sign in sign up
<<
Home , Cranial neural crest, Ectoderm, Gastrulation, Neural crest, Neural fold, Neural plate

REVIEWS

INDUCTION OF THE : A MULTIGENE PROCESS

Anne K. Knecht and Marianne Bronner-Fraser In the , the neural crest is an important population of cells that gives rise to diverse derivatives, including the peripheral and the craniofacial skeleton. Evolutionarily, the neural crest is of interest as an important innovation in . Experimentally, it represents an excellent system for studying fundamental developmental processes, such as induction. Classical embryologists have identified interactions between tissues that lead to neural crest formation. More recently, geneticists and molecular biologists have identified the genes that are involved in these interactions; this recent work has revealed that induction of the neural crest is a complex multistep process that involves many genes.

NEURAL TUBE A central question in developmental is how a sin- ancestral vertebrates, such as lampreys (BOX 1). The cells A cylindrical structure that runs gle — the egg — can give rise to many tissues. A prin- that form neural crest are induced at the border between through the midline of the cipal mechanism for generating such complexity is the , which forms the central nervous sys- embryo; it expands in the induction, whereby one set of cells influences another, tem, and the non-neural , which forms the to form the and in the trunk to form the . causing those cells to adopt a different fate. Although (FIG. 1). As the neural plate folds over itself to induction is usually represented by a single arrow in dia- form the , border regions (NEURAL FOLDS) from NEURAL FOLDS grams, recent advances in the molecular of devel- opposite sides of the ectoderm come together and later Tips of invaginating ectoderm opment have shown that induction involves several sig- fuse. In this way, neural crest progenitors come to lie in, that will close to form the dorsal nals. These include inducers, which activate a new genetic and/or immediately adjacent to, the dorsal neural tube1. portion of the neural tube. programme that leads to conversion to a different cell During or after neural tube closure (depending on the type; competence factors, which control the time at which species), neural crest cells leave the neural tube and cells can respond to these inducers; maintenance factors, migrate throughout the body, where they differentiate which maintain the induced developmental programme, into , , and many other perhaps through many intermediate stages; and cell- types of cell (BOX 2). This review focuses only on the survival or proliferation signals, which regulate growth. early formation of neural crest progenitors; neural crest Likewise, recent research has identified several genes that migration and differentiation are discussed elsewhere1. are involved in the induction of the neural crest, includ- Although neural crest progenitors lie in the neural ing, in particular, secreted growth factors of the plate border, this region does not give rise exclusively morphogenetic protein (BMP), fibroblast to neural crest. If a single cell in the neural folds is (FGF) and Wingless/INT-related (WNT) families. labelled before neural tube closure, then the labelled However, the precise functions of these genes remain derivatives can later be found in the neural crest, 2 Division of Biology 139-74, unclear. This review describes recent findings about the neural tube and epidermis . Even after neural tube clo- California Institute of molecular basis of neural crest induction, with the goal of sure, cells in the dorsal neural tube in the chick2 and in Technology, Pasadena, elucidating the steps and signals that are involved. the frog3 can generate both neural tube cells and neural California 91125, USA. crest cells. After emigration from the neural tube, Correspondence to M.B.-F. Early development of neural crest e-mail: neural crest cells do not normally contribute to the [email protected] The neural crest is a transient, migratory population of neural tube, but if these cells are injected into the ven- doi:10.1038/nrg819 cells found in all , including the tral neural tube, they can adopt the fates of their

NATURE REVIEWS | GENETICS VOLUME 3 | JUNE 2002 | 453 REVIEWS

Box 1 | Evolution of the neural crest regions of the neural folds that have the potential to form neural crest5,6. For simplicity, this review refers Vertebrate evolution has been intimately linked to the evolution of two embryonic cell to the induction of Slug as neural crest induction, but populations — the neural crest and the cranial ectodermal PLACODES. These cell with the understanding that this might actually repre- populations together give rise to many of the defining characteristics of vertebrates, sent an early step in an ongoing process. including a well-defined head with teeth and paired sensory organs. Both neural crest and placodes are migratory populations that form at the border between neural plate Models of neural crest induction and epidermis. They generate some of the same cell types, such as sensory neurons. Neural induction. An obvious first step in the forma- Although they share many characteristics, there are also some differences between them; tion of the neural-plate border is the formation of the for example, placodes are confined to the head, whereas neural crest cells arise from most of the anteroposterior axis. Unlike placodes, neural crest cells form melanocytes neural plate. During , ectoderm is induced to form neural tissue by signals from a specialized and AUTONOMIC NEURONS, and produce mineralized matrices like bone. These differences might reflect differences in their evolutionary origin74. region of known as ‘the organizer’. Neural crest evolved soon after the split of cephalochordates (amphioxus) and A model of the molecular basis of neural induction, vertebrates. Amphioxus, the closest living relative of the vertebrates, shares which is referred to as the neural default model some characteristics with vertebrates, such as segmented muscles, but lacks definitive (reviewed in REF. 7), has been developed during the neural crest75,76. By contrast, structures that are derived from the neural crest, such as past decade on the basis of experiments carried out in PHARYNGEAL DENTICLES, are present in the earliest vertebrate fossils, and the most basal the laevis. According to this model, the extant vertebrates, and lampreys, have well-developed structures that are default fate of ectoderm is to form neural tissue; how- derived from the neural crest. Lampreys are jawless that represent the most ever, before neural induction, all of the ectoderm pro- primitive extant vertebrates for which it is feasible to obtain embryos, and these clearly duces the growth factor BMP4, which suppresses the have neural crest cells77. formation of neural tissue and promotes the forma- Because of the genome-wide duplications associated with vertebrate evolution, one tion of epidermis. To overcome this suppression and, possible way to explain the evolution of neural crest is that, in the vertebrate lineage, new therefore, to generate the neural plate, the organizer genes were formed by duplication and this facilitated diversification of gene function, secretes BMP antagonists, such as , which led to the origin of a new . However, many homologues of vertebrate neural and , which bind to BMPs and prevent crest markers have been cloned in both amphioxus and lampreys, which indicates that the signalling through their receptors7. same complement of genes exists in both species, even though the former lacks a definitive Although this model is well supported by experi- neural crest77–79.Although the functions of these genes have not been studied sufficiently ments in Xenopus, it remains unclear how well it applies for us to be certain that they act the same way in different species, in general, it seems more to other vertebrates, and whether other factors might be likely that the evolution of the neural crest was accompanied by the use of old genes in new involved. Mouse knockouts of BMPs and BMP antago- ways rather than by the invention of new genes for a new cell population. Further 8 molecular characterization and embryological analysis are necessary to gain a better nists were largely uninformative , probably owing to understanding of the evolutionary origin of neural crest cells and placode cells. functional redundancy between genes. In the chick, Streit et al.9 questioned whether BMP antagonism is necessary for neural induction for the following reasons. First, they showed that BMP4 and BMP7 expression in neighbours and form and motor neurons4. the chick disappears from the ectoderm before gastrula- Such results show that, although induction at the tion (HAMBURGER–HAMILTON STAGE 2)9.However,Faure et PLACODES neural plate border leads to the formation of neural al.10 found that BMP signalling, which is manifested by Thickenings in the vertebrate cranial ectoderm that invaginate crest cells, there are many steps in between, in which activation of the downstream signalling molecule and form parts of cranial cells show remarkable flexibility in their cell-fate deter- SMAD1, is reactivated throughout the ectoderm at stage sensory ganglia and paired mination. 3 of chick development and is subsequently downregu- sensory organs. Because of this, it is difficult to apply traditional lated in the prospective neural plate at stage 4. So, as in developmental terms, such as ‘commitment’ or ‘specifi- Xenopus, reduction of BMP signalling correlates with AUTONOMIC NEURONS cells of the peripheral cation’,to neural crest. A tissue is operationally defined neural induction. nervous system that innervate as specified for a certain fate if it continues to adopt Second, Streit et al. also found that the addition of the viscera, smooth muscles and that fate when explanted away from other external sig- BMPs to prospective neural plate in stage-3 chick exocrine glands. nals. By this definition, the neural folds are specified to embryos does not block neural induction, and applica- form neural crest, as explants from this region produce tion of chordin to ectoderm cannot induce neural tis- PHARYNGEAL DENTICLES 2 Dense structures on the surface neural crest derivatives ; however, as described above, sue. However, these experiments might have been of pharyngeal arches of early not all cells in this region form neural crest, as the carried out too late in development, after the initial vertebrates that are thought to neural, epidermal and neural crest lineages are not yet neural specification. Although neural tissue is not be pressure sensitive. segregated. Furthermore, if early neural folds represent specified in Xenopus before gastrulation, Wilson et al.11

GASTRULATION a heterogeneous population of cells, including both showed that explants that are cultured from chick Morphogenetic movement that neural and epidermal cells, signalling might occur embryos before gastrulation are already specified to transforms a single-layered between cell types in explants, and therefore the express neural markers, and Streit et al.12 found the embryo into an embryo with prospective neural crest is not truly isolated. Given expression of a neural marker before gastrulation. At three germ layers. these issues, it is difficult to say when neural crest these early stages, the entire shows BMP sig- 10 HAMBURGER–HAMILTON induction really occurs. In recent experiments, nalling , which supports the idea that BMP inhibition STAGES researchers have focused on expression of the tran- is not necessary for initial neural specification. In Stages that describe the age of scription factor Slug as one of the earliest indicators of culture, however, prospective neural explants down- chick embryos; stage 2 refers to neural crest induction, owing to its expression in regulate BMP4 expression11, as in Xenopus. the time before gastrulation.

454 | JUNE 2002 | VOLUME 3 www.nature.com/reviews/genetics REVIEWS

a which it seems to maintain16. This model is, however, np epi supported by BMP-pathway mutants in (swirl/bmp2b, snailhouse and somitabun) that have nf either reduced or expanded domains of neural crest progenitors, depending on the precise alteration of the b 17 epi BMP-signalling levels . Also, injection of BMP4 RNA into Xenopus embryos leads to a reduction in neural crest18, whereas injection of BMP antagonists causes np neural crest expansion6,18. Importantly, the expanded neural crest domain remains contiguous with its nor- mal domain and cannot be induced throughout the c ectoderm, which indicates that other signals limit the territory that forms neural crest. epi Although these in vivo results indicate an impor- nt tant role for BMP signalling in neural crest induction, the mesoderm is also affected in these mutants and, for this reason, in vitro experiments are necessary to determine whether BMP alone is sufficient and d whether its action is direct. In Xenopus ectodermal nc cells in vitro, intermediate BMP levels can induce epi some genes that are normally expressed in the ante- nt rior regions of neural plate border (XAG1 (REF. 19) and cpl1 (REF. 20)). Furthermore, in vivo, some genes that Figure 1 | Regions that form neural crest during mark the neural plate border in Xenopus, such as in a hypothetical vertebrate embryo. a | Open 21 neural plate (np) stage, after gastrulation. b | Closing neural folds Snail and Pax3 (REF. 22), are expressed at the develop- (nf). c | Closed neural tube (nt). d | Migrating neural crest (nc). ing border as early as mid-gastrula stages, which indi- Non-neural epidermis (epi) is shown in grey; np and nt in cates that this border forms during neural induction. orange; and np border, nf and migrating nc are shown in yellow. However, in Xenopus, Slug is not expressed until later6,23 and, in explants of Xenopus ectoderm, inter- mediate levels of BMP signalling alone are insufficient Interestingly, this downregulation occurs in the to induce robust expression of Slug 18. Marchant et al.24 absence of detectable BMP antagonists; instead, these observed the induction of Slug in explants by interme- explants express FGF3, and FGF signalling is required diate BMP signalling, but Villanueva et al.25 suggested for BMP downregulation11. Several studies in the chick that this result was due to overly large explants, which have shown that FGF can act as a neural inducer12,13 and probably incorporated extra signals. In conclusion, that FGF signalling is required for neural induction12. although some aspects of the neural plate border Furthermore, members of the WNT family have also might be specified during neural induction, and inter- been shown to have a role in avian neural induction. mediate BMP signalling is probably an important first WNT3A and WNT8C are expressed in prospective epi- step towards forming the neural crest, the induction , and WNT3A can block neural induction in of cells with the potential to form neural crest appar- explants by blocking their ability to downregulate BMPs ently requires additional signals. in response to FGFs14. In summary, although neural induction is likely to be more complex than the neural Two-signal model of neural crest induction. Although default model — involving other signals, such as FGFs intermediate levels of BMP signalling alone cannot and WNTs — the absence of BMP activity in prospec- induce neural crest in Xenopus ectodermal explants, sev- tive neural tissue remains a key feature of the developing eral groups have found that the addition of a second sig- neural plate that is conserved in both the frog and the nal in the same assay could induce Slug. First, bFGF chick. (FGF2)6, and later eFGF18, were shown to induce Slug in A variation on the neural default model is that combination with noggin or chordin. The following neural induction divides the ectoderm into not two but members of the Wnt family were also found to induce three fates: epidermis, neural plate and neural plate Slug expression when combined with a BMP antagonist: border (reviewed in REF. 15). If diffusion of BMP antag- Wnt1 and Wnt3a26, Wnt7b27 and Wnt8 (REF.18).Many onists through the ectoderm creates a gradient of BMP of these genes are endogenously expressed at the appro- activity, in which high activity specifies epidermis and priate stages in candidate tissues that induce neural crest low activity specifies neural plate, then intermediate (see the next section for further discussion): eFGF 28 and levels of BMP activity could specify the neural plate Wnt8 (REF.29) are expressed in , and border. Because the role of BMPs in neural induction Wnt7b is present throughout the ectoderm27. FGFs EPIBLAST remains controversial in the chick and in the mouse, it might act indirectly through Wnts; eFGF can induce A term for the embryonic layer is unclear how this model relates to these vertebrates, Wnt8 expression and its ability to induce Slug in in chicks, mice and from which the embryo proper especially given that BMP4 is most highly expressed in Xenopus ectodermal explants can be strongly inhibited arises during gastrulation. early chick embryos at the neural plate border, by an inhibitor of Wnt signalling18. Overexpression of

NATURE REVIEWS | GENETICS VOLUME 3 | JUNE 2002 | 455 REVIEWS

Box 2 | Derivatives of neural-crest cells

Neural crest cells migrate throughout the body and differentiate into many different cell types (see REF.1 for more details). Although neural crest cells are pluripotent, differences exist between cells that are generated from different anteroposterior levels: neural crest cells in the trunk form melanocytes and several and cell types, whereas neural crest cells in the cranial (the embryonic head region) also have the potential to form mesenchymal derivatives, such as cartilage, bone and . Cranial Trunk

Neurons and glia Dorsal Pigment cells of cranial ganglia

Neural tube

Cartilage and bone Sensory neurons and glia

Ventral

Connective tissue Sympatho-adrenal cells

the Wnt-signalling components frizzled 3 (REF.30) and Regardless of whether the second signal (Wnt, FGF β-catenin18 can also induce Slug expression in this assay, and/or retinoic acid) posteriorizes anterior neural plate and the effects of β-catenin are cell autonomous, which border or directly induces neural crest in cooperation indicates that Wnt signalling might be a direct inducer. with intermediate BMP signalling, it is clear that these Recently, retinoic acid was also shown to induce two signals are sufficient to mediate neural crest induc- Slug expression in combination with a BMP inhibitor tion. However, are these candidates necessary for neural in Xenopus ectodermal explants25. Retinoic acid31, crest induction in vivo? For some factors, this has been FGF2 (REF. 32) and Wnt3a33 have all been shown to tested genetically. Mouse embryos that lack both Wnt1 function as posteriorizing signals in the anteroposte- and Wnt3a had a reduced amount of neural crest, which rior patterning of the neural tube. Because of this, it indicates that these genes are not required for neural has been proposed that these signals function in crest induction but might be important for the prolifer- explant assays by posteriorizing the tissue that is ation of neural crest precursors (consistent with their induced by intermediate BMP signalling and that relatively late expression in the dorsal neural tube)34.For resembles anterior neural plate border15,25. This idea is other genes, loss-of-function experiments have been supported by evidence that the endogenous anterior carried out in non-genetic organisms (that is, in neural plate border, which normally does not express Xenopus and in the chick) by overexpressing antagonists Slug or form neural crest, can be induced to express or dominant-negative constructs to block endogenous Slug by FGFs or retinoic acid25. However, this model signalling. For all three candidate second signalling cannot account for the ability of Wnt signalling to pathways, such experiments have resulted in the inhibi- induce neural crest in ectodermal explants that overex- tion of Slug expression, which indicates that FGF,Wnt press Slug, as Slug does not induce anterior neural bor- and retinoic acid signalling are all required to some der18. Moreover, it is not clear that the source of the sig- extent for neural crest formation18,25,26,30,35. A caveat of nal that induces neural crest in vivo lies in the posterior. these experiments is that they generally inhibit sig- The epidermis is also a source of signals that induce nalling by several members of the targeted gene family. MORPHOLINO ANTISENSE neural crest (see the next section for further discus- A more-specific approach is to use MORPHOLINO ANTISENSE OLIGONUCLEOTIDES sion), and the activities of epidermal signals might be OLIGONUCLEOTIDES to block translation of a particular Modified antisense separable from posteriorizing signals. Epidermis can gene. When frizzled 3 (REF.30) and its proposed adaptor oligonucleotides that are induce explants of the anterior neural plate to express protein Kermit36 were depleted in this way in Xenopus designed to block translation by pairing with the translation start Slug, but cannot induce Pax3, another neural crest embryos, Slug expression was reduced, which verifies a site in the 5′ untranslated region. marker that is regulated by posteriorizing signals22. crucial role for Wnt signalling in neural crest induction.

456 | JUNE 2002 | VOLUME 3 www.nature.com/reviews/genetics REVIEWS

Sources of inducing signals before neural induction, the experiments described in There is a continuing debate about which tissue in the this section involved epidermis and neural plate that embryo provides the signals that induce the neural was explanted after neural induction was complete. crest. After gastrulation, the region that forms the Therefore, if intermediate BMP signalling in the two- neural crest is in contact with three tissues: the neural signal model is interpreted as an extension of neural plate, the epidermis and the underlying mesoderm induction, then that signalling will be complete before (PARAXIAL MESODERM, which forms the ); all three these later experiments have begun. It might be that the tissues have been proposed to participate in neural neuralized ectoderm in chick neural plate explants acts crest induction. like Xenopus explants that have been neuralized by BMP antagonism, so that this first signal is bypassed in the Paraxial mesoderm. One of the earliest neural crest chick and induction requires only the second signal. experiments in showed that paraxial meso- Alternatively, it could be that neural crest induction in derm can induce ectoderm to form neural crest37.More the chick is fundamentally different, or that the later recently, this experiment has been verified in vitro; experiments in the chick exploit a later phase in neural explants of Xenopus prospective paraxial mesoderm and crest induction (FIG. 2). ectoderm were combined and later found to express Recently, researchers have sought to identify the epi- Slug 18,24,38. In the chick, paraxial mesoderm can also dermal inducer. Liem et al.44 found that both BMP4 induce explants of neural plate to form melanocytes, a and BMP7, which are expressed in the epidermis before neural crest derivative2. neural tube closure, can mimic this epidermal signal, It is not surprising that paraxial mesoderm has such inducing chick ventral neural plate explants to express activity, as this tissue expresses candidate inducing signals, Slug and produce migratory neural crest cells. such as eFGF, WNT8 and BMP4. The crucial question is Furthermore, the inducing ability of epidermis in these whether paraxial mesoderm is necessary for neural crest cultures could largely be blocked by adding the BMP induction. Both Bonstein et al.38 and Marchant et al.24 antagonists noggin and follistatin, which indicates that tested this by excising the prospective paraxial mesoderm BMPs might be necessary components of the epider- from Xenopus embryos at the onset of gastrulation; both mal signal45. However, the timing of this BMP require- found that expression of Slug was greatly reduced, which ment might be later than expected for the epidermal indicates a possible requirement for paraxial mesoderm signal. Selleck et al.46 showed that noggin cannot block in vivo. However, with such a large-scale, early dissection, the formation of neural crest during the open neural it remains possible that neural crest reduction is due to plate phase in chick embryos, when BMPs are indirect effects of the treatment, such as changes in the expressed weakly in the epidermis and the epithelial lateral ectoderm that this tissue would normally underlie. signal has been shown to act2. Instead, noggin was Also, the requirement for paraxial mesoderm remains to found to block neural crest induction at a later stage, in be shown in other vertebrates, in which such dissections the closing neural tube, when the BMPs are expressed are nearly impossible, except perhaps through genetic in the dorsal neural folds. So, Selleck et al. proposed manipulations. that neural crest induction involves at least two phases: an early phase, in which neural crest is induced by an –neural plate interactions. Another source of epidermal signal but is insensitive to BMP signalling; inducing signals is the interaction between neural plate and a later phase, in which neural crest progenitors are and epithelium. This possibility was first indicated by sensitive to BMP4, which is by then present in the dor- experiments in which explants of epidermis or neural sal neural folds. According to this model, the role of plate were grafted ectopically in embryos; BMP4 is to maintain neural crest progenitors that have neural crest cells were observed wherever new bound- been induced by an earlier signal. aries were created between neural and non-neural tis- Why, then, were BMPs able to mimic the earlier sues39,40. This result was also shown recently in similar signal in explant experiments? Isolated explants might grafting experiments in zebrafish embryos41. Moury and be sufficiently labile to allow the later BMP signal to Jacobson42 further showed in amphibians that the bypass the early BMP-independent phase or, alterna- induced neural crest was formed from both neural and tively, the culture medium that was used might have epidermal tissues, which indicates the possibility of reci- contained other factors. Liem et al.44 cultured their procal signalling from both tissues. In these experi- explants in a medium that contained additives (N3 ments, paraxial mesoderm remained a possible source supplement that contains insulin and other hor- of inducing signals; however, more recent in vitro stud- mones). When BMP4 protein was used in a medium ies have verified that Slug expression can be induced by without such factors, García-Castro et al.47 found that PARAXIAL MESODERM combining only neural plate and epidermis, in the this signal was unable to induce neural crest. By con- Mesoderm that is adjacent to the 2 43 neural tube and that is destined chick and in Xenopus . trast, medium that had been conditioned with to form somites. It is important to note that the zebrafish experiments Wingless protein, a Drosophila homologue of Wnt are done at later developmental stages than the Xenopus that is active in vertebrates, can induce neural crest in SOMITES experiments described above, which makes it difficult to the absence of additives to the medium. Moreover, Mesodermal balls of cells interpret these results in terms of the Xenopus two- injection of broad-spectrum Wnt inhibitors was adjacent to the neural tube that will differentiate into the muscle, signal model. Although Xenopus experiments have gen- shown to inhibit neural crest formation in vivo, vertebrae and dermis. erally used relatively naive ectoderm that was explanted and nuclear localization of β-catenin, which is an

NATURE REVIEWS | GENETICS VOLUME 3 | JUNE 2002 | 457 REVIEWS

Xenopus Chick

ect np ect ect np ect Open neural plate

Closing neural folds

Closed neural tube BMP BMP Wnt1/3a/7b Wnt1/3a

Wnt BMP

Figure 2 | Dynamic expression patterns of Wnts and BMPs in the ectoderm of frog and chick. In the frog, bone morphogenetic protein (BMP) is expressed strongly in the ectoderm (ect) at all stages of neurulation. Wnt7b (wingless-related 7b) also is expressed in the ectoderm. After neural tube closure, BMP and other Wnts (for example, Wnt1 and Wnt3a) are expressed in the dorsal neural tube. In the chick, Wnt6 is expressed in the ectoderm at all stages during neurulation. By contrast, BMP is expressed in the ectoderm only at low levels (light green) and transiently when the neural plate (np) is open. As the neural folds elevate, BMP is downregulated in the ectoderm but upregulated in the neural folds. After neural tube closure, BMP is expressed in the dorsal neural tube, as are Wnt1 and Wnt3a.

indicator of active Wnt signalling, was observed in tory neural crest or neural tube neurons. In zebrafish neural folds at the open neural-plate stage47.A mem- embryos, pre-migratory neural crest and ROHON–BEARD ber of the Wnt family therefore represents a strong SENSORY NEURONS are thought to form from a common candidate for the early epidermal signal. Wnt6, in par- progenitor pool. Zebrafish Delta mutants (deltaA mis- ticular, is expressed in the epidermis at the appropri- sense ; dIAd×2) have increased numbers of ate stages47. However, because Wnts are also expressed Rohon–Beard cells and decreased neural crest cells, in prospective paraxial mesoderm, it remains possible which indicates that Delta is necessary to promote the that this mesoderm is the usual source of Wnts and decision to form neural crest51. Interestingly, this that epidermis merely mimics this activity. Future requirement is restricted to neural crest in the trunk, experiments that involve conditional inactivation of with cranial regions being unaffected. genes in specific tissues will be necessary to resolve Experiments by Endo et al.52 in the chick have indi- this issue, although it is also possible that signalling cated an earlier role for Notch signalling in neural crest from many tissues is required. induction in the cranium, as perturbations of Notch sig- nalling in early embryos strongly reduce Slug expression. Other signals in neural crest induction The authors propose that this effect is mediated indirectly Research in many systems has focused on the role of by BMP4, which they found is genetically downstream of BMPs and Wnts in neural crest induction. These signals Delta. Both overactivation and inhibition of Notch sig- probably act at many different times in neural crest nalling reduce BMP4 expression in the epidermis, indi- development. For example, Wnt1 and Wnt3a are candi- cating that intermediate levels of Notch signalling might dates to regulate proliferation of neural crest progeni- be required to maintain epidermal BMP4, which might tors30, and BMP4 has been shown to regulate neural be necessary to induce neural crest. However, in these crest emigration from the neural tube48. However, other experiments, BMP4 is also downregulated in the dorsal signals, including Notch/Delta, Noelin and possibly neural tube, so a loss of Slug expression might reflect the Narrowminded, are also likely to be involved in neural later impairment of neural crest maintenance rather than crest formation. an early effect on neural crest induction.

Notch and Delta. Signalling by the cell-surface protein Noelin. The neural tube is competent to produce neural Delta, and its Notch, has been implicated in crest for a limited period of time. Factors that control neural crest development49. Notch signalling has been this competence are largely unknown, with the possible ROHON–BEARD SENSORY studied mainly as a mechanism that allows cells with exception of the large secreted factor Noelin, which can NEURONS equal developmental potential to adopt two different prolong the period of neural crest production53. Early-differentiating neurons in the dorsal neural tube of fish and fates (reviewed in REF.50). Such a process occurs in the However, this molecule is largely uncharacterized and amphibians. dorsal neural tube, where cells can form either migra- the mechanism of its action is as yet unknown.

458 | JUNE 2002 | VOLUME 3 www.nature.com/reviews/genetics REVIEWS

Narrowminded. A recent zebrafish screen identified the specification of neural crest; neural crest is expanded mutant narrowminded, which has reduced numbers of by overexpression of Slug 18 and blocked by injection neural crest cells at early stages of development54. of dominant-negative Slug constructs55.However, Rohon–Beard cells are also absent, which further sup- these studies also showed that Slug overexpression, ports the contention that these cells share a common like intermediate BMP signalling, cannot generate progenitor with neural crest cells. The narrowminded neural crest unless it is combined with FGFs or Wnts. mutation has been shown to function cell autonomously, So, although Slug can bypass the requirement for which indicates that it might have a role in receiving or intermediate BMP signalling, neural crest formation interpreting the signals that induce neural crest. requires other genes that are induced by FGFs and/ Molecular characterization of the mutation will provide or Wnts. further clues to its actual function. Although early inhibition of Slug blocks neural crest formation, its later inhibition blocks neural crest Downstream transcription factors migration55. This result agrees with those of experi- Inductive signals bring about the expression of tran- ments in the chick, in which application of antisense scription factors that mediate further neural crest oligonucleotides against Slug allows formation of development. Overexpression or inhibition experi- neural crest precursors but blocks their emigration ments have indicated roles in neural crest formation from the neural tube5. In these experiments, antisense for the following classes of transcription factors: oligonucleotides were unable to block neural crest for- Slug5,18,55,Pax56,57,Fox58–60,Zic61–65,Sox66,67 and Meis68. mation. However, in recent experiments, Slug overex- However, for most of these, the precise function pression in the chick neural tube caused increased remains unclear. Although some genes are likely to be production of neural crest71, as seen in Xenopus. direct downstream targets of neural crest induction, Interestingly, this effect was restricted to the cranial others are thought to be involved in later aspects of region. Chick Slug might have more restricted activi- neural crest development. For example, Pax3 might ties because of the later initiation of Slug expression in maintain the neural crest progenitors in an undifferen- closing neural folds and not at the borders of the open tiated state in the dorsal neural tube69, and Foxd3 neural plate, as in Xenopus. Furthermore, in the might cause these uncommitted progenitors to mouse, Slug is expressed in migratory neural crest but become neural crest instead of interneurons58. not in pre-migratory cells72,73; instead, the related gene The functions of Slug are more thoroughly under- Snail is expressed in the mouse in a pattern that is stood. Because the of the Xenopus Slug gene similar to that of Slug expression in the chick. This contains a functional binding site for Lef and reversal and redundancy might explain the absence of β-catenin, which are downstream effectors of Wnt a neural crest in mice that lack Slug 72.In signalling, Slug might, therefore, be a direct target of summary, Slug, and probably other factors as well, Wnt induction70. Accordingly, early perturbation might have different roles in neural crest development of Slug expression in Xenopus affects the early in different species.

Table 1 | Steps in neural-crest induction Stage Process Genes/proteins Induction of neural plate and neural-plate border BMPs and BMP antagonists (Solid line indicates the strong inhibition of BMPs by their antagonists, which come from the organizer region and which induce the neural plate. Dotted line represents a weaker antagonism that induces the neural plate border.)

Induction of neural crest potential as assayed by Wnts, FGFs and retinoic acid from Slug expression epidermis or lateral mesoderm Maintenance, proliferation and non-differentiation BMPs, Wnt1, Wnt3a of neural crest precursors and Pax3 in the dorsal neural tube

Subdivision of neural crest/neural tube lineage Notch/Delta, Foxd3

Emigration of neural crest BMPs, Slug

The proposed stage at which each process functions is indicated by the diagram in the left column, whereas genes/proteins that are implicated in each process are listed in the right column. Although processes are separated for clarity, they might actually occur simultaneously in the embryo. BMP, bone morphogenetic protein; FGF, .

NATURE REVIEWS | GENETICS VOLUME 3 | JUNE 2002 | 459 REVIEWS

Multistep model of neural crest induction involves Notch signalling and Foxd3. Finally, BMPs Research in the past decade has made great strides and Slug are again important as neural crest cells emi- towards understanding the molecular basis of neural grate from the neural tube. crest induction. Although much remains unclear, a Although this model is useful for thinking about the picture is nevertheless emerging of a continuing steps that are involved in neural crest induction, it is not process that involves several signals that act at different intended to be complete or inarguably accurate. Too stages. A model that lists these proposed stages and sig- much remains to be studied and too many controversies nals is described in TABLE 1. Briefly, the process begins remain unresolved. One gaping hole is the absence of with induction of the neural plate and its border, solid data about when each step of the process occurs mediated, at least in part, by inhibition of BMP sig- (many might occur simultaneously) and when each gene nalling. This border region acquires the potential to acts. Another problem is how to reconcile conflicting form neural crest as a result of additional signalling data from different species; although species differences (by Wnts, FGFs or retinoic acid) from the epidermis, do exist, it is likely that much of the process is conserved, paraxial mesoderm or both. Neural crest potential but it is difficult to compare results when experiments must then be maintained, perhaps through the activity use different tissues at different stages. Finally, more of BMPs in the dorsal neural tube. Multipotent precur- genetic experiments will be necessary to clarify the highly sors proliferate in response to Wnt1 and Wnt3a, but complex interactions between these candidate genes. are prevented from differentiating prematurely by Hopefully, the next decade will prove to be just as fruitful Pax3. Eventually, these precursors adopt the fate of as the past one in making progress towards understand- either neural crest or dorsal neurons in a process that ing this complicated genetic pathway.

1. Le Douarin, N. M. & Kalcheim, C. The Neural Crest 2nd 18. LaBonne, C. & Bronner-Fraser, M. Neural crest induction in 33. McGrew, L. L., Lai, C. J. & Moon, R. T. Specification of the edn (Cambridge Univ. Press, Cambridge, UK, 1999). Xenopus: evidence for a two-signal model. Development antero-posterior neural axis through synergistic interaction 2. Selleck, M. A. J. & Bronner-Fraser, M. Origins of the avian 125, 2403–2414 (1998). of the Wnt signaling cascade with noggin and follistatin. neural crest: the role of neural plate–epidermal 19. Wilson, P. A. & Hemmati-Brivanlou, A. Induction of Dev. Biol. 172, 337–342 (1995). interactions. Development 121, 525–538 (1995). epidermis and inhibition of neural fate by Bmp-4. Nature 34. Ikeya, M., Lee, S. M., Johnson, J. E., McMahon, A. P. & 3. Collazo, A., Bronner-Fraser, M. & Fraser, S. E. Vital dye 376, 331–333 (1995). Takada, S. Wnt signaling required for expansion of neural labelling of Xenopus laevis reveals 20. Knecht, A. K. & Harland, R. M. Mechanisms of crest and CNS progenitors. Nature 389, 966–970 (1997). multipotency and novel pathways of migration. dorsal–ventral patterning in noggin-induced neural tissue. 35. Mayor, R., Guerrero, N. & Martinez, C. Role of FGF and Development 118, 363–376 (1993). Development 124, 2477–2488 (1997). noggin in neural crest induction. Dev. Biol. 189, 1–12 4. Ruffins, S., Artinger, K. B. & Bronner-Fraser, M. Early 21. Essex, L. J., Mayor, R. & Sargent, M. A. Expression of (1997). migrating neural crest cells can form ventral neural tube Xenopus snail in mesoderm and prospective 36. Tan, C. et al. Kermit, a frizzled interacting protein, regulates derivatives when challenged by transplantation. Dev. Biol. ectoderm. Dev. Dyn. 198, 108–122 (1993). frizzled 3 signaling in neural crest development. 203, 295–304 (1998). 22. Bang, A. G., Papalopulu, N., Kintner, C. & Goulding, M. D. Development 128, 3665–3674 (2001). 5. Nieto, M. A., Sargent, M. G., Wilkinson, D. G. & Cooke, Expression of Pax-3 is initiated in the early neural plate by 37. Raven, C. R. & Kloos, J. Induction by medial and lateral J. Control of cell behavior during vertebrate development posteriorizing signals produced by the organizer and by pieces of the roof, with special reference to the by Slug, a gene. Science 264, 835–839 posterior non-. Development 124, determination of neural crest. Acta Neerl. Morphol. 5, (1994). 2075–2085 (1997). 384–362 (1945). 6. Mayor, R., Morgan, R. & Sargent, M. G. Induction of the 23. Linker, C., Bronner-Fraser, M. & Mayor, R. Relationship 38. Bonstein, L., Elias, S. & Frank, D. Paraxial-fated mesoderm prospective neural crest in Xenopus. Development 121, between gene expression domains of Xsnail, Xslug, and is required for neural crest induction in Xenopus embryos. 767–777 (1995). Xtwist and cell movement in the prospective neural crest of Dev. Biol. 193, 156–168 (1998). 7. Wilson, P. A. & Hemmati-Brivanlou, A. Vertebrate neural Xenopus. Dev. Biol. 224, 215–225 (2000). 39. Rollhäuser-ter Horst, J. Artificial neural crest formation in induction: inducers, inhibitors, and a new synthesis. 24. Marchant, L., Linker, C., Ruiz, P., Guerrero, N. & Mayor, R. amphibia. Anat. Embryol. 157, 113–120 (1979). Neuron 18, 699–710 (1997). The inductive properties of mesoderm suggest that the 40. Moury, J. D. & Jacobson, A. G. Neural fold formation at 8. Hogan, B. L. Bone morphogenetic proteins in neural crest cells are specified by a BMP gradient. Dev. Biol. newly created boundaries between neural plate and development. Curr. Opin. Genet. Dev. 6, 432–438 (1996). 198, 319–329 (1998). epidermis in the axolotl. Dev. Biol. 133, 44–57 (1989). 9. Streit, A. et al. Chordin regulates 25. Villanueva, S., Glavic, A., Ruiz, P. & Mayor, R. 41. Woo, K. & Fraser, S. E. Specification of the fate in development and the stability of induced neural cells, but Posteriorization by FGF, Wnt, and retinoic acid is required the zebrafish. Dev. Biol. 197, 283–296 (1998). is not sufficient for neural induction in the chick embryo. for neural crest induction. Dev. Biol. 241, 289–301 (2002). 42. Moury, J. D. & Jacobson, A. G. The origins of neural crest Development 125, 507–519 (1998). 26. Saint-Jeannet, J.-P., He, X., Varmus, H. E. & Dawid, I. B. cells in the axolotl. Dev. Biol. 141, 243–253 (1990). 10. Faure, S., Santa Barbara, P., Roberts, D. J. & Whitman, Regulation of dorsal fate in the neuraxis by Wnt-1 and 43. Mancilla, A. & Mayor, R. Neural crest formation in Xenopus M. Endogenous patterns of BMP signaling during early Wnt-3a. Proc. Natl Acad. Sci. USA 94, 13713–13718 laevis: mechanisms of Xslug induction. Dev. Biol. 177, chick development. Dev. Biol. 244, 44–65 (2002). (1997). 580–589 (1996). 11. Wilson, S. I., Graziano, E., Harland, R., Jessell, T. M. & 27. Chang, C. & Hemmati-Brivanlou, A. Neural crest induction 44. Liem, K. F., Tremml, G., Roelink, H. & Jessell, T. M. Dorsal Edlund, T. An early requirement for FGF signaling in the by Xwnt7B in Xenopus. Dev. Biol. 194, 129–134 (1998). differentiation of neural plate cells induced by BMP- acquisition of neural cell fate in the chick embryo. Curr. 28. Isaacs, H. V., Tannahill, D. & Slack, J. M. W. Expression of a mediated signals from epidermal ectoderm. Cell 82, Biol. 10, 421–429 (2000). novel FGF in the Xenopus embryo. A new candidate 969–979 (1995). 12. Streit, A., Berliner, A. J., Papanayotou, C., Sirulnik, A. & inducing factor for mesoderm formation and 45. Liem, K. F., Tremml, G. & Jessell, T. M. A role for the roof Stern, C. D. Initiation of neural induction by FGF signalling anteroposterior specification. Development 114, 711–720 plate and its resident TGFβ-related proteins in neuronal before gastrulation. Nature 406, 74–78 (2000). (1992). patterning in the dorsal spinal cord. Cell 91, 127–138 13. Storey, K. G. et al. Early posterior neural tissue is induced 29. Christian, J. L., McMahon, J. A., McMahon, A. P. & Moon, (1997). by FGF in the chick embryo. Development 125, 473–484 R. T. XWnt-8, a Xenopus Wnt-1/int-1 related gene 46. Selleck, M. A. J., García-Castro, M. I., Artinger, K. B. & (1998). responsive to mesoderm-inducing growth factors, may play Bronner-Fraser, M. Effects of Shh and Noggin on neural 14. Wilson, S. et al. The status of Wnt signalling regulates a role in ventral mesodermal patterning during crest formation demonstrate that BMP is required in the neural and epidermal fates in the chick embryo. Nature embryogenesis. Development 111, 1045–1055 (1991). neural tube but not ectoderm. Development 125, 411, 325–330 (2001). 30. Deardorff, M. A., Tan, C., Saint-Jeannet, J.-P. & Klein, P. 4919–4930 (1998). 15. Mayor, R. & Aybar, M. J. Induction and development of A role for frizzled 3 in neural crest development. 47. García-Castro, M. I., Marcelle, C. & Bronner-Fraser, M. Wnt neural crest in Xenopus laevis. Cell Tissue Res. 305, Development 128, 3655–3663 (2001). in the ectoderm functions as a neural crest inducer. Science 203–209 (2001). 31. Papalopulu, N. & Kintner, C. A posteriorising factor, retinoic (in the press). 16. Streit, A. & Stern, C. D. Establishment and maintenance acid, reveals that anteroposterior patterning controls the 48. Sela-Donenfeld, D. & Kalcheim, C. Regulation of the onset of the border of the neural plate in the chick: involvement timing of neuronal differentiation in Xenopus of neural crest migration by coordinated activity of BMP4 of FGF and BMP activity. Mech. Dev. 82, 51–66 (1999). . Development 122, 3409–3418 (1996). and Noggin in the dorsal neural tube. Development 126, 17. Nguyen, V. H. et al. Ventral and lateral regions of the 32. Lamb, T. M. & Harland, R. M. Fibroblast growth factor is a 4749–4762 (1999). zebrafish gastrula, including the neural crest progenitors, direct neural inducer, which combined with noggin 49. Coffman, C. R., Skoglund, P., Harris, W. A. & Kintner, C. R. are established by a bmp2b/swirl pathway of genes. Dev. generates anterior–posterior neural pattern. Development Expression of an extracellular of Xotch diverts cell Biol. 199, 93–110 (1998). 121, 3627–3636 (1995). fate in Xenopus embryos. Cell 73, 659–671 (1993).

460 | JUNE 2002 | VOLUME 3 www.nature.com/reviews/genetics REVIEWS

50. Artavanis-Tsakonas, S., Rand, M. D. & Lake, R. J. Notch 61. Nakata, K., Nagai, T., Aruga, J. & Mikoshiba, K. Xenopus 73. Sefton, M., Sanchez, S. & Nieto, M. A. Conserved and signaling: cell fate control and signal integration in Zic3, a primary regulator both in neural and neural crest divergent roles for members of the Snail family of transcription development. Science 284, 770–776 (1999). development. Proc. Natl Acad. Sci. USA 94, 11980–11985 factors in the chick and mouse embryo. Development 125, 51. Cornell, R. A. & Eisen, J. S. Delta signaling mediates (1997). 3111–3121 (1998). segregation of neural crest and spinal sensory neurons 62. Nakata, K., Nagai, T., Aruga, J. & Mikoshiba, K. Xenopus 74. Baker, C. V. H. & Bronner-Fraser, M. The origins of the neural from zebrafish lateral neural plate. Development 127, Zic family and its role in neural and neural crest crest. II. An evolutionary perspective. Mech. Dev. 69, 13–29 2873–2882 (2000). development. Mech. Dev. 75, 43–51 (1998). (1997). 52. Endo, Y., Osumi, N. & Wakamatsu, Y. Bimodal functions of 63. Kuo, J. S. et al. Opl: a zinc finger protein that regulates 75. Maisey, J. G. and tails: a phylogeny. Notch-mediated signaling are involved in neural crest neural determination and patterning in Xenopus. Cladistics 2, 201–256 (1986). formation during avian ectoderm development. Development 125, 2867–2882 (1998). 76. Peterson, K. J. A phylogenetic test of the calcichordate Development 129, 863–873 (2002). 64. Brewster, R., Lee, J. & Ruiz í Altaba, A. Gli/Zic factors scenario. Lethaia 28, 25–38 (1995). 53. Barembaum, M., Moreno, T. A., LaBonne, C., Sechrist, J. pattern the neural plate by defining domains of cell 77. Kuratani, S., Nobusada, Y., Horigome, N. & & Bronner-Fraser, M. Noelin-1 is a secreted glycoprotein differentiation. Nature 393, 579–583 (1998). Shigetani, Y. of the lamprey and evolution of the involved in generation of the neural crest. Nature Cell Biol. 65. Nakata, K., Koyabu, Y., Aruga, J. & Mikoshiba, K. A novel vertebrate jaw: insights from molecular and developmental 2, 219–225 (2000). member of the Xenopus Zic family, Zic5, mediates neural perspectives. Phil. Trans. R. Soc. Lond. B 356, 1615–1632 54. Artinger, K. B., Chitnis, A. B., Mercola, M. & Driever, W. crest development. Mech. Dev. 99, 83–91 (2000). (2001). Zebrafish narrowminded suggests a genetic link between 66. Dutton, K. A. et al. Zebrafish colourless encodes 78. Holland, L. Z. & Holland, N. D. Evolution of neural crest and formation of neural crest and primary sensory neurons. and specifies non-ectomesenchymal neural crest fates. placodes: amphioxus as a model for the ancestral Development 126, 3969–3979 (1999). Development 128, 4113–4125 (2001). vertebrate? J. Anat. 199, 85–98 (2001). 55. LaBonne, C. & Bronner-Fraser, M. Snail-related 67. Spokony, R. F., Aoki, Y., Saint-Germain, N., Magner-Fink, 79. Neidert, A. H., Panopoulou, G. & Langeland, J. A. Amphioxus transcriptional repressors are required in Xenopus for both E. & Saint-Jeannet, J.-P. The Sox9 is goosecoid and the evolution of the head organizer and the induction of the neural crest and its subsequent required for development in Xenopus. prechordal plate. Evol. Dev. 2, 303–310 (2000). migration. Dev. Biol. 221, 195–205 (2000). Development 129, 421–432 (2002). 56. Epstein, D. J., Vekemans, M. & Gros, P. splotch (Sp2H), a 68. Maeda, R. et al. Xmeis1, a protooncogene involved in Acknowledgements mutation affecting development of the mouse neural tube, specifying neural crest cell fate in Xenopus embryos. The authors thank T. Moreno for help with the figures, and shows a deletion within the Paired homeodomain of Pax-3. Oncogene 20, 1329–1342 (2001). M. García-Castro and M. Albrecht for critical reading of the Cell 67, 767–774 (1991). 69. Reeves, F. C., Burdge, G. C., Fredericks, W. J., Rauscher, manuscript. 57. Mansouri, A., Stoykova, A., Torres, M. & Gruss, P. F. J. & Lillycrop, K. A. Induction of antisense Pax-3 Dysgenesis of cephalic neural crest derivatives in Pax7−/− expression leads to the rapid morphological differentiation mutant mice. Development 122, 831–838 (1996). of neuronal cells and an altered response to the mitogenic Online links 58. Dottori, M., Gross, M. K., Labosky, P. & Goulding, M. The growth factor bFGF. J. Cell Sci. 112, 253–261 (1999). winged-helix transcription factor Foxd3 suppresses 70. Vallin, J. et al. Cloning and characterization of three DATABASES interneuron differentiation and promotes neural crest cell Xenopus Slug promoters reveal direct regulation by Lef/β- The following terms in this article are linked online to: fate. Development 128, 4127–4138 (2001). catenin signaling. J. Biol. Chem. 276, 30353–30358 LocusLink: http://www.ncbi.nlm.nih.gov/LocusLink 59. Kos, R., Reedy, M. V., Johnson, R. L. & Erickson, C. A. The (2001). BMP4 | BMP7 | Foxd3 | Pax3 | Snail | snailhouse | somitabun | winged-helix transcription factor Foxd3 is important for 71. del Barrio, M. G. & Nieto, M. A. Overexpression of Snail swirl/bmp2b | Wingless | Wnt1 | Wnt3a | WNT3A establishing the neural crest lineage and repressing family members highlights their ability to promote chick melanogenesis in avian embryos. Development 128, neural crest formation. Development 129, 1583–1593 FURTHER INFORMATION 1467–1479 (2001). (2002). Encyclopedia of Life Sciences: http://www.els.net 60. Sasai, N., Mizuseki, K. & Sasai, Y. Requirement of FoxD3- 72. Jiang, R., Lan, Y., Norton, C., Sundberg, J. P. & Gridley, T. Neural crest: origin, migration and differentiation | class signaling for neural crest determination in Xenopus. The Slug gene is not essential for mesoderm or neural Vertebrate embryo: patterning the neural crest lineage Development 128, 2525–2536 (2001). crest development in mice. Dev. Biol. 198, 277–285 (1998). Access to this interactive links box is free online.

NATURE REVIEWS | GENETICS VOLUME 3 | JUNE 2002 | 461