Opposing Nodal and BMP Signals Regulate Left–Right Asymmetry in the Sea Urchin Larva

A left-right patterning study in developing sea urchin shows that the opposing roles of Nodal and BMP signaling in patterning the left-right axis are conserved in deuterostomes.


Introduction
One of the most fascinating features of bilaterian development is the consistent left-right (LR) asymmetry of their internal organs. In the past two decades, studies of molecular pathways controlling LR asymmetry have broadened our knowledge regarding the dissimilar and conserved mechanisms among different animal models [1]. It has been suggested that the mechanisms of initial symmetry breaking are not conserved across different vertebrate classes [2]. For example, a leftward fluid flow generated by the posteriorly tilted nodal cilia initiates left-sided gene expression in the mouse embryo [3]. On the other hand, differential activities of ion pumps in early stage embryos are important for establishing LR asymmetry in the chick, frog, and zebrafish [4]. Although detailed symmetry-breaking mechanisms vary among different vertebrate species, the common outcome is the expression of nodal, which encodes a transforming growth factor b (TGFb) ligand, in the left lateral plate mesoderm (LPM) [5]. The Nodal pathway not only regulates LR asymmetry in vertebrates but also controls the formation of the left-sided adult rudiment in sea urchins [6] and body chirality in snails [7].
Bone morphogenic protein (BMP), another TGFb family member, is also involved in LR patterning [8]. In opposition to conserved nodal expression on the left side, BMP transcripts or activities are observed on the right side of the node or LPM. For example, bmp4 is expressed on the right side of Hensen's node in the chick embryo and initiates a right-sided signaling cascade [9]. Although bmp genes are expressed symmetrically in the LPM of the mouse and chick embryos, BMP activity is greater on the right side due to the presence of BMP antagonists on the left side [10][11][12][13]. A BMP/ALK2/Smad-mediated signaling pathway is also proposed to be active on the right side of the Xenopus embryo [14]. Therefore, right-sided BMP opposing left-sided Nodal seems to be a conserved feature. The role of BMP in LR patterning is largely unknown in invertebrates, even though right-sided expressed dpp-bmp2/4 plays important roles in shell formation and coiling in gastropods [15,16].
Sea urchins belong to the phylum Echinodermata, which is characterized by a pentaradiate adult body plan. In indirect developing sea urchins, the adults are derived from bilaterally symmetric larvae. The transition from a bilaterally symmetric to pentasymmetric body plan relies on a LR asymmetrical control that results in the formation of an adult rudiment on the left side of the larva ( Figure S1). During gastrulation, a coelomic pouch (CP) composed of veg2 mesoderm and small micromeres (Smm) forms at the archenteron tip and later separates into left and right pouches in the bilateral gastrula [17]. The first morphological signature of LR asymmetry in the pluteus larva is the extension of a duct-like structure, the hydroporic canal (HC), from the left CP to the aboral ectoderm where the hydropore forms [18]. The ciliated HC is thought to be an excretory organ that contributes to normal body width maintenance in the larva [19,20] and later differentiates into a portion of the adult water vascular system [21]. The differentiated left coelom together with the invaginated left oral ectoderm, called the primary vestibule, develop into an adult rudiment with pentaradial symmetry [22,23].
Our knowledge of the molecular mechanisms concerning LR patterning in sea urchins is relatively limited. It has been shown that sea urchin LR axis specification depends on cell interactions [24]. A series of microsurgery experiments revealed that the positioning of the adult rudiment on the left side is directed by signals from the right side [25]. Duboc et al. further demonstrated that nodal expression on the right side, which is reversed when compared to vertebrates, prevents the formation of the adult rudiment [6]. In addition, signals emitted from the micromeres also regulate LR asymmetry [26], although the identity of the micromere-derived signal remains unknown. It is also not known whether positive signals or a default pathway are required for the left-sided structure development.
In this study, we focused on the role of the BMP pathway and examined the molecular basis of LR asymmetry in the sea urchin embryo. We found that bmp genes are symmetrically expressed in skeletogenic micromeres, but BMP signaling is asymmetrically activated in the left CP-derived HC. Through cell lineage analysis, we detected active BMP signaling in veg2 descendants but not in the Smm. We further provided evidence that BMP signaling is required for left-sided structure development and the expression of several left-sided marker genes. We also demonstrate that rightsided Nodal signaling restricts BMP activity and is involved in the asymmetrical separation and apoptosis of the Smm. We discuss these findings in the context of Nodal and BMP signaling in patterning LR asymmetry in the sea urchin embryo.

pSmad1/5/8 Was Detected on the Left Side of the Larva
To study the role of BMP signaling in LR asymmetry in sea urchins, we first examined the expression patterns of genes related to the BMP signaling pathway. The sea urchin genome contains three bmp ligand genes: bmp2/4, bmp3, and bmp5-8 [27]. Bmp2/4 is initially transcribed in the oral ectoderm at the blastula stage, but the Bmp2/4 ligand translocates to the aboral side and plays key roles in the aboral ectoderm gene regulatory network [28][29][30][31][32][33]. The expression patterns of sea urchin bmp3 and bmp5-8 have not been elucidated. Therefore, we performed quantitative PCR and discovered that the bmp3 transcripts are not detectable during the first 3 d of development, whereas bmp5-8 was expressed in the egg and during this period (unpublished data). In situ hybridization demonstrated that bmp2/4 expression shifted from the oral ectoderm to the aboral skeletogenic mesenchyme cells during gastrulation and remained in a few cells at the apex of the pluteus larva ( Figure 1A). This expression pattern is similar to Pl-bmp2/4 from sea urchin Paracentrotus lividus; however, the timing for the expression domain shift occurs later in this species because its oral ectodermal expression can still be observed in the gastrula stage [27,31]. The bmp5-8 transcripts were ubiquitously detected in the egg and later in the whole ectoderm at the blastula and early gastrula stages. Similar to bmp2/4, bmp5-8 expression also shifted to the aboral skeletogenic cells at the late gastrula and pluteus stages ( Figure 1A). Both the bmp2/4 and bmp5-8 genes were bilaterally expressed during all analyzed stages. We further examined the expression patterns of BMP receptors (alk1/2, alk3/6, bmpr2, and acvr2) and did not observe asymmetrical LR expression (unpublished data). We then investigated BMP signaling activity by monitoring the phosphorylation and nuclear translocation of its downstream mediator Smad1/5/8 (pSmad). Immunostaining with a pSmad antibody showed that BMP signaling was activated in the aboral ectoderm at the blastula stage, as previously reported [29,31]. During gastrulation, pSmad was bilaterally detected in the skeletogenic, ectodermal, and archenteron cells on the aboral side of the embryo ( Figure 1B,C). In addition, we observed an LR asymmetric pSmad pattern with a stronger signal on the left CP of the pluteus larva ( Figure 1D-G). Detailed observation revealed that the staining was restricted to the HC but was absent at the base of the left CP ( Figure 1H,I). These results suggest that although bmp2/4 and bmp5-8 are symmetrically expressed, BMP signaling is activated on the left side in the pluteus larva and may play a role in controlling leftsided development in sea urchins.

pSmad-Positive Cells in the Left CP Are veg2 Descendants
The CPs consist of two cell lineages: Smm and veg2 descendants [34]. Because only some cells in the left CP were labeled with pSmad, we examined the lineage of these cells by using BrdU pulse-chase labeling at the one-cell stage to mark the slow cellcycling Smm [35]. At the gastrula stage, BrdU-labeled Smm were located in the roof of the archenteron, while the aboral side of the archenteron tip was stained with pSmad in a graded manner from the tip to the base ( Figure 2A). Later at the pluteus stage, pSmad was detected in the cells constituting the HC and did not overlap with the BrdU-labeled Smm ( Figure 2B). These results indicate that the pSmad-positive cells in the left CP were derived from the veg2 cells and differentiated into the HC. Smm lacking pSmad staining were located in the roof of the archenteron at the gastrula stage and later constituted the base of the left CP in the pluteus larva.

Differential Gene Expression at the Archenteron Tip and the Left CP
Several genes are known to be expressed at the tip of the archenteron during gastrulation and later in the CPs, but the cell lineages in which these genes are expressed are unknown. Therefore, we examined the expression patterns of several known left-sided markers in conjunction with pSmad staining that labeled veg2 descendants or a Drosophila Vasa antibody that labeled Smm similarly to BrdU ( Figure S2A) [36]. We detected soxE transcripts [6,37] at the gastrula stage on the aboral side of the archenteron tip (aboral tip) where pSmad staining was strongest ( Figure 2C) and Vasa protein was absent ( Figure 2D). At the pluteus stage, soxE was expressed in the HC of the left CP but not in Vasa-positive Smm ( Figure 2E), which was similar to the pSmad pattern. Comparable expression patterns were observed for the left-sided markers pax6, six1/2, and genes encoding Six1/2 cofactors eya [38] Author Summary Bilateral animals that are externally symmetric in appearance often have internal organs that are asymmetric with respect to the left and right sides of their bodies. Two signals, Nodal and BMP, have been shown to establish this asymmetry during vertebrate embryogenesis. We investigate here whether the same mechanisms that establish left-right patterning in vertebrates are conserved in invertebrate animals, specifically in the California purple sea urchin. This sea urchin passes through various stages in its lifecycle before developing to adulthood, including a feeding larva stage in which the tissue that goes on to form the adult, the so-called adult rudiment, forms on the left side. Previous studies have shown that right-sided Nodal signaling in sea urchins prevents the formation of the adult rudiment. In this study, we show that BMP signaling is activated on the left side and is required for the development of this left-sided structure. We also demonstrate that Nodal signaling blocks BMP activity and induces apoptosis of the tissue that forms the adult rudiment on the right side. We propose that the roles of Nodal and BMP signals in establishing left-right asymmetry are evolutionarily conserved. and dach ( Figure 2F-I). However, dach was also expressed on the oral side of the archenteron tip at the late gastrula stage ( Figure  S2B,C).
We also reexamined germ-line-associated genes known to be expressed in the Smm, such as vasa, seawi, and nanos2 [37]. Indeed, these genes were expressed in Vasa protein-positive Smm located in the archenteron roof at the gastrula stage ( Figure 2J,K). Vasa and seawi transcripts were also detected on the aboral side of the archenteron tip ( Figure S2D, arrows). At the pluteus stage, vasa and seawi continued to be expressed in the Smm in both CPs, although nanos2 expression quickly disappeared on the right side ( Figure 2L-N). We also noted that vasa and seawi mRNA signals were stronger on the left side.
Three forkhead genes, foxC, foxF, and foxY, were reported to be expressed in the archenteron tip during gastrulation and later in the CPs [39][40][41]. We detected foxC transcripts on the oral side of the archenteron tip at the gastrula stage ( Figures 2O and S2E). This expression pattern marked a novel domain (oral tip) that is not derived from the Smm that contain Vasa protein and nanos2 transcript and does not express soxE ( Figures 2P and S2F). As the CP budded out from the archenteron tip and separated into the left and right CP, foxC expression decreased in the oral tip and was detected in the Smm ( Figures 2Q,R and S2F). The foxF expression pattern was broader than that of foxC at the gastrula stage. In addition to the expression in the oral tip, foxF was also weakly expressed in the aboral tip ( Figure 2S,T). At the pluteus stage, foxF transcripts remained in the oral tip cells that constitute part of the CPs adjacent to the esophagus (esophageal domain) (Figures 2U,V and S2F). The third forkhead gene, foxY, was expressed in the Smm from the blastula stage to the gastrula stage; however, transcripts were not detected at the pluteus stage ( Figure S2G).
Collectively, there are at least three domains on the archenteron tip at the gastrula stage: the veg2-derived oral and aboral tips and the Smm located on the roof ( Figure 2W). Each of these three domains expressed a set of genes. FoxC, foxF, and dach were expressed in the oral tip; the aboral tip contained pSmad signal and expressed soxE, pax6, six1/2, eya, dach, vasa, seawi, and foxF; and the roof contained Vasa protein and expressed vasa, seawi, nanos2, and foxY. During CP formation, the pSmad signal on the right side disappeared. After the two CPs separated, the left CP was distinguishable from the right CP based on the pSmad signal and the larger expression domains of vasa and seawi ( Figure 2X). At the pluteus stage, the left CP could also be distinguished into three domains based on gene expression patterns ( Figure 2Y). The HC was derived from the left aboral tip of the archenteron and was marked with pSmad and expressed the same set of genes, except vasa, seawi, and foxF. The cells at the oral tip of the archenteron moved to a position that lay next to the esophagus and continued to express foxF. These cells were at the same location as the myoblasts that were labeled with an anti-actin antibody, and they may have later extended processes and differentiated into esophageal muscles [42]. The Smm located in the roof of the archenteron at the gastrula stage moved to the base of the CP in the pluteus larva, which represented the third domain that expressed foxC and germline markers vasa, seawi, and nanos2.

BMP Signaling Is Required for Left-Sided Structure Formation and Gene Expression in veg2 Descendants
Given that pSmad was detected in the HC, we next examined whether BMP signaling is required for HC and rudiment formation. To bypass the early function of BMP signaling in oral-aboral (OA) axial patterning, we used pharmacological treatments to perturb BMP signaling prior to LR axis establishment. A small molecule, dorsomorphin (DM), which selectively inhibits BMP type I receptors and blocks Smad phosphorylation, was previously identified in a screen for compounds that perturb dorsoventral (DV) axis formation in zebrafish [43]. To test whether DM inhibited BMP signaling in the sea urchin, we treated the embryos from fertilization to the mesenchyme stage and performed immunoblot analysis using the pSmad antibody ( Figure  S3A and Text S1). We observed a dose-dependent reduction of pSmad ( Figure S3B). DM also reduced the expression level of hox7, a downstream target gene of BMP signaling [33], as well as pSmad staining intensity ( Figure S3C). Therefore, DM inhibits BMP signaling in the sea urchin. Conversely, treating embryos with recombinant mouse BMP4 protein (mBMP4) as an exogenous source of BMP ligand expanded hox7 expression and pSmad signal ( Figure S3C). When the embryos were treated with DM or mBMP4 after the OA axis was established ( Figure S3H), we observed defects in CP and HC formation. In DM-treated embryos, the levels of bmp2/4 transcript and pSmad signal were attenuated in the aboral skeletogenic cells at the late gastrula stage ( Figure 3A), indicating that bmp2/4 expression is controlled by its own signaling. The hydropore was not observed in treated pluteus larva, and these embryos also lacked left-sided pSmad staining and ciliated HC ( Figure 3B). We observed that the pSmad signal persisted in the aboral skeletogenic and ectodermal cells, indicating that DM did not completely abolish BMP signaling. After the inhibitor was washed out from the culture, 79% of the DM-treated embryos recovered to form rudiments on the left side at the advanced rudiment (AR) stage ( Figure 3C). This was significantly lower than that of the controlled embryos (98%). Eighteen percent of the treated embryos formed rudiments on both sides ( Figure 3B, bottom panels), and 4% had right-sided rudiments. Ectopic mBMP4 treatment resulted in random positioning of a single undivided CP with soxE expression in 89% of the embryos ( Figure 3D, open arrowheads). In 11% of mBMP4-treated embryos, two HCs with bilateral soxE expression formed ( Figure 3D, solid arrowheads, and 3E). These results indicate that BMP signaling is required for proper CP differentiation and left-side HC formation.
We further analyzed LR marker gene expression patterns after BMP signaling was perturbed. The expressions of all examined genes that were normally expressed in the aboral veg2 descendants, including soxE, pax6, six1/2, eya, and dach, were diminished in DMtreated embryos but remained in the single CP when BMP signaling was elevated ( Figure 3F). Smm genes, such as vasa, seawi, nanos2, and foxC, continued to be expressed at the tip of the archenteron in DM-treated and in the single CP in mBMP4treated embryos ( Figure 3G). These results are in agreement with the lineage analysis demonstrating that BMP signaling acts on aboral veg2 descendants but not on Smm. Surprisingly, we also discovered that inhibiting and elevating BMP signals both resulted in the loss of right-sided gene expression, including nodal and its downstream targets lefty, pitx2, and not ( Figure 3H) [6,39,44]. The requirement of BMP signals for nodal expression is likely indirect because pSmad was not detected in the right lateral ectoderm where nodal is expressed ( Figure 1C). We further examined nodal expression at the late gastrula stage when its right-sided expression began to distinguish whether BMP signals are required for initiation or maintenance of nodal expression. The results showed that all embryos lost their right-sided nodal expression when BMP signaling was blocked ( Figure 3I). The expression of nodal either disappeared (84%) or was retained in the oral ectoderm (16%). These results indicate that BMP is required for nodal expression initiation, although the mechanism remains unknown. Although DM has been used as a selective BMP signaling inhibitor [43,45], it also inhibits a panel of kinases in vitro [46]. Therefore, to specifically inhibit BMP signaling and bypass its early function, we treated the embryos with vivo-morpholinos (vMOs), which are antisense morpholino oligonucleotides linked to eight guanidinium head groups for effective cellular membrane penetration [47]. vMOs have been shown to be effective in a variety of systems, including mice, chick embryo, adult zebrafish, and cultured cells [47][48][49]. We first tested the efficacy and specificity of BMP2/4 vMO in sea urchin embryos. We observed that BMP2/4 vMO effectively blocked green fluorescent protein (GFP) expression in a dose-dependent manner when the embryos were injected with mRNA containing the vMO binding site fused to the GFP sequence ( Figure  S3D-F and Text S1). The effect was specific because GFP fluorescence was not attenuated by the control or non-specific vMO. When embryos were treated at the 1-cell stage, BMP2/4 vMO also blocked expression of the downstream target hox7 but had little effect on the non-target chordin ( Figure S3G, MB). When the vMO was added later at the mesenchyme blastula stage, similar effects were observed ( Figure S3G, LG). Therefore, vMOs are effective in the sea urchin embryos and can be used at various developmental stages. When embryos were treated with BMP2/4 vMO from the mesenchyme blastula stage to the late gastrula stage, pSmad staining at the HC disappeared, but vasa expression remained in the Smm ( Figure 3J). In addition, the expression of nodal and the leftsided genes soxE, pax6, six1/2, and eya disappeared, which was similar to the effects induced by DM ( Figure 3I,J). However, the effects of DM and BMP2/4 vMO on dach expression were different ( Figure 3F,J). Dach expression was absent in DMtreated embryos, but its transcripts remained on the archenteron tip in BMP2/4 vMO-treated embryos. Given that dach is expressed in both the oral and aboral archenteron tip at the late gastrula stage ( Figure S2B), we suspected that its oral tip expression domain was not affected by BMP2/4 vMO, and the complete disappearance following DM treatment might be due to a non-specific drug effect. We also examined the effects of BMP2/4 vMO on adult rudiment formation. Similar to the effects of DM, most embryos treated with BMP2/4 vMO recovered and formed rudiments on the left side ( Figure 3K). A small but significant portion of the treated embryos developed bilateral rudiments, presumably due to the requirement of nodal expression by BMP signaling. Taken together, these results supported the hypothesis that BMP signaling is required for proper LR patterning and the expression of the left-sided genes that are normally expressed in aboral veg2 descendants.

Nodal Signaling Inhibits BMP Signaling and Its Downstream Target Gene Expression
Because Nodal is known to function on the right side to prevent the rudiment formation in sea urchins [6], we further examined the relationship between Nodal and BMP signaling during LR axis establishment. At 42 h post-fertilization (hpf), nodal expression was in the oral ectoderm and pSmad staining and soxE expression were bilateral at the tip of archenteron ( Figure 4A). At 48 hpf (late gastrula stage), nodal expression shifted to the right lateral ectoderm and pSmad staining and soxE transcripts started to restrict on the left side. By 54 hpf, nodal expression remained in the right lateral ectoderm, whereas pSmad and soxE expression were detected in the differentiating left CP. Therefore, the appearance of nodal transcripts on the right side seems to correlate with decreased BMP signaling on the right side. We then examined whether the disappearance of BMP signaling on the right side was regulated by Nodal signals. When Nodal signaling was inhibited with small molecule compounds SB-431542 and SB-505124, pSmad staining became bilateral, and two ciliated HCs also formed, which resulted in two hydropores on the aboral ectoderm ( Figures 4B and S4A, arrows). In addition, elevating Nodal signaling with recombinant human Activin (hActivin) prevented HC formation. These results suggest that Nodal signals are upstream of BMP signaling and normally block BMP action on the right side.
We further analyzed the effects of Nodal signaling on LRrelated genes. When Nodal signaling was blocked, nodal, lefty, pitx2, and not were downregulated, confirming that Nodal signaling is upstream of the right-sided genes ( Figure S4B). However, enhancing Nodal signaling with hActivin did not lead to bilateral expression of right-sided genes. This result is different from the previous study in P. lividus sea urchin [6], suggesting that the transcriptional regulation of the nodal gene might differ in the two species. Consistent with the idea that Nodal signals block BMP signaling, we observed that all left-sided BMP signaling downstream target genes were expressed bilaterally in both CPs when Nodal signaling was blocked, whereas hActivin inhibited their expression ( Figures 4C and S4C, arrowheads). Interestingly, Nodal signaling had different effects on genes that are expressed in the Smm. The expression of nanos2 became bilateral when Nodal signaling was blocked. In the hActivin-treated embryos, nanos2 expression was diminished, and foxC and foxF transcripts were reduced. The LR asymmetric vasa and seawi signals became symmetric when Nodal signaling was blocked (Figures 4D and  S4D). Similar effects were also observed when the embryos were treated with Nodal vMO (Figures 4E and S4E). These results suggest that right-sided Nodal signaling represses nanos2 expression and controls the asymmetric expression of vasa and seawi.

Nodal Signaling Regulates Asymmetrical Separation and Induces Small Micromere Apoptosis
Although we showed that Nodal signaling prevented left-sided development by inhibiting BMP signaling in veg2 descendants, the effects of Nodal signaling on the Smm are unclear. A detailed analysis of nodal and pitx2 transcripts revealed that in addition to the expression in the right lateral ectoderm, both genes were expressed in the Vasa-positive Smm in the right CP at the late gastrula and the pluteus stages ( Figure 5A,B). Another Nodal downstream gene, not, exhibited a similar expression pattern, except that it was not detected in the ectoderm ( Figure 5C). These data suggest that Nodal signaling is received by the Smm on the right side. At the early pluteus stage, we frequently observed that BrdU-or Vasa-positive Smm were asymmetrically partitioned into the left and right CPs and eventually disappeared from the right side. Of the eight Smm at this stage, in most cases, five and three cells were partitioned into the left and right CP, respectively  Figure 5D). This 5:3 ratio was also observed when Smm were labeled with histone H1cs antiserum [50]. Because the right-sided Smm receive Nodal signals, we next determined whether Nodal signaling controls their asymmetrical separation. We classified the separation as asymmetrical (L$5; R#3) or symmetrical (L = 4; R = 4) (Figure 5Ea). Inhibiting Nodal signaling significantly increased the rate of symmetrical separation from 8% to 56% (Figure 5Eb), indicating that Nodal signaling regulates asymmetrical Smm separation. These results are consistent with the observation that the asymmetric expression patterns of Smmexpressed genes, such as vasa and seawi, became symmetric after Nodal signaling was blocked ( Figure 4D,E).
When labeling Smm with the Vasa antibody, we observed that the number of the Vasa-positive Smm on the right side decreased to two or even one in some of the pluteus larva ( Figure 5D). This observation suggests that the disappearance of the Smm on the right side is a gradual process. It was proposed that nanos-depleted Smm undergo apoptosis [51,52]. Because nanos2 expression was blocked by Nodal signaling in the right-sided Smm ( Figure 4D), we wondered whether Nodal signaling induces Smm to undergo apoptosis. When the larvae were stained with TUNEL, we observed that one to three cells in the right CP were apoptotic, whereas no apoptotic cells were detected in the left CP ( Figure 5F). Furthermore, inhibition of Nodal signaling prevented apoptosis in the CPs, and hActivin treatment resulted in apoptotic cells in both CPs ( Figure 5G, arrows). These effects correlate well with nanos2 expression following Nodal signaling perturbation. These data suggest that Nodal signaling induces apoptosis in the right-sided Smm, possibly by controlling nanos2 expression. In addition to apoptotic cells in the right CP, we also observed TUNEL-positive cells in the aboral ectoderm of pluteus larva ( Figure 5G). These signals were attenuated and increased when Nodal signaling was blocked and elevated, respectively, suggesting that Nodal signaling is also involved in aboral ectodermal cell apoptosis.

The Molecular Pathways in LR Patterning
Based on the lineage and perturbation analyses, we provided a schematic representation of the molecular pathways in LR patterning. Figure 6 represents the relationships between Nodal and BMP signals in controlling genes expressed in the right or left CP from two lineages, Smm and veg2 descendants, at the early pluteus stage. We showed that although bmp genes are expressed in aboral skeletogenic cells, pSmad staining was detected in the leftsided HC at the pluteus stage. These cells express soxE, pax6, six1/ 2, eya, and dach. The esophageal domain of the left CP expresses foxF. The initial bilateral pSmad signal at the tip of the archenteron in the late gastrula stage becomes restricted to the left side due to the inhibition by right-sided Nodal signaling, which also regulates its downstream genes in the right CP. Moreover, the initiation of nodal expression on the right side indirectly depends on BMP signaling, and a right lateral ectoderm (RLE) input may also be involved in the spatial regulation of nodal expression. Taken together, these data suggest that BMP signaling is both upstream and downstream of Nodal signaling.

Discussion
Most sea urchin adult tissues derive from the rudiment developed from the left CP. Although it is known that both Smm and the veg2 mesoderm contribute to the CPs, past studies were unable to clearly identify genes that are specifically expressed in either lineage. It was also unknown which of the two lineages contributed to the left CP-derived HC. In addition to identifying several lineage-specific genes in the CP and the HC, we also provided evidence to demonstrate that BMP signals act in the left CP together with Nodal signaling to regulate LR patterning. Given that left-sided nodal expression is a conserved feature in chordates [53,54] and right-sided BMP signaling is observed in several vertebrate species, the opposing Nodal and BMP signals regulating LR asymmetry is likely a conserved mechanism in deuterostomes. However, the mechanisms controlling LR asymmetry in the sea urchin are reversed compared to chordates, using the convention that the mouth is located on ventral sides of embryos. Thus, our study reinforces the possibility that DV inversion occurred in the chordate lineage. Below, we discuss other important findings from this study.

Opposing Nodal and BMP Signals Control LR Axis Patterning
We demonstrated that elevating either Nodal or BMP signaling resulted in the loss of the other signal. This mutual antagonism between Nodal and BMP signaling has been observed during LR patterning in vertebrates. Nodal signaling inhibits BMP signals in the left LPM of mouse embryos by activating the expression of chordin and noggin genes, which encode BMP antagonists [10]. BMP signaling also has been shown to block Nodal signals in the right LPM of mouse, chick, and zebrafish embryos by activating the expression of lefty genes that encode Nodal antagonists [8,55]. The inhibition of BMP signals by Nodal signaling has also been observed in sea urchin embryos during DV (aboral-oral) axis establishment. Nodal signaling in the oral ectoderm is required for the expression of chordin, which restricts BMP signals in the aboral ectoderm [31]. However, we could not detect any asymmetrical LR expression of genes encoding BMP antagonists, such as chordin, noggin, follistatin, dan, or gremlin in the sea urchin embryo (unpublished data). The second molecular mechanism to explain the mutual antagonism between Nodal and BMP signaling is the direct competition between the two signals for the limited amount of the common effector Smad4. In the mouse embryo, BMP signaling has been shown to set a repressive threshold for Nodal signaling in the LPM by limiting Smad4 availability [56].

Micromere-Derived Signals Control LR Asymmetry in Sea Urchin Embryos
The repressive role of Nodal signaling on BMP in the sea urchin embryo is evident given that increasing or blocking Nodal signaling results in the loss of or bilateral pSmad staining in CPs, respectively. However, the effects of BMP signaling on Nodal are complicated because increasing and blocking BMP signaling both result in the loss of nodal expression. These results suggest that BMP signaling is required for right-sided nodal expression in the sea urchin embryo. This positive role of BMP signaling on nodal gene expression has also been observed in vertebrates. In the absence of mouse embryonic BMP4, nodal expression is lost in the left LPM [57]. In chick embryos, implanting either bmp2expressing cells [58] or BMP-soaked beads [59] in the LPM increases nodal expression. During the late segmentation stages of zebrafish embryos, BMP4 signaling is required to activate the expression of the nodal-related gene cyclops in the left LPM [60]. Therefore, BMP signaling can be a positive or negative regulator of Nodal signaling depending on the developmental stages and tissue layers during LR patterning in vertebrates.
Although we observed LR asymmetrical BMP signaling with pSmad staining in the CPs in the sea urchin, bmp genes are transcribed in the skeletogenic mesenchyme cells near the aboral apex of the larva. These observations suggest that BMP ligands are secreted from these micromere-derived skeletogenic cells to control LR asymmetry. It was previously shown that when the micromere lineage was removed from embryos of sea urchin Hemicentrotus pulcherrimus, the LR placement of the rudiment was randomized [26]. BMP could be the micromerederived signal that regulates larval LR polarity, although other signaling molecules may also be involved in this process.

Nodal, Nanos, and Apoptosis in Small Micromeres
In sea urchin embryos, Smm form during the fifth cleavage and are regulated by a set of conserved germline lineage genes, including vasa, nanos2, and seawi [37]. Adults from Smm-deleted embryos formed small gonads without gametes [61]. Therefore, Smm are required for germline specification in sea urchins. Smm lineage fate is maintained by Nanos [52], which is also required for Smm descendant survival [51]. Studies in fly and vertebrates have also shown that Nanos has a conserved role in maintaining germline identity by preventing apoptosis [62][63][64]. In this study, we showed that Nodal signaling acts on the Smm partitioned in the right CP. Nodal signaling perturbation affected both nanos2 expression and cell death. These results suggest that Nodal signaling in the right CP represses nanos2 expression in the Smm, which leads to cell death. To our knowledge, the role of Nodal signaling in repressing the expression of germline lineage genes such as nanos has not been reported in other systems. Intriguingly, Nodal is capable of inducing apoptosis in adult ovary during follicular degeneration and human trophoblast cells during normal placentation [65,66]. Therefore, there may be a conserved role for Nodal to induce apoptosis in extraembryonic tissues. In the case of sea urchins, Nodal signaling-induced apoptosis in the right Smm is essential for normal development, and a lack of Nodal results in bilateral rudiments that give rise to a juvenile composed of two conjoined urchins [6].

Animals, Embryos, and Treatments
Adult sea urchins (Strongylocentrotus purpuratus) and their gametes were obtained as previously described [29]. Rhodomonas lens provided by Pat Leahy was used to feed the larva (see Text S1). BMP or Nodal signaling perturbation was performed by treating embryos with inhibitors, recombinant proteins, or vivo-morpholinos (vMOs) and culturing them in the dark. Unless otherwise indicated, the concentrations of the reagents used in this study were as follows: dorsomorphin (16 mM; Sigma), SB-431542 (5 mM; Tocris Cookson), SB-505124 (2 mM; Sigma), mouse BMP4 (250 ng/ml; R&D Systems), and human Activin AB (120 ng/ml; R&D systems). Solvents (DMSO or 0.1% BSA) were added as controls. The sequences of the BMP and Nodal vMOs (Gene Tools) are the same as the previously published regular MOs [33]. vMOs were diluted at 1:100 to 5 mM from stock solution in phosphate-buffered saline into 500 ml of culture. The treatment times are summarized in Figures S3H and S4E. To keep the larva feeding normally and viable, the treatments were washed out no later than EPL stage (72 hpf). Note that at higher concentrations (over 20 mM), the vMOs precipitate in seawater and are toxic to the embryos.

In Situ Hybridization and Immunostaining
The primers used to construct the clones for probe synthesis were designed based on gene models [67] and are listed in Table  S1. In situ hybridization (ISH) and immunostaining were performed as previously described [29]. The primary antibodies used in this study were rabbit anti-pSmad1/5/8 (1:200; Cell Signaling), mouse anti-acetylated a-tubulin (1:1,000; Sigma), and Schematic representation of the molecular pathways in LR patterning, created using the BioTapestry program [69]. BMP signal from the aboral skeletogenic mesenchyme cells (aboral PMC) is required for nodal expression in the right lateral ectoderm (RLE) through an unknown mechanism (red dashed line). Nodal signal at the RLE activates its downstream genes in the RLE and right CP (RCP). Nodal signal also induces apoptosis in the rightsided Smm and presumably does so by repressing nanos2 expression. Right-sided BMP signaling is blocked by Nodal and only functions in aboral veg2 descendants in the left CP (LCP) to activate its downstream genes. The genes shown in black, light gray, or dark gray are expressed, not expressed, or expressed weakly, respectively, in their specific domains at the early pluteus stage. The arrows do not necessarily indicate direct interactions. doi:10.1371/journal.pbio.1001402.g006 rabbit anti-DmVasa (1:300) [68]. The nuclei were counterstained with Hoechst 33342 (1:1,000; Invitrogen), and the cytoplasmic membrane was visualized with CellMask Deep Red (1:2,000; Invitrogen). The embryos were imaged using a Leica TCS-SP5 AOBS inverted confocal system.

BrdU Labeling and TUNEL Assay
After removing the fertilization envelope, 1-cell-staged embryos were incubated with 50 mM 5-bromo-2-deoxyuridine (BrdU; Sigma) for 1 h and then washed twice with 500 mM thymidine. For double labeling, the biotin-avidin system was used to detect the BrdU signal. Antigen retrieval of the incorporated BrdU was conducted by DNA denaturation using 1 N HCl in PBST for 30 min. To block endogenous biotin, the embryos were incubated with 0.01% avidin and 0.001% biotin sequentially. The embryos were then immunolocalized with biotinylated anti-BrdU antibody (1:1,000; Abcam) and detected with HiLyte Fluor Streptavidin (1:400; AnaSpec). Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) was performed by using the In Situ Cell Death Detection Kit (TMR red; Roche) for 40 min at 37uC. DM treatment from 42 to 48 hpf was sufficient to block HC formation, whereas the vMO had no effect when applied in the same period. The different effects might be due to the natures of the two blocking mechanisms. DM inhibits BMP receptor kinase activity and blocks BMP signaling immediately after it penetrates cells. On the other hand, the vMO blocks translation of bmp2/4, and BMP signaling could still be active until the remaining BMP2/4 is degraded. (TIF) Figure S4 Effects of Nodal signaling on LR asymmetry. (A) pSmad and acetylated a-tubulin staining in SB-431542-treated embryos revealed bilateral HC (arrows) in EPL. (B) Expression of right-sided genes (indicated by arrows) following Nodal signaling perturbation. (C and D) Expression of LR marker genes after SB-431542 treatments. The numbers in the bottom left-hand corners of the photos indicate the phenotype ratios. (E) The embryos were incubated for the indicated time and concentration, and the effects on the oral-aboral (OA) axis and HC formation in over 90% of the embryos are listed. The black bar shows the treatment time used in most experiments. The effect of SB inhibitors and Nodal vMO was different in that Nodal vMO did not cause OA defects when treated during MB. The difference might also be due to the differential inhibitory mechanisms: SB inhibitors directly block signaling, whereas vMO blocks translation of the ligand. (TIF) Table S1 Gene IDs and primers used to construct clones for probe synthesis in this study.