Developmental Biology 241, 289–301 (2002)doi:10.1006/dbio.2001.0485, available online at http://www.idealibrary.com on

Posteriorization by FGF, Wnt, and Retinoic AcidIs Required for Neural Crest Induction

Sandra Villanueva,* ,† Alvaro Glavic,* Pablo Ruiz,* and Roberto Mayor* ,1

*Millennium Nucleus in Developmental Biology and Facultad de Ciencias, Universidadde Chile, and †Pontificia Universidad Catolica de Chile, Santiago, Chile

The neural crest is a unique cell population induced at the lateral border of the neural plate. Neural crest is not producedat the anterior border of the neural plate, which is fated to become forebrain. Here, the roles of BMPs, FGFs, Wnts, andretinoic acid signaling in neural crest induction were analyzed by using an assay developed for investigating theposteriorization of the neural plate. Using specific markers for the anterior neural plate border and the neural crest, theposterior end of early neurula embryos was shown to be able to transform the anterior neural plate border into neural crestcells. In addition, tissue expressing anterior neural plate markers, induced by an intermediate level of BMP activity, wastransformed into neural crest by posteriorizing signals. This transformation was mimicked by bFGF, Wnt-8, or retinoic acidtreatment and was also inhibited by expression of the dominant negative forms of the FGF receptor, the retinoic acidreceptor, and Wnt signaling molecules. The transformation of the anterior neural plate border into neural crest cells was alsoachieved in whole embryos, by retinoic acid treatment or by use of a constitutively active form of the retinoic acid receptor.By analyzing the expression of mesodermal markers and various graft experiments, the expression of the mutant retinoicacid receptor was shown to directly affect the ectoderm. We thereby propose a two-step model for neural crest induction.Initially, BMP levels intermediate to those required for neural plate and epidermal specification induce neural folds with ananterior character along the entire neural plate border. Subsequently, the most posterior region of this anterior neural plateborder is transformed into the neural crest by the posteriorizing activity of FGFs, Wnts, and retinoic acid signals. We discussa unifying model where lateralizing and posteriorizing signals are presented as two stages of the same inductive processrequired for neural crest induction. © 2001 Elsevier Science

Key Words: neural crest; Xenopus; BMP gradient; FGF; Wnts; retinoic acid; Xslug.

INTRODUCTION

The neural crest is a unique set of ectodermal-derivedcells that undergo extensive and coordinated movements,giving rise to numerous and diverse cell types, includingmuch of the peripheral nervous system, the cranofacialskeleton, and pigment cells (for reviews on neural crestdevelopment see LaBonne and Bronner-Fraser, 1999; Mayoret al., 1999; Christiansen et al., 2000; Mayor and Aybar,2001).

The neural crest originates at the border between theneural plate and the future epidermis. This border corre-sponds to the neural folds, which surround the entire neuralplate, but only the medial and posterior portions of the foldsgive rise to neural crest cells, while the anterior neural fold

1 To whom correspondence should be addressed. Fax: (56 2) 276

3802. E-mail: rmayor@uchile.cl.

0012-1606/01 $35.00© 2001 Elsevier ScienceAll rights reserved.

differentiates as forebrain. Within the neural folds, pre-sumptive neural crest cells are distinguishable from epider-mal and neural plate precursors by virtue of their distinctpatterns of gene expression. For example, the zinc fingerand helix–loop–helix transcription factors Xsnail, Xslug,and Xtwist provide early markers for the region from whichneural crest cells arise in frogs, fish, and chicks (Thiesenand Bach, 1993; Nieto et al., 1994; Mayor et al., 1995;Linker et al., 2000). The Xsnail gene is expressed initially atthe midgastrula stage, in an arc above the dorsal marginalzone which can be precisely identified as a distinct band ofcells surrounding the prospective neural plate cells (Mayoret al., 1993; Essex et al., 1993). At the end of gastrulation,the anterior transverse neural fold ceases to express Xsnail,which is then expressed exclusively in the medial andposterior portions of the neural folds, regions subsequentlyfated to become neural crest. On the other hand, the

expression of Xslug and Xtwist begins at the end of gastru-

289

290 Villanueva et al.

lation and does not occur in the anterior transverse neuralfold. Consequently, this region does not form part of theneural crest but does contribute to the structure of theforebrain (Mayor et al., 1995; Hopwood et al., 1989). Theloss of Xsnail expression in the anterior transverse neuralfold is a consequence of gene down-regulation rather thancell movement (Linker et al., 2000). Taken together, theseobservations suggest that these neural fold marker genes areinitially activated along the entire neural plate border, laterbecoming restricted to the medioposterior portion of theneural fold, which then generates prospective neural crestcells. It should be mentioned that Xsnail, Xtwist, and Xslugare not specific markers for the neural crest as they are alsoexpressed in mesoderm (Linker et al., 2000; Mayor et al.,2000).

Several, but not necessarily substitutive, hypotheses havebeen proposed to explain how the neural crest is induced.Pioneering studies in Urodeles first suggested that theinduction of the neural crest could be mediated by under-lying tissue, independent from the neural plate (Raven andKloos, 1945). More recent experiments show that the dor-solateral mesoderm, which underlies the prospective neuralcrest at the midgastrula stage, is able to induce neural crestmarkers in competent ectoderm, whereas Xenopus em-bryos, from which presumptive nonaxial mesoderm hasbeen removed, fail to express neural crest markers (Bonsteinet al., 1998; Marchant et al., 1998).

Several experiments also favor a second hypothesis whichproposes that neural crest formation requires interactionsbetween the neural plate and the epidermis (Moury andJacobson, 1989, 1990; Dickinson et al., 1995; Selleck andBronner-Fraser, 1995; Mayor et al., 1995; Mancilla andMayor, 1996). More recently, it has been suggested thatlevels of BMP signaling, intermediate to those that specifyneural plate and epidermis, may play a role in determiningthe fate of the neural plate border as well as the neural crest(Wilson et al., 1997; Marchant et al., 1998). There is alsoevidence that this may be true in zebrafish (Neave et al.,1997; Nguyen et al., 1998). To date, it has been assumedthat this BMP activity gradient is generated through theinteraction of BMPs produced by the ectoderm and BMP-binding molecules produced by the Organizer (reviewed inWeinstein and Hemmati-Brivanlou, 1999). Consequently,this gradient should generate a maximum of BMP activityin the ventral ectoderm and a minimum in the most dorsalectoderm, or dorsal midline, and should specify neural crestat an intermediate concentration. Although such a gradientcould explain why the neural crest is induced at the borderof the neural plate, this hypothesis does not account for theanterior–posterior localization of the crest. Indeed, theneural crest is not induced in the most anterior neural fold.

The anterior–posterior pattern of the neural plate hasbeen the subject of many studies. Two models have beenproposed to explain the specification of this anterior–posterior axis (reviewed in Slack and Tannahill, 1992;Gamse and Sive, 2000). The qualitative model proposes that

the different anterior–posterior positions are determined by

© 2001 Elsevier Science. A

different signals arising from corresponding regions of theunderlying mesoderm. On the other hand, the activation/transformation model (Nieuwkoop and Albers, 1990) sug-gests that two inducer factors act sequentially during gas-trulation to induce the anterior–posterior specification.Initially, the dorsal ectoderm would receive a signal fromthe Organizer, determining the induction of the anteriorneural plate. Subsequently, a posteriorizing factor wouldconvert part of the previously induced anterior neural tissueto more posterior neural fates (transformation). Severallines of evidence favor this second model (Sive et al., 1989;Holowacz and Sokol, 1999; Blumberg et al., 1997; Papa-lopulu and Kinter, 1996; McGrew et al., 1995). Much of theevidence that supports this model has come from theidentification of neural inducers and molecules able totransform neuralized tissue. Several neural-inducing pro-teins, such as Noggin, Chordin, Follistatin, Xnr3, Cerberus,and Gremlin, bind to BMPs and induce the anterior neuralplate (reviewed in Harland and Gerhart, 1997), functionswhich could correspond to the “activator” signal describedin the activation/transformation model.

Various molecules have been proposed as putative media-tors of the transformation phase. For instance, four mem-bers of the fibroblast growth factor (FGF) family, namelyeFGF, FGF3, FGF8, and FGF9, expressed in the posteriordorsal, lateral, and ventral mesoderm of Xenopus embryos,are known to be involved in the posteriorization of theneural plate (Isaac, 1997; Holowacz and Sokol, 1999). Reti-noic acid also appears to confer posterior positional infor-mation. Tissue culture experiments have suggested thatretinoic acid acts in a graded manner to induce differentanterior–posterior levels (Simeone et al., 1990).

Similarly, Wnt signaling induces posterior markers inXenopus embryos during gastrulation (Wodarz and Nusse,1998), at which stage, Xwnt3a and Xwnt8 are expressed inthe posterior dorsal, lateral, and ventral mesoderm. Wnt3acan induce the expression of posterior marker genes inneuralized ectoderm (McGrew et al., 1995). Conversely, atruncated Xwnt8 that blocks Wnt signaling prevents theinduction of posterior markers in neuralized ectoderm(Bang et al., 1999). In summary, there is strong evidence tosupport a role for FGFs, Wnts, and retinoic acid in theposteriorization of the neural plate. However, whether theneural crest is also posteriorized via a similar cellular ormolecular mechanism has yet to be determined.

In this study, we show for the first time that the anteriorneural fold, which does not express neural crest markers,can be transformed into a neural crest fate by the addition ofposteriorizing signals. In order to study the anterior–posterior pattern of the neural plate border, a novel assaywas employed. Pieces of anterior neural fold were dissectedfrom late-gastrula embryos and either conjugated with themost posterior region of gastrula embryos or treated withFGF, Xwnt8, or retinoic acid. Under these conditions, theexpression of anterior neural markers was inhibited and theinduction of neural crest markers was observed. In addition,

weakly neuralized animal caps, which expressed only ante-

ll rights reserved.

291Posteriorization of the Neural Crest

rior neural markers due to the injection of a dominantnegative form of the BMP4 receptor, were also transformedto express neural crest markers by conjugating with theposterior region of a late-gastrula embryo or by treatingwith posteriorizing molecules. Finally, the use of dominantnegative mutants of the FGF receptor, a retinoic acidreceptor (RAR), and a Wnt signaling molecule showed thateach of these signals was required for the specification ofneural crest cells. In this paper, we discuss a model wherebythe mediolateral position of the neural crest is specified bya BMP signaling gradient, while the anterior–posterior pat-tern is specified by FGF, retinoic acid, and Wnt signaling.

MATERIAL AND METHODS

Embryos and Micromanipulations

Xenopus embryos were obtained by artificial fertilization, dejel-lied in 2% cysteine (Smith and Slack, 1983), reared in 10% normalamphibian medium (NAM; Slack, 1984), and staged according toNieuwkoop and Faber (1967). Explants, conjugates, and dissectionswere carried out by using eyebrow knives, as described in Mancillaand Mayor (1996).

RNA and RLDx Injection

Dejellied one- or two-cell embryos were placed in 75% NAMwith 5% Ficoll, injected with either differing amounts of RNA, asindicated, or with 10 nl of 25 mg/ml rhodamine dextran (RLDx,Molecular Probes), and subsequently reared at 14–16°C. Fluores-cence was viewed and photographed by using epifluorescenceoptics (Zeiss).

Capped RNAs were synthesized from linearized plasmids byusing an appropriate RNA polymerase (Boehringer Mannheim) inthe presence of 500 mM 59-mGpppG-39 cap analog. Injected RNAswere as follows: DBMPR, a dominant negative of the type II BMPreceptor (Sasai et al., 1995); Xwnt-8 (Christian et al., 1991);RARa1405, a dominant negative of RARa1 (Blumberg et al., 1997);XFD, a dominant negative of FGF-1 receptor (Amaya et al., 1991);dnTCF-3, a dominant negative of TCF-3 (Molenaar et al., 1996);VP16-xRARa1, a constitutively active form of RARa1 (Blumberg etal., 1997); and b-galactosidase (b-gal) used as a lineage tracer.

Treatment with FGF and Retinoic Acid

Embryos or animal caps taken from stage 9.5 embryos or theanterior neural fold taken from stage 12.5 embryos were incubatedwith 0.1 ng/ml bFGF (Gibco) or several concentrations (1025 to 1028

M) of all-trans-retinoic acid for 9–12 h, washed, and analyzed asdescribed under Results. It should be mentioned that the minimalretinoic acid concentration able to induce neural crest in animalcaps varied between retinoic acid batches.

Whole-Mount in Situ Hybridizationand b-Gal Staining

Antisense RNA probes were prepared for Xslug (neural crestmarker; Mayor et al., 1995); Xag (anterior neural fold marker; Blitz

and Cho, 1995); Xbra (mesodermal marker; Smith et al., 1991);

© 2001 Elsevier Science. A

cpl-1 (anterior neural fold marker; Richter et al., 1998); goosecoid(anterior mesodermal marker; Blumberg et al., 1991); and Xsox-2(pan neural plate marker; Dr. R. M. Grainger, personal communi-cation). Specimens were prepared, hybridized, and stained by themethod of Harland (1991) with the modifications described byMancilla and Mayor (1996). Double in situ hybridization wasperformed as described by Lamb and Harland (1995), using thefollowing alkaline phosphatase substrates: NBT/BCIP (BoehringerMannheim), which produces a purple color; and BCIP, whichproduces a green color. b-Gal staining was developed by usingX-gal, as described in Whiting et al. (1991).

RESULTS

Posteriorizing Signals Induce Neural Crest CellsPrevious studies have indicated that the induction of the

neural crest occurs when a threshold concentration ofBMP4 is reached in the ectoderm of Xenopus or zebrafishembryos (Marchant et al., 1998; Nguyen et al., 1998). Ourown work in animal caps has shown that the expression ofan intermediate level of DBMPR, a dominant negative formof the BMP receptor, induces the activation of neural crestmarkers, while neural plate markers are activated at higherlevels of the same mutant receptor (Figs. 1A and 1B;Marchant et al., 1998). However, neural crest markers wereusually found to be expressed in larger animal caps but notso in smaller ones. Whether the size of the animal caps wasimportant in determining the expression of neural crestmarkers was therefore investigated. Embryos were injectedat the one-cell stage with 35 pg of DBMPR mRNA, thethreshold concentration found to induce neural crest mark-ers in the absence of neural plate markers. Both large andsmall animal caps were dissected at stage 9.5 and culturedto the equivalent of stage 17 for the analysis of neural crest(Xslug) and neural plate (Xsox-2) markers and to stage 11 forthe mesodermal marker Xbra. No mesodermal (0%, n 5 38)or neural plate markers (Fig. 1E; 0%, n 5 29) were found tobe expressed in either large or small animal caps, confirm-ing previous results (Marchant et al., 1998). However, Xslugexpression was observed in 85% of large animal cap and 0%of small animal caps (Fig. 1C; n 5 62). As a control that thenumber of cells was not the factor required to induce Xslugexpression in the large animal caps, groups of three or foursmall animal caps were conjugated together in order tomake a large animal cap, but no Xslug expression wasdetected (Fig. 1D; 0%, n 5 6). These results suggest that, inorder to induce neural crest cells, another factor is presentand active in larger animal caps in addition to the BMPthreshold.

In this context, we postulated that the application of thisadditional signal to small animal caps expressing theDBMPR threshold concentration could achieve the induc-tion of neural crest markers. Here, we describe the posteri-orizing signals in the context of the activation/transformation model; the initial analysis of these signals issummarized in Fig. 2A. Small animal caps from embryos

injected at the one-cell stage with 35 pg of DBMPR mRNA

ll rights reserved.

292 Villanueva et al.

were dissected at stage 9.5 and cultured in isolation orcombined with the most posterior fifth of late-gastrula

FIG. 1. Effect of the size of neuralized caps on neural crest indusignaling specify neural fold (NF) tissue, lower levels specify the neof DBMPR mRNA, corresponding to the neural crest threshold cosmall animal caps were dissected at stage 9.5 and cultured to the eqcaps expressed Xslug, while no small caps displayed Xslug expresgroups of three or four caps, cultured as in (C); no Xslug expressionexpression (0%, n 5 16).FIG. 2. Posteriorization of nonneuralized ectoderm. (A) Embryos acaps were dissected at stage 9.5 and either cultured alone (B and Cthe one-cell stage with RLDx (E–H). (I–K) The neurula embryo segm(C, F, J) was analyzed when caps reached the stage 19 equivalent oXag (96%, n 5 40) but no Xslug (0%, n 5 42) or Xbra (0%, n 5 26)(20% of explants expressed Xag, n 5 43) and an induction in the expAlthough conjugates also expressed Xbra (G), the expression was orithe quenching in the intensity of the fluorescence by the in situ sign 5 38) but did express Xbra (100%, n 5 42).

embryos, until the caps reached the equivalent of stage 11

© 2001 Elsevier Science. A

or 19 for the analysis of mesodermal or neural markers,respectively. Gastrula embryos were previously injected

. (A, B) Summary of previous results: intermediate levels of BMPplate (NP), and higher levels specify the epidermis (E). (C–E) 35 pgtration, were injected at the one-cell-stage embryo. (C) Large andlent of stage 17, when Xslug expression was analyzed: 85% of large(n 5 62). (D) Small animal caps were dissected and conjugated inbserved (0%; n 5 6). (E) Big animal caps as (C), analyzed for Xsox-2

one-cell stage were injected with 35 pg of DBMPR mRNA. Animalonjugated with the posterior fifth of stage 13 embryos injected atwere also cultured alone. The expression of Xag (B, E, I) and Xslug

ge 11 for Xbra expression (D, G, K). (B–D) Injected caps expressedession was detected. (E, F) In conjugates, a down-regulation of Xagn of Xslug (83% of explants expressed Xslug; n 5 42) was observed.ed from the neurula embryo (labeled tissue, dotted area in H). NoteI–J) Posterior pieces did not express Xag (0%, n 5 33) or Xslug (0%,

ctionuralncenuivasionwas o

t the) or centsr staexprressioginatnal. (

with RLDx as a lineage label, and the posterior segments

ll rights reserved.

anterior neural fold. (A) The anterior neural fold/plate was dis-

293Posteriorization of the Neural Crest

© 2001 Elsevier Science. A

were cultured for the same time period as the animal caps.As expected, isolated caps expressed anterior neural platemarkers such as Xag and cpl-1, but not neural crest markerssuch as Xslug or mesodermal markers such as Xbra (Figs.2B–2D). Conversely, the isolated pieces of posterior gastrulaexpressed neither Xag nor Xslug, but as expected, expressedXbra given their posterior mesoderm content (Fig. 2K).

sected from stage 12.5–embryos, avoiding the inclusion of meso-derm or epidermal tissue. The tissue was cultured alone (B, C) orconjugated with the posterior fifth of stage 13 embryos, injectedwith RLDx at the one-cell stage (D–G). (H and I) The posteriorpieces of the neurula embryos were also cultured alone. The neuralplate was also conjugated with the posterior fifth of stage 13embryos (J, K). All explants were cultured to the stage 19 equivalentwhen the expression of Xag (B, D, and H), Xslug (C, E–G, I, K), andXsox2 (J) was determined. (B, C) The anterior neural fold/plateexpressed Xag (90%, n 5 50) but no expression of Xslug (0%, n 536) was detected. (D–G) In the conjugates, Xslug expression wasobserved in many cases (85% of Xslug expression, n 5 45), whilethat of Xag was down-regulated (12% of conjugates expressed Xag,n 5 48). Xslug-expressing cells came from the anterior neuralfold/plate (unlabeled tissue; F, G), whereas the posterior neurulasegments expressed neither Xag (H, 0%, n 5 38) nor Xslug (I, 0%,n 5 40). The conjugates of neural plate and the posterior fifth of astage 13 embryo expressed Xsox2 (J, 100%, n 5 25), but noinduction of Xslug expression was observed (K, 0%, n 5 25).FIG. 4. Posteriorizing molecules involved in neural crest induc-tion. (A) Embryos at the one-cell stage were injected with 35 pg ofDBMPR mRNA. Animal caps were dissected at stage 9.5 andcultured in isolation (B) or with 1026 M of retinoic acid (C), 0.1ng/ml bFGF (D), or conjugated with stage 9.5 animal caps takenfrom embryos injected at the one-cell stage with 300 pg ofpCSKAXwnt-8 (E). Animal caps were then cultured to the stage 18equivalent and the expression of Xslug was determined. (B) NoXslug expression was detected in injected but untreated animalcaps (0%, n 5 40), whereas Xslug expression was clearly observedin caps treated with bFGF (C, 50%, n 5 30), retinoic acid (D, 45%,n 5 22), or Xwnt-8 (E, 70%, n 5 33). No expression of Xsox-2 orXbra was observed in the caps. (F) The anterior neural fold wasdissected from stage 12.5–embryos, treated with 1026 M of retinoicacid (G) or with 0.1 ng/ml of bFGF (H), and cultured until theequivalent of stage 18, and the expression of Xslug was analyzed.Expression of Xslug was observed in 57% (G, n 5 7) or 63% (H, n 58) of the explants. Arrows indicate Xslug expression.FIG. 5. Inhibition of posteriorizing signals blocks neural crestinduction in vitro. (A) Embryos at the one-cell stage were coin-jected with 35 pg of DBMPR mRNA and the dominant negatives fordifferent posteriorizing signals, as indicated below. Animal capswere dissected at stage 9.5 and conjugated with the posterior fifthof stage 13 neurula embryos. Conjugates were cultured until theanimal cap reached the equivalent of stage 19, when the expressionof Xslug was analyzed. (B) Conjugates of control animal capsinjected only with DBMPR mRNA displayed clear Xslug expression(85%, n 5 32), while Xslug expression was inhibited in conjugateanimal caps taken from embryos coinjected with 3 ng of RARa1405

mRNA (C; 22% of explants expressed Xslug, n 5 28), 0.2 ng of XFDmRNA (D; 21% of explants expressed Xslug, n 5 31), or 0.5 ng of

FIG. 3. Neural crest markers induced by posteriorization of the dnTCF-3 mRNA (E; 52% of explants expressed Xslug, n 5 28).

ll rights reserved.

294 Villanueva et al.

However, the conjugation of injected animal caps withposterior late gastrula down-regulated the expression of Xag(Fig. 2E, 20% of the explants expressed Xag) and induced theexpression of Xslug (83% of the explants expressed Xslug,Fig. 2F). Moreover, expression of Xslug was never observedin the fluorescent part of the conjugates, indicating that thecells expressing Xslug originated from the animal cap.

In order to rule out the possibility that the posteriorgastrula sections may induce the formation of mesoderm inthe injected caps, leading to the induction of neural crestmarkers via the mesoderm, the following experiments wereperformed. First, conjugates expressing Xbra were analyzedunder fluorescence microscope to identify the origin of theXbra-expressing cells. No animal caps (nonfluorescent)were seen to express the mesodermal marker Xbra (Figs. 2Gand 2H, 0%, n 5 42). Second, analysis of Xbra was alsocarried out in animal caps conjugated with vegetal cellstaken from stage 8 embryos and cultured until the capsreached the equivalent of stage 11. No Xbra-expressing cellswere observed in these conjugates (not shown). Theseresults indicate that animal caps taken at stage 9.5 were notcompetent to be responsive to the potential mesodermalinducers present in the posterior part of the late gastrula.Thus, the transformation of injected animal caps expressinganterior neural fold markers into caps expressing neuralcrest markers, achieved through the juxtaposition withposterior gastrula sections, suggests that neural crest induc-tion is likely to require the presence of posteriorizingsignals.

If the above conclusion is true, it should be possible todirectly transform a segment of anterior neural fold, fated tobecome forebrain, into neural crest by conjugating it withthe posteriorizing signal source. This experiment was car-ried out and is described in Fig. 3A. Sections of anteriorneural fold were dissected from early neurula embryos,taking care to avoid including mesoderm or ventral epider-mis in the explants. These pieces were conjugated withposterior portions of the same stage, embryos previouslyinjected with RLDx (at the one-cell stage). As a control,anterior and posterior explants were cultured in isolation.As expected, anterior explants were found to express Xagbut not Xslug, and posterior explants did not express eitherof these markers (Figs. 3B, 3C, 3H, and 3I). However, theexpression of Xag was seen to be down-regulated in theconjugates (Fig. 3D; 12% of the explants expressed Xag n 548), whereas a clear induction of Xslug was observed (Fig.3E; 85% of Xslug n 5 45). In order to determine the originof the tissue expressing the neural crest marker, conjugateswere examined under a fluorescence microscope. TheXslug-expressing cells were seen to originate from theanterior neural fold (unlabeled tissue). These findings showthat the neural crest can be induced by posteriorization ofthe anterior neural fold. In order to analyze whether theinduction of neural crest markers by the posterior region ofa neurula was a specific transformation of the neural fold orwas simply an effect of posteriorization of neural plate,

which in turn expressed the neural crest markers, an

© 2001 Elsevier Science. A

additional experiment was carried out. The posterior part ofa neurula embryo was conjugated with the anterior neuralfold or with the neural plate, and the expression of Xslugand Xsox-2 was analyzed (Figs. 3J and 3K). As it wasdescribed before when anterior neural fold was used, expres-sion of Xslug was observed; however, when neural plate wasused in the conjugates, no Xslug expression was observed(Fig. 3K), although these conjugates expressed Xsox-2(Fig. 3J).

Posteriorizing Molecules Induce Neural Crest Cells

As mentioned earlier, the involvement of three signalingmolecules has been described in the posteriorization of theneural plate: FGF, Wnts, and retinoic acid. As the posteri-orization of the anterior neural fold was found to contributeto the induction of the neural crest, the dependency of thisprocess on the same signals that posteriorize the neuralplate was subsequently analyzed.

Embryos were injected at the one-cell stage with 35 pg ofDBMPR mRNA, and small animal caps were dissected atstage 9.5 and either cultured in 38% NAM or 38% NAMcontaining 0.1 ng/ml bFGF or 1026 M retinoic acid orconjugated with caps taken from embryos injected with 300pg pCSKAXwnt-8 DNA (Fig. 4A). The caps and conjugateswere cultured to a stage 10 or 19 equivalent, after which theexpression of Xslug, Xbra, and Xsox-2 was determined. NoXbra or Xsox-2 expression was detected, indicating theabsence of mesodermal contamination in the cap explantsand that the amount of DBMPR mRNA injected was insuf-ficient to cause cap neuralization. Xslug expression was notdetected in injected caps cultured alone (Fig. 4B). However,injected caps treated with FGF, retinoic acid, or Xwnt-8showed a clear expression of Xslug (Figs. 4C–4E). Anequivalent experiment, in which the segment of anteriorneural fold described in Fig. 3A was treated with either FGFor retinoic acid, produced a similar expression of the neuralcrest marker Xslug (Figs. 4F–4H). These results show thatany of these posteriorizing signals alone, under the presentexperimental conditions, is sufficient to induce neural crestin a tissue otherwise fated to become the anterior border ofthe neural plate.

As shown in Fig. 5A, the posteriorizing assay was thenused to analyze whether a lack of these signals inhibited theinduction of neural crest cells. Embryos were injected at theone-cell stage with either 35 pg DBMPR mRNA or thedominant negative forms of the FGF-1 receptor or theretinoic acid receptor, XFD (Amaya et al., 1991) andRARa1405 (Blumberg et al., 1997), respectively, or with thedominant negative form of TCF3 (dnTCF3), a component ofthe Wnt signaling pathway (for a review, see Barker andClevers, 2000). At the late blastula stage, animal caps weredissected and conjugated with the posterior part of earlyneurula-stage embryos, as described before. Conjugateswere cultured until the equivalent of stage 19, after whichthe expression of Xslug was analyzed. As previously stated,

conjugates containing DBMPR mRNA-injected caps ex-

ll rights reserved.

295Posteriorization of the Neural Crest

pressed the neural crest marker Xslug (Fig. 5B, 83% ofexplants expressed Xslug; n 5 34), while this expressionwas inhibited in animal caps coinjected with eitherRARa1405 (Fig. 5C, 22% of explants expressed Xslug; n 528), XFD (Fig. 5D, 21% of explants expressed Xslug; n 5 31),or dnTCF3 (Fig. 5E, 52% of explants expressed Xslug; n 528). These results show that no neural crest cells areinduced when each of the posteriorizing signals is inhibited,suggesting the involvement of all these signals in theinduction process. It is important to note, however, that theinhibition achieved by blocking the Wnt signaling pathwaywas weaker than that observed for the absence of FGF orretinoic acid signaling; this will be commented on furtherunder Discussion.

Retinoic Acid Signaling in Neural Crest Induction

Previous studies have proposed FGF and Wnt signals to beinvolved in neural crest induction (Mayor et al., 1995, 1997;Saint-Jeannet et al., 1997; LaBonne and Bronner-Fraser,1998). However, those signals were not shown to functionas posteriorizing agents and were instead proposed to act inconjunction with neural inducer molecules. In addition, therole of retinoic acid in neural crest induction has yet to befully investigated. The results described in this study indi-cate that retinoic acid is not only sufficient but alsonecessary for the induction of neural crest cells, as deter-mined by using the animal cap and posteriorizing assays.Further experiments using retinoic acid in whole embryoswere therefore carried out.

Early-gastrula embryos treated with varying concentra-tions of retinoic acid for 3–6 h were cultured until stage 17.The expression of the anterior neural fold markers cpl-1(Fig. 6A) and Xslug (Fig. 6D) was then analyzed. Severalexpression patterns were observed depending on the reti-noic acid concentration employed. At 1 mM, the mostfrequent outcome was a reduction in the expression of cpl-1and its localization to the most anterior part of the embryo(Fig. 6B). On the other hand, Xslug was up-regulated in themost anterior neural fold (asterisk in Fig. 6E). An anteriormovement in the posterior limit of Xslug expression wasalso observed (Fig. 6E), and sometimes this concentrationinduced a complete absence of Xslug (white arrow in Fig.6H). At 10 mM, complete inhibition of cpl-1 expression wasobserved (Fig. 6C), while Xslug was either detected at verylow levels or not detected at all (Fig. 6F). These results canbe interpreted to show retinoic acid functioning as a poste-riorizing signal and determining the outcome of neuralfolds as anterior folds or neural crest. However, retinoicacid is also known to affect mesodermal patterning and themesoderm is involved in neural crest induction.

In order to assess this alternative explanation, double insitu hybridization experiments were performed in treatedembryos, using neural fold and mesodermal markers, todetermine whether the embryos that displayed an affectedpattern in the neural folds also possessed an abnormal

pattern in the mesoderm. These experiments are summa-

© 2001 Elsevier Science. A

rized in Figs. 6G–6J. The most extreme phenotype observedfor Xslug was the complete inhibition of its expression (Fig.6H, 25% of explants expressed Xslug; n 5 48). However,even in those cases, the position of the mesoderm was notaffected, as shown by the normal localization of gsc in thesame embryos (arrowheads in Figs. 6G and 6H). The sameextreme inhibitory phenotype was obtained for cpl-1 (whitearrow in Fig. 6J), in which case the position of the notochordwas normal, as detected by Xbra expression (arrowhead inFigs. 6I and 6J). Taken together, these results suggest thatretinoic acid is involved in the transformation of theanterior neural fold into neural crest cells without signifi-cant effects on the mesoderm. In order to test this conclu-sion further, several mutant forms of RARs were used.

Although only the results for RARa are described here,the dominant negative and constitutively active forms ofboth a and g RAR were tested, as no difference wasobserved between these two kinds of retinoic acid recep-tors. One blastomere of two-cell-stage embryos was in-jected with 1 ng of xRARa1405 mRNA (the dominant nega-tive form of RARa) together with b-gal mRNA, as a markerfor the injected blastomere. Embryos thus treated exhibitedcpl-1 expression in a more posterior position (Fig. 7A), whilethe effect on Xslug expression was more complex. A clearmovement toward the posterior end was observed at theposterior limit of Xslug expression, whereas no effect wasobserved at the anterior limit (Fig. 7C; 35%, n 5 125). Innormal embryos, the posterior border of cpl-1 expressioncorresponds with the anterior border of Xslug expression. Ifthe embryos injected with xRARa1405 showed a posteriorrelocation in the expression of cpl-1 but no effect at theanterior border for the expression of Xslug, an overlappingin the expression of these genes should be expected; to testthis possibility, double in situ hybridization was performed.A clear overlap in the expression of these two markers wasobserved at the injected side (Fig. 7E). These results showthat inhibition of RAR is able to profoundly modify theanterior–posterior expression pattern of the neural folds andthat the anterior neural fold and the posterior limit of theneural crest are repressed by RA activity.

Next, the effect of a constitutively active form of RAR(VP16-xRARa1) was assessed by injecting 500 pg of VP16-xRARa1 mRNA into one blastomere of two-cell-stage em-bryos and analyzing cpl-1 and Xslug expression at theneurula stage. The injection of VP16-xRARa1 produced ananterior movement in cpl-1 expression (Fig. 7B, 67%, n 589) and sometimes complete inhibition. Upon analysis ofXslug expression, two phenotypes were most frequent: 75%of the injected embryos (n 5 167) showed no Xslug expres-sion in the injected side (Fig. 7D), whereas 16% displayed ashift toward the anterior fold. Expression of Xslug at theanterior neural fold was observed more frequently when theembryos were injected at the one-cell stage (Fig. 7F, 36%,n 5 58). Thus, the most common phenotypes induced usingVP16-xRARa1, an activated form of the RAR, were similar

to those produced by direct treatment with retinoic acid.

ll rights reserved.

rs.

296 Villanueva et al.

Nonetheless, the effect was more localized by mRNAinjection than by retinoic acid application in solution.

As the results shown in Figs. 8G–8J suggest that the

FIG. 6. Effect of retinoic acid treatment on neural fold patternconcentrations of retinoic acid for 3–6 h, washed in 10% NAM, andanterior neural fold marker cpl-1 (A–C, arrowheads; I and J, arrows)the mesodermal markers gsc (G and H, arrowheads) and Xbra (I and(B, E, H, J) Embryos treated with retinoic acid at 1026 M or (C and Fand Xslug (E, 56%, n 5 88) expression was observed, with up-reguin E). Some embryos did not express Xslug (H, 75%, n 5 48) or cpposterior position of gsc expression (arrowhead in H, 75% normalexpression, n 5 53). At 1025 M, the percentage of embryos displayexpression (F, 71% with normal expression, n 5 56) was higher. Thearrows and arrowheads indicate marker expression, whereas whiteFIG. 7. Effect on neural fold patterning of mutant retinoic acid rinjected with 3 ng of xRARa1405 (A, C, E) or 1 ng of VP16-xRARa1together with b-gal mRNA as a marker. Embryos were culturedanalyzed. Injection of xRARa1405 produced a posterior movement iXslug expression shifted toward a more posterior position, the ahybridization for Xslug (purple) and cpl-1 (green) was performed t(arrow). Injection of VP16-xRARa1 produced an anterior shift in thof Xslug (D; 48%, n 5 49). However, when the injection was perfoXslug at the anterior neural fold (arrow in F; 33%, n 5 42). The antbars indicating the posterior limits in the expression of the marke

retinoic acid effect was not mesoderm-dependent, an addi-

© 2001 Elsevier Science. A

tional experiment was performed to rule out this possibil-ity. Embryos injected at the one-cell stage with 1 ng ofeither xRARa1405 or VP16-xRARa1 mRNA were cultured to

Early-gastrula embryos (stage 10.5) were incubated with varyingured until sibling embryos reached stage 17. The expression of theneural crest marker Xslug (D–F arrowheads; G and H, arrows), androwheads) was then determined. (A, D, G, and I) Control embryos.025 M. At 1026 M, an anterior movement in cpl-1 (B; 75%, n 5 87)

of Xslug in the most anterior neural fold in some cases (asterisk, 77%, n 5 53), in which cases no effect was seen on the anterioression, n 5 48) or Xbra expression (arrowhead in J, 100% normalnhibited cpl-1 (C; 18% with normal expression, n 5 35) or Xslugrior ends of the embryos are shown in an upward orientation. Blackws and arrowheads indicate inhibited marker expression.or expression. Single blastomeres of two-cell-stage embryos were) or injected at the one-cell stage with 0.5 ng of VP16-xRARa1 (F),age 17 and the expression of cpl-1 (A, B, E) and Xslug (C–F) wasexpression of cpl-1 (A, 92%, n 5 67). While the posterior limit of

ior limit was not affected (C; 58%, n 5 76). (E) Double in situess the overlapping expression of these genes in the injected arearession of cpl-1 (B; 67%, n 5 48) and inhibition in the expressionat the one-cell stage, more embryos showed ectopic expression ofends of the embryos are shown in an upward orientation, with the

ing.cult

, theJ, ar

) at 1lationl-1 (Jexpring iantearro

ecept(B, Dto stn thenter

o asse exp

rmederior

stage 10 and a piece of ectoderm was taken and grafted onto

ll rights reserved.

297Posteriorization of the Neural Crest

normal embryos (Fig. 8A). Ectoderms taken from embryosinjected with b-gal mRNA were used as controls. Whencontrol ectoderm was grafted onto the prospective neuralcrest region, no expression of cpl-1 was observed (Fig. 8B).However, when the graft contained xRARa1405, clear ex-

FIG. 8. Effect of mutant retinoic acid receptors is not mesoderm-either VP16-xRARa1 or xRARa1405 and b-gal mRNA and culturedstage 12 embryos. The arrows and dotted lines indicate the positio(B) Dorsal view of an embryo with a control graft in the region ofgrafted segment (0%, n 5 15). (C) Dorsal view of an embryo graftedneural crest: cpl-1 expression was observed in the graft; notice thegrafts express Xslug, n 5 40). (D) Anterior view of an embryo with aof cpl-1 was seen (90%, n 5 20). (E) Anterior view of an embryo grawhere an inhibited expression of cpl-1 was noted. Notice that the rthe grafts express cpl-1; n 5 10).FIG. 9. Model of neural crest induction. Dotted area in (A) anmolecules. (A) Early-gastrula stage. A gradient of BMP activity isactivities of BMP intermediate to those required for the specificatiofold (yellow). (B) Early to midgastrula stage. Signals from the laterotransforming the anterior neural plate border into the future neurainto the posterior region of the embryo, where it continues to prodof these signals is required for the complete specification of the neA, anterior; P, posterior.

pression of cpl-1 was observed in the graft (Fig. 8C). Given

© 2001 Elsevier Science. A

that VP16-xRARa1 injection exerted an inhibitory effect oncpl-1 at the anterior neural fold (Fig. 7B), the graft express-ing this construct was placed in the anterior neural fold.When control ectoderm was grafted onto that region, nor-mal expression of cpl-1 was observed (Fig. 8D, 90%, n 5 20).

dent. (A) Embryos were injected at the one-cell stage with 1 ng ofage 10, after which a portion of ectoderm was grafted onto normalthe grafts. The grafts were recognized by the green b-gal staining.prospective neural crest: No cpl-1 expression was observed in the

a segment containing xRARa1405 in the region of the prospectivelap in the Xslug expression and b-gal staining (arrow; 50% of the

rol graft in the anterior neural fold region, where normal expressionith VP16-xRARa1-expressing ectoderm in the anterior neural fold,n of b-gal staining (green) does not shows cpl-1 expression (20% of

represents the Spemann organizer, which secrets BMP-bindingblished in the ectoderm. The neural plate border is specified at

neural plate or epidermis. This tissue is specified as anterior neuralral mesoderm, such as Wnt8 and eFGF, spread into the ectoderm,t. (C) Late-gastrula stage. The lateroventral mesoderm has moved

Wnt8, eFGF, and possible retinoic acid (RA). The continued effectcrest cells. An, animal pole; Vg, vegetal pole; D, dorsal; V, ventral;

depento stn ofthewithover

contft wegio

d (B)esta

n ofventl cresuce

ural

Grafting of VP16-xRARa1-expressing ectoderm, however,

ll rights reserved.

298 Villanueva et al.

caused an inhibition on cpl-1 expression in 80% of thegrafts (Fig. 8E, n 5 10), a result similar to that obtained bydirect injection of VP16xRARa1 mRNA at the two-cellstage. This experiment shows that the effect of the mutantRAR forms was localized directly on the neural folds andwas not dependent on a modification of the mesoderm, seento be normal in the grafted embryos. Taken together, theseresults point to an important role for retinoic acid and itsreceptors for the correct anteroposterior specification of theneural crest cells.

DISCUSSION

Molecules Involved in Neural Crest InductionThe participation of several molecules in the induction of

the neural crest has been previously reported. We haveshown here that FGF, Wnts, and retinoic acid signals areinvolved in the induction of these tissues. BMPs have alsobeen proposed for the dorsal specification of the neural tubevia two different mechanisms. It has been shown thatneural crest markers are induced by adding BMPs to neuralplate explants (Liem et al., 1995). Recently, several inde-pendent reports have highlighted the importance of particu-lar concentrations of BMPs in the specification of the neuralplate border. Levels of BMP intermediate to those requiredto induce neural plate and epidermis have been shown toinduce the anterior neural plate border (Wilson et al., 1997),as well as neural crest cells in Xenopus and in Zebrafishembryos (Morgan and Sargent, 1997; Marchant et al., 1998;LaBonne and Bronner Fraser, 1998; Nguyen et al., 1998; thisreport). Streit and Stern (1999) also showed that the neuralplate border of chick embryos develops by a combination ofsignals where BMPs and their inhibitors are involved.

FGF has also been implicated in neural crest induction(Mayor et al., 1995, 1997; LaBonne and Bronner-Fraser,1998). Addition of FGF to weakly neuralized animal capshas been shown to activate the expression of neural crestmarkers, whereas the use of a dominant negative form ofthe FGF receptor inhibited the induction of the neural crestin treated embryos. The role of FGF is nonetheless contro-versial, as it has also been implicated in neural plateinduction (Launay et al., 1996). However, in Xenopus, thedirect participation of FGF as a neural inducer was ruled outmore recently (Holowacz and Sokol, 1999; Ribisi et al.,2000; Curran and Grainger, 2000). Wnts and some elementsof the Wnt signaling pathway have also been implicated inneural crest induction (Saint-Jeannet et al., 1997; LaBonneand Bronner-Fraser, 1998; Chang and Hemmati-Brivanlou,1998). Based on the timing of Wnt expression, Wnt factorswere suggested to participate in the maintenance of theneural crest differentiation program, rather than its initialspecification (Saint-Jeannet et al., 1997).

We have reported here that FGF and Wnt signalingparticipate in the neural crest induction process, confirm-ing previous results. Evidence is also presented regarding

the role of retinoic acid in neural crest induction. The latter

© 2001 Elsevier Science. A

is a novel finding, as this molecule has previously only beenproposed to participate in neural crest differentiation (Bara-nowitz, 1989; Dupe et al., 1999). It has also been shown thattreatment of Xenopus embryos with retinoic acid affectsthe development of cranial nerves derived from neural crest(Papalopulu et al., 1991). Additionally, it has been shownthat, upon retinoic acid treatment, Pax-3, a gene expressedat the posterior border of the neural plate and within theneural plate, is induced in neuralized ectoderm (Bang et al.,1997). However, in the present study, a novel mechanismfor the participation of these molecules in neural crestinduction is explored.

Models for Neural Crest Induction

The model for neural crest induction currently acceptedtoday relates to the interaction between neural plate andepidermis (review in LaBonne and Bronner-Fraser, 1999). Adifferent, but not necessarily substitutive model, is basedon the induction of specific ectodermal tissues via a BMPgradient (reviewed in Mayor et al., 1999). Two independentreports have shown the difficulty of inducing neural crestmarkers by lowering BMP activity in the ectoderm, point-ing to the need for an additional signal (Wilson et al., 1997;LaBonne and Bronner-Fraser, 1998). On the other hand, twoother reports have shown that neural crest induction waspossible under similar conditions (Morgan and Sargent,1997; Marchant et al., 1998). In the latter case, the level ofneural crest induction was not very high. The present studyprovides a possible explanation for this discrepancy, namelythe size of the animal cap employed, as the neural crest isinduced only when large animal caps are dissected fromembryos treated with the anti-BMPs molecules. Conse-quently, another factor in addition to the BMP thresholdconcentration needs to be present in the animal cap forneural crest markers to be expressed.

LaBonne and Bronner-Fraser (1998) analyzed an addi-tional signal required for neural crest induction. Theseauthors concluded that FGF and Wnt signals could inducethe neural crest in weakly neuralized animal caps, andproposed a function for these molecules in what theynamed a “lateralization” process. We confirmed the partici-pation of FGFs and Wnts signals on neural crest inductionand added retinoic acid as another additional signal. Wefound that all these signals can work as posteriorizingsignals based on an assay developed by Cox and Hemmati-Brivanlou (1995). Our results show that the anterior neuralplate border, which expresses Xag, can be transformed byposteriorizing signals into a tissue that expressed neuralcrest markers. In addition, anterior neural plate markers,induced by an intermediate level of BMP activity, couldalso be transformed into neural crest by posteriorizingsignals. Our evidence therefore indicates that the signalsinvolved in the posteriorization of the neural plate exert aposteriorizing effect on the neural fold, such as FGF, Wnt-8,or retinoic acid, were sufficient to transform anterior neural

fold into neural crest cells in weakly neuralized animal

ll rights reserved.

299Posteriorization of the Neural Crest

caps. Neural crest cells were also induced in pieces of theanterior neural fold treated with either FGF or retinoic acid.However, blocking any of these signals via dominant nega-tive forms inhibited the induction of neural crest cells inthe in vitro posteriorizing assay. Although each posterior-izing signal was able to induce the neural crest, the absenceof a single agent inhibited the induction process. A possibleexplanation for this discrepancy could relate to the concen-trations used in the assay, which are probably much abovethe levels reached in a normal embryo. It is possible that, inorder to induce the neural crest, the participation of eachposteriorizing signal must occur in an additive manner, sothat independently or under the inhibition of one of thesignals, these would be too weak to posteriorize the em-bryo. However, the addition in vitro of sufficiently elevatedconcentrations of each agent could rescue the embryo fromthe absence of other signals. Alternatively, FGF, retinoicacid, and Wnts signaling could not act independently. Theycould work serially; thus, blocking any of the signals wouldinhibit induction, but overexpression of any could hyperac-tivate the pathway leading to neural crest induction.

The outcomes using dominant negatives in whole em-bryos are also compatible with a role for these agents asposteriorizing signals, especially in the case of retinoic acidas analyzed here. The anterior neural fold was transformedinto neural crest by addition of retinoic acid to the embryosor by expression of a constitutively active form of itsreceptor. Inhibition of retinoic acid signaling in dominantnegative mutants caused the opposite result, whereby theregion of the neural crest expressed markers for the anteriorneural plate border, and the posterior limit of the neuralcrest moved to a more posterior position. It is interesting tonote that the region of the neural crest least affected by bothexpression of the dominant negative form of RAR andretinoic acid application was the anterior border. Thisobservation suggests that the different posteriorizing agentshave different roles in the specification of differentanterior–posterior positions in the neural plate border.Thus, retinoic acid may be more important in the specifi-cation of the posterior limit of the neural crest and theposterior limit of the anterior neural fold, while signalssuch as Wnts or FGF could play more prominent roles at theanterior limit of the neural crest cells. The use of adominant negative form of the FGF receptor completelyblocked the expression of neural crest markers, a findingwhich is concordant with a different role for FGF andretinoic acid in the posteriorization process of the neuralplate border (Mayor et al., 1997). A similar inhibitory effectwas also observed for the injection of a dominant negativemutant of Xwnt-8 in treated embryos (LaBonne andBronner-Fraser, 1998).

A remaining question was the spatial localization of theFGFs, Wnts, and retinoic acid signaling molecules, whichmust be present in the right place in order to function as aposteriorizing agents. At the late-gastrula and early-neurulastages, several members of the FGF and Wnt families have

been found to be expressed in the posterior region of the

© 2001 Elsevier Science. A

embryo (Isaac, 1997; McGrew et al., 1995; Christansen etal., 1991). Although it has been difficult to determine thedistribution of retinoic acid in the embryo, RARs andconverting enzymes have been found in the presumptivehindbrain and the posterior region of the embryo duringXenopus gastrulation (Kolm and Sive, 1997).

A Model of Neural Crest Induction

We present a unifying model of neural crest inductionbased on previous reports and on our observations describedhere (Fig. 9). At the early-gastrula stage, a gradient of BMPactivity is established in the ectoderm, which specifies theneural plate, the neural plate border, and the epidermis atprogressively higher concentrations of BMP (Fig. 9A). Theneural plate border, induced at a precise location within themediolateral axis of the ectoderm, has an anterior character(yellow in Fig. 9A). Later, between early and midgastrulastage, signals presumably originating from the ventrolateralmesoderm transform a region of the anterior neural plateborder into prospective neural crest cells (Fig. 9B). A role forthis mesoderm in neural crest induction has been shown(Bonstein et al., 1998; Marchant et al., 1998). The spread ofthese molecules from the mesoderm into the ectoderm (Fig.9B) consequently locates them only in large animal caps,explaining why the neural crest was not induced whensmall animal caps were used. These signals could corre-spond to Wnt8 and eFGF, as it is known that they areexpressed in the ventrolateral mesoderm, and could corre-spond to the lateralizing signals described by LaBonne andBronner-Fraser (1998). However, the neural crest is notspecified at this stage, as this does not occur until the end ofgastrulation (Mayor et al., 1995; Mancilla and Mayor, 1996).Thus, additional signals are required for the final inductionof the neural crest. Finally, as gastrulation proceeds, theventrolateral mesoderm becomes localized to the posteriorregion of the embryo, where it continues to produce Wnt8,eFGF, and possibly retinoic acid, as well as other, as yetunknown, posteriorizing agent that generate an anterior–posterior gradient of these morphogenes. This gradientwould be required for the final specification of the neuralcrest in the most posterior region of the neural plate border(Fig. 9C). Thus, the lateral–posterior regions of the neuralplate border receive the lateralizing/posteriorizing signalsfor an extended period of time, finally specifying them asneural crest. In contrast, the anterior neural plate borderdoes not receive such signals or these are inhibited by otheragents produced by the anterior regions of the embryo, suchas cerberus or dkk1, two known Wnts inhibitors (Bouw-meester et al., 1996; Glinka et al., 1998), and, as a conse-quence, does not develop as neural crest cells. It is temptingto speculate that the anterior–posterior differences withinthe neural crest could be controlled by a similar mecha-

nism.

ll rights reserved.

300 Villanueva et al.

ACKNOWLEDGMENTS

We are very grateful to Angela Nieto and Miguel Allende forcomments on the manuscript, B. Blumberg for the mutated formsof the retinoic acid receptor, to R. Moon for the Wnts reagents, toE. Amaya for XFD, to K. Cho for the dominant negative of the BMPreceptor, and to H. Sive, R. Grainger, and J. Smith for the Xag,Xsox-2, and Xbra clones. This investigation was supported bygrants from Fondecyt (1990570, 2010132), the Human FrontierScience Program (RG0042/98), and the Iniciativa Cientıfica Mile-nio (P99-137F).

REFERENCES

Amaya, E., Musci, T. J., and Kirschner, M. W. (1991). Expression ofa dominant negative mutant of the FGF receptor disrupt meso-derm formation in the Xenopus embryo. Cell 66, 257–270.

Bang, A., Papalopulu, N., Goulding, M., and Kintner, C. (1997).Expression of Pax-3 in the lateral neural plate is dependent on aWnt-mediated signal from posterior nonaxial mesoderm. Dev.Biol. 212, 366–380.

Bang, A. G., Papalopulu, N., Goulding, M. D., and Kintner, C.(1999). Expression of Pax-3 in the lateral neural plate is depen-dent on a Wnt-mediated signal from posterior nonaxial meso-derm. Dev. Biol. 212, 366–380.

Baranowitz, S. A. (1989). Regeneration, neural crest derivatives andretinoids: A new synthesis. J. Theor. Biol. 140, 231–242.

Barker, N., and Clevers, H. (2000). Catenins, Wnt signaling andcancer. BioEssays 22, 961–965.

Blitz, I. L., and Cho, K. W. (1995). Anterior neuroectoderm isprogressively induced during gastrulation: The role of the Xeno-pus homeobox gene orthodenticle. Development 121, 993–1004.

Blumberg, B., Wright, C. V., De Robertis, E. M., and Cho, K. W.(1991). Organizer-specific homeobox genes in Xenopus laevisembryos. Science 253, 194–196.

Blumberg, B., Bolodo, J., Moreno, T., Kintner, C., Evans, R., andPapalopulu, N. (1997). An essential role for retinoid signaling inanteroposterior neural patterning. Development 124, 373–379.

Bonstein, L., Elias, S., and Frank, D. (1998). Paraxial fated meso-derm is required for neural crest induction in Xenopus embryos.Dev. Biol. 193, 156–168.

Bouwmeester, T., Kim, S., Sasai, I., Lu, B., and De Robertis, E. M.(1996). Cerberus is a head-inducing secreted factor expressed inthe anterior endoderm of Spemann’s organizer. Nature 382,595–601.

Chang, C., and Hemmati-Brivanlou, A. (1998). Neural crest induc-tion by Xwnt7B in Xenopus. Dev. Biol. 194, 129–134.

Christian, J. L., McMahon, J. A., McMahon, A. P., and Moon, R. T.(1991). Xwnt-8, a Xenopus Wnt-1/int-1-related gene responsiveto mesoderm-inducing growth factors, may play a role in ventralmesodermal patterning during embryogenesis. Development111, 1045–1055.

Christiansen, J. H., Coles, E. G., and Wilkinson, D. G. (2000).Molecular control of neural crest formation, migration anddifferentiation. Curr. Opin. Cell Biol. 12, 719–724.

Cox, W. G., and Hemmati-Brivanlou, A. (1995). Caudalization ofneural fate by tissue recombination and bFGF. Development 121,4349–4358.

Curran, K. L., and Grainger, R. M. (2000). Expression of activated

MAP kinase in Xenopus laevis embryos: Evaluating the roles of

© 2001 Elsevier Science. A

FGF and other signaling pathways in early induction and pattern-ing. Dev. Biol. 228, 41–56.

Dickinson, M. E., Selleck, M. A. J., McMahon, A. P., and Bronner-Fraser, M. (1995). Dorsalization of the neural tube by the non-neural ectoderm. Development 121, 2099–2106.

Dupe, V., Ghyselinck, N. B., Wendling, O., Chambon, P., and Mark,M. (1999). Key roles of retinoic acid receptors alpha and beta inthe patterning of the caudal hindbrain, pharyngeal arches andotocyst in the mouse. Development 126, 5051–5059.

Essex, L. J., Mayor, R., and Sargent, M. G. (1993). Expression ofXenopus snail in mesoderm and prospective neural fold ecto-derm. Dev. Dyn. 198, 108–122.

Gamse, J., and Sive, H. (2000). Vertebrate anteroposterior pattern-ing: The Xenopus neurectoderm as a paradigm. BioEssays 22,976–986.

Glinka, A., Wu, W., Delius, H., Monaghan, A. P., Blumenstock, C.,and Niehrs, C. (1998). Dickkopf-1 is a member of a new family ofsecreted proteins and functions in head induction. Nature 391,357–362.

Harland, R. M. (1991). In situ hybridization: An improved whole-mount method for Xenopus embryos. Methods Cell Biol. 36,685–695.

Harland, R., and Gerhart, J. (1997). Formation and function ofSpemann’s organizer. Annu. Rev. Cell Dev. Biol. 13, 611–667.

Holowacz, T., and Sokol, S. (1999). FGF is required for posteriorneural patterning but not for neural induction. Dev. Biol. 205,296–308.

Hopwood, N. D., Pluck, A., and Gurdon, J. B. A. (1989). XenopusmRNA related to Drosophila twist is expressed in response toinduction in the mesoderm and the neural crest. Cell 59, 893–903.

Isaacs, H. (1997). New perspectives on the role of the fibroblastgrowth factor family in amphibian development. Cell. Mol. Life53, 350–361.

Kolm, P. J., and Sive, H. L. (1995). Regulation of the Xenopus labialhomeodomain genes, HoxA1 and HoxD1: Activation by retin-oids and peptide growth factors. Dev. Biol. 167, 34–49.

Kolm, P. J., and Sive, H. L. (1997). Retinoids and posterior neuralinduction: a reevaluation of Nieuwkoop’s two-step hypothesis.Cold Spring Harbor Symp. Quant. Biol. 62, 511–521.

LaBonne, C., and Bronner-Fraser, M. (1998). Neural crest inductionin Xenopus: Evidence for a two-signal model. Development 125,2403–2414.

LaBonne, C., and Bronner-Fraser, M. (1999). Molecular mecha-nisms of neural crest formation. Annu. Rev. Cell Dev. Biol. 15,81–112.

Lamb, T. M., and Harland, R. M. (1995). Fibroblast growth factor isa direct neural inducer, which combined with noggin generatesanterior–posterior neural pattern. Development 121, 3627–3636.

Launay, C., Fromentoux, V., Shi, D., and Boucaut, J. (1996). Atruncated FGF receptor blocks neural induction by endogenousXenopus inducers. Development 122, 869–880.

Liem, K. F., Tremmi, G., Roelink, H., and Jessell, T. (1995). Dorsaldifferentiation of neural plate cells induced by BMP-mediatedsignals from epidermal ectoderm. Cell 82, 969–979.

Linker, C., Bronner-Fraser, M., and Mayor, R. (2000). Relationshipbetween gene expression domains of Xsnail, Xslug, and Xtwistand cell movement in the prospective neural crest of Xenopus.Dev. Biol. 224, 215–225.

Mancilla, A., and Mayor, R. (1996). Neural crest formation inXenopus laevis: mechanism of Xslug induction. Dev. Biol. 177,

580–589.

ll rights reserved.

301Posteriorization of the Neural Crest

Marchant, L., Linker, C., Ruiz, P., Guerrero, N., and Mayor, R.(1998). The inductive properties of mesoderm suggest that theneural crest are specified by a BMP gradient. Dev. Biol. 198,319–329.

Mayor, R., Essex, L. J., Bennett, M. F., and Sargent, M. G. (1993).Distinct elements of the xsna promoter are required for meso-dermal and ectodermal expression. Development 119, 661–671.

Mayor, R., Morgan, R., and Sargent, M. G. (1995). Induction of theprospective neural crest of Xenopus. Development 121, 767–777.

Mayor, R., Guerrero, N., and Martinez, C. (1997). Role of FGF andnoggin in neural crest induction. Dev. Biol. 189, 1–12.

Mayor, R., Young, R., and Vargas, A. (1999). Development of neuralcrest in Xenopus. Curr. Top. Dev. Biol. 43, 85–113.

Mayor, R., Guerrero, N., Young, R. M., Gomez-Skarmeta, J. L., andCuellar, C. (2000). A novel function for the Xslug gene: Controlof dorsal mesendoderm development by repressing BMP-4. Mech.Dev. 97, 47–56.

Mayor, R., and Aybar, M. (2001). Induction and development ofneural crest in Xenopus laevis. Cell Tissue Res. 305, 203–209.

McGrew, L. L., Lai, C. J., and Moon, R. T. (1995). Specification ofthe antero-posterior neural axis through synergistic interactionof the Wnt signaling cascade with noggin and follistatin. Dev.Biol. 172, 337–342.

Molenaar, M., Van de Wetering, M., Oosterwegel, M., Peterson-Maduro, J., Godsave, S., Korinek, V., Roose, J., Destree, O., andClevers, H. (1996). XTcf-3 transcription factor mediates beta-catenin-induced axis formation in Xenopus embryos. Cell 86,391–399.

Morgan, R., and Sargent, M. G. (1997). The role in neural patterningof translation initiation factor eIF4AII; induction of neural foldgenes. Development 124, 2751–2760.

Moury, J. D., and Jacobson, A. G. (1989). Neural fold formation atnewly created boundaries between neural plate and epidermis inthe axolotl. Dev. Biol. 133, 44–57.

Moury, J. D., and Jacobson, A. G. (1990). The origins of neural crestcells in axolotl. Dev. Biol. 141, 243–253.

Neave, B., Holder, N., and Patient, R. (1997). A graded response toBMP4 spatially coordinates patterning of the mesoderm andectoderm in the zebrafish. Mech. Dev. 62, 183–195.

Nguyen, V. H., Schmid, B., Trout, J., Connors, S. A., Ekker, M., andMullins, M. C. (1998). Ventral and lateral regions of the zebrafishgastrula, including the neural crest progenitors, are establishedby a bmp2/swirl pathway of genes. Dev. Biol. 199, 103–110.

Nieto, M. A., Sargent, M. G., Wilkinson, D. G., and Cooke, J.(1994). Control of cell behavior during vertebrate developmentby Slug, a zinc finger gene. Science 264, 835–839.

Nieuwcoop, P. D., and Faber, J. (1967). “Normal Table of Xenopuslaevis.” 2nd ed. Daudin, North Holland, Amsterdam.

Nieuwcoop, P. D., and Albers, B. (1990). The role of competence inthe cranio-caudal segregation of the central nervous system. Dev.Growth Differ. 32, 23–31.

Papalopulu, N., and Kintner, C. (1996). A posteriorising factor,retinoic acid, reveals that anteroposterior patterning controls thetiming of neuronal differentiation in Xenopus neuroectoderm.Development 122, 3409–3418.

Raven, C. P., and Kloos, J. (1945). Induction by medial and lateralpieces of the archenteron roof, with special reference to thedetermination of the neural crest. Acta Neeland. Morphol.

Norm. Pathol. 55, 348–362.

© 2001 Elsevier Science. A

Ribisi, S., Jr., Mariani, F. V., Aamar, t. M., Frank, D., and Harland,R. M. (2000). Ras-mediated FGF signaling is required for theformation of posterior but not anterior neural tissue in Xenopuslaevis. Dev. Biol. 227, 183–196.

Richter, K., Grunz, H., and Dawid, I. B. (1988). Gene expression inthe embryonic nervous system of Xenopus laevis. Proc. Natl.Acad. Sci. USA 85, 8086–8090.

Saint-Jeannet, J. P., He, X., Varmus, H. E., and Dawid, I. B. (1997).Regulation of dorsal fate in the neuraxis by Wnt-1 and Wnt-3a.Proc. Natl. Acad. Sci. USA 94, 13713–13718.

Sasai, Y., Lu, B., Steinbeisser, H., and De Robertis, E. M. (1995).Regulation of neural induction by the Chd and Bmp-4 antagonis-tic patterning signals in Xenopus. Nature 376, 333–336.

Selleck, M. A., and Bronner-Fraser, M. (1995). Origins of the avianneural crest: The role of neural plate epidermal interactions.Development 121, 525–538.

Selleck, M. A., Gracıa-Castro, M. I., Artinger, K. B., and Bronner-Fraser, M. (1998). Effects of Shh and Noggin on neural crestformation demonstrate that BMP is required in the neural tubebut not ectoderm. Development 125, 4919–4930.

Simeone, A., Acampora, D., Arcioni, L., Bancinelli, E., and Mavilio,F. (1990). Sequential activation of Hox 2 genes by retinoic acid inhuman embryonal carcinoma cells. Nature 34, 763–766.

Sive, H. L., Hattori, K., and Weintraub, H. (1989). Progressivedetermination during formation of the anteroposterior axis inXenopus laevis. Cell 58, 171–180.

Slack, J. M. W. (1984). Regional byosinthetic markers in the earlyamphibiam embryo. J. Embryol. Exp. Morphol. 80, 289–319.

Slack, J. M., and Tannahill, D. (1992). Mechanism of anteroposte-rior axis specification in vertebrates. Lessons from the amphib-ians. Development 114, 285–302.

Smith, J. C., and Slack, J. M. W. (1983). Dorsalization and neuralinduction: Properties of the organizer in Xenopus laevis. J.Embryol. Exp. Morphol. 78, 299–317.

Smith, J. C., Price, B. M., Green, J. B., Weigel, D., and Herrmann,B. G. (1991). Expression of a Xenopus homolog of Brachyury (T) isan immediate-early response to mesoderm induction. Cell 67,79–87.

Streit, A., and Stern, C. (1999). Establishment and maintenance ofthe border of the neural plate in the chick: Involvement of FGFand BMP activity. Mech. Dev. 82, 51–66.

Thiesen, H. J., and Bach, C. (1993). DNA recognition of C2H2zinc-finger proteins. Evidence for a zinc-finger-specific DNArecognition code. Ann. NY Acad. Sci. 684, 246–249.

Weinstein, D. C., and Hemmati-Brivanlou, A. (1999). Neural in-duction. Annu. Rev. Cell Dev. Biol. 15, 411–433.

Whiting, J., Marshall, H., Cook, M., Krumlauf, R., Rigby, P. W. J.,Scott, D., and Alleman, R. K. (1991). Multiple spatially specificenhancers are required to reconstruct the pattern of Hox-2.6 geneexpression. Genes Dev. 5, 2048–2059.

Wilson, P. A., Lagna, G., Sizuki, A., and Hemmati-Brinvalou, A.(1997). Concentration-dependent patterning of the Xenopus ec-toderm by BMP4 and its signal transducer Smad1. Development124, 3177–3184.

Wodarz, A., and Nusse, R. (1998). Mechanisms of Wnt signaling indevelopment. Annu. Rev. Cell Dev. Biol. 14, 59–88.

Received for publication July 16, 2001Revised September 17, 2001

Accepted October 4, 2001

Published online December 13, 2001

ll rights reserved.