Neural crest (NC) cells arise bilaterally at the border of the nonneural ectoderm and the neural plate, in the region fated to give rise to the neural folds, in an A–P range from the MHB region through the spinal cord. During neurulation, the NC cells arrive to the most dorsal regions of the embryo, where they delaminate, undergoing an epthethelial–mesenchymal transition. NC cells then begin to emigrate throughout the embryo in distinct paths, going ventrally and then to both the anterior and posterior of the body (Figure 1.1). NC cells undergo differentiation to a wide range of cell lineages including facial cartilage and bone, smooth heart muscle, peripheral nervous system neurons, glial cells, and melanocytes.

NC induction is a complex process, likely dependent on multiple temporal and spatial signaling inputs during early development. NC induction is initiated in early gastrula stages and may proceed to later neurula stages. Experiments in amphibians and chick showed that paraxial mesoderm induces NC in the adjacent neural folds region (Bonstein et al., 1998; Selleck and Bronner-Fraser, 1995). During gastrulation and early neurulation, the paraxial mesoderm underlies the presumptive neural crest and its presence is required for proper NC induction. It is not clear if paraxial mesoderm is required in zebrafish or mouse embryos for neural crest induction. In zebrafish mutants lacking paraxial mesoderm, there is no parallel loss of NC fates (Ragland and Raible, 2004). Additional experiments in amphibian and chick embryos have also shown that cell–cell interactions between the nonneural epidermal ectoderm and the neural plate is required to induce NC (Dickinson et al., 1995; Mancilla and Mayor, 1996; Selleck and Bronner-Fraser, 1995). Lineage trace studies in amphibians show that NC can be derived from either the epidermal or the neural plate fated cells (Moury and Jacobson, 1990). Studies in Xenopus suggest that a Wnt-signal from the paraxial mesoderm is required at two time intervals, an early gastrula-dependent Wnt-signal that acts with BMP antagonism and a later neural-stage signal that acts in concert with BMP signaling to maintain NC cell fates (Steventon et al., 2009). This study proposes that the need for these two time windows may explain some of the discrepancies seen in Xenopus and chick explant studies, with regard to the necessity of BMP signaling in NC formation.

The role of Wnt signaling in NC induction has been shown in all major vertebrate developmental biology model systems (Figures 8.1 and 8.2) (Brault et al., 2001; Chang and Hemmati-Brivanlou, 1998; Dorsky et al., 1998; Garcia-Castro et al., 2002; Ikeya et al., 1997; LaBonne and Bronner-Fraser, 1998; Lewis et al., 2004; Saint-Jeannet et al., 1997). Initial studies in Xenopus showed that the ectopic expression of various canonical Wnt ligands could induce NC in ectodermal explants in addition to spreading the NC field in developing embryos (Chang and Hemmati-Brivanlou, 1998; LaBonne and Bronner-Fraser, 1998; Saint-Jeannet et al., 1997). In Xenopus and chick embryos, overexpression of canonical Wnt pathway components such as frizzled, Lrp6, or β-catenin expanded NC fates (Deardorff et al., 2001b; LaBonne and Bronner-Fraser, 1998; Saint-Jeannet et al., 1997; Tamai et al., 2000). Pathway inhibition, by overexpressing inhibitory Wnt ligands, gsk-3 or dominant-negative forms of the LRP6 and TCF proteins perturbed NC induction (Deardorff et al., 2001a; Garcia-Castro et al., 2002; LaBonne and Bronner-Fraser, 1998; Lewis et al., 2004; Saint-Jeannet et al., 1997; Tamai et al., 2000). In zebrafish, canonical Wnt pathway perturbations using dominant-negative TCF protein or Wnt8 ligand knockdown prevented NC induction (Lewis et al., 2004). Zebrafish mutants lacking involuting mesoderm, which includes paraxial mesoderm, make NC; however, Wnt ligands expressed in the early noninvoluted and remnant ventral mesoderm tissues may suffice to induce NC (Ragland and Raible, 2004). In a similar manner, ectopic Wnt3a expression induced NC in Xenopus embryos, in which paraxial mesoderm was ablated (Elkouby et al., 2010).

FIGURE 8.1

Wnt/β-catenin activity is required for neural crest induction. (A) Slug expression is inhibited in Xenopus embryos knocked down for the Frz3 Wnt receptor and is rescued by overexpression of mouse Frz3 protein (right injected side, red β-gal (more...)

FIGURE 8.2

FIGURE 8.2: Wnt/β-catenin activity induces neural crest formation. (A) Slug expression is ectopically expanded to the anterior and lateral on the injected side (right, blue β-gal staining) of Wnt3a-overexpressing Xenopus embryos (Saint-Jeannet (more...)

Similar to the observations in lower vertebrates,Wnt1/ Wnt3a knockout mice had highly reduced levels of NC (Ikeya et al., 1997). However, these embryos also suffer from severe mesodermal perturbations, so the effect could be indirectly related to the loss of inducing mesoderm tissues. In mice, Wnt activity can expand subsets of NC cells by promoting cell proliferation and survival; nevertheless, strong β-catenin activity promotes cell fate specification in NC precursors and does not appear to be required for cell population survival and expansion (Lee et al., 2004a). In mice, β-catenin activity was ablated in the MHB, Wnt1-positive region, by a Wnt1-Cre-mediated deletion. β-Catenin elimination in the Wnt1 expression domain strongly inhibited NC formation, but this phenotype was not seen in the Wnt1 knockout mice (Brault et al., 2001; McMahon et al., 1992). This result suggests that more global canonical Wnt signaling is required for NC induction in the mouse neural plate. The Wnt1 ligand itself is not sufficient to induce NC, and other regionally expressed Wnt ligands must participate in NC induction. In chick embryos, ectopic expression of Wnt or dominant-negative Wnt ligands leads to either expansion or suppression of NC fates in the forming neural tube (Garcia-Castro et al., 2002). In chick embryos, Wnt6 is expressed in the neural plate borders, and it was originally suggested to be the candidate ligand (Garcia-Castro et al., 2002); further studies showed that Wnt6 likely promotes NC induction through noncanonical Wnt signaling (Schmidt et al., 2007). There is a thus a great deal of experimental evidence supporting a role for Wnt in NC induction, but for the most part, the specific details as to the exact ligand and inducing tissue are lacking in most vertebrate species.

It has been difficult to pinpoint the specific timing and position of the critical Wnt ligand source/s that induce NC. There are a number of open questions. Is Wnt expression in the paraxial mesoderm tissue crucial for NC induction? Is Wnt expression in the neural plate border required for NC induction? Is Wnt expression in the neural plate obligatory for NC induction? What are the transcription factors acting upstream to Wnt that activate its expression? What are the transcription factors acting downstream to Wnt signaling that specify the NC in the neural folds region?

In Xenopus, Wnt3a expressed in the paraxial fated mesoderm is required for inducing neural crest, in addition to other posterior neural cell fates (Elkouby et al., 2010). Wnt3a secreted from the paraxial mesoderm induces expression of the TALE-class homeobox protein Meis3 in the adjacent neural ectoderm. Meis3 protein suffices to specify NC, hindbrain, and primary neuron cell fates when embryos are depleted of canonical Wnt activity by either Dkk-1 protein expression or the Wnt3a morpholino. Meis3 expression precedes Wnt3a expression in the neural plate, but Wnt3a is expressed later in the lateral neural plate region bordering the folds, where it could have a role in NC fate maintenance.

Wnt8 is another early mesoderm expressed candidate that could participate in NC induction. While inhibition of Wnt8 protein activity disrupts NC induction in Xenopus and zebrafish embryos, it appears that this may be the result of mesoderm perturbation and not a direct induction effect mediated by the ligand (Bang et al., 1999; Erter et al., 2001; Hoppler et al., 1996; Lekven et al., 2001; Lewis et al., 2004). No conclusive evidence has shown the uncoupling of NC induction from perturbation of paraxial mesoderm formation in Wnt8 knockdown embryos. In Xenopus, Brachyury gene expression was normal in Wnt8 morphants, but neither Wnt3a gene expression nor paraxial mesoderm formation was determined (Hong et al., 2008). In Xenopus, Wnt3a gene expression may actually lie downstream to Wnt8 (see Chapter 12, “The Role of Mesoderm and Specific Wnt Ligands in Neural Patterning”) because Wnt3a morphant embryos express Wnt8 but do not make NC (Elkouby et al., 2010). However, the independent reduction of either mesodermally expressed Wnt3a or Wnt8 protein could suffice to inhibit NC induction (Elkouby et al., 2010; Hong et al., 2008). The Wnt7a ligand is expressed ubiquitously in Xenopus gastrula and neurula stage ectoderm, but knockdown studies have not addressed its endogenous temporal and regional role in NC specification (Chang and Hemmati-Brivanlou, 1998).

In Xenopus embryos, the frizzled-3 (Fz-3) receptor is expressed in the early neural plate and NC regions (Shi et al., 1998). In Fz-3 morphant embryos, NC development is strongly reduced (Deardorff et al., 2001b). This study did not examine the role for Fz-3 in regulating posterior neural cell specification of the spinal chord, hindbrain, or MHB. Thus, it is still an open question if Fz-3protein acts as a Wnt receptor specifically inducing NC or a general Wnt receptor for inducing multiple posterior neural cell fates.

The disheveled (Dvl) protein acts as a mediator of both canonical and noncanonical Wnt pathways. In vertebrate, there are three homologues, Dvl1–3. In Xenopus, Dvl1 and Dvl2 proteins are expressed in multiple tissues, including the early NC and paraxial mesoderm. In Dvl1 and Dvl2 morphant embryos, NC formation is seriously perturbed (Gray et al., 2009). These Dvl morphant embryos also had somite segmentation phenotypes, so it is possible that the Dvl knockdown effect NC is secondary, being mediated via the incorrect formation of the paraxial mesoderm. In mice, Dvl1 knockout did not disrupt NC formation, but Dvl2 and Dvl3 knockout mice lost NC derived cardiac cells (Etheridge et al., 2008; Hamblet et al., 2002). It was suggested that Dvl-protein functional redundancy could mask more robust phenotypes in these mice; it is also still an open question whether the NC phenotypes are a result of inhibiting canonical or noncanonical (or both) Wnt pathways in Dvl knockout mice (Etheridge et al., 2008).

Inhibition of endogenous anteriorly expressed Wnt pathway antagonist proteins leads to an expansion of the NC. Dkk-1 null mice as well as Xenopus embryos in which endogenous Dkk-1 protein is blocked by an antibody have similar phenotypes; NC cells are ectopically spread to the anterior neural folds region (Carmona-Fontaine et al., 2007). In a similar manner, blocking the TCF-3 Wnt antagonist protein in morphant zebrafish embryos causes the same phenotype (Carmona-Fontaine et al., 2007). These Wnt-antagonist proteins maintain anterior cell fate integrity in the embryo, and in their absence, residual Wnt signaling caudalizes the anterior neural folds to NC fates.

In addition to the induction of NC cells, canonical Wnt signaling is also required by NC for their later delamination and subsequent migration. In chick, Wnt1 (in contrast to Wnt3a) gene expression is activated in the dorsal neural tube in a BMP-dependent manner. This BMP activation of Wnt1 expression is required to drive a G1/S transition in premigratory NC cells that is required for their delamination and migration (Burstyn-Cohen et al., 2004).

A number of key transcription factors that mark early-induced premigratory NC, are the slug (Snail family), FoxD3 (Winged-helix), and Sox9 (SoxE HMG family) proteins. In addition to serving as markers for the earliest induced NC, these proteins are all functionally required for proper NC differentiation in vertebrates (Cheung et al., 2005; Kos et al., 2001; LaBonne and Bronner-Fraser, 2000; Lee et al., 2004b; Sasai et al., 2001). Wnt appears to act by either direct or indirect mechanisms to regulate their expression. A functional β-catenin/TCF-binding site is present in the Xenopus slug promoter region, which is active after initial NC induction and appears to regulate slug expression for NC maintenance (Vallin et al., 2001). However, in chick, β-catenin cannot activate slug transgene expression in the NC region (Sakai et al., 2005). In the mouse Sox9 gene, a β-catenin-dependent enhancer element regulates NC-specific expression (Bagheri-Fam et al., 2006).

Upstream to FoxD3, slug and Sox9 are additional transcription factors that also appear to lie downstream or act concomitantly with Wnt signaling to specify neural crest. These transcription factors are the Meis3 and gbx2 homeobox genes, as well as Pax3, Pax7, and Zic1 genes that are expressed in the NC and neural plate region. In Xenopus, Pax3 and Zic1 proteins act in concert with canonical Wnt signaling to induce NC (Monsoro-Burq et al., 2005; Sato et al., 2005). A two-step Wnt-activity model was suggested. An early Wnt signaling event is required for initiating Pax3 expression (Bang et al., 1999), whereas Zic1 expression is activated by BMP antagonism. An additional Wnt signaling event is required for Pax3 and Zic1 proteins to activate expression of early NC markers like slug and FoxD3 (Monsoro-Burq et al., 2005; Sato et al., 2005). In chick, it was suggested that Pax7 and Pax3 may have similar roles because Pax gene expression is reduced when Wnt activity is inhibited (Basch et al., 2006). In Xenopus, Pax7 protein was shown to be important for NC induction by specifying paraxial mesoderm and does not appear to have a direct role in the neuroectoderm (Maczkowiak et al., 2010).

The Xenopus, Gbx2 homeobox gene is expressed in late gastrula-stage neural ectoderm in a region that overlaps the presumptive NC (Li et al., 2009). Gbx2 is a direct target of Wnt/β-catenin signaling and its activity is required for proper NC induction (see Chapter 5, “Induction of the Hindbrain”). Gbx2 protein is required for proper activation of Pax3 expression and it also likely interacts with to Zic1 protein to specify NC. Gbx1/2 proteins are also involved in specifying the MHB and anterior hindbrain regions (Kikuta et al., 2003; Rhinn et al., 2009; Wassarman et al., 1997) (see Chapter 4, “Induction of the Midbrain–Hindbrain Border”).

The Meis3 Tale-class homeobox protein is also required for NC induction (Gutkovich et al., 2010). Like, Gbx2, it is also a direct target of β-catenin/ Wnt signaling (see Chapter 5, “Induction of the Hindbrain”) and it is also expressed in posterior neural plate regions in late gastrula-stage embryos (Elkouby et al., 2010). Meis3 knockdown caused a reduction in Gbx2 expression (Elkouby et al., 2010), but the reciprocal regulation of Meis3 gene expression by Gbx2 has not been examined. Supporting a role for downstream Meis3 protein function in NC induction, Meis3 gene expression is lost in Wnt3a, Zic1, or Pax3 morphant embryos. The cross-talk between these two Wnt/β-catenin direct-target homeobox proteins, Meis3 and gbx2, and their interactions with other transcription factors is a crucial nexus in regulating induction of NC and other posterior neural cell fates in the developing embryo.

We suggest a model in which two signals (1) Wnt3a/ Wnt8 and (2) FGF8 from the paraxial mesoderm activate transcription factor expression in the neural folds/plate region (Figure 8.3). Intermediate BMP levels in the neural folds region enables FGF8 activation of Msx1 expression. Msx1 protein participates in activating Pax3 expression in the neural folds. In parallel, the paraxial mesoderm source of the Wnt-signal directly activates expression of the Gbx2 and Meis3 homeobox protein genes. These two genes may mutually control one another’s expression in the neural plate. Gbx2 protein is also required to properly activate Pax3 gene expression. BMP antagonism in the neural plate is required to activate Zic1 gene expression in the neural plate. Zic1 and Pax3 proteins are required along with Wnt3a/8 signaling to activate subsequent Meis3 gene expression. As a result of Meis3 and Gbx2 expression, posterior neural cell fates are induced in the neural plate and NC is induced in the neural folds region.

FIGURE 8.3

The gene regulatory network and tissue interactions regulating neural crest induction by the Wnt/β-catenin pathway. NC induction is initiated by two signals (1) Wnt3a/Wnt8 and (2) FGF8 from the paraxial mesoderm. These signals activate transcription (more...)