|Home | About | Journals | Submit | Contact Us | Français|
The oral-aboral (OA) axis in the sea urchin is specified by the TGFβ family members Nodal and BMP2/4. Nodal promotes oral specification, whereas BMP2/4, despite being expressed in the oral territory, is required for aboral specification. This study explores the role of Chordin (Chd) during sea urchin embryogenesis. Chd is a secreted BMP inhibitor that plays an important role in axial and neural specification and patterning in Drosophila and vertebrate embryos. In L. variegatus embryos, Chd and BMP2/4 are functionally antagonistic. Both are expressed in overlapping domains in the oral territory prior to and during gastrulation. Perturbation shows that, surprisingly, Chd is not involved in OA axis specification. Instead, Chd is required both for normal patterning of the ciliary band at the OA boundary and for development of synaptotagmin B-positive (synB) neurons in a manner that is reciprocal with BMP2/4. Chd expression and synB-positive neural development are both downstream from p38 MAPK and Nodal, but not Goosecoid. These data are summarized in a model for synB neural development.
The sea urchin is a powerful model organism for studying embryonic axis specification, particularly with the recently sequenced genome (Davidson et al., 2002; Sodergren et al., 2006). Nascent networks for the specification of the larval DV axis, known as the oral-aboral (OA) axis, have been described in various sea urchin species (Amore et al., 2003; Bradham and McClay, 2006; Duboc et al., 2004). In Lytechinus variegatus embryos, the earliest known event in oral specification is the asymmetric nuclearization of p38 MAPK (Bradham and McClay, 2006). Inhibition of p38 during early sea urchin development results in aboralized embryos that fail to express Nodal, a crucial activator of the oral program (Bradham and McClay, 2006; Duboc et al., 2004).
While Nodal is necessary for oral specification, BMP2/4 appears to play an important role in aboral specification in several sea urchin species (Angerer et al., 2000; Duboc et al., 2004). BMP2/4 is expressed downstream from Nodal in the oral ectoderm, but is somehow required for the specification of the aboral ectoderm, since BMP mRNA injection promotes aboralization, while BMP inhibition promotes oral specification (Angerer et al., 2000; Duboc et al., 2004). Further, embryos injected with BMP morpholino antisense oligo (MO) strongly resemble Nickel-treated embryos, whose ectoderm possesses only oral character (Duboc et al., 2004; Hardin et al., 1992). In order to better understand how this signaling ligand promotes specification of cell fates in a population that is spatially distinct from those that express it, we investigated the function of Chordin, a dedicated BMP inhibitor.
Chordin (Chd) is a well-studied protein that is required for axial and neural specification and patterning in a wide range of animals. Chd was first identified independently in Drosophila and Xenopus, where it is required for the specification of the secondary (dorsal-ventral, DV) body axis and the associated neural field (Ferguson and Anderson, 1992; Sasai et al., 1994). Chd is a large secreted protein that inhibits BMP from transducing signals by binding BMP and preventing its interaction with its receptor (Piccolo et al., 1996). Thus, embryonic regions that express Chd are protected from BMP signaling and thereby acquire a distinct fate. Perturbation analysis and functional assays were used to characterize the role of Chordin in sea urchin embryos. The results show that, surprisingly, Chordin is required not for OA specification, but rather for the patterning of the ciliary band and the development of synaptotagmin B-positive neurons.
SpChordin was cloned as a 400 bp fragment in an S. purpuratus EST project (Poustka et al., 2003). This fragment was used to probe an arrayed cDNA library (Poustka et al., 2004), from which a full length SpChordin clone was identified. LvChordin was cloned from an arrayed cDNA library (Bradham and McClay, 2006) using SpChordin as a probe. The GenBank accession numbers for these sequences are EU048249 (Lv) and EU015401 (Sp). Phylogenetic analysis was performed as described (Bradham et al., 2006).
LvBMP2/4, LvNodal, LvGsc, and LvTbx2/3 constructs were previously described (Angerer et al., 2000; Bradham and McClay, 2006; Gross et al., 2003). Morpholino anti-sense oligos (MOs) (Gene Tools Inc.) specific for LvChd (5′-CGGCGTAAAGTGTGATGCGGTACAT) and LvBMP (5′-GACCCCAATGTGAGGTGGTAACCAT) overlap AUG start codons to block translation. The MO specific for LvNodal was described (Bradham and McClay, 2006). Two MOs were used for Gsc: a start site MO (5′-ACGTCAGGAAGGTGATAGTCCATCG), and a splice blocking MO (5′-TAAGTCTTACCTCGACTCGTTCCTC). The latter was based on the sequence for the splice junction at the 5′ end of the second intron of LvGsc. Each Gsc MO produced the same phenotype upon injection, and each was rescued by co-injection with Gsc mRNA (not shown). SB203580 was obtained from Calbiochem, and was used at 20 μM. All other reagents were obtained from Sigma or Fisher unless otherwise noted.
Lytechinus variegatus sea urchins were obtained from Sea Life Inc. (FL), or the Duke University Marine Lab. Eggs were harvested and injected as described (Bradham and McClay, 2006) using in vitro translated mRNAs produced with the mMessage mMachine kit (Ambion) or MOs. SB203580 treatments (20 μM) and 2 cell-stage injections were performed as described (Bradham and McClay, 2006). For all injections and drug treatments, dose-response experiments were performed to minimize the dose of each reagent in order to optimize specificity. For mRNAs, injection doses (pg/pL) were: Chd 0.125; BMP2/4 0.022; Nodal 0.42; Gsc 0.162; except where otherwise noted. For MOs, injection doses (mM) were: Chd 1.0; BMP 0.57; Nodal 0.57; Gsc 1.33 (splice blocking) or 1.6 (start site). For phenotypic counts (Fig. 2C, Fig. 5), embryos were photographed in groups, then counted from the images.
In situ hybridizations were performed as described (Bradham and McClay, 2006).
Ciliary band staining was performed with 1° antibody 295 as described (Bradham and McClay, 2006). For neural labeling, immunostaining was performed with 1° antibodies anti-synaptotagmin B at 1:50 (Burke et al., 2006; Nakajima et al., 2004) and anti-serotonin at 1:1000 (Sigma). Embryos were fixed in 4% paraformaldehyde in sea water without methanol, PBS with 5% normal goat serum and 0.1% Triton X-100 was used throughout, and embryos were incubated in 1° antibody for 48 hours; otherwise the staining protocol and confocal microscopy were as described (Bradham and McClay, 2006).
RNA isolation, cDNA synthesis and QPCR analysis were performed as described (Poustka et al., 2007). QPCR primers for SpUbiquitin, SpZ12-1, SpBMP2/4, SpChd, and SpNodal have been described (Poustka et al., 2007). The primers for SpNoggin were forward 5′GCAGCTGATCTCGCTGAATCT and reverse 5′CAGTTCGCCTTCCACATCAAC. Average CT values were normalized to ubiquitin, then converted to transcript numbers using the known values for SpZ12-1 (Minokawa et al., 2004) with the formula QSpGOI = QSpZ12 * 1.96−ΔCT.
To evaluate the role of Chordin (Chd) during sea urchin development, Chd orthologs were cloned from L. variegatus (EU048249) and S. purpuratus (EU015401). Chd proteins are characterized by four cysteine-rich (CR) domains. Alignment of these four domains from various Chd protein sequences demonstrates that this architecture is well-conserved across vertebrates, urchins, and flies (Fig. 1A). Phylogenetic analysis positions sea urchin Chds between vertebrate Chds and Drosophila Chd (SOG) (Fig. 1B), consistent with the evolutionary classification of sea urchins as non-chordate deuterostomes.
LvChd expression was detected beginning at hatched blastula stage (Fig. 1C). LvChd expression is initially asymmetric, and is restricted to the oral territory throughout development. During gastrulation, the LvChd expression domain becomes progressively narrowed to a stripe between the oral and anal openings (Fig. 1C6–10) which appears to correspond to the portion of the ciliary band at the oral-aboral (OA) boundary that is opposite from the apical plate.
A network model for the specification of the oral region in L. variegatus has previously been described (Bradham and McClay, 2006). In both L. variegatus and S. purpuratus embryos, p38 MAPK activity is required for the oral expression of the signaling ligand Nodal (Bradham and McClay, 2006; Bradham and McClay, 2007). LvNodal in turn induces the oral expression of the transcriptional repressor LvGsc, which both represses aboral genes and relieves repression of oral genes (Bradham and McClay, 2006). Both Nodal and Gsc play similar roles in other urchin species (Amore et al., 2003; Angerer et al., 2001; Duboc et al., 2004; Flowers et al., 2004), suggesting that these elements of the oral specification network are conserved across various sea urchin species.
To determine the position of LvChd in the oral network, Chd expression was assessed in perturbed embryos. When p38 was inhibited with SB203580 (SB) treatment, Chd expression was blocked (Fig. 1D2), indicating that Chd expression is downstream from p38 activity. This result was corroborated by QPCR analysis, which showed an approximate 128 fold average net difference for LvChd expression between control and SB-treated embryos at late gastrula (not shown). Similar QPCR results were obtained in SB-treated S. purpuratus embryos (not shown). LvChd expression was also responsive to Nodal perturbation, as injection of a morpholino-substituted antisense oligonucleotide (MO) specific for LvNodal inhibited LvChd expression (Fig. 1D3), while LvNodal misexpression resulted in robust LvChd expression throughout the ectoderm, excluding only the apical and vegetal plates (Fig. 1D4). Thus LvChd expression is also downstream from LvNodal signaling. The onset of expression of LvNodal at late blastula (Bradham and McClay, 2006) is prior to the onset of Chd at hatched blastula, consistent with Chd expression occurring downstream from Nodal. Interestingly, neither LvGsc-specific MO (Fig. 1D5) nor LvGsc mRNA (Fig. 1D6) injection had a significant impact on LvChd expression level or domain, indicating that LvChd expression is independent of LvGsc-mediated repression. Together, these data indicate that in L. variegatus embryos, Chd expression is driven by the pathway p38 > Nodal > Chd, and that signaling branches downstream from Nodal to independently induce Chd and Gsc.
LvBMP2/4 is also expressed in the oral ectoderm (Fig. 1E1) as in previous reports (Angerer et al., 2000; Duboc et al., 2004). Comparison of the domains of expression for LvChd and LvBMP2/4 at late gastrula/early prism stages (Fig 1D1 top and 1E1) indicates that the domain of Chd expression is more restricted than BMP2/4, suggesting that, at this stage, active BMP2/4 signaling is likely excluded from most of the oral territory. LvBMP2/4 expression requires p38 activity (Fig. 1E2) and, consistent with previous reports (Duboc et al., 2004), LvNodal expression (Fig. 1E3). Like LvChd, LvBMP2/4 expression is essentially unaffected by LvGsc perturbation (Fig. 1E5, 6). However, LvNodal mRNA injection did not strongly drive LvBMP2/4 expression (Fig. 1E4), unlike the effect of LvNodal on LvChd (Fig. 1D4). In fact, the domain of LvBMP2/4 expression appears contracted in LvNodal-injected embryos compared with control embryos (Fig. 1E1, 1E4). The simplest interpretation of this result is that LvBMP2/4 expression requires an additional, Nodal-independent input that uses AND logic together with the Nodal-dependent input.
When LvChd was perturbed either by injection of an LvChd-specific MO or by mRNA injection, the resultant embryos had similar phenotypes (Fig. 2A2, A3), although Chd mRNA-injected embryos were somewhat elongated along the animal-vegetal axis. In either case, endoderm and mesoderm appear to be normally specified based on morphological appearance of the gut, pigment and skeletogenic mesenchyme, although arm-like extensions were not observed. The LvChd MO-induced phenotype was rescued to morphological normality by co-injection with Chd mRNA, demonstrating the specificity of this reagent (Fig. 2B).
Chd inhibits BMP signaling in vertebrates and flies (Ferguson and Anderson, 1992; Hammerschmidt et al., 1996b; Piccolo et al., 1996); however, in the cnidarian Nematostella, NvChd does not appear to be a strong BMP antagonist (Rentzsch et al., 2006). To test whether LvChd antagonizes LvBMP2/4, embryos were injected with LvBMP2/4 mRNA either alone or together with LvChd mRNA to determine whether Chd could rescue the BMP2/4-induced phenotype. LvBMP2/4 mRNA injection alone produced embryos with a distinctly wrinkled ectoderm and an apparently normal archenteron (Fig. 2C2), though invagination of the gut was delayed (see Fig. 3B3, B6, B9). This phenotype is consistent with that obtained in other urchin species (Angerer et al., 2000). Co-injection of LvBMP2/4 and LvChd mRNAs shifted the phenotype to the LvChd overexpression phenotype, with a smooth ectoderm and triangular shape (Fig. 2C4) in a dose-responsive manner (Fig. 2C5). LvChd mRNA injection at a moderate dose was able to completely rescue the LvBMP2/4 phenotype and produce normal-appearing pluteus larvae (Fig. 2C3 and C5, red bars). Although the rescue likely reflects the titration of exogenous molecules, this does not belie the outcome, particularly since it is the protein products, rather than the injected mRNAs, that are antagonistic. These results demonstrate that LvChd functionally antagonizes LvBMP2/4, strongly suggesting that LvChd protein is capable of binding and inhibiting LvBMP2/4, in a manner parallel to Chd’s function in vertebrates and flies with respect to BMP inhibition.
The above results suggested that Chd may promote oral specification, since it is antagonistic to BMP2/4, which is known to be required for aboral specification (Angerer et al., 2000; Duboc et al., 2004). To test this, oral and aboral gene expression was evaluated in Chd-perturbed embryos (Fig. 3A). The genes tested were Gsc, a transcriptional repressor expressed in the oral ectoderm (Angerer et al., 2001) (Fig. 3A1–3), FoxA, a transcriptional repressor expressed in both the gut and the future mouth subregion of the oral territory (Oliveri et al., 2006) (Fig. 3A4–6), and the aborally-expressed transcription factor Tbx2/3 (Croce et al., 2003; Gross et al., 2003) (Fig. 3A7–9). The results show that neither Chd MO nor Chd mRNA injection impacted the expression of any of these genes, indicating that LvChd does not participate in specifying the OA axis.
This unexpected result prompted a similar analysis of LvBMP2/4-perturbed embryos (Fig. 3B). Here, the genes tested were Nodal and Gsc, both expressed in the oral ectoderm, and Tbx2/3, which is aborally expressed. The morphology of LvBMP2/4 MO-injected embryos was consistent with the phenotype previously published for P. lividus embryos (Duboc et al., 2004) and was rescued to morphological normality by co-injection of LvBMP2/4 mRNA (not shown), demonstrating the specificity of the L. variegatus-specific MO. LvBMP2/4 MO-injected embryos expressed LvNodal in an asymmetric manner as expected (Duboc et al., 2004) (Fig. 3B2). These embryos expressed LvGsc symmetrically in the non-apical ectoderm (Fig. 3B5), and did not express LvTbx2/3 (Fig. 3B8). This shows that LvBMP2/4 is required for aboral specification, consistent with previous findings (Angerer et al., 2000; Duboc et al., 2004). When LvBMP2/4 mRNA was injected (Fig. 3B3, B6, B9), none of the genes were detected, indicating that ectopic LvBMP2/4 blocks both oral and aboral specification. To maximize specificity, LvBMP2/4 mRNA was injected at the lowest level that produced a phenotype, but even at this minimal dose (0.022 pg/pL), blockade of all three markers was observed. This result can potentially be explained by the block to Nodal both at this stage (Fig. 3B3) and at the onset of LvNodal expression (not shown) in BMP2/4 mRNA-injected embryos, since Nodal is required for both oral and aboral specification (Duboc et al., 2004). In any case, these results collectively indicate that while LvBMP2/4 perturbation impacts OA specification, LvChd perturbation does not.
Comparison of Chd and BMP2/4 in situ results (Fig 1D1, 1E1) suggested that BMP signaling occurs at the boundary of the oral territory (and beyond it). The boundary is composed of the ciliary band, a strip of ciliated cells between the oral and aboral territories (Cameron et al., 1993). The ciliary band is first detected by immunofluorescence staining at late gastrula to prism stage in L. variegatus embryos, becoming prominent in pluteus stages (Fig. 4A1). LvChd MO injection resulted in little or no ciliary band staining (Fig. 4A2), while LvChd mRNA-injected embryos display aberrant establishment of the ciliary band (Fig. 4A3). LvBMP2/4 MO injection resulted in relocation of the ciliary band to the vegetal ectoderm, with some irregular extension of this territory animally (Fig. 4A4); conversely, LvBMP2/4 mRNA injection blocked formation of the ciliary band (Fig. 4A5). Thus, LvChd and LvBMP2/4 perturbation appear to reciprocally impact the ciliary band, consistent with the observed functional antagonism between these proteins.
The results for BMP perturbation agree with results in other sea urchin species. In BMP2/4-injected S. purpuratus embryos, little ciliary band staining was detected (Angerer et al., 2000) and SpHnf6 expression (a ciliary band marker) was detected only in a small apical region (Yaguchi et al., 2006). In contrast, XNoggin-injected S. purpuratus embryos show an enlarged and disorganized ciliary band (Angerer et al., 2000); the ciliary band (as measured by tubulin expression) is also disorganized in BMP2/4 MO-injected P. lividus embryos (Duboc et al., 2004).
The nervous system of the sea urchin larva is positioned mainly in parallel and adjacent to the ciliary band (Nakajima et al., 2004). Chd-mediated antagonism of BMP promotes neural versus ectodermal specification in flies and vertebrates (Francois et al., 1994; Hammerschmidt et al., 1996a; Holley et al., 1995; Khokha et al., 2005; Kuroda et al., 2004; Oelgeschlager et al., 2003; Sasai et al., 1995); therefore, the effect of LvChd and LvBMP2/4 perturbation on neural development was examined next. Two distinct groups of neurons have been identified in sea urchin larva: serotonin-producing, and non serotonin-producing (Nakajima et al., 2004). The serotonin-positive subset arise at the apical plate (Fig. 4B1, green), an ectodermal territory distinct from the oral and aboral regions (Takacs et al., 2004; Yaguchi et al., 2006; Yaguchi et al., 2007). The remaining neurons appear mainly at the oral edge of the ciliary band, and are labeled with anti-synaptotagmin B (synB), a pan-neural marker in sea urchin embryos (Burke et al., 2006; Nakajima et al., 2004). Additional synB-positive neural fibers encircle the mouth (Fig. 4B1, red).
LvChd MO injection blocked nearly all synB expression; some scattered, weak staining was detected that may be serotonergic neurons (Fig. 4B2). LvChd mRNA injection (Fig. 4B3) and LvBMP2/4 MO injection (Fig. 4B4) each induced excessive, disorganized synB-positive neural fibers. Finally, like LvChd MO, LvBMP2/4 mRNA injection resulted in a loss of synB neurons (Fig. 4B5). Thus, LvChd and LvBMP2/4 have opposing effects on synB-positive neural development, consistent with the functionally antagonistic relationship between these two proteins. In S. purpuratus embryos, XNoggin mRNA injection resulted in few or no serotonergic neurons (Yaguchi et al., 2006), consistent with the results herein; however, SpBMP2/4 overexpression produced extra serotonergic neurons, though this effect was observed only in late plutei (Yaguchi et al., 2006).
Although LvChd appeared to be necessary for development of synB-positive neurons, the results with LvChd mRNA injection were not clear as to whether LvChd induced excessive synB neurons or merely excessive fibers and abnormal pathfinding from the normal number of synB neurons. To address that distinction, synB neurons were labeled in embryos at early pluteus stage, when these neurons are first readily detectable in L. variegatus (Fig. 4C1). LvChd mRNA-injected embryos clearly show excessive numbers of neural cell bodies at this stage (Fig. 4C2), suggesting that exogenous Chd, and therefore inhibition of BMP2/4, is indeed capable of promoting the development of synB neurons. To confirm this result, the ability of LvChd to rescue neurogenesis in LvBMP2/4 mRNA-injected embryos was tested (Fig. 4D). LvBMP2/4 mRNA injection alone blocked synB expression (Fig. 4D1). Co-injection of LvBMP2/4 and a moderate dose of LvChd (0.125 pg/pl) rescued the normal pattern of synB expression (Fig. 4D2), while a higher dose of LvChd (0.25 pg/pl) induced excessive numbers of synB-positive neurons (Fig. 4D3). Thus, exogenous Chd rescues development of synB-positive neurons in BMP2/4 mRNA-injected embryos. Collectively these results indicate that LvChd-mediated LvBMP2/4 inhibition promotes synB-positive neural development.
To functionally corroborate the previous findings, the requirement of LvChd for the development of the OA axis, the ciliary band, and synB-positive neurons was assessed with a patterning assay (Bradham and McClay, 2006). The assay is based on inhibiting p38 MAPK activity by treating with SB up to early gastrula stage, which results in aboral specification of the embryonic ectoderm (Bradham and McClay, 2006). SB is then washed out the to allow the skeleton to form, providing a readout for the patterning of the axis. These ‘SB-wash’ embryos display aberrant skeletal patterning (Fig. 5B1, B2) (Bradham and McClay, 2006). Embryos treated with SB and then washed at early gastrula stage recover the expression of Nodal and Gsc, although the ectoderm remains specified as aboral (Bradham and McClay, 2006), and produces an expanded ciliary ‘band’ that covers the animal half of the embryo (Fig. 5B3), as well as serotonergic and synB-positive neurons (Fig. 5B4).
If, in this context, Nodal mRNA (which promotes oral specification) is injected into one blastomere at the 2-cell stage to mimic its normally asymmetric expression, then OA patterning is rescued and morphologically normal plutei are produced in the majority of the embryos (Fig. 5C1, C2) (Bradham and McClay, 2006). Normal patterning of the ciliary band (Fig. 5C3) and synB-positive neurons (Fig. 5C4, red) was also rescued, while the development of serotonergic neurons was suppressed (Fig. 5C4, green). Since development of the serotonergic neurons is inhibited by SpNodal (Yaguchi et al., 2006), this result is not surprising. Thus, rescue of the oral territory by Nodal mRNA rescues normal embryonic patterning, including patterning of the ciliary band and the synB-positive neurons, although ser neurons are absent.
To test whether Chd is required downstream from Nodal for any of these rescue effects, Chd MO was co-injected with Nodal mRNA in one blastomere at the 2-cell stage in this context. A similar percentage of the resultant embryos had normal axial patterning (Fig. 5D1, D2). These embryos had delayed skeletogenesis compared with Nodal alone, an effect consistently observed with LvChd MO (data not shown). In these embryos, relatively normal development of both the ciliary band (Fig. 5D3) and serotonergic neurons (Fig. 5D4, green) was evident. In contrast, few or no synB-positive neurons developed (Fig. 5D4, red). The rescue of serotonergic neurons is consistent with Yaguchi et al.’s findings that SpBMP mRNA injection promotes serotonergic neural development (Yaguchi et al., 2006). Together, these data indicate that Chd-mediated BMP2/4 repression is required in particular for the development of synB-positive, non-serotonergic neurons. In addition, they suggest that Chd is not strictly required for ciliary band specification and patterning, nor for serotonergic neural development. Indeed, the absence of Chd (and therefore the presumed increase in BMP signaling) protected serotonergic neural development from the inhibitory effect of Nodal.
When Chd was inhibited in whole embryos, the ciliary band was almost undetectable (Fig. 4), whereas inhibition of Chd in the patterning assay did not affect the ciliary band (Fig. 5), making the role of Chd in specifying this structure ambiguous. Chd is expressed downstream from p38 and Nodal but not Gsc (Fig. 2D). If the specification of the ciliary band is downstream from Chd, then this specification event should also depend on p38 and Nodal but not Gsc. To test that prediction, the impact of perturbing p38, Nodal, and Gsc on the ciliary band specification development was determined (Fig. 6). In the absence of either p38 or Nodal signaling, the ciliary band expanded to occupy the non-vegetal ectoderm (Fig. 6A2, A3). In contrast, misexpression of Nodal reversed this pattern, limiting this territory to the vegetal ectoderm (Fig. 6A4). Gsc MO-injected embryos possess a ciliated band that is restricted to narrow stripe encircling the embryo (Fig. 6A5). When Gsc was misexpressed, the ciliary band was almost undetectable (Fig. 6A6), indicating that Gsc somehow has an inhibitory effect on this structure.
Together these data demonstrate that the specification of the ciliary band is independent of p38, Nodal, and Gsc, since inhibition of each of genes resulted in embryos with clear ciliary bands (Fig. 6A2, A3, A5). Thus, the specification of the ciliary band must also be independent of Chd, despite the absence of the ciliary band in both Chd MO and BMP-injected embryos. It is therefore more likely that these inputs affect the later development, patterning and/or differentiation of this structure and not its specification.
Although OA specification is dependent on LvGsc (Bradham and McClay, 2006) and independent of LvChd (data herein), it is important to determine whether the reciprocal relationship holds true for neural development. Indeed, since Chd is downstream from p38 and Nodal, but not Gsc, the prediction is that neural development will have the same dependencies. Inhibition of p38 blocked most synB-positive neural development. A few synB-positive neurons were still detected near the apical plate, as marked by serotonin labeling (Fig. 6B2). Thus, while a small number of apically-localized synB neurons were detected, no ‘peripheral’ synB neurons were observed. Note that this result is distinct from SB-wash embryos (Fig. 5B), in which peripheral synB neural development recovers during the wash phase. A role for p38 in neural development is consistent with findings in other systems (Choi et al., 2001; Hansen et al., 2000; Ishii et al., 2001; Iwasaki et al., 1999; Morooka and Nishida, 1998; Takeda et al., 2000).
LvNodal MO injection had a similar effect on neural development (Fig. 6B3), in that a few synB-positive neurons were present near the apical plate, while no peripheral synB neurons were detected. For L. variegatus embryos, this is consistent with p38 and Nodal functioning upstream of Chd for synB-positive neural development. In S. purpuratus embryos, Nodal was inhibited by antivin mRNA injection, which resulted in a profound increase in synB-positive neurons (Yaguchi et al., 2006), unlike the findings herein with LvNodal MO. The reason for this discrepancy is currently unclear and could reflect a species difference.
LvNodal mRNA injection induced a striking synB neural pattern with few or no serotonergic neurons (Fig. 6B4), in agreement with S. purpuratus results (Yaguchi et al., 2006). SynB-positive neural fibers were present at the circumferential mouth and the apex of the embryo (corresponding to the tip of the oral hood in controls), as well as in a ring at the edge of the vegetal ectoderm (Fig. 6B4), adjacent to the ciliary ‘band’ territory in these embryos (Fig. 6A4). Thus, LvNodal-injected embryos display what appears to be largely normal synB-positive neural patterning, given the oralized context.
Both synB- and serotonin-positive neurons were present in LvGsc MO- (Fig. 6B5) and LvGsc mRNA- (Fig. 6B6) injected embryos, suggesting that LvGsc plays little or no role in neural specification and development. Thus neural development in L. variegatus embryos is Chd-dependent and Gsc-independent. Interestingly, LvGsc MO-injected embryos show prominent neural cell bodies with fewer extended processes (Fig. 6B5), suggesting that an aboral climate interferes with neurite extension. Injection of LvGsc mRNA produced synB-positive fibers around the boundary of the embryo and in an interior circumferential ring, likely corresponding to a circumferential mouth, since Gsc is an oralizer. In S. purpuratus embryos, SpGsc overexpression blocked specification of serotinergic neurons (Yaguchi et al., 2006); in L. variegatus, serotonergic neurons were observed but were not localized to the animal pole, but rather arrayed about the ring of synB-positive neurons that may reflect a circumferential mouth (Fig. 6B5). Collectively, these results indicate that in L. variegatus embryos, the pathway p38> Nodal> Chd-| BMP-| synB-neural promotes development of synB-positive neurons, which occurs independently of Gsc.
This study evaluates the role of Chordin (Chd), a well-studied BMP antagonist, in sea urchin embryogenesis. The results show that Chd is not involved in oral-aboral (OA) specification, but rather is required for appropriate patterning of the ciliary band at the OA boundary and for the development of synB-positive neurons. Since LvGsc functionally suffices to rescue oral specification downstream from p38 (Bradham and McClay, 2006) but plays no role in neural development, this suggests that the oral and neural development pathways diverge downstream of p38 and Nodal into the Chd pathway for neural and the Gsc pathway for oral.
The defects that are reciprocal between LvChd and LvBMP2/4 perturbations in whole embryos are ciliary band patterning and synB-positive neural development. Notably, these structures arise along the boundary between the oral and aboral territories, suggesting that the endogenous function of Chd-mediated BMP2/4 antagonism may be to organize that boundary. A boundary-establishing function has been described for BMP in fly and chick (Masucci et al., 1990; Posakony et al., 1990; Shen and Dahmann, 2005; Smith et al., 2000), providing a precedent for this idea. The notion that this antagonism organizes the OA boundary is additionally supported by the expression patterns for LvChd and LvBMP2/4 in the oral territory. These domains overlap, with the LvBMP2/4 region extending beyond the LvChd region, suggesting that a gradient of BMP2/4 signaling exists at the aboral border of the oral territory. It seems unlikely that this antagonism is directly responsible for specifying the ciliary band, since it clearly specified independently of p38 and Nodal (Fig. 6A2, A3). Nevertheless, a role for Chd and BMP2/4 in organizing/positioning/patterning the ciliary band and associated structures is consistent with the results.
LvChd expression is downstream from p38 and LvNodal, but not LvGsc. Similarly, ‘peripheral’ synB-positive neural development requires p38 and LvNodal, but not LvGsc. Thus, the results indicate that the pathway p38> Nodal> Chd-| BMP-| neural promotes the development of synB-positive neurons. LvBMP2/4 expression also requires p38 and LvNodal but not LvGsc; however, the results obtained with LvNodal mRNA-injected embryos are strikingly different for LvChordin and LvBMP2/4. LvChordin is strongly induced by LvNodal mRNA, whereas LvBMP2/4 expression appears contracted rather than expanded by LvNodal, indicating that the regulation of LvBMP2/4 must include other unknown components. These results are depicted in a network model for the ectoderm in sea urchins (Fig. 7). This model is an extension of the previously published ectodermal network for L. variegatus (Bradham and McClay, 2006). LvChd is depicted downstream from Lvp38 and LvNodal, while LvBMP2/4 is shown downstream from LvNodal as well as an additional unknown input (yellow). LvChd relieves LvBMP2/4-mediated repression of synB-positive neural development. This ectodermal network model is consistent with that determined for S. purpuratus embryos, in which the Nodal-independent input into BMP2/4 is HNF6 (http://sugp.caltech.edu/endomes/#EctodermNetwork).
It is worth noting that the neural phenotype of both LvChd mRNA- and LvBMP2/4 MO-injected embryos (Fig. 4B3, 4B4) resembles neither embryos with orally-specified ectoderm, such as LvNodal- or LvGsc mRNA-injected embryos, which possess orderly arranged neural fibers (Fig. 6B4, B6), nor embryos with an aborally-specified ectoderm, such as LvGsc MO-injected embryos (Fig. 6B5), in which the neurons extend only a few short processes. Instead, LvChd mRNA- (and LvBMP2/4 MO-) injected embryos show a disruption of neural patterning that is distinct from either of these examples, consistent with conclusion that LvChd perturbation fails to impact specification of the OA axis. However, BMP MO-injected embryos are oralized and also have disrupted neural patterning. One difference between embryos injected with BMP MO and Nodal- or Gsc-injected embryos is the ciliary band, which is disrupted with BMP MO, versus absent from the non-vegetal ectoderm, or nearly so, with both Nodal or Gsc. This difference in the ciliary band may explain the dissimilar neural patterning results with BMP MO and other oralizing perturbations.
Interestingly, the loss of serotonergic neurons in the LvNodal-injected embryos in the patterning assay (Fig. 5C1) and their rescue by inclusion of LvChd MO (Fig. 5D1) indicates that the balance between BMP2/4 and Chd is critical for the development both neural types: LvBMP2/4 signaling is necessary for the development of serotonergic neurons, whereas protection from LvBMP2/4 (via Chd) is required for synB-positive neural development. This suggests that the two neural subtypes might initially arise in spatially adjacent regions. A full understanding of these relationships will require determining both the timing and location of neural specification. Work from other species suggests that neural specification occurs early in development, but in sea urchin embryos, current tools only allow the detection of differentiated neurons as they arise relatively late in development.
Endomesodermal signals prevent the specification of the ectodermal territory as apical plate by restricting this region animally via the activity of FoxQ (Yaguchi et al., 2008; Yaguchi et al., 2006). Thus, loss of endomesodermal β-catenin-dependent signaling respecifies the ectoderm as apical. In the presence of endomesoderm, loss of p38 or Nodal function respecifies the ectoderm as ciliary band (data herein). This effect has previously been observed with Nodal perturbation in other sea urchin species (Duboc et al., 2004; Yaguchi et al., 2006), and has led to the proposal that the default state of the ectoderm in the absence of Nodal is the ciliated fate (Duboc et al., 2004).
Comparison of the effects of Chd or BMP perturbation on neural and ciliary band development (Fig. 4) suggests a relationship between these structures, particularly in that the loss of the ciliary band correlates with loss of neural development, while disorganization of one correlates with disorganization of the other structure. However, the results shown in Figure 6 argue against these correlations. Gsc-misexpressing embryos have almost no ciliary band and yet have a well-developed neural system, indicating that a strict correlation between neural and ciliary band development cannot be drawn. In Nodal-misexpressing embryos, synB-positive neurons arise within the oralized ectoderm and along its boundary with the vegetal ciliary band, whereas in p38-inhibited and Nodal knockdown embryos, the aboralized ectoderm expresses the ciliary band epitope except in the vegetal ectoderm; no synB neurons arise along this boundary, though the apical plate territory is preserved. This indicates that a boundary of ciliated band and ectoderm is insufficient to direct neural development, and further suggests that an aboral environment is incompatible with synB neural development. However, Gsc MO-injected embryos are aboralized, yet possess a reasonably restricted ciliary band and synB-positive neural cell bodies, belying an incompatibility between aboral ectodermal specification and synB-positive neural development. Overall, the results indicate that little or no correlation can be drawn between synB-positive neural development and either ciliary band specification/patterning or ectodermal specification. The exception may be that a disorganized (as opposed to a relocalized) ciliary band appears to correlate with disorganized synB-neural patterning (Fig. 4). However, this observation is limited to only two perturbations; it is possible that contradictory findings will emerge in the future.
BMP2/4-perturbed embryos show a disruption of OA specification in 4 different sea urchin species ((Angerer et al., 2000; Duboc et al., 2004) and data herein). All these studies show essentially the same result: that BMP2/4 is required for aboral specification. Since a critical role for BMP/Chd antagonism in dorsal/ventral axis specification is well-established in a variety of organisms (Francois et al., 1994; Hammerschmidt et al., 1996a; Holley et al., 1995; Kuroda et al., 2004; Oelgeschlager et al., 2003; Sasai et al., 1995), this finding fits well with expectations. However, when LvChd was perturbed, no impact on OA specification could be detected. The expression of all tested oral and aboral markers was normal in LvChd-perturbed embryos; moreover, a functional assay demonstrates clearly that loss of LvChd does not impact the OA axis, and instead inhibits only synB-positive neural development. Gsc is sufficient to rescue the OA axis in the patterning assay (Bradham and McClay, 2006); since Chd expression is independent of Gsc, this is internally consistent. LvBMP-morphant embryos have OA effects, and LvChd and LvBMP2/4 are functionally antagonistic, leaving open the question as to why a discrepancy exists with respect to axial specification when comparing Chd- and BMP2/4-perturbed embryos.
Loss of Chd in vertebrate embryos often has a weak phenotype compared to the effects of BMP overexpression or a constitutively activated BMP receptor (Bachiller et al., 2000; Bachiller et al., 2003; Dal-Pra et al., 2006; Khokha et al., 2005; Oelgeschlager et al., 2003). Usually, the explanation provided for this discrepancy is redundancy with Noggin (Nog) and Follistatin (Fol), two other proteins that also bind and inhibit BMPs during DV specification (Hemmati-Brivanlou et al., 1994; Iemura et al., 1998; Smith and Harland, 1992; Zimmerman et al., 1996). However, although both genes have been identified in the sea urchin genome (Lapraz et al., 2006), neither Nog nor Fol is expressed during sea urchin development based on array analysis (Samanta et al., 2006; Wei et al., 2006). The lack of embryonic SpNog expression was confirmed by QPCR analysis (not shown). Thus, redundancy with other BMP inhibitors is unlikely to explain why Chd fails to impact OA specification.
First, why doesn’t Chd knockdown result in aborally specified embryos? Loss of Chd should allow BMP2/4 signaling within the presumptive oral ectoderm, preventing its specification as such. A simple albeit speculative explanation is that Nodal dominates over BMP and prevents it from signaling within the oral territory. A mutually antagonistic relationship between Nodal and BMP has been shown in frogs (Hansen et al., 1997; Haramoto et al., 2006; Haramoto et al., 2004; Yeo and Whitman, 2001) and mice (Yang and Klingensmith, personal communication), suggesting that such antagonism may be generalizable. The mechanism for Nodal-mediated BMP inhibition in Xenopus is interesting, as it occurs independently of transcription, Nodal signaling, and even Nodal processing (Yeo and Whitman, 2001); the most recent reports indicate that BMP is bound and inhibited by the Nodal prodomain but not the mature protein (Haramoto et al., 2006; Haramoto et al., 2004). This antagonism is somehow mutual, since ectopic BMP or activated BMPR interferes with Nodal signaling, assessed both by SMAD2 activation and neuralizing function (Hansen et al., 1997; Yeo and Whitman, 2001). In sea urchin embryos, Nodal expression precedes BMP expression in the oral territory ((Bradham and McClay, 2006) and data not shown), and Nodal is positively autoregulatory (Nam et al., 2007; Range et al., 2007). In embryos injected with BMP2/4, Nodal mRNA is absent at all timepoints (Fig 3B2 and data not shown), indicating that LvBMP2/4 is capable of antagonizing LvNodal. In embryos injected with Nodal mRNA, the BMP expression domain is contracted (Fig. 1E3), potentially reflecting the reciprocal antagonism. However, it is not known whether LvBMP2/4 is autoregulatory and thus whether BMP2/4 expression is a reasonable measure of BMP2/4 signaling activity in sea urchins. An unambiguous answer to this question will eventually require a reliable readout for BMP2/4 signaling.
Next, why doesn’t Chd mRNA injection result in orally specified embryos, particularly since Chd and BMP2/4 are mutually antagonistic? Misexpression of Chd should inhibit BMP2/4 globally and thereby inhibit aboral specification. One explanation is that the dose of Chd mRNA is too low to completely inhibit BMP; however, the same dose of Chd mRNA was capable of neutralizing exogenous BMP2/4 in the rescue experiments, so this does not seem likely. An alternate possibility invokes Tolloid (Tld), a metalloprotease that cleaves Chd and thereby promotes BMP signaling (Ferguson and Anderson, 1992; Piccolo et al., 1997). If a Tld-like metalloprotease were expressed in the aboral compartment, it would prevent both endogenous and exogenous Chd from inhibiting aborally-relocated BMP2/4.
An aborally-restricted Tld-like gene might also provide an explanation for how BMP2/4 protein arrives in the aboral territory. Work performed in Drosophila embryos support a model in which Chd functions not only to locally inhibit BMP signaling activity, but also to transport BMP to a spatially distant region of the embryo (Shimmi et al., 2005; Wang and Ferguson, 2005). This transport was shown to require twisted gastrulation (Tsg) and Tld in addition to Chd (Shimmi et al., 2005). The assembly of Chd, Tsg and BMP moves dorsally toward the domain of Tld expression, where Tld-mediated cleavage of Chd releases BMP (Shimmi et al., 2005). Thus BMP signaling abruptly becomes significantly stronger in this dorsal-most region (Shimmi and O’Connor, 2003; Shimmi et al., 2005; Wang and Ferguson, 2005). Tld forms a sink for Chd: it is expressed in a region spatially distinct from Chd, thereby ‘defining’ a transport vector (Little and Mullins, 2006). For this reason, the co-expression of BMP and Chd in an overlapping spatial domain in the urchin may not be the most relevant issue. Instead, the relationship between the expression territories of Chd and Tld may actually be more important. If Tld were expressed in the aboral region, spatially opposed to Chd, this could suffice to drive Chd-mediated transport of BMP from the oral to the aboral compartment.
Several Tld-like genes have been identified in the sea urchin genome, although none of them possess the signature domain structure of vertebrate or Drosophila Tlds (Angerer et al., 2006; Hwang et al., 1994). SpBMP1, the closest phylogenetic homolog for human Tld, is spatially ubiquitous (Angerer et al., 2006; Hwang et al., 1994), inconsistent with defining a transport vector. Another metalloprotease, SpAN, induces axial defects when perturbed; however, like BMP1, SpAN is broadly expressed in the embryo, with no suggestion of an OA asymmetry (Angerer et al., 2006; Lepage et al., 1992; Reynolds et al., 1992). It will be of great interest to identify the Tld-like protease that specifically targets Chd in sea urchin embryos. The spatiotemporal dynamics of its expression in combination with functional analysis is likely to clarify many of the questions raised in this study.
We thank Robert Angerer for providing the LvBMP ORF, Robert Burke for the 1e11 antibody, Shunsuke Yaguchi for his immunostaining protocol, and Lynne Angerer, Robert Burke, Jenifer Croce, and Yu-Ping Yang for critical comments on this manuscript. Support for this work was obtained from NIH (HD 14483, GM 61464) and DOE (DRM), and from Max-Planck Gesellschaft z.F.d. Wiss. e.V. and the EU Network of Excellence Marine Genomics contract No 505430 (AJP).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.