injection also down-regulates two genes expressed in the aboral ectoderm, Tbx2/3 and Spec1. In L. variegatus embryos, Tbx2/3 is expressed in the aboral territories of several embryonic tissues and is repressed by overexpression of LvBMP2/4, suggesting that Tbx2/3 is a downstream component of the AO axis program (Gross et al., 2003). Because we observe an up-regulation of BMP2/4 by SpNodal overexpression these results are consistent with repression of the aboral ectoderm and expansion of the oral ectoderm. Conversely, SpNodal antisense morpholino injections generated embryos where pigment cell numbers in the epithelium are increased, BMP2/4 levels are decreased, and Tbx2/3 and Spec1 transcript levels are up-regulated, consistent with expanded aboral ectoderm. The SpNodal morpholino phenotype is morphologically very similar to that described for P. lividus (Duboc et al., 2004); however, they report that the P. lividus homolog of Tbx2/3 is not expressed in the P. lividus nodal morpholino-injected embryos at early stages and another aboral ectoderm marker 29D is restricted to a region surrounding the blastopore. These results led them to conclude that aboral ectoderm is not specified in P. lividus morpholino-injected embryos. However, identification of the aboral marker 29D around the blastopore is consistent with at least some aboral ectoderm specification. Our observation that SpNodal morpholino did not repress all aboral ectoderm formation may also be explained by partial knockdown of SpNodal function. The presence of maternal SpNodal transcripts suggests SpNodal protein is present in S. purpuratus eggs. If SpNodal protein is maternally synthesized, its function would not be affected by morpholino antisense injection, explaining the differences we observe. Development of a nodal antibody is necessary before it will be possible to determine whether SpNodal is completely eliminated in our morpholino antisense experiments.

Since Ni+2 has been reported to influence multiple cellular activities (Kasprzak et al., 2003), the precise mechanism of Ni+2 action on sea urchin AO axis specification is unclear. Although our data indicate nodal signaling influences AO patterning in a manner similar to Ni+2 treatment, how nodal signaling relates to the action of Ni+2 is also unclear. Because the respiratory and redox state of blastomeres is thought to influence AO axis specification, one attractive hypothesis is that NiCl2 alters the respiratory asymmetry and redox status of blastomeres. It has been proposed that differential gene expression along the AO axis depends initially on globally distributed maternal transcription factors that are asymmetrically modified due either to physiological asymmetries (such as redox state) or in response to intercellular signaling (Coffman and Davidson, 2001). Because NiCl2 can activate hypoxia-inducible genes (Salnikow et al., 1997), one possible scenario is that NiCl2 alters the inherent respiratory asymmetry of the early embryo, which may in turn influence the activation of signaling components or maternal transcription factors. Alternatively, DAN, a member of the Cerberus/DAN/gremlin family of TGF-β antagonists, has been shown to be a Ni-binding protein whose activity is inhibited by Ni+2 (Kondo et al., 1995). If a homolog of Cerberus/DAN/gremlin is present in the early sea urchin embryo and regulates nodal's function, its inhibition by Ni+2 could also explain the similar phenotypes observed with Ni+2-treatment and nodal/activin injections.

In vertebrates, nodal signaling has been shown to play important roles in setting up the embryonic axes, induction of endoderm and mesoderm, patterning of the nervous system, and determination of left–right asymmetries (Schier and Shen, 2000; Shier, 2003). A single nodal gene has been identified in the mouse and chicken while multiple nodals have been identified in Xenopus (Xnr 1-6) and Zebrafish (squint, Cyclops, southpaw). Nodal-related genes have also been identified in two nonvertebrate chordates: amphioxus (Yu et al., 2002) and the ascidian (Morokuma et al., 2002). As we observe in the sea urchin, nodal expression in both these species occurs primarily in the developing ectoderm during early development. The failure to identify nodal-related ligands in Caenorhabditis elegans and Drosophila have led to the proposal that nodals constitute a subgroup of TGF-βs that is restricted to chordates (Schier, 2003). The identification of a nodal homolog in the sea urchin argues that nodal signaling is evolutionarily conserved in all deuterostomes and that the sea urchin may be similar to a bilateral ancestral deuterostome. The identification and characterization of all nodal pathway components in the sea urchin will help provide clues not only into how the AO axis is specified but also how the nodal signaling pathway evolved.



L. variegatus were obtained from Beaufort Biologicals, Duke University Marine Laboratory (Beaufort, NC) and S. purpuratus and L. pictus were obtained from Charles Hollohan (Santa Barbara, CA). Embryos were cultured in Millipore Filtered Artificial Sea Water (MFASW) at a 1% concentration (1 ml of settled eggs in 100 ml of MFASW) for all treatments except microinjection. L. variegatus embryos were cultured at room temperature, and S. purpuratus and L. pictus were cultured at 15°C. Microinjected embryos were cultured in freshly prepared M.B.L. Formula (Cavanaugh, 1975) artificial sea water (MBL AFSW).


Human activin βB cDNA was obtained from B. Gumbiner (Sloan-Kettering Memorial Cancer Institute) and Xenopus activin βB cDNA was from C. Chang and A. Hemmati-Brivanlou (Rockefeller University). Xnr-2 cDNA was from O. Wessely and E. De Robertis (UCLA). To generate capped RNAs for microinjection, all cDNAs were linearized and in vitro transcribed using the Mmessage Machine according to the manufacturer's instructions (Ambion, Inc., Austin, TX). The in vitro transcribed RNAs and morpholino antisense oligonucleotide solutions were diluted in sterile 40% glycerol and filtered before injection through a 0.22-μm Millipore syringe filter (Millipore, Billerica, MA). The capped RNAs were microinjected into fertilized eggs at approximately 2 pl/egg. Microinjected embryos were cultured in MBL AFSW.

Activin and Nickel Treatments

Recombinant human activin-A was obtained from A.F. Parlow, the National Hormone and Pituitary Program (Harbor-UCLA Medical Center, Torrance, CA) or from R & D Systems (Minneapolis, MN). Embryos were cultured in different concentrations of activin or NiCl2 diluted in MFASW.


Embryos were fixed in −20°C methanol (MeOH) for a minimum of 20 min or stored in MeOH at −20°C indefinitely without significant loss of immunoreactivity. MeOH-fixed embryos were washed in phosphate buffered saline (PBS) with 0.1% Tween 20 (PBST) for 3 × 5 min and blocked in 3% normal goat serum in PBST for a minimum of 30 min. After blocking, embryos were incubated in the appropriate dilution of primary antibody for 1H-ON and then washed for 3 × 5 min in PBST. The embryos were then incubated for a minimum of 30 min in the appropriate dilution of secondary antibody, washed 3 × 5 min in PBST and mounted in 40% glycerol:PBS. Immunofluorescent staining was viewed on a Zeiss Axiovert 100 microscope equipped with epifluorescence and differential interference contrast optics and photographed by using a DAGE video camera.


Monoclonal antibodies recognizing different sea urchin embryonic tissues or cell types were from D. McClay (Duke University, Durham, NC) and included: EctoV (oral ectoderm and foregut), CBA (ciliary band), and Endo-1 (endoderm of the mid- and hindgut). 6A9 is a monoclonal antibody that is specific to primary mesenchyme cells, and was from C. Ettensohn (Carnegie-Mellon, Pittsburgh, PA). Polyclonal anti-MHC recognizes the circumesophageal muscle cells that surround the foregut) and was a gift from G. Wessel (Brown University, Providence, RI). Secondary antibodies were goat anti-mouse fluorescein isothiocyanate (used at 1:100 dilution) and goat anti-rabbit tetrarhodamine isothiocyanate (used at 1:200 dilution), both from Cappel, MP Biomedicals (Irvine, CA). Monoclonal antibody supernatants were used at a 1:2 dilution. Polyclonal anti-MHC was used at a 1:200 dilution. All antibody dilutions were done in PBST.

PCR Cloning and Macroarray Screening

Degenerate primers were designed from sequence comparisons of activin and nodal from various vertebrate species using the consensus-degenerate hybrid oligonucleotide (CODEHOP) program available at The primers were as follows: VNF1-forward 5′-ACT CTA GAA ACG CCT ACM GNT GYG ARG G-3′ and VNR1-reverse 5′-ACG AAT TCA CAT CCA CAC TCC TCN ACD ATC AT-3′. These primers were used in PCR reactions with genomic DNA isolated from S. purpuratus sperm. The PCR product was radiolabeled with 32P and used as probe to screen 7 H and 14 H S. purpuratus cDNA macroarray filters (Davidson Laboratory, California Institute of Technology, Pasadena, CA). 5′-RACE was performed using the GeneRacer Kit (Invitrogen Corporation, Carlsbad, CA). RACE primers used were the 5′ GeneRacer primer and a gene-specific primer (Fig. 7, double arrow) to SpNodal1 (Fig. 6A). Seminested RACE was performed twice using the 5′ Nested GeneRacer primer and the same gene-specific primer as in the first round of amplification. In addition, a partial SpNodal sequence was identified in a “unigene” screen (Poustka et al., 1999, 2003) that encoded 615 bp of coding sequence and a portion of the 5′-UTR. This 5′ clone (SpNodal5; Fig. 6A) was inserted in frame into the longest macroarray-derived clone (SpNodal1; Fig. 6A) to yield SpNodal for microinjection.


Total RNA was isolated from eggs and embryos by using TriAzol reagent (Gibco BRL, NY). The resulting RNAs were DNAse treated at 37°C and quantified by using an Eppendorf BioPhotometer (Brinkmann, Westbury, NY). cDNA was transcribed with random hexamers and 2 μg of total RNA using the Superscript II RT Kit (Invitrogen), according to the manufacturer's instructions. Both RT-PCR and quantitative PCR were performed to determine SpNodal mRNA transcript levels in embryos at different embryonic stages.

For quantitative PCR, cDNAs were diluted to 1 μg/μl and a dilution curve (0.1 μg; 0.01 μg; 0.001 μg; and 0.0001 μg) was established using 64-cell stage S. purpuratus cDNA and SpNodal-specific (5′-GACATC ACCAAGATCGTCAAACAG-3′ and 5′-ATTCTTGTCGTGGTCAGCTTCT-3′) and S. purpuratus mitochondrial rRNA (SpMito; Angerer et al., 2001) specific (5′-ACTCTCTCCTCGGAGCTATA-3′ and 5′-GTATAATTTTTGCGTATTCGGC-3′) primers. Control samples were run without RT to exclude genomic DNA contamination. SpNodal and SpMito quantitative PCR products were quantified by using a Bio-Rad SYBRGreen Supermix (Bio-Rad Laboratories, Hercules, CA). Each mitochondrial 64-cell stage standard was run in duplicate; SpNodal PCR reactions were run in triplicate on cDNA from each cell stage. The number of copies of SpNodal cDNA was determined for each reaction, using the calculated mole fraction of target DNA in the genome, the moles of template in the SpNodal standards, and Avogadro's number.

Primers used for RT-PCR were as follows: BMP2/4 Forward 5′-CAG GCCTACTATTGTCGC-3′, Reverse 5′-GGTACTAGTGCTGGGTTG-3′; Tbx2/4 Forward 5′-CCGAGGCCGCCGAGG TCAG-3′, Reverse 5′-CTTGTTCAGGCC AGGTCCGTTCAG-3′; Spec1 Forward 5′-GAGATGTTGATGGGGATTGC-3′, Reverse 5′-GGATGATTGCTTTGATTTTC-3′.

SpNodal Perturbations

To generate SpNodal RNA for injections SpNodal cDNA was in vitro transcribed using Sp6 and capped as described above. The capped RNA was injected into fertilized eggs at varying concentrations ranging from 1.32 pg/pl to 3.96 pg/pl. To block SpNodal function, an antisense morpholino oligonucleotide was designed against SpNodal sequence (Gene Tools, Philomath, OR) and microinjected into fertilized eggs at varying concentrations ranging from 1.6 to 2.66 pM/pl. The morpholino spanned nucleotides 5′-AGT GACGACATCGTTCCAGCAAAGC-3′, which were located 28 nucleotides upstream of the start of translation (Fig. 7). A standard control morpholino, 5′-CCTCTTACCTCCAGTTACAATTTATAT-3′ (Gene Tools, Philomath, OR), was also injected at comparable concentrations. To further test the specificity of the antisense morpholino for endogenous SpNodal, a construct was designed from the SpNodal clone (Fig. 7), which lacked the sequence recognized by the morpholino, this sequence (ΔSpNodal) was used to generate capped RNA that was coinjected along with the antisense SpNodal morpholino oligonucleotides in rescue experiments.

In Situ Hybridization

In situ hybridization was performed as described in Arenas-Mena et al. (2000). SpNodal sense and antisense cDNA was in vitro transcribed from SpNodal5 cDNA (Fig. 6A) by using digoxigenin-UTP and the Megascript Kit (Ambion Inc.) according to manufacturer's instructions.


The authors thank Drs. Peter Cserjesi (LSUHSC) and Oliver Wessley (LSUHSC) for their critical comments and helpful discussions during the course of this work. We also thank Drs. Charles Ettensohn (Carnegie-Mellon), David McClay (Duke University), and Gary Wessel (Brown University) for generously providing antibodies and Tung-chin Chiang and Dr. John McLachlan (Tulane University School of Medicine) for the use of their Real Time PCR machine. J.M.V. was funded by a grant from the National Science Foundation.