The neural crest (NC) is a transient and multipotent embryonic cell population, unique to vertebrates, which arises at the border between the neural plate and epidermis, posteriorly to the diencephalon and migrates extensively, giving rise to a wide variety of cell types, including neurons and glia of the peripheral nervous system, craniofacial cartilage and bone, and pigment cells of the skin (Le Douarin and Dupin, 2003; Sauka-Spengler and Bronner-Fraser, 2008; Prasad et al., 2012). NC cells are induced during gastrulation by the integrated action of signals from the non-neural ectoderm, the neural plate, and the underlying paraxial mesoderm (Basch et al., 2004; Steventon et al., 2005; Sauka-Spengler and Bronner-Fraser, 2008). Studies in chick, Xenopus, and zebrafish embryos have identified major signaling pathways critical for NC induction, including BMP, Wnt, FGF, and Notch pathways (Sauka-Spengler and Bronner-Fraser, 2008; Stuhlmiller and Garcia-Castro, 2012). It has been suggested that the intermediate levels of BMP signal, which are essential for NC induction, are established by the activity of BMP antagonists such as chordin and noggin from the paraxial mesoderm (Marchant et al., 1998; Tribulo et al., 2003). However, the levels of BMP signal alone are not sufficient for NC induction and other signaling pathways are also involved (Chang and Hemmati-Brivanlou, 1998; LaBonne and Bronner-Fraser, 1998; Garcia-Castro et al., 2002). In Xenopus, FGF signals have been shown to be necessary for NC induction by the paraxial mesoderm (Monsoro-Burq et al., 2003). Wnt signal, thought to originate from the paraxial mesoderm in fish and frog or the ectoderm in chick, has been shown to regulate NC induction independently of its activity in the antero-posterior neural patterning (Bang et al., 1999; Garcia-Castro et al., 2002; Monsoro-Burq et al., 2003; Lewis et al., 2004; Wu et al., 2005; Li et al., 2009). While there is evidence that the major NC-inducing signals (FGF, Wnt, and BMP) act in parallel to regulate NC formation (Monsoro-Burq et al., 2003, 2005), FGF8a signal has been also reported to induce NC in a Wnt-dependent manner in the mesoderm (Hong et al., 2008). Therefore, it remains to be understood whether and how the NC-inducing signaling pathways further interact with each other to specify NC at the restricted territory of ectoderm, the neural fold.
The NC-inducing signals activate a genetic cascade of transcription factors in the prospective NC cells, first up-regulating neural plate border (NPB) specifiers (Zic1, Msx1, Pax3, and Dlx3) that, in turn, may trigger the induction of the NC specifiers (Snail, Slug, FoxD3, Sox9, and Sox10) (Heeg-Truesdell and LaBonne, 2004; Sauka-Spengler and Bronner-Fraser, 2008; Prasad et al., 2012). Intermediate levels of BMP activity are critical for the expression of Msx1, Pax3, and Zic1 in the NPB (Tribulo et al., 2003; Sato et al., 2005). Similarly, FGF and Wnt signaling are also required for the expression of Msx1 and Pax3 (Monsoro-Burq et al., 2005; Sato et al., 2005). Of note, Wnt induction of Slug expression depends on the presence of Pax3 but not Msx1, whereas NC induction by FGF requires both Pax3 and Msx1 activities (Monsoro-Burq et al., 2005). In addition, Msx1 activates Pax3 and Zic1 cell autonomously and acts upstream of Pax3 in NC induction (Monsoro-Burq et al., 2005). AP2a acts not only as the earliest NPB specifier to activate Pax3 but also as a NC specifier to regulate later NC specification (de Croze et al., 2011). Gbx2, a direct target gene of Wnt signaling, has also been shown to act as an early factor in the genetic cascade to regulate NC formation upstream of Msx1 and Pax3 (Li et al., 2009). NC specifiers regulate the adhesive properties, shape, and motility of NC progenitors, inducing their segregation from the dorsal neuroepithelium through the epithelial to mesenchymal transition (EMT) process and subsequent migration to many destinations in the body (Sauka-Spengler and Bronner-Fraser, 2008; Theveneau and Mayor, 2012). These transcription factors also control cell-cycle, cell survival, and multipotency maintenance of NC cell precursors (Crane and Trainor, 2006; Sauka-Spengler and Bronner-Fraser, 2008; Prasad et al., 2012). However, our current knowledge of the intricate gene regulatory network for NC specification, migration, and differentiation into diverse derivatives still appears to be preliminary. Thus, more connections and hierarchical relationships between the transcription factors involved in NC formation remain to be established.
Sp5 is a member of the Sp1 transcription factor family with a highly conserved Cys2His2 zinc finger DNA binding domain at the C-terminus and functions as a transcriptional activator or repressor in a context-dependent manner (Harrison et al., 2000; Tallafuss et al., 2001; Treichel et al., 2001; Safe and Abdelrahim, 2005; Suske et al., 2005; Fujimura et al., 2007). Sp5 is a direct target gene of Wnt/β-catenin signaling, and its proximal promoter contains five TCF/LEF binding sites that mediate direct regulation of its expression by Wnt signaling (Tallafuss et al., 2001; Fujimura et al., 2007). Its expression has been found to be elevated in several human tumors including colon cancer and gastric cancer (Safe and Abdelrahim, 2005; Chen et al., 2006). Sp5 displays a dynamic expression pattern in developing mouse embryos, being expressed in the primitive streak throughout gastrulation and later restricted to the developing brain, the spinal cord, pharyngeal region, and somites (Harrison et al., 2000; Treichel et al., 2001). Mice homozygous for a targeted mutation in the Sp5 gene show no overt abnormalities, possibly due to the functional compensation by other members of the same family (Harrison et al., 2000). However, the enhancement of the T/+ phenotype in double mutant mice (Sp5lacZ/Sp5lacZ, T/+) suggests a genetic interaction between Sp5 and Brachyury (Harrison et al., 2000). In zebrafish, Sp5 (previously called bts1) and its paralog, Sp5-like (Sp5l) have been shown to act redundantly in dorsoventral mesoderm patterning downstream of Wnt/β-catenin signaling (Weidinger et al., 2005). Sp5l is also required for hindbrain patterning and tail development (Thorpe et al., 2005; Weidinger et al., 2005). The Xenopus ortholog of Sp5, designated as XSPR-1, has been also reported to be expressed in the non-involuting marginal zone, forebrain, otic vesicles, and the midbrain/hindbrain boundary in early embryogenesis (Ossipova et al., 2002), but its developmental function in Xenopus has not been analyzed.
In the current study, we present a novel role of Sp5 in NC formation in Xenopus early embryogenesis. Sp5 regulates NC induction via NPB specifiers, Msx1 and Pax3, downstream of Wnt and FGF pathways. In contrast, it does not affect neural development. We thus suggest that Sp5 functions as one of the earliest factors in the genetic cascade to mediate the activities of Wnt and FGF signaling only in NC formation.
Xenopus Sp5 is Expressed in the Prospective Neural Crest Domain
BLAST searches revealed that Xenopus laevis has two duplicated orthologs of Sp5, previously called XSp6 and XSPR-1. These two paralogs are over 94% identical with each other at the amino acid level (data not shown). In addition, XSp6 shows 65 and 28% amino acid identities to mouse Sp5 and Sp6, respectively, suggesting that it is a Xenopus ortholog of Sp5 but not of Sp6. To avoid confusion in gene nomenclature, we will re-designate XSp6 and XSPR-1 as Sp5.a and Sp5.b, respectively, hereafter by following new guidelines for Xenopus gene naming, which are found on the Xenbase website. To unravel the biological function of Xenopus Sp5, we first sought to examine its expression pattern in early embryogenesis. However, the developmental expression pattern of Sp5.b has been described previously (Ossipova et al., 2002), which suggests its possible role in NC development. Thus, we further investigated the expression of Sp5 at the time of NC formation by performing in situ hybridization with anti-Sp5.a probe. At the late gastrula stage (stage 12) when NC induction occurs, Sp5.a was observed in the dorso-lateral marginal zone, being excluded from the most dorsal region (Fig. 1A, B). This broad domain includes the area from which NC arises (Steventon et al., 2009). During the early neurula stage (stage 14), Sp5.a was expressed more markedly in a posterior-dorsal domain and was absent from the dorsal midline and anteriormost region of embryo (Fig. 1C), which resembles the expression pattern of the NC markers, Snail2 and Gbx2 (Li et al., 2009). Sections of the early neurulae revealed that Sp5.a is expressed in the ectodermal layer but not in the mesoderm (Fig. 1D, E). As neurulation proceeds, Sp5.a expression became more restricted to the NC area lateral to the neural fold (Fig. 1F). Double staining in situ hybridization with an NC marker, Sox10, confirmed that Sp5.a and Sox10 transcripts clearly co-localize in this presumptive NC region (Fig. 1G). At the tadpole stages, Sp5.a was detectable in discrete domains within the brain as well as NC-related structures including the branchial arches, otic vesicle, and dorsal midline in the fin (Fig. 1H). Overall, the spatial expression pattern of Sp5.a is similar to that of Sp5.b, confirming that Sp5 may be implicated in NC formation in Xenopus early embryo.
Depletion of Sp5 Disrupts Formation of Neural Crest Derivatives
In order to test whether Sp5 is essential for NC formation, we employed a MO-mediated knockdown approach. Since Sp5.a and Sp5.b might functionally substitute for each other, we designed two kinds of MOs (MO1 and MO2) that could target simultaneously to the sequences upstream of or around translation initiation sites of both Sp5.a and Sp5.b mRNA. As analyzed by western blotting, MO1 and MO2, but not control MO, could block the production of both Sp5.a and Sp5.b proteins (Fig. 2A), confirming the specificity and efficiency of the MOs. Injection of Sp5 MO1 resulted in severe defects in craniofacial structures such as the eyes and in pigmentation or dorsal fin along the trunk (58%, n=65; Fig. 2B). These defective phenotypes could be rescued to some degree by co-injection of Sp5.a mRNA (62%, n=60), which does not contain the MO target site and is resistant to its inhibition. To further corroborate the specificity of the MO phenotypes, we investigated whether the two MOs would display synergistic effects in disrupting the craniofacial structures. Thus, we titrated the MOs down to levels that did not elicit the phenotypes on their own. As shown in Figure 2C, injection of either MO1 or MO2 at such suboptimal doses did not yield any phenotypes, whereas co-injection of the MOs at the same reduced concentrations resulted in defective eye formation and pigmentation (93%, n=45), phenocopying the defects in the embryos injected with a high level of MO1 alone. Therefore, these results support the specific effects of the MOs on formation of craniofacial structures. Histological analysis also revealed that Sp5 MO1, but not control MO, could disrupt formation of the lens and multilayered retina in the eye and of head mesenchyme in the injected side of embryo (Fig. 2D). We further examined the effect of the MOs on formation of a NC derivative, craniofacial cartilage. As visualized by alcian blue staining, unilateral injection of Sp5 MO1, but not control MO, could cause a severe loss or reduction of craniofacial skeletal elements in the injected side as compared with the uninjected control side of embryo (Fig. 2E). Co-injection of the two MOs at below threshold levels could also exhibit synergy in disrupting cartilage formation (Fig. 2E). The gross morphology analysis of craniofacial cartilages revealed that NC-derived cartilages such as Meckel's and ceratobranchial ones were more markedly atrophied or absent. Together, these results suggest that Sp5 is critical for development of NC-related structures.
Sp5 Promotes Neural Crest Formation
We next performed a gain-of-function analysis to test whether Sp5 could regulate NC formation. We overexpressed Sp5 mRNA in the NC region by targeting the injection toward the neural plate border (NPB) and investigated NC marker expression at the mid-neural stages with in situ hybridization analysis. As shown in Figure 3A, injection of Sp5.a could expand the expression of NC markers, Sox9 and Slug, in the injected side (43%, n=37 and 51%, n=55, respectively), compared with the uninjected control side of embryo. We also examined whether Sp5.b might have the same effect on NC formation as Sp5.a. Like Sp5.a, injection of Sp5.b mRNA in the prospective NC region could also result in the enlargement of the Slug (44%, n=25 for Sp5.a; 54%, n=28 for Sp5.b) or Sox10 (50%, n=28 for Sp5.a; 37%, n=30 for Sp5.b)-positive areas in the injected side (Fig. 3B). Markers of the NPB and NC begin to be expressed during and shortly after gastrulation (Stuhlmiller and Garcia-Castro, 2012). Moreover, Sp5 is expressed in the prospective NC regions from gastrula through tadpole stages as shown above. Thus, we investigated the time window in which Sp5 is able to promote NC specification. For this, we used a dexamethasone (DEX)-activated form of Sp5.a (Sp5.a-GR), which allows conditional activation of Sp5 function at specific stages. Without DEX treatment, injected Sp5.a-GR mRNA had no effect on the expression of a NPB marker, Msx1, and a NC marker, Slug (Fig. 3C). In contrast, Sp5.a-GR-injected embryos had expanded expression of those markers (88%, n=32 for Msx1; 47%, n=32 for Slug) when DEX was treated from stage 10 onward but not from stage 13 or 15 (Fig. 3C). Together, these results demonstrate that Sp5.a and its paralog, Sp5.b, can enhance NC induction in vivo during gastrula stages.
Sp5 is Essential for Neural Crest Induction
We next assessed the effect of knockdown of Sp5 on the initial steps of neural crest development. Unilateral injection of Sp5 MO1, but not control MO, could result in complete loss or reduction of the expression of NC markers, Sox10 (81%, n=62), Sox9 (73%, n=78), or Slug (44%, n=68) in the injected side, compared with the uninjected control side of embryos (Fig. 4A). Furthermore, injection of either MO1 or MO2 at suboptimal levels had no effect on the expression of Sox10, whereas co-injection of the two MOs at the same reduced doses could interfere with its expression (63%, n=30; Fig. 4B), confirming a specificity of the MOs in suppression of NC markers. NC can be induced ectopically in the animal cap tissue by injecting Wnts and anti-BMP neuralizing factors such as noggin and chordin (Mayor et al., 1995; Chang and Hemmati-Brivanlou, 1998). Using this animal cap assay, we tested whether Sp5 is also required for ectopic NC formation. Co-expression of noggin and XWnt8 could induce the expression of Sox9 and Slug markers in animal caps (Fig. 4C). Co-injection of Sp5 MO1, but not control MO, could abrogate this ectopic expression, which could be reversed by co-injecting Sp5.a mRNA immune to MO inhibition (Fig. 4C). Notably, injection of Sp5 MO1 could expand Zic1 expression in the neural fold (Fig. 4A). Since Zic1 plays a dual role in specification of both NC and preplacodal regions, which depends on a specific level of its activity (Hong and Saint-Jeannet, 2007), we assumed that Sp5 might function as a posteriorizing factor of the neural fold to inhibit formation of the anterior preplacodal region by repressing Zic1 expression. To test this possibility, we examined the effect of Sp5 on the expression of a Zic1-induced preplacodal marker, Six1. As shown previously (Brugmann et al., 2004), inhibition of BMP signaling could up-regulate Six1 expression in the animal cap tissue and this effect was impeded by co-injection of Sp5.a (Fig. 4D), suggesting an inhibitory role of Sp5 in anteriorization of the neural fold. Taken together, these data demonstrate that Sp5 is indispensable for the early specification of the NC cells both in vivo and in vitro.
Sp5 is Not Involved in Neural Induction and Patterning
The paralog of Sp5, Sp5-like (Sp5l, previously named XSPR-2) is 46% identical with Sp5.a at the amino acid level (data not shown) and exhibits distinct expression pattern in early embryogenesis (Ossipova et al., 2002). Sp5l has been shown to act downstream of Wnt signaling in neural patterning in zebrafish (Weidinger et al., 2005). Wnt signaling pathway regulates the antero-posterior patterning of neuroectoderm in vertebrate embryos (Kiecker and Niehrs, 2001). Since Sp5 as well as Sp5l is a target gene of Wnt signaling (Tallafuss et al., 2001; Weidinger et al., 2005), we next asked whether Sp5 would also function in neural development. Unilateral injection of MO1 or MO2 did not affect the expression of a pan-neural marker, Sox2, and a neuronal differentiation marker, N-tubulin (Fig. 5A–D), suggesting that Sp5 is not critical for neural induction and primary neurogenesis. In addition, the anterior-posterior patterning of neural tissue was not altered in the Sp5-depleted embryos as evidenced by no effect on the expression of forebrain markers Otx2 and Bf1, a midbrain-hindbrain marker En2 and spinal cord marker HoxB9 in the injected side of embryos (Fig. 5E–J, M, N). Notably, the expression of a hindbrain marker, Krox20, by rhombomeres three (r3) and five (r5) was slightly shifted posteriorly on the injected side, whereas only Krox20 expression in r5 was strongly reduced by the MOs (84%, n=51 for MO1; 62%, n=30 for MO2; Fig. 5K, L). In Xenopus, r5 generates NC that migrates laterally, such that the Krox20 stripe in r5 extends more laterally than that in r3 (Bradley et al., 1993). Thus, the loss of Krox20 expression in r5 indicates the depletion of the NC derived from the fifth rhombomere. We conclude that Sp5 is implicated in NC formation only but not in neural induction and patterning.
Sp5 Regulates the Expression of NPB Specifiers During NC Induction
The gain-of-function of Sp5 expanded Msx1 expression along the neural fold (Fig. 3C), suggesting that Sp5 acts upstream of NPB specifiers in the genetic cascade to regulate NC formation. Thus, we further analyzed its regulation of the expression of NPB markers. NPB markers can also be induced in the ectodermal tissue by a combination of BMP inhibition with Wnt or FGF signaling (Tribulo et al., 2003; Monsoro-Burq et al., 2005). Overexpression of Sp5.a could induce the expression of Msx1 and Pax3 in the animal cap tissues where BMP signaling was attenuated by injecting a dominant negative (DN) BMP4 receptor (Fig. 6A, lane 5), suggesting that Sp5 mediates the activity of Wnt and/or FGF signals to induce the expression of the NPB markers. Msx1 expression lateral to the neural plate of early embryo depends on Wnt signal (Monsoro-Burq et al., 2005). Thus, targeted expression of GSK3, an inhibitor of Wnt signaling, abolished the in vivo expression of Msx1 in the neural fold (69%, n= 36; Fig. 6B), which could be reversed by co-expression of Sp5.a (51%, n=35; Fig. 6B). Furthermore, injection of Sp5 MO1, but not Co MO, could reduce Msx1 expression in the neural plate border (62%, n=29; Fig. 6B). These results confirm a critical role of Sp5 as a downstream effector of Wnt signaling in the regulation of Msx1 expression. Although Msx1 and Xvent-1, both of which are target genes of BMP signaling (Gawantka et al., 1995; Suzuki et al., 1997), became abrogated in the animal caps expressing a DN BMP4 receptor (Fig. 6A, lane 4), co-injection of Sp5 recovered the expression of Msx1 only but not of Xvent-1 (lane 5), indicating that Sp5 could up-regulate Msx1 expression independently of BMP signaling. Consistently, the knockdown of Sp5 had no effect on BMP4 induction of Msx1 (Fig. 6C, lanes 4 and 6) and down-regulated the endogenous expression of Msx1 in the ectodermal tissue but not that of Xvent-2 (Fig. 6C, lane 7), which is another BMP target gene. Notably, injection of Sp5.a alone could up-regulate the expression of Msx1 only but not of Pax3 (Fig. 6A, lane 6), and Sp5.a induction of Msx1 expression, however, did not occur in the animal caps treated with a protein synthesis inhibitor, cycloheximide (CHX) (Fig. 6D, lanes 6 and 7), implying its indirect control of Msx1 transcription. As Sp5 induces the expression of Msx1 and Pax3 as shown above, we next examined whether they might function downstream of Sp5 in the NC genetic cascade. As expected, Sp5 MO1 could markedly decrease the expression of Sox10 in the injected side (83%, n=50; Fig. 6E), and this reduction could be recovered by co-injection of Msx1 or Pax3 (60%, n=45 for Msx1; 73%, n=53 for Pax3; Fig. 6E). In summary, these results suggest that Sp5 regulates NC specification by up-regulating the expression of NPB specifiers.
Neural Crest Induction by FGF or Wnt Signal Requires Sp5
Sp5 has been shown to be a target gene of Wnt signals in mouse and zebrafish (Tallafuss et al., 2001; Fujimura et al., 2007). In addition, FGF and Wnt/β-catenin signaling are able to activate the expression of its paralog Sp5l during gastrulation (Zhao et al., 2003; Weidinger et al., 2005). Since FGF and Wnt signal are key upstream inducers of NC, we tested whether Sp5 might also be their target gene in Xenopus and act as an essential mediator of their activities in NC specification. Real-time PCR analysis using animal cap tissues revealed that Xenopus Sp5 can be induced by FGF8a or Wnt8 signal (Fig. 7A). Given that previously FGF8a signal has been shown to be a potent inducer of Wnt8 expression in animal cap tissue and in whole embryos (Hong et al., 2008), it is possible that FGF8a signal could induce Sp5 expression indirectly via activation of Wnt signaling. As expected, co-injection of dominant negative (DN) XWnt8 or DN Tcf3, which inhibits Wnt signaling, could interfere with FGF8a induction of Sp5 (Fig. 7B), suggesting the requirement of Wnt signaling for FGF8a to activate Sp5 expression.
We next investigated whether NC formation by FGF and Wnt signals might depend on Sp5 activity. Unilateral overexpression of FGF8a mRNA could augment strongly the expression of Sox10 along the neural fold in the injected side of embryos (82%, n=33; Fig. 8A), which was impeded by co-injection of Sp5 MO1 (90%, n=30; Fig. 8B). The inhibitory effect of MO1 could be rescued by co-expressing Sp5.a mRNA without the MO target site (58%, n=38; Fig. 8C), suggesting that Sp5 is specifically required for FGF8a induction of NC. Consistently, injection of DN FGFR4, which blocks FGF8 signaling through FGFR4, could abrogate Sox10 expression in the NC region (70%, n=59; Fig. 8G, an arrow) and this inhibitory effect could be recovered by co-injection of Sp5.a (63%, n=71; Fig. 8K), thereby indicative of Sp5 mediation of FGF signal in NC induction.
We also tested the requirement for Sp5 during NC induction by XWnt8. As XWnt8 can influence strongly the dorso-ventral patterning of mesoderm in Xenopus embryos (Christian and Moon, 1993), we injected XWnt8 DNA (XWnt8-CSKA), which starts to be expressed after the onset of zygotic transcription, to overcome this early effect. Like FGF8a RNA, XWnt8 DNA could expand markedly the expression of Sox9 in the injected side as compared with the uninjected control side of embryos (77%, n=35; Fig. 8D). Co-injection of Sp5 MO1 could attenuate the potent XWnt8 enhancement of Sox9 expression (69%, n=36; Fig. 8E), which was reversed by co-expressing Sp5.a mRNA (56%, n=36; Fig. 8F), indicating the specific effect of the MO on XWnt8 induction of NC. We further analyzed whether Sp5 could recover the depletion of NC caused by inhibition of Wnt signaling. Targeted injection of DN XWnt8, GSK3, or DN Tcf3 mRNA led to the loss or significant reduction of Sox10 expression in the injected side (DN XWnt8, 75%, n=64; GSK3, 80%, n=50; DN Tcf3, 58%, n=53; Fig. 8H–J). The suppressive effects on NC induction of these inhibitors of Wnt signaling could be rescued efficiently by co-expressing Sp5.a (49%, n=69, 59%, n=44 and 71%, n=58, respectively; Fig. 8L–N). Of note, Sp5 recovery of DN Tcf3 inhibition of Sox10 expression suggests that Sp5 acts as a target gene of Wnt signal to mediate its activity in NC formation. In summary, these data demonstrate that Sp5 is a critical downstream mediator of FGF and Wnt signaling for NC specification.
In this study, we have demonstrated that Sp5 plays a critical role in NC formation in Xenopus. During gastrulation, its developmental expression begins to be detectable in the dorso-lateral marginal zone of early embryo, which includes NC territory and persists in the restricted NC areas such as the lateral neural plate border in neurulae and branchial arches and dorsal fin in tadpole. As a result, Sp5-deficient embryos exhibit morphological defects in craniofacial cartilage, pigmentation, and fin structure. Gain-of-function of Sp5 expands the expression of NC markers such as Sox9, Sox10, and Slug, whereas its knockdown inhibits both endogenous and ectopically-induced expression of those markers. Sp5 is also able to induce the expression of NPB specifiers, Msx1 and Pax3. Moreover, our analysis using an inducible Sp5 construct reveals that the time window of its ability to promote the expression of NPB and NC markers is limited to gastrula stages at which NC induction is initiated. The NC-inducing signals (Wnt, FGF, and BMP) trigger a genetic cascade of transcription factors, first activating the expression of NPB specifiers, which could subsequently control the induction of NC markers (Sauka-Spengler and Bronner-Fraser, 2008; Prasad et al., 2012). Sp5 is a direct target gene of Wnt/β-catenin signaling (Fujimura et al., 2007) and mediates the activity of this pathway in NC induction (Fig. 8). Therefore, we propose that Sp5 might function as one of the earliest factors in the genetic cascade to regulate NC specification (Fig. 8O).
Xenopus Sp5 can induce, albeit indirectly, Msx1 expression by itself in the animal cap tissue, and its depletion interferes with Msx1 expression in the neural plate border as well as in the naive ectoderm, suggesting that Sp5 may function to induce another factor indispensable for Msx1 expression. Pax3 expression can be induced by Sp5 only when BMP signaling is attenuated (Fig. 6A), indicating that a factor induced by BMP down-regulation might be required for this induction. Since Msx1 can also induce Pax3 expression in neuralized animal cap tissue and in vivo (Monsoro-Burq et al., 2005), it is possible that Sp5 may control Pax3 expression via induction of Msx1. Although the expressions of Msx1 and Pax3 appear to be regulated differentially by Sp5, both of them could recover NC induction inhibited by loss of the Sp5 function. Furthermore, Msx1 acts upstream of Pax3 for NC induction (Monsoro-Burq et al., 2005). Given these findings, it seems likely that Sp5, Msx1, and Pax3 act in a linear regulatory cascade for NC specification.
Msx1 has been shown to be a direct downstream target of BMP signaling (Suzuki et al., 1997; Gonzalez et al., 1998). BMP induction of Msx1 expression can occur in the absence of Sp5 function, suggesting that Sp5 regulates Msx1 expression in parallel to BMP pathway. Wnt and FGF signals are also required for Msx1 expression in the neural fold (Monsoro-Burq et al., 2005), though it remains unknown whether Msx1 is a direct target of these signals. Like Sp5, Wnt8, FGF3, and FGF8 ligands are expressed in the dorso-lateral marginal zone of embryos during gastrula stages when NC induction begins (Monsoro-Burq et al., 2003; Hong et al., 2008). Our study shows that Sp5 acts as a critical inducer of Msx1 expression to regulate NC specification downstream of Wnt8 and FGF8 signals. These results suggest that Sp5 could mediate the induction of Msx1 expression by Wnt and FGF signals during the initial stages of NC specification. Consistently, Sp5 can recover Msx1 expression abrogated by GSK3-mediated inhibition of Wnt signaling. Recent evidence shows that FGF8a induces indirectly the NC through the activation of Wnt8 expression in the paraxial mesoderm (Hong et al., 2008). In addition, FGF8a can induce indirectly Sp5 expression via Wnt signaling as shown in our data. Collectively, these suggest that FGF8a and Wnt8 pathways function as part of the same signaling cascade to generate NC through Sp5.
Zic1 acts as one of the NPB specifiers to regulate NC and preplacodal ectoderm (PE) formation (Hong and Saint-Jeannet, 2007). A specific level of Zic1 activity in the ectoderm alone promotes PE fate at the expense of NC specification (Hong and Saint-Jeannet, 2007), whereas the combined activity of Zic1 and Pax3 or Gbx2 induces NC (Monsoro-Burq et al., 2005; Sato et al., 2005; Hong and Saint-Jeannet, 2007; Li et al., 2009). We have shown that loss of the Sp5 function decreases the expression of NC markers but expands that of Zic1 in the neural fold. Given that NC and PE fates are exclusive from each other (Hong and Saint-Jeannet, 2007), knockdown of Sp5 could enhance PE development at the expense of NC fate. In support of this, Sp5 could down-regulate the expression of a preplacodal marker, Six1. While Zic1 and its target gene, Six1, can be activated in vivo and in vitro by attenuation of BMP signaling (Mizuseki et al., 1998; Brugmann et al., 2004; Sato et al., 2005; Hong and Saint-Jeannet, 2007), Sp5 MO-induced expansion of Zic1 expression appears not to involve BMP inhibition because Sp5 has no effect on this pathway (Fig. 6). The synergistic activity of Zic1 and Pax3 in inducing NC fate can be achieved only when these two factors reach a proper balance (Hong and Saint-Jeannet, 2007). Furthermore, their cooperative induction of NC differentiation requires Wnt signaling (Monsoro-Burq et al., 2005; Sato et al., 2005). Thus, it is tempting to speculate that Sp5 functions as a mediator of Wnt signaling to maintain their balanced levels by down-regulating Zic1 expression for NC determination. Since Sp5 works as a transcriptional repressor as well as activator (Fujimura et al., 2007), it is possible that it could reduce directly Zic1 expression. Zic1 expression has been also shown to be up-regulated in the embryos lacking Pax3 function (Sato et al., 2005; Hong and Saint-Jeannet, 2007). Of note, depletion of Sp5 expands the lateral expression domain of Zic1 only where Pax3 and Zic1 are co-expressed in the neural fold, without affecting its most anterior expression. Therefore, it is possible that Sp5 would down-regulate indirectly Zic1 expression via Pax3.
Wnt/β-catenin signaling regulates both NC specification and antero-posterior patterning of neural plate (Kiecker and Niehrs, 2001; Sauka-Spengler and Bronner-Fraser, 2008; Stuhlmiller and Garcia-Castro, 2012). However, these two events induced by Wnt signal can be dissociated from each other in Xenopus (Monsoro-Burq et al., 2003; Wu et al., 2005). It has been shown that NC induction requires posteriorization of the neural fold, which is independent of antero-posterior neural plate patterning (Aybar and Mayor, 2002; Li et al., 2009). In support of this, Gbx2, a homeobox gene regulated directly by Wnt signaling, acts as a posteriorizing factor of the neural fold for NC induction without any effect on neural patterning (Li et al., 2009). Likewise, loss of the Sp5 function does not affect neural induction and antero-posterior patterning in Xenopus. Notably, the expression of a hindbrain marker, Krox20, became abrogated by Sp5 depletion only in rhombomere five (r5) from which NC derives but not in rhombomere three (r3). Sp5 also has a repressive effect on PE development but not on NC fate, suggesting its role in posteriorization of the neural fold for NC induction. Unlike Sp5, its paralog Sp5l appears to function as a mediator of Wnt signaling to posteriorize the CNS in zebrafish (Weidinger et al., 2005). Xenopus Sp5l, previously named XSPR-2, exhibits a different expression pattern from that of Sp5 in early embryogenesis (Ossipova et al., 2002). In particular, it is expressed on the dorsal ridge of the neural fold along the antero-posterior axis of embryo, suggesting its possible role in neural patterning. Together, these findings indicate that Wnt signaling employs respective regulatory factors for NC induction and neural patterning that occur simultaneously during early development.
Specific knockdown of Sp5 in Xenopus embryos reveals its developmental role in neural crest formation. This finding is challenged by the observation that Sp5 knockout mice do not display overt developmental phenotypes (Harrison et al., 2000). It is possible that the embryonic cells of each species may have different requirements for Sp5. However, loss-of-function of Sp5 in the T/+ genetic background could enhance the T/+ phenotype (Harrison et al., 2000), suggesting a critical role for Sp5 in mesoderm formation in early development. In addition, the extent of functional redundancy or compensation among family members would be another likely explanation for this discrepancy. Xenopus has three orthologs of mammalian Sp5, named Sp5.a, Sp5.b, and Sp5l. However, human and mouse have no ortholog of amphibian Sp5l as revealed by BLAST searches of the genomes. The murine Sp5 is expressed in the ectodermal and mesodermal layers of primitive streak during gastrulation and later in the midbrain, spinal cord, notochord, somites, pharyngeal region, otic vesicles, and tailbud (Harrison et al., 2000; Treichel et al., 2001). Xenopus Sp5 and Sp5l are differentially expressed during various stages of development (Ossipova et al., 2002): Sp5 is predominantly found in the epithelial and sub-epithelial layers during gastrulation, whereas Sp5l expression is restricted to the presumptive mesoderm of the involuting marginal zone. During neurula and tailbud stages, Sp5 is expressed in the forebrain, mid-hindbrain boundary, neural crest domain, otic vesicles, and branchial arches. Sp5l transcripts are detectable in the notochord, lateral plate mesoderm, neuroectoderm along the neural fold in neurulae, and exclusively in the tip of the tail, being excluded from the head region, at the tailbud and tadpole stages.
These expression patterns of Sp5 and Sp5l suggest that both of them recapitulate different aspects of the expression of mouse Sp5. Both zebrafish Sp5 and Sp5l, which are also called bts1 and spr2, respectively, are expressed in the mesoderm precursors during early gastrulation (Tallafuss et al., 2001; Zhao et al., 2003). Notably, spr2 morphants did not exhibit any obvious defects in dorsoventral mesoderm patterning, whereas the combined knockdown of both bts1 and spr2 resulted in synergistic defects in this process (Zhao et al., 2003; Weidinger et al., 2005). These findings indicate that Sp5 and Sp5l share a functional redundancy when expressed in the same tissue. In Xenopus, Sp5 and Sp5l have distinct expression patterns from gastrulation through the tadpole stages as compared above. Thus, depletion of Sp5 appears to have a strong inhibitory effect on NC induction but not on neural induction and patterning. Zebrafish Sp5l has been shown to be involved in the anterior-posterior neural patterning (Weidinger et al., 2005). As mentioned above, mouse has no counterpart of frog and fish Sp5l in its family of Sp1-like proteins, but it is possible that like Sp5l, another member of the family such as Sp8 may compensate for Sp5 in the Sp5 knockout mice. Sp8 shows similar expression pattern to that of Sp5, being restricted to the primitive streak during gastrulation and, later, to the spinal cord, otic vesicles, mid-hindbrain boundary, and tailbud (Treichel et al., 2003). Like Sp5, it is also induced by Wnt/β-catenin signaling. Thus, this compensation by Sp8 may account for the absence of phenotypes in the Sp5-null mice.
In conclusion, we have identified a novel role of Sp5 as an early regulator in the NC genetic cascade, thus providing an insight into how the NC-inducing signals are connected to transcription factors specifying the NPB and how NC induction and neural patterning are regulated separately by the same signaling pathway. Future experiments are warranted to elucidate the more precise mechanism by which Sp5 has different regulatory effects on the expression of NPB specifiers including Msx1, Pax3, and Zic1. In addition, it will be necessary to investigate in detail the cross-regulation of these factors.
Embryo Manipulation and Lineage Tracing
In vitro fertilization, embryo culture, and microinjection were carried out as described previously (Sive et al., 1989). Developmental stages of embryos were determined according to the Nieuwkoop and Faber's normal table of development (Nieuwkoop and Faber, 1994). For lineage tracing, β-galactosidase mRNA (LacZ, 50 pg) was co-injected and its activity was visualized with the X-Gal substrate (Duchefa Biochemie).
Constructs, Morpholino Oligonucleotides, and RNA Synthesis
The complete coding region of Xenopus laevis Sp5.a (GenBank Accession No. NM_001096636) was amplified by PCR and inserted into the BamHI and XhoI sites of pCS2+ vector (Sp5.a-CS2+). Sp5.a-Myc and Sp5.b-Myc (GenBank Accession No. NM_001085859) constructs were generated by subcloning the PCR products encompassing their respective coding regions and MO target sites (5′ UTR) into the BamHI site of pCS2+-Myc vector. To produce the dexamethasone (DEX)-inducible construct, the coding region of the human glucocorticoid receptor ligand-binding domain (hGR) was fused in frame to the C-terminus of Sp5.a and subcloned into the BamHI and XhoI sites of pCS2+ vector (Sp5.a-GR-CS2+). For sense mRNA synthesis, Sp5.a-CS2+, Sp5.a-GR-CS2+, Sp5.a-Myc-CS2+, and Sp5.b-Myc-CS2+ constructs were linearized with NotI and transcribed with SP6 RNA polymerase. The following constructs were described previously: DN BMP4 receptor (Suzuki et al., 1994), BMP4 (Maeno et al., 1994), Msx1 (Maeda et al., 1997), Pax3 (Monsoro-Burq et al., 2005), FGF8a (Shim et al., 2005), DN FGFR4 (Hongo et al., 1999), Noggin (McGrew et al., 1997), GSK3 (Yost et al., 1996), XWnt8 (Christian et al., 1991), XWnt8-CSKA (Christian and Moon, 1993), DN XWnt8 (Hoppler et al., 1996) and DN Tcf3 (Brannon et al., 1997). Capped mRNAs were in vitro synthesized using the mMessage mMachine kit (Ambion, Austin, TX). Anti-sense morpholino oligos were obtained from Gene Tools (Philomath, OR). Xenopus Sp5 MOs had the following sequence: MO1, 5′-ACTGGTGGTGATACATGAATCAGTT-3′; MO2, 5′-TAGGACAGCCACGGCAGCCATAGTC-3′. Control MO was a standard morpholino oligo from Gene Tools whose sequence is 5′-CCTCTTACCTCAGTTACAATTTATA-3′.
RT-PCR and Quantitative RT-PCR
For quantitative and conventional RT-PCR analyses, total RNA was extracted from whole embryos and animal cap explants using TRI Reagent (Molecular Research Center) and treated with RNase-free DNase I to remove genomic DNAs. RNA was transcribed using M-MLV reverse transcriptase (Promega, Madison, WI) at 37°C for 1 hr. The numbers of PCR cycles for each primer set were determined empirically to maintain amplification in the linear range. The sequences of PCR primers used here are as follows: Msx1 (Suzuki et al., 1997); Xvent-1 (Gawantka et al., 1995); Xvent-2 (Onichtchouk et al., 1996); Sox9 (Monsoro-Burq et al., 2003); Slug (Mizuseki et al., 1998); ODC (Bassez et al., 1990); Pax3 (de Croze et al., 2011); Six1 (Pandur and Moody, 2000). Real-time RT-PCR was performed using the primers for Sp5.a, 5′-TGCCCTGGTGGAGCATTCAGC-3′ (forward), 5′-GTTTGGCAGT TGGGGCACCT-3′ (reverse) and the LightCycler 480 DNA SYBR Green I Master kit (Roche Diagnostics, Indianapolis, IN) on LightCycler (Roche Diagnostics). The cycling conditions were as follows: denaturation at 95°C (30 s), annealing at 60°C (30 s), and extension at 72°C (30 s). In each case, ODC was used as an internal control, and each bar on the histograms has been normalized to the level of ODC expression. The histograms in Figure 7 are representative results of three independent experiments.
In Situ Hybridization and Western Blotting
Whole-mount in situ hybridization was performed as described in Harland (1991). Anti-sense digoxigenin or fluorescein-labeled RNA probes were in vitro synthesized using template cDNA encoding Msx1 (Maeda et al., 1997), Zic1 (Hong and Saint-Jeannet, 2007), Sox9 (Spokony et al., 2002), Sox10 (Aoki et al., 2003), Slug (Mayor et al., 1995), Sox2 (Mizuseki et al., 1998), Otx2 (Blitz and Cho, 1995), and Krox20 (Bradley et al., 1993). For anti-sense Sp5.a probe, Sp5.a-pGEM T vector was linearized with SpeI and transcribed with T7 RNA polymerase. BM purple (Roche) and Fast Red (Roche) were used as substrates for the alkaline phosphatase. Western blotting analysis was performed according to a standard protocol with anti-Myc (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA) and anti-β-actin (1:1,000, Santa Cruz) antibodies.
Histology and Cartilage Staining
For histology, embryos were fixed in MEMFA at stage 40 and embedded in Paraplast (Leica, Exton, PA). Sections were cut at thickness of 10 μm on a rotary microtome and stained with eosin (Sigma, St. Louis, MO). For cartilage staining, embryos were fixed in MEMFA at stage 45, dehydrated in ethanol and stained three nights in 0.04% alcian blue/30% acetic acid in ethanol. Embryos were washed extensively with ethanol and rehydrated in 2% KOH solution. Finally, the embryos were washed in 20% glycerol/2% KOH for at least 1 hr and dehydrated through a glycerol series into 80% glycerol. Cranial cartilages were manually dissected out and photographed.
We are grateful to Chang-Soo Hong for providing DNA constructs.