Patterning of the dorsal-ventral axis in echinoderms: insights into the evolution of the BMP-chordin signaling network

Abstract

Formation of the dorsal-ventral axis of the sea urchin embryo relies on cell interactions initiated by the TGFbeta Nodal. Intriguingly, although nodal expression is restricted to the ventral side of the embryo, Nodal function is required for specification of both the ventral and the dorsal territories and is able to restore both ventral and dorsal regions in nodal morpholino injected embryos. The molecular basis for the long-range organizing activity of Nodal is not understood. In this paper, we provide evidence that the long-range organizing activity of Nodal is assured by a relay molecule synthesized in the ventral ectoderm, then translocated to the opposite side of the embryo. We identified this relay molecule as BMP2/4 based on the following arguments. First, blocking BMP2/4 function eliminated the long-range organizing activity of an activated Nodal receptor in an axis rescue assay. Second, we demonstrate that BMP2/4 and the corresponding type I receptor Alk3/6 functions are both essential for specification of the dorsal region of the embryo. Third, using anti-phospho-Smad1/5/8 immunostaining, we show that, despite its ventral transcription, the BMP2/4 ligand triggers receptor mediated signaling exclusively on the dorsal side of the embryo, one of the most extreme cases of BMP translocation described so far. We further report that the pattern of pSmad1/5/8 is graded along the dorsal-ventral axis and that two BMP2/4 target genes are expressed in nested patterns centered on the region with highest levels of pSmad1/5/8, strongly suggesting that BMP2/4 is acting as a morphogen. We also describe the very unusual ventral co-expression of chordin and bmp2/4 downstream of Nodal and demonstrate that Chordin is largely responsible for the spatial restriction of BMP2/4 signaling to the dorsal side. Thus, unlike in most organisms, in the sea urchin, a single ventral signaling centre is responsible for induction of ventral and dorsal cell fates. Finally, we show that Chordin may not be required for long-range diffusion of BMP2/4, describe a striking dorsal-ventral asymmetry in the expression of Glypican 5, a heparin sulphated proteoglycan that regulates BMP mobility, and show that this asymmetry depends on BMP2/4 signaling. Our study provides new insights into the mechanisms by which positional information is established along the dorsal-ventral axis of the sea urchin embryo, and more generally on how a BMP morphogen gradient is established in a multicellular embryo. From an evolutionary point of view, it highlights that although the genes used for dorsal-ventral patterning are highly conserved in bilateria, there are considerable variations, even among deuterostomes, in the manner these genes are used to shape a BMP morphogen gradient.

Figure 1. Position of echinoderms in phylogeny of bilateria and establishment of D/V polarity during early development of the sea urchin embryo.

(A) Simplified phylogenetic tree of bilateria. Echinoderms together with hemichordates form a sister group of the chordates. (B) Scheme describing the early development of the sea urchin embryo together with the fate maps and gene expression territories. D/V polarity first becomes apparent at the beginning of gastrulation by the flattening of the presumptive ventral side, the bending of the gut, and the asymmetrical positioning of the skeletogenic mesenchyme cells that build the spicules. D/V polarity is further accentuated when four arms grow on the ventral side while the mouth opens on the ventral ectoderm and the larva lengthens along the dorsal side to become the pluteus larva.

Figure 2. Rescue of the D/V axis of nodal morpholino injected embryos by an activated Nodal receptor requires BMP2/4.

(A) Scheme of the experiment. Eggs were injected with a morpholino directed against nodal together with a red fluorescent dextran (RLDX). At the eight-cell stage, embryos were then re-injected into one randomly chosen blastomere with the alk4/5/7 mRNA together with a green fluorescent dextran (FLDX) as a lineage tracer. At the 16-cell stage, embryos were sorted according to the type of blastomere targeted by the injection (animal or vegetal) and allowed to develop in separate dishes. Fluorescent and DIC observations were made 72 h after fertilization. Only embryos derived from injection into an animal blastomere are presented here. See Figure S1 for the results of vegetal blastomere injections. (B) (i, ii, vi, vii) Typical morphology of a 48 h rescued embryo derived from injection of alk4/5/7QD mRNA into an animal blastomere. Although development of these embryos is slightly delayed, they display a perfectly normal D/V axis (i) and look like control embryos (xi). (iii–v, viii–x) injection of alk4/5/7QD mRNA in an animal blastomere at the eight-cell stage fails to restore the D/V axis of embryos previously co-injected with the nodal and BMP2/4 morpholinos. These embryos never elongate and look like BMP2/4 morpholino injected embryos (xiv). In these embryos, all cells inherited the morpholino(s), as evidenced by the RLDX red fluorescence (viii). Although the dorsal side is not rescued, the ventral ectoderm is specified as indicated by the expression of goosecoid in the clone of Alk4/5/7QD injected cells (v,x) (100%, n = 17). In (v, x) presence of the FLDX lineage tracer was revealed using an anti-fluorescein antibody conjugated to alkaline phosphatase and using Fast red as substrate. Therefore, the lineage tracer appears in red. Embryos injected simultaneously with the nodal and BMP2/4 morpholinos (xv) are radialized and indistinguishable from embryos injected with the nodal morpholino alone (xii). (ii, vi, vii) are animal views observed under an optical fluorescence microscope. (iv, viii, ix) are latitudinal confocal optical sections of a representative embryo. Red (vi, viii): RLDX; green (vii,ix): FLDX, (ii, iv) show overlays.

Figure 3. BMP2/4 and Alk3/6 are required for specification of the dorsal ectoderm.

Embryos injected with morpholino oligonucleotides directed against BMP2/4 (0.4 mM) (A, G) or alk3/6 (0.8 mM) (C, I) are strongly radialized. They nevertheless retain D/V polarity since the gut is always positioned asymmetrically and bends towards the side of the embryo where a mouth opens (arrow in G and J). A characteristic feature of this phenotype is the presence of ectopic spicules on the side opposite to the mouth (white arrows) associated with a thickened epithelium (black arrows) that resembles the epithelium of the ciliary band (A, C). Injection of lower doses of the BMP2/4 morpholino oligonucleotide (0.25 mM) appeared to affect preferentially development of the dorsal region leaving the ventral region relatively unaffected (B, H). The resulting embryos looked like plutei but had striking truncations of the dorsal side. Similarly, injection of the alk3/6 morpholino at 0.6 mM prevented development of the dorsal side but allowed partial development of ventral arms (D, J). Embryos overexpressing alk3/6QD (E, K) or BMP2/4 (F, L) mRNA are also strongly radialized. When observed at 48 h (E, F), they show a typical elongated shape and are covered by a thin and wrinkled ectoderm characteristic of the dorsal ectoderm. Starting at 72 h, however, they contain very long unbranched spicules (K, L) reminiscent of the spicules normally found on the dorsal side in wild type embryos (R). The embryos in (A–D) are viewed from the animal pole while the embryos in (E–J) are viewed from the side. The embryos in (K, L) are viewed from the vegetal pole. (M–T) Rescue experiment. While all the embryos injected with the alk3/6 morpholino alone developed with a radialized phenotype (P, Q), about half of the embryos injected with both the alk3/6 morpholino and a 300 mg/ml solution of synthetic alk3/6 mRNA containing nine mismatches in the sequence recognized by the morpholino (alk3/6mm) developed into pluteus larvae, like control embryos (R) and embryos injected with the alk3/6mm mRNA (S, T). The remaining embryos were either only partially rescued (30%) or not rescued to a significant extent.

Figure 4. Knockdown of BMP2/4 or Alk3/6 suppresses specification of the dorsal ectoderm and expands the ciliary band territory without affecting the expression of ventral marker genes.

Expression of the dorsal marker gene tbx2/3 is abolished in the bmp2/4 (B) or alk3/6 (C) morphants. The goosecoid gene is expressed in the ventral ectoderm of control embryos (D, G). goosecoid expression is unaffected in bmp2/4 (E, H) or alk3/6 morphants (F, I). In control late gastrulae, hnf6 expression is restricted to a belt of ectodermal cells at the boundary between the ventral and dorsal ectoderm that constitutes the presumptive neurogenic ciliary band (J, M, P). In both the bmp2/4 (K, N, Q) or alk3/6 (L, O, R) morpholino injected embryos, hnf6 expression dramatically expands into the presumptive dorsal ectoderm, filling most of this territory. At 48 h, Delta expression labels individual neurons of the facial ectoderm and ciliary band (S). In both the bmp2/4 and alk3/6 morphants, strong ectopic expression of Delta is detected in the presumptive dorsal ectoderm. The sm30 gene, which encodes a spicule matrix protein , is preferentially expressed in the bilateral clusters of skeletogenic mesenchyme cells that form at the level of the presumptive ciliary band (V) . In both the alk3/6 and bmp2/4 morphants, clusters of skeletogenic mesenchyme cells expressing sm30 are present under the thickened ectoderm on the presumptive dorsal side (W, X). These phenotypes are consistent with the idea that the default fate of the dorsal ectoderm in the absence of BMP signaling may be a neurogenic ciliary band-like fate. Embryos in (C, G–I, M–R) are seen from the side while embryos shown in (A, B, D–F, J–L, S–X) are seen from the vegetal pole.

Figure 5. Kinetics, pattern, and BMP2/4 dependence of pSmad1/5/8 during development of the sea urchin embryo.

(A) Time course analysis of pSmad1/5/8 signaling analyzed by Western blot. 60, 60-cell stage; eB, early blastula; hB, hatching blastula; sB, swimming blastula; MB, mesenchyme blastula; lMB, late mesenchyme blastula; veG, very early gastrula; Pr, prism stage; Pl, pluteus stage. The major antigen recognized by the anti-pSmad1/5/8 antibody accumulates at mesenchyme blastula and gastrula stages. (B) pSmad1/5/8 pattern during embryogenesis. The pSmad1/5/8 signal first appears at the late blastula stage, immediately before ingression of the primary mesenchyme cells, in roughly half of the embryo (i, ii, v). At mesenchyme blastula stage, strong nuclear staining is observed in half of the embryo, both in the ectoderm and in skeletogenic mesenchyme cells (iii, vi). At gastrula stages, the staining is still restricted to half of the embryo and encompasses all three germ layers (iv, vii). Note that some nuclei within the clusters of skeletogenic mesenchyme cells are labeled (white solid arrowheads). (viii) Reconstructed 3D animal view of a representative control embryo using 40 section images and using the “Volume viewer” ImageJ plugin. (ix) Representative single section image corresponding to a latitudinal section of a control embryo, dorsal side is on the right. (x) Surface plotting of the image in (ix) using the “Interactive 3D Surface Plot v2.32” ImageJ plugin. The curve is centered on the dorsal side of the embryo. All the images are colored with false colors. Low signals are colored in blue, high signals are colored in white. (C) pSmad1/5/8 staining is dependent on BMP2/4 and Alk3/6. (i, iv) Control mesenchyme blastula embryos displaying asymmetric nuclear pSmad1/5/8 immunostaining on one side of the embryo. This staining is greatly reduced in alk3/6 morpholino injected embryos (ii, v). In these embryos only residual patches of fluorescent nuclei are detected (white arrows). In bmp2/4 morpholino injected embryos, no staining is detectable. The embryos in (i–iii) are seen from the animal pole while the embryos in (iv–vi) are seen from the side.

Figure 6. Ventral Nodal signals induce BMP2/4 translocation to the dorsal side where it activates tbx2/3 and msx in nested patterns.

(A) (i) Scheme of the experiment. pSmad1/5/8 immunostaining at the mesenchyme blastula stage in control embryo (ii), nodal morpholino injected embryo (iii), and embryos injected with nodal morpholino at the one-cell stage then injected into one blastomere with alk4/5/7QD mRNA at the eight-cell stage (v, vi). The nodal morpholino was injected together with a red fluorescent dextran and the alk4/5/7QD mRNA with a green fluorescent dextran. Both tracers can be seen in the fully rescued embryo in (iv). (B) Combined in situ hybridization and pSmad1/5/8 immunostaining reveal that BMP signaling is restricted to the dorsal side of the embryo. (i, ii) In situ hybridization with a probe for the ventral marker gene goosecoid (blue) confirms that goosecoid transcripts are expressed on the opposite side of the pSmad1/5/8 immunostaining (red). In contrast, tbx2–3 transcripts (blue) are always found on the same side as the pSmad1/5/8 immunostaining signal (red) (iii, iv). (C) Nested expression of dorsal marker genes along the D/V axis. Double in situ hybridizations with nodal and tbx2/3 (i, ii, v, vi) or nodal and msx (iii, iv, vii, viii) probes. tbx2/3 is expressed in the whole dorsal ectoderm territory adjacent to the ciliary band. In contrast, msx is expressed in a subdomain of the dorsal ectoderm more distant from the ciliary band. Expression of both msx and tbx2/3 is regulated by BMP2/4 (Lepage et al. unpublished data). The embryos in (i–iv) are viewed from the animal pole. The embryos in (v–viii) are viewed from the side. (v) and (vii) are surface views. (D) Activation of target genes by the BMP2/4 morphogen gradient. msx and tbx2/3 are expressed in nested patterns centered on the dorsal midline.

Figure 7. Sea urchin chordin is an early zygotic transcript expressed in the ventral ectoderm downstream of Nodal signaling.

(A) Structure of the Chordin protein and cDNA. VWC, von Willebrand factor type C; CHRD, Chordin domain. (B) Northern blot analysis of chordin transcripts during sea urchin development. Embryonic stages are egg (O), 16 cells (16), 60 cells (60), early blastula (eB), swimming blastula (sB), mesenchyme blastula (mB), early gastrula (eG), late gastrula (lG), prism (Pr), and pluteus (Pl). (Li), embryos treated with lithium; (Ni), embryos treated with nickel; (Diss), dissociated embryos. Loading control is 28S mRNA. chordin is zygotically expressed starting at the swimming blastula stage. Its expression increases at the mesenchyme blastula stage, then decreases progressively during gastrulation up to the pluteus stage. chordin transcripts are not detected in dissociated or lithium treated embryos but are overexpressed in nickel treated embryos. (C) Spatial expression analysis of the chordin and bmp2/4 transcripts by in situ hybridization. Embryonic stages are: 128 cell-stage (i), swimming blastula (ii–iv, xi, xii), mesenchyme blastula (v, vi), gastrula (vii, viii, xiii), late gastrula (ix, xiv), early pluteus (x, xv). (v) is a double in situ hybridization with chordin and GATA1/2/3, which is expressed in the non-skeletogenic precursors on the ventral side of the vegetal plate, while (iv) is a double in situ hybridization with chordin and bmp2/4 probes showing that the two genes are expressed in a highly similar pattern. chordin expression begins at the swimming blastula stage and is strictly restricted to the ventral ectoderm up to the early pluteus stage where it is only maintained in a subdomain of the ciliary band. Note that at the pluteus stage, BMP2/4 expression in the ventral ectoderm has faded while strong expression is initiated in the dorsal-most skeletogenic mesenchyme cells (xv). (D) In situ hybridization of chordin transcripts at the mesenchyme blastula stage in a control embryo (i), or in an embryo injected with either the nodal morpholino (ii), nodal mRNA (iv), alk4/5/7QD mRNA (v), or treated with the Nodal receptor inhibitor SB431542 (iii).

Figure 8. Chordin is required for patterning the D/V axis and for preventing pSmad1/5/8 signaling on the ventral side.

(A) Blocking Chordin function using antisense morpholino oligonucleotides severely perturbs D/V polarity. About 80% of the embryos injected with Mo1 chordin at either 1 or 1.5 mM developed with a radialized phenotype with numerous ectopic spicules (i, vi). The remaining embryos were polarized along the D/V axis but failed to elongate and conserved a round shape (ii, vii). In some experiments, however, injection of the same doses of morpholino produced a milder but very reproducible phenotype characterized by a luge-like shape due to the parallel growth of the body rod spicules, and the presence of a proboscis in the animal region in place of the oral arms (iii, iv, viii, ix). Injection of Mo2 chordin at 0.6 mM produced predominantly the luge-like phenotype (iv, ix). The strong and mild phenotypes were observed in multiple experiments, one phenotype or the other prevailing in a given experiment depending on the batch of embryos. Treatments with low doses (1 μM) of the Alk4/5/7 inhibitor SB431542 phenocopy the mild chordin morpholino phenotype (v, x). In situ hybridization performed on these SB431542 treated embryos showed that expression of chordin is strongly reduced compared to controls (Lepage et al. unpublished data). Overexpression of chordin mRNA disrupted D/V patterning and caused complete (xii, xiii) or partial (xiv, xv) radialization. Here some variability was observed within a given experiment, with about half of the embryos developing with the strong phenotype and the other half with the milder phenotype (n>500; 5 experiments). Strongly affected embryos showed numerous ectopic spicules, a straight archenteron, and never elongated while less affected embryos acquired a bilateral symmetry but failed to elongate. (B) pSmad1/5/8 immunostaining of mesenchyme blastula stage embryos. While control embryos showed a restricted dorsal nuclearization of pSmad1/5/8 (i), embryos injected with the Mo1 chordin morpholino showed a broad expansion of the nuclear staining to the whole circumference of the ectoderm (ii). Embryos injected with Mo2 chordin also showed an expansion of the pSmad staining (iii), but it appeared less dramatic than that observed with Mo1 chordin. Conversely, pSmad1/5/8 staining was either not detectable or strongly reduced in most embryos (80%, n = 10) injected with the chordin mRNA (iv). (C) In situ hybridization showing tbx2/3 transcripts expressed on the dorsal side in control embryos at mesenchyme blastula (i) and gastrula (v). The expression domain of tbx2/3 was strongly expanded in most (30/40) embryos injected with a morpholino directed against chordin (ii, vi) covering the whole circumference of the ectoderm, while in the remaining embryos it appeared only slightly expanded (iii). This expansion of tbx2/3 in chordin morpholino injected embryos is still visible at the gastrula stage but is less dramatic since most embryos at this stage show a ventral restriction of tbx2/3 expression (vii). Injection of chordin mRNA causes the opposite effect and eliminates tbx2/3 expression (n = 29) at the mesenchyme blastula stage (iv).

Figure 9. BMP2/4 is widely diffusible in the absence of Chordin.

(A) Scheme of the experiment. At the two-cell stage, embryos were injected into one blastomere with either mRNA encoding the activated Alk3/6 receptor, alk3/6QD, or with the bmp2/4 mRNA together with a green fluorescent dextran (FLDX) as a lineage tracer. To prevent chordin expression the embryos were treated continuously with the Nodal receptor inhibitor SB431542 starting soon after fertilization and up to the mesenchyme blastula stage when they were fixed and immunostained with the anti-pSmad1/5/8 antibody. (B) (i, ii) Control embryos showing asymmetrical nuclear pSmad1/5/8 staining in the dorsal half of the embryo. Embryos treated with SB431542 did not show any pSmad1/5/8 signal in any cell (iii, iv). Embryos injected at the two-cell stage with mRNA encoding Alk3/6QD, the activated BMP receptor, and treated with SB431542 displayed strong nuclear staining in half of the embryo. This staining strictly co-localized with cells that received the alk3/6QD mRNA (v, ix). Embryos injected at the two-cell stage with mRNA encoding BMP2/4 and treated with SB431542 showed nuclear pSmad1/5/8 staining in all cells of the ectoderm (xiii–xx). Note that the territory showing nuclear pSmad includes as many cells that received the bmp2/4 mRNA as cells that did not receive the mRNA (xiii, xv). The pSmad signal is indicated by the red fluorescence (i, iii, ix, xi, xvii, xix), position of the nuclei is revealed by Hoechst staining (ii, iv, vi, viii, xiv, xvi), and the lineage tracer is detected by the green fluorescence of the fluoresceinated dextran (x, xii, xviii, xx). Sagittal confocal optical sections of representative embryos (i–iv, vii, viii, xi, xii, xv, xvi, xix, xx). Latitudinal confocal optical section of representative embryos (v, vi, ix, x, xiii, xiv, xvii, xviii).

Figure 10. Glypican 5 expression is progressively restricted to dorsal cells and is regulated by BMP2/4-Alk3/6 signaling.

(A) Expression pattern of glypican 5 during embryogenesis. (i) egg; (ii) 60-cell stage; (iii) early blastula; (iv) swimming blastula; (v, vi) mesenchyme blastula; (vii, viii) gastrula. (B) After an initial phase of expression in a belt of cells around the equator (iv), glypican 5 expression becomes strongly asymmetric (v–viii). (B) The dorsal expression of glypican 5 was abolished in bmp2/4 (ii, vi) or alk3/6 (iii, vii) morphants. In contrast, glypican 5 was ectopically expressed throughout the ectoderm in bmp2/4 (iv, viii) or alk3/6QD (unpublished data) overexpressing embryos.

Figure 11. A model for morphogen gradient formation in the sea urchin embryo.

(A) chordin and bmp2/4 are both expressed in the ventral ectoderm downstream of Nodal while the presumed Chordin protease Tolloid/BMP1 , the BMP2/4 receptor alk3/6, and glypican 5 are expressed widely throughout the ectoderm at blastula stages. BMP2/4 ligands expressed on the ventral side of the embryo are immediately complexed with Chordin and cannot bind to their receptors. Therefore, on the ventral side of the embryo, no free BMP2/4 ligand can bind to BMP receptors since an excess of Chordin is able to constantly reform Chordin/BMP2/4 complexes. (B) The Chordin/BMP2/4 complexes produced on the ventral side of the embryo can nevertheless diffuse towards the dorsal side of the embryo where Tolloid/BMP1 releases free BMP2/4 ligands that, in the absence of Chordin, can bind to their receptors and trigger Smad1/5/8 phosphorylation and nuclearization. BMP signaling on the dorsal side up-regulates expression of glypican 5, which may facilitate BMP2/4 mobility and BMP2/4 binding to its receptor, thereby reinforcing BMP2/4 signaling via a positive feedback loop.

Figure 12. Schematic diagram comparing the expression domains of BMP2/4 and Chordin homologues in different experimental models.

The relative expression patterns of BMP and chordin in organisms belonging to the main clades of metazoan (protostomes and deuteurostomes together with diploblastic cnidarians as outgroup) are depicted. O, oral; Ab, aboral; D, dorsal; V, ventral. While in most organisms chordin and BMP are expressed in mutually exclusive patterns, in the sea urchin and in cnidarians these genes are co-expressed on the same side of the embryo.