15
RESEARCH ARTICLE Migratory Patterns and Developmental Potential of Trunk Neural Crest Cells in the Axolotl Embryo Hans-Henning Epperlein, 1 * Mark A.J. Selleck, 2 Daniel Meulemans, 3 Levan Mchedlishvili, 4 Robert Cerny, 5 Lidia Sobkow, 4 and Marianne Bronner-Fraser 3 Using cell markers and grafting, we examined the timing of migration and developmental potential of trunk neural crest cells in axolotl. No obvious differences in pathway choice were noted for DiI-labeling at different lateral or medial positions of the trunk neural folds in neurulae, which contributed not only to neural crest but also to Rohon-Beard neurons. Labeling wild-type dorsal trunks at pre- and early-migratory stages revealed that individual neural crest cells migrate away from the neural tube along two main routes: first, dorsolaterally between the epidermis and somites and, later, ventromedially between the somites and neural tube/notochord. Dorsolaterally migrating crest primarily forms pigment cells, with those from anterior (but not mid or posterior) trunk neural folds also contributing glia and neurons to the lateral line. White mutants have impaired dorsolateral but normal ventromedial migration. At late migratory stages, most labeled cells move along the ventromedial pathway or into the dorsal fin. Contrasting with other anamniotes, axolotl has a minor neural crest contribution to the dorsal fin, most of which arises from the dermomyotome. Taken together, the results reveal stereotypic migration and differentiation of neural crest cells in axolotl that differ from other vertebrates in timing of entry onto the dorsolateral pathway and extent of contribution to some derivatives. Developmental Dynamics 236:389 – 403, 2007. © 2006 Wiley-Liss, Inc. Key words: DiI; fluorescent dextrans; GFP transgenic embryos; trunk neural crest; migration; dorsal fin; Rohon-Beard cells; lateral line; axolotl Accepted 31 October 2006 INTRODUCTION The neural crest is a transient migra- tory population of cells found in all vertebrate embryos (Knecht and Bronner-Fraser, 2002). These cells mi- grate extensively along defined path- ways and give rise to diverse deriva- tives, including neurons and glial cells of the peripheral nervous system, pig- ment and adrenomedullary cells, the craniofacial skeleton and, in lower vertebrates, mesenchyme of the dor- sal fin (Hall and Ho ¨rstadius, 1988; Le Douarin and Kalcheim, 1999). Neural crest cells arise in a prospective neu- ral fold area of the ectoderm at the border between prospective neural plate and epidermis (Moury and Ja- cobson, 1989; Selleck and Bronner- Fraser, 1995). Because the neural folds are contiguous with neural plate and epidermis (Moury and Jacobson (1990), it is not clear where they “end” and neural tissue or epidermis begin. 1 Department of Anatomy, TU Dresden, Dresden, Germany, 2 Department of Cell and Neurobiology, University of Southern California Keck School of Medicine, Los Angeles, California 3 Division of Biology and Beckman Institute, California Institute of Technology, Pasadena, California 4 Max-Planck-Institute of Molecular Cell Biology and Genetics, Dresden, Germany 5 Department of Zoology, Charles University Prague, Prague, Czech Republic Grant sponsor: DFG; Grant number: EP8/7-1; Grant sponsor: NIH; Grant number: NS051051; Grant sponsor: James H. Zumberge Research and Innovation Fund; Grant sponsor: Donald E. and Delia B. Baxter Foundation. *Correspondence to: Hans H. Epperlein, Department of Anatomy, TU Dresden, Fetscherstr. 74, 01307 Dresden, Germany. E-mail: [email protected] DOI 10.1002/dvdy.21039 Published online 19 December 2006 in Wiley InterScience (www.interscience.wiley.com). DEVELOPMENTAL DYNAMICS 236:389 – 403, 2007 © 2006 Wiley-Liss, Inc.

Migratory patterns and developmental potential of trunk ... · RESEARCH ARTICLE Migratory Patterns and Developmental Potential of Trunk Neural Crest Cells in the Axolotl Embryo Hans-Henning

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

RESEARCH ARTICLE

Migratory Patterns and DevelopmentalPotential of Trunk Neural Crest Cells in theAxolotl EmbryoHans-Henning Epperlein,1* Mark A.J. Selleck,2 Daniel Meulemans,3 Levan Mchedlishvili,4

Robert Cerny,5 Lidia Sobkow,4 and Marianne Bronner-Fraser3

Using cell markers and grafting, we examined the timing of migration and developmental potential of trunkneural crest cells in axolotl. No obvious differences in pathway choice were noted for DiI-labeling atdifferent lateral or medial positions of the trunk neural folds in neurulae, which contributed not only toneural crest but also to Rohon-Beard neurons. Labeling wild-type dorsal trunks at pre- and early-migratorystages revealed that individual neural crest cells migrate away from the neural tube along two main routes:first, dorsolaterally between the epidermis and somites and, later, ventromedially between the somites andneural tube/notochord. Dorsolaterally migrating crest primarily forms pigment cells, with those fromanterior (but not mid or posterior) trunk neural folds also contributing glia and neurons to the lateral line.White mutants have impaired dorsolateral but normal ventromedial migration. At late migratory stages,most labeled cells move along the ventromedial pathway or into the dorsal fin. Contrasting with otheranamniotes, axolotl has a minor neural crest contribution to the dorsal fin, most of which arises from thedermomyotome. Taken together, the results reveal stereotypic migration and differentiation of neural crestcells in axolotl that differ from other vertebrates in timing of entry onto the dorsolateral pathway andextent of contribution to some derivatives. Developmental Dynamics 236:389–403, 2007.© 2006 Wiley-Liss, Inc.

Key words: DiI; fluorescent dextrans; GFP transgenic embryos; trunk neural crest; migration; dorsal fin; Rohon-Beardcells; lateral line; axolotl

Accepted 31 October 2006

INTRODUCTIONThe neural crest is a transient migra-tory population of cells found in allvertebrate embryos (Knecht andBronner-Fraser, 2002). These cells mi-grate extensively along defined path-ways and give rise to diverse deriva-tives, including neurons and glial cells

of the peripheral nervous system, pig-ment and adrenomedullary cells, thecraniofacial skeleton and, in lowervertebrates, mesenchyme of the dor-sal fin (Hall and Horstadius, 1988; LeDouarin and Kalcheim, 1999). Neuralcrest cells arise in a prospective neu-ral fold area of the ectoderm at the

border between prospective neuralplate and epidermis (Moury and Ja-cobson, 1989; Selleck and Bronner-Fraser, 1995). Because the neuralfolds are contiguous with neural plateand epidermis (Moury and Jacobson(1990), it is not clear where they “end”and neural tissue or epidermis begin.

1Department of Anatomy, TU Dresden, Dresden, Germany,2Department of Cell and Neurobiology, University of Southern California Keck School of Medicine, Los Angeles, California3Division of Biology and Beckman Institute, California Institute of Technology, Pasadena, California4Max-Planck-Institute of Molecular Cell Biology and Genetics, Dresden, Germany5Department of Zoology, Charles University Prague, Prague, Czech RepublicGrant sponsor: DFG; Grant number: EP8/7-1; Grant sponsor: NIH; Grant number: NS051051; Grant sponsor: James H. Zumberge Researchand Innovation Fund; Grant sponsor: Donald E. and Delia B. Baxter Foundation.*Correspondence to: Hans H. Epperlein, Department of Anatomy, TU Dresden, Fetscherstr. 74, 01307 Dresden, Germany.E-mail: [email protected]

DOI 10.1002/dvdy.21039Published online 19 December 2006 in Wiley InterScience (www.interscience.wiley.com).

DEVELOPMENTAL DYNAMICS 236:389–403, 2007

© 2006 Wiley-Liss, Inc.

Furthermore, single cell lineage anal-ysis in several species suggests that asingle neural fold precursor can giverise not only to neural crest but also toneural tube and epidermal derivatives(Selleck and Bronner-Fraser, 1995).Accordingly, Chibon (1966) concludedthat Rohon-Beard neurons, missingafter neural fold ablation, were neuralcrest derived.

A major unresolved question in neu-ral crest development concerns the re-lationship between neural crest andneural tube with respect to the alloca-tion of different lineages and path-ways of migration. This question hasbeen extensively studied in the chickembryo, which is amenable to cultureand microsurgical manipulation incombination with numerous cell-marking techniques for following neu-ral crest migration (Le Douarin, 1969;Le Douarin and Kalcheim, 1999; Ser-bedzija et al., 1989; Elena de Bellardand Bronner-Fraser, 2005). The mi-gratory behavior of avian neural crestappears to be highly organized, suchthat the first cells to exit the neuraltube migrate along a ventromedialpathway through the somites, whilelater migrating cells are restricted to adorsolateral pathway between the epi-dermis and somites where they formpigment cells (Le Douarin and Teillet,1974; Guillory and Bronner-Fraser,1986; Erickson et al., 1992; Serbedzijaet al., 1989). In the chick, it has beenproposed that melanocytes are speci-fied before entering the dorsolateralmigratory route (Reedy et al., 1998)and that distinct neurogenic and mel-anogenic sublineages may diverge be-fore or soon after neural crest cell em-igration from the neural tube (Luo etal., 2003).

Despite the wealth of informationon avian neural crest migration, it re-mains unknown whether ordered mi-gratory behavior is common to all ver-tebrate embryos, or whether it isavian specific. Some differences in mi-gratory behavior from those observedin chick have been noted in amphibi-ans like Xenopus as well as in mam-mals like the mouse. In Xenopus, neu-ral crest cells including presumptivepigment cells migrate preferentiallyon the ventromedial route, betweenthe somite and neural tube/notochordand not laterally or between thesomites (Macmillan, 1976; Krotoski et

al., 1988; Collazo et al., 1993). In themouse, no timing differences exist be-tween entry onto the dorsolateral andventromedial pathways (Serbedzija etal., 1990). Thus, species-specific differ-ences exist in some aspects of neuralcrest migration. There is reason tothink that migratory behavior maydiffer between axolotl (Ambystomamexicanum) and chick as well. In theaxolotl, some premigratory neuralcrest cells, such as presumptive pig-ment cells, appear to be committed totheir fate prior to the onset of migra-tion. Therefore, it is possible that sub-sequent migration may follow differ-ent rules. It is unknown whether thepremigratory crest is entirely “mo-saic” and contains only unipotent cellsor a mixture of cells with differentranges of developmental potential.

Here, we address this issue bystudying the migratory patterns anddevelopmental potential of early dif-ferentiating trunk neural crest cells inthe axolotl such as pigment cells, Ro-hon-Beard cells, or components of thelateral line. As a first step, we haveperformed a series of cell-marking andgrafting studies using dark wild-typeand white mutant embryos, the latterhaving defective pigment cell migra-tion (Dalton, 1953; Keller et al., 1982;Lofberg et al., 1985). The results showthat neural crest cells follow both adorsolateral and a ventromedial mi-gratory route similar to other verte-brates. However, the time course isdifferent than in other anamniotes oramniotes, since the dorsolateral routeis taken first. We confirm that neuralcrest and Rohon Beard cells share acommon lineage within the neuralfolds and find that only few axolotlneural crest cells contribute to thedorsal fin, which also has a contribu-tion from the dermomyotome (Sobkowet al., 2006). Furthermore, we showthat anterior but not mid- or posteriortrunk neural folds contribute to thelateral line system. These results re-veal interesting similarities and dif-ferences in the migratory pathwaysand derivatives between axolotl neu-ral crest and that of other vertebrates.

RESULTSDiI labeling was performed at premi-gratory stages (15–19; 25), early (33),and late migratory (35) stages to ana-

lyze the migratory patterns and deriv-atives of trunk neural crest cellsemerging from the neural folds/tube(Fig. 1A–E). DiI injections were per-formed at different positions along themediolateral axis (Fig. 1A,F) prior tothe onset of neural crest migration.Along the rostrocaudal axis, anterior,middle, or posterior levels of the dor-somedial neural folds or neural tubewere injected (Fig. 1B–E;F). In thetrunk of early tailbud stage embryos,the neural crest can be morphologi-cally identified as a wedge of cells thatforms on the dorsal aspect of the neu-ral tube and expresses snail tran-scripts (Fig. 1G). Later, an elevatedneural crest string is present on top ofthe neural tube (Fig. 1H). Once neuralcrest cells migrate, they move alongtwo primary pathways: a dorsolateralpathway underneath the epidermisthat is primarily thought to give riseto pigmented derivatives (melano-phores and xanthophores) and a ven-tromedial pathway followed by pre-cursors to sensory, sympathetic,adrenomedullary, and glial cells. It issometimes difficult to discriminate be-tween laterally migrating neuralcrest–derived pigment precursors,epidermis, or neural tube on the onehand and between pigment precursorsand components of the lateral line onthe other. To best illustrate this point,we examined dark instead of whitemutant embryos since they havemany more laterally migrating cells(Figs. 1I–L). DiI-labeled cells can bedistinguished in transverse sectionsafter one anterior (Fig. 1I) or threemidtrunk injections (Fig. 1J) into thetrunk neural fold from epidermal orneural tube cells by means of theirposition (Fig. 1K). Pigment precursorsand components of the lateral linesuch as neurons (ganglionic cells),glial cells, and lateral line nerves canbe distinguished through the use ofspecific markers. We found that glialcells are present only in the vicinity ofthe lateral line primordium or nerve(Fig. 1L). They follow these structurescaudally from the posterior cranialneural crest situated at a level caudalto the otic placode (Northcutt et al.,1994; Schlosser, 2002). They do notmigrate laterally from mid- or caudaltrunk neural crest regions.

390 EPPERLEIN ET AL.

DiI Labeling at DifferentLateral or Medial Positionsof the Trunk Neural FoldsAt neurula stages, the exact positionof the prospective neural crest withinthe neural fold is not known. To ad-dress whether different aspects of theneural folds of neurulae differ in de-velopmental potential, neural foldswere labeled laterally or mediallyalong their anterior/posterior axis(Fig. 1F) prior to the onset of neuralcrest migration (stages 15–19; Figs.1A, 2A,D; Table 1). Embryos werefixed 2 or 4 days post-injection and the

Fig. 1.

Fig. 2. A–F: Lateral or medial fold injections (average: stage 16). Prior to the onset of neural crest migrationin dark (D/D) neurulae (stages 15–19), lateral (A) or medial (D) positions of trunk neural folds were labeled withthree focal injections of DiI. Embryos were fixed 2 days (B,E) and 4 days (C,F) after the injection. In eithercase, after 4 days many more labeled cells were observed on the lateral than on the medial route whereasthe distance traversed by neural crest cells was similar on both routes (see also Table 1). The pink numbersindicate DiI-labeled cells in the dorsal fin (df), the green numbers indicate sessile neural crest (nc) cells/derivatives on the neural tube, and the red and blue numbers indicate neural cress cells/derivatives on thelateral (lat) and medial (med) route, respectively.

Fig. 1. DiI labeling and identification of neuralcrest cells in the axolotl trunk. DiI labeling wascarried out at stages 15–19, 25, 33, and 35 toanalyze the migratory potential of trunk neuralcrest. A: At stage 16 (mean stage used in thisschematic to represent stages 15–19), three fo-cal DiI injections were made into the lateral ormedial aspects of the trunk neural folds (F).B: At stage 16, the left trunk neural fold wasinjected anteriorly or posteriorly in a middle po-sition (F). At stages 25 (C), 33 (D), and 35 (E),three positions along the dorsal midtrunk werelabeled. F: Red dots indicate positions of DiIlabeling in the left trunk neural fold at stage 16.G: In transverse sections through the trunk ofearly tailbuds (stage 23), the neural crest (arrow)can be morphologically identified as wedge ontop of the neural tube where it stains with thesnail riboprobe. H: In transverse sectionsthrough the trunk of later tailbuds (stage 34), anelevated neural crest string is present on top ofthe neural tube (DiI labeling of the neural crest,dapi staining, anti-fibronectin staining). I–L: DiIlabeling of neural fold in dark (D/D) neurulae andidentification of neural crest cells in larvae inrelation to epidermis and neural tube. I, J showleft trunks in larvae (stage 37); in I, one leftanterior trunk neural fold, and in J, three leftmidtrunk neural fold positions were injectedwith DiI. In I, labeled cells are scattered in theupper anterior trunk and the middle lateral lineprimordium is stained. The arrow marks theinjection site. In J, labeled cells are foundthroughout the lateral trunk, on top of the dorsalfin, and sparsely in the middle lateral line pri-mordium. Injection sites are not more visible. K:Transverse section through the midtrunk of thelarva in I counterstained with anti-fibronectin; incontrast to DiI-labeled epidermal cells (arrow),labeled neural crest cells (arrowheads) arefound more internal to the epidermal basementmembrane. L: Transverse section through themidtrunk of the larva in J counterstained withanti-fibronectin and dapi. Arrowheads point toDiI-labeled neural crest cells on the lateral andventromedial pathway. The arrow indicates themiddle lateral line primordium; the two labeledcells in its vicinity are very likely glial cells be-coming associated with the middle lateral linenerve. m, middle lateral line primordium; nc,neural crest; nt, neural tube; not, notochord.

TRUNK NEURAL CREST CELLS IN THE AXOLOTL EMBRYO 391

distribution of DiI-labeled cells deter-mined (stages 35 or 38; Fig. 2B,C,E,F;Table 1). Because DiI labeling of theneural folds marks epidermis andneural tube in addition to neuralcrest, neural crest cells were carefullydistinguished from epidermal andneural tube using morphological crite-ria (Fig. 1K).

Two days following lateral injec-tions into neural folds of wild type em-bryos, the leading edge of neural crestcells had migrated !450 "m awayfrom the injection site and !1,100 "mafter 4 days and was found equally faralong the dorsomedial and ventrolat-eral pathways (Fig. 2B,C; Table 1).Following medial injections, neuralcrest cells had migrated 600–650 "mafter 2 days and 1,150/1,050 "m after4 days (Fig. 2E,F; Table 1). Thus, no

major differences were noted for me-dial versus lateral neural fold injec-tions with respect to the distance tra-versed by neural crest cells away fromthe injection site. However, more mi-grating cells were observed along thedorsolateral pathway after 4 days(Fig. 2C,F).

DiI Labeling of DifferentRostrocaudal Axial LevelsUsing scanning electron microscopyand dopa-histochemistry, we havepreviously shown that delamination ofneural crest cells starts around stage28 in the anterior trunk of wild-typeembryos, while dorsolateral migrationof prospective pigment cells com-mences by stage 31. Initiation of mi-gration proceeds gradually along the

rostrocaudal axis such that by stage35, migrating pigment cells havespread throughout the flank. It waspreviously assumed that only pigmentcells migrate along the dorsolateralpathway (Lofberg et al. 1980; Epper-lein and Lofberg, 1990).

Here, we investigated whetherother neural crest populations mi-grated along the dorsolateral pathwayusing DiI labeling to mark the neuralfolds of white mutant embryos, whichhave a defect in pigment cell migra-tion and thus have only a few melano-phores that could obscure recognitionof migrating non-pigmented neuralcrest derivatives. A focal DiI injectionwas made into one dorsal middle areaof either the anterior or posteriortrunk neural fold on the left side of aneurula (Fig. 3AB). Previously, the

TABLE 1. Trunk Neural Crest Cell Migration In Axolotl Based on DiI Infectionsa

Groups Figure GenotypeStage atinjection N

Site/number ofinjections

Time/stageafterinjection

Neural crest cells

Migrateddistance("m)

df nc lat med lat med

1 2B D/D 18/19 5 Lateral fold/3 2d/35 01 02 09 10 450 4502 2C D/D 15 5 Lateral fold/3 4d/38 06 04 51 35 1100 10503 2E D/D 17/18 5 Medial fold/3 2d/34–35 02 00 16 12 600 6504 2F D/D 15 5 Medial fold/3 4d/38 09 02 58 24 1150 1050

14 3B d/d 16 5 Anterior fold/1 4d/38 02 04 91 22 800 120015 3C d/d 16 5 Posterior fold/1 4d/38 15 15 23 26 800 9505 6B D/D 25 5 Dorsal mid trunk/3 1d/32–33 00 14 05 00 300 06 6C D/D 25 5 Dorsal mid trunk/3 2d/36 00 14 55 07 700 4007 6E d/d 25 5 Dorsal mid trunk/3 1d/33 00 00 01 09 350 2008 6F d/d 25 5 Dorsal mid trunk/3 2-3d/36–37 05 03 14 28 450 6509 7B D/D 33/34 5 Dorsal mid trunk/3 1d/36 02 07 24 07 750 600

10 7C D/D 33/34 5 Dorsal mid trunk/3 2d/37 03 17 40 44 750 65011 7E d/d 33/34 5 Dorsal mid trunk/3 1d/36 00 13 03 02 200 20012 7F d/d 33/34 5 Dorsal mid trunk/3 2d/37 02 22 07 31 450 65013 8B D/D 35/36 5 Dorsal mid trunk/3 4d/39–40 24 17 22 79 700 110016 8C D/D 35/36 5 Dorsal mid trunk/3 4d/39–40 16 09 32 62 800 1050

4A D/D 16 5 Anterior fold/1 4d/384A D/D 16 10 Anterior fold/1 5d/395A d/d 16 10 Anterior fold/1 2d/325A d/d 16 5 Anterior fold/1 4d/385A d/d 16 5 Anterior fold/1 5d/394B d/d 16 12 Posterior fold/1 2d/324B d/d 16 15 Posterior fold/1 4d/384B d/d 16 15 Posterior fold/1 5d/39– D/D 16 5 Posterior fold/1 4d/38– D/D 16 10 Posterior fold/1 5d/39

aFigures 2,3,6–8 show a quantitative evaluation of DiI-labeled cells. Cell counting in each of 16 groups is based on approximately20 transverse sections (100-"m thick) through the trunk of embryos until stage 34 and 30 sections until stage 39/40. Five specimenswere sectioned per group. Labeled cells in sections through all specimens of a group were counted and mirror-imaged to the left sideof a summary section that represented the group (Figs. 2,3,6–8; Table 1). Figures 4A,B and 5A contain data on anterior or posteriortrunk neural fold injections with Dil, which were qualitatively evaluated. D/D, wild-type (dark); d/d, white mutant; df, dorsal fin;nc, neural crest; lat, lateral route; med, med at route; n, number of specimens investigated.

392 EPPERLEIN ET AL.

postotic posterior lateral line placodeand associated neural crest have beenproposed to contribute to the lateralline system in the trunk (Northcutt etal. 1994; Schlosser, 2002; Collazo etal., 1994).

Following anterior injections intowhite neurulae (Fig. 3A,B; stage 16,axial level of prospective 2nd somite),the majority of labeled cells migratedalong the dorsolateral rather than theventromedial pathway (e.g., 91 vs. 22cells for summing 5 representative an-terior injections; Table 1). After poste-rior injections (Fig. 3A,C, stage 16, atthe most posterior trunk neural foldlevel), dorsolateral migration wassimilar to ventromedial migration (23vs. 26 labeled cells summing 5 repre-sentative posterior injections; Table1), but lateral migration was fourfoldless than after anterior injections.

Interestingly, the front of dorsolat-erally migrating neural crest cells ap-pears to stagnate around the middlelateral line nerve regardless of the siteof injection. Following anterior injec-tions, there appears to be an addi-tional dorsolateral group of labeledcells at stage 38 (Fig. 3B) correspond-ing to the position of the dorsal lateralline, which by this stage is present inthe anterior but not yet in the poste-rior trunk (see fig. 16 in Northcutt etal., 1994). Because the number of dor-solateral melanophores in white lar-vae is !30 (see “Quantification of DiI-Labeling and Control” in theExperimental Procedures section), themuch higher number of labeled cellsobserved in white larvae for those in-jected anteriorly at stage 38 can onlybe ascribed to the presence of manymore non-melanophores. This raisesthe intriguing question if mid and pos-terior trunk neural fold cells mightcontribute glia or neurons to the lat-eral line.

Anterior But Not Mid orPosterior Trunk NeuralFolds Contribute to Neuronsand Glia of the Lateral LineIn fish and aquatic amphibians, thelateral line system forms part of theVIIIth cranial nerve complex. In thetrunk of axolotl larvae, three lines ofneuromasts develop, one dorsally, onemedially, and one ventrolaterally(Smith et al., 1990). All three are de-

rived from the lateral line placode pri-mordium situated posterior to the oticplacode (Northcutt et al., 1994), re-ceive afferent and efferent innerva-tion, and grow along invariant path-ways (Schlosser, 2002).

The primordium contains prospec-tive lateral line receptors (neuromastand ampullary primordia) and sen-sory neurons of lateral line gangliaand glia (Schlosser, 2002; Gilmour etal., 2002). By stage 37, the middle lat-eral line nerve extends to the tail re-gion, and by stage 41, the dorsal andmiddle lateral lines fuse above the clo-aca (Northcutt et al., 1994).

To address the source of neuralcrest cells that appear to cease migra-tion in the vicinity of the dorsal ormiddle lateral line nerve, focal injec-tions of DiI were applied into anterior(Fig. 4A) or posterior (Fig. 4B) neuralfolds of dark and white neurulae(stage 16; for exact level of injections,see Fig. 3A). This results in randomdistribution of labeled cells under-neath the epidermis of developing lar-vae. For anterior injections, labeledcells were occasionally found along themiddle lateral line (dark larva atstage 37; Fig. 4A) probably depositingthe neuromast primordia (Northcutt,personal communication). In contrast,DiI-labeled cells were not found alongthe dorsal or middle lateral linenerves after posterior injections(white larva at stage 41; Fig. 4B).These results suggest that only ante-rior trunk neural fold/neural crestcells contribute to the lateral line. Togain better cellular resolution of neu-ral crest contributions to the lateralline, we used a second labeling ap-proach of grafting trunk neural foldsfrom GFP-positive transgenic donorembryos into white hosts. Becausewhite mutants have few melano-phores compared with wild-type, pig-ment cells tend not to obscure the GFPsignal. For grafts of the posteriortrunk neural fold region, we observedonly GFP# pigment cells, leaving thegraft and no fluorescent cells migrat-ing toward or associating with lateralline nerves (data not shown), suggest-ing that the posterior trunk fails tocontribute to the lateral line.

Next, small mid trunk fragments orentire trunk neural folds from GFP#transgenic neurulae (stage 17) weregrafted in place of small left midtrunk

fragments or the entire left trunk neu-ral fold, from post-otic region to thetail, of white mutant embryos. Bystage 35/36, the primordium of themiddle lateral line was labeled inter-mittently (Fig. 4C). By stage 41, GFP-positive glia cells and neuromastswere observed intermittently alongthe middle nerve (not visible) on boththe ispilateral side bearing the graft(Fig. 4D) as well as the contralateralside (Fig. 4E). In addition, a few GFP-positive pigment and epidermal cellswere randomly distributed in the dor-solateral trunk (not shown). In somecases, only the right, unilateral neuralfold graft was used to replace the en-tire host since this often aided in vi-sualizing the GFP-labeled cells on theleft side better (no left epidermis la-beled). Using this approach, the leftdorsal, middle, and ventral lateralline nerves become clearly labeled inlarvae at stage 40/41 (Fig. 4F). GFP-positive neural crest cells were ob-served associated with all three lat-eral line nerves (Fig. 4G). However,only surrounding the ventral nervewas there a swarm of GFP-positivecells (Fig. 4G) whose origin from themid-trunk neural crest was question-able. In transverse sections, theyproved to be localized more deeply, inthe lateral plate mesoderm (Fig. 4H),and, thus, represent neural crest–de-rived enteric neurons/glial cells mi-grating on the ventromedial route andturning to the surface from under-neath the myotomes. These findingsconfirm our DiI labeling and suggestthat GFP-positive glial cells associ-ated with lateral line nerves at trun-cal levels arise from anterior neuralfolds from behind the otic placode.

Immunohistochemistry was used toexamine the phenotype of labeled neu-ral crest cells traveling along thenerves. Five days after an anterior in-jection when neuromasts are presentalong the dorsal and middle lateralline nerves, the nerves themselves areonly lightly labeled with DiI, appear-ing as a fine continuous line (whitelarva at stage 39; Fig. 5A,B). In con-trast, Schwann cells scattered alongthe nerves were brightly stained withDiI. Anti-tubulin staining of trans-verse sections through this larva re-vealed the nerves as two spots on ei-ther side of the dorsal trunk (Fig. 5C).Anti-tubulin staining was observed

TRUNK NEURAL CREST CELLS IN THE AXOLOTL EMBRYO 393

intermixed with DiI staining of puta-tive glial cells on the left side (Fig.5C). The glia character of these DiI-positive cells (Fig. 5A–C) was shownby anti-GFAP staining of microwave-treated paraffin sections through themidtrunk of stage 41 larvae (Fig. 5D,blue circle) adjacent to anti-tubulinstaining of the lateral line nerve (Fig.5E, condensed brown structure).

DiI-Labeling After NeuralTube Closure RevealsDifferences in Timing ofNeural Crest Entry OntoLateral Versus VentromedialRoutesWe next investigated the choice ofpathway and timing of migration of

Fig. 3. A–C: Anterior or posterior trunk neuralfold injections (stage 16). One focal DiI injectionwas made into the left trunk neural fold in amiddle position either at an anterior or posteriorposition of a white (d/d) neurula (stage 16) inorder to investigate the contribution of neuralcrest cells to the lateral line system. Followinganterior injections, labeled cells greatly favoredthe lateral pathway over the medial pathway (B).In white embryos injected posteriorly, manyfewer labeled cells migrated laterally (C). In ei-ther type of injection, the front of lateral migra-tion stagnated in the flank at the level of themiddle lateral line nerve (mll); following anteriorinjections, labeled cells seemed to be groupedalso around the dorsal lateral line nerve (dll).

Fig. 4.

Fig. 5.

394 EPPERLEIN ET AL.

neural crest cells by labeling the dor-sal neural tube after its closure butprior to the onset of neural crest emi-gration. Results were compared be-tween wild-type and white mutantembryos. DiI was injected at threesites along the middle of the trunkneural tube of stage 25 wild-type andwhite mutant embryos (Fig. 6A,D)and the results examined 1 or 2–3days later in fixed embryos (stages32–33 or 36–37). As in DiI labeling oflateral and medial trunk neural folds,we also carefully distinguished be-tween neural crest and epidermalcells when counting neural crest cellson the dorsolateral and ventromedialpathways.

In dark embryos, the front of neuralcrest cells on the lateral route hadmoved about 300 "m after 1 day (5cells) and 700 "m after 2–3 days (55cells; Figs. 6B,C; Table 1). On the ven-tromedial pathway, labeled cells wereabsent after 1 day but had movedmaximally 400 "m away from the in-jection site after 2 days (7 cells), sug-gesting that this pathway opens sec-ondarily (Fig. 2A,B; Table 1). In thewhite embryo, dorsolateral migrationwas reduced (1 cell after one day, 14cells after 2–3 days; Fig. 6E,F) com-

pared to the wild type. Labeled cells inthe mutant appeared stagnant intheir migration !450 "m distant fromthe injection site. Again, their site ofarrest corresponded to the position ofthe middle lateral line nerve.

Taken together, these results sug-gest that neural crest cells in wildtype embryos labeled at stage 25 mi-grate first on the lateral route wherethey are more numerous than thosemigrating ventromedially. In whitemutants, in contrast, lateral migra-tion is reduced whereas ventromedialmigration (stage 36–37, Fig. 6F) re-sembles that in the wild type (stage36, Fig. 6C; stage 37, Fig, 7C).

DiI-Labeling at Early andLate Migration StagesTo determine how long neural crestmigration persists on different migra-tory pathways, cell marking was car-ried out at progressively later stagesranging from early (stage 33; Fig.7A–F) to late migration (stage 35; Fig.8A–C). Each embryo received focal in-jections of DiI at three rostrocaudallocations on the dorsum of the neuraltube and was examined one day (stage

36), two days (stage 37), or four days(stage 40) post-injection.

In wild-type embryos labeled duringearly migratory stages (Fig. 7A),many more labeled cells were ob-served on the dorsolateral than theventromedial pathway one day afterinjection (Fig. 7B; Table 1) but num-bers were approximately equal by twodays after injection (Fig. 7C; Table 1).The front of dorsolaterally migratingcells was about 750 "m away from theinjection site at both 1 and 2 days.Ventromedial migration was about600 "m away from the injection siteafter 1 day and 650 "m after 2 days.In white mutant embryos (injectionsites Fig. 7D), few labeled cells werepresent on either route after 1 day(Fig. 7E; Table 1). After 2 days, a fewmore labeled cells were observed onthe dorsolateral pathway but ap-peared to stagnate in their migration(Fig. 7F; Table 1). Those on the ven-tromedial pathway, however, had in-creased considerably in number andtraveled 650 "m. Thus, in dark em-bryos dorsolateral cells migrate firstand are more frequent than those onthe ventromedial pathway similar toembryos injected at stage 25. In whitemutant embryos, dorsolateral migra-

Fig. 4. Contribution of neural crest cells to the lateral line. A: Left side of a dark axolotl larva at stage 37, 4 days after injection of DiI into an anteriorlocation of the left trunk neural fold. DiI likely labels the main trunk lateral line placode depositing the neuromast primordia for the middle lateral line(m). B: Left tail region of a white axolotl larva at stage 41, 5–6 days after injection of DiI into a posterior location of the left trunk neural fold. Neuromastsof the middle lateral line nerve are encircled. A few labeled cells spread laterally from the injection site but do not associate with the middle or dorsallateral line nerves. C: Left side of a white larva (stage 35/36) bearing a trunk neural fold graft from a GFP# transgenic neurula on its left side. The dorsalneural tube and covering epidermis are heavily labeled. The primordium of the middle lateral line nerve (m) migrates posteriorly and has reached a levelabove the cloaca. The intermittent labeling along the nerve is very likely due to glial cells migrating posteriorly. From the labeled neural tube, no labeledneural crest cells were observed to migrate laterally and associate with the middle lateral line nerve. D: Lateral aspect of left midtrunk of white larva(stage 41) whose entire left trunk neural fold was replaced with a GFP# fold. Elongated fluorescent glial cells (arrows) were present intermittently alongthe middle lateral line nerves of the ipsi- (D) and contralateral side (E). Arrowheads in E point to GFP# centers of neuromasts where possibly supportivecells are stained. From the GFP# grafted neural fold (dorsal, outside the image), no lateral migration of labeled neural crest cells was observed towardslateral line nerves. F: Left side of a white larva (stage 40/41) following removal of both trunk neural folds and replacement of the right fold with a GFP#fold. The dorsal and middle lateral line nerve and the advancing ventral nerve are labeled. G: Enlarged inset from F showing the enlarged middle (m)and ventral (v) nerve. The advancing tip of the ventral nerve is surrounded by a swarm of GFP# cells. Only those at the nerve or in close vicinity toit seem to be glial cells. GFP# cells further apart are very likely enteric neurons/glial cells that migrated into the lateral plate on the ventromedial route.H: Transverse section through larva in F,G in the plane indicated by a white vertical line in G; GFP# cells in the epidermis were observed only in theposition of the ventral lateral line (v). All other GFP# cells were lying in the lateral plate mesoderm (arrowheads) and had reached this position aftermigrating on the ventromedial pathway below the myotomes (my). Immunostaining with primary anti-fibronectin and secondary Cy3-conjugatedantibody to reveal basement membranes; dapi staining.

Fig. 5. Identity of lateral line nerves and glia. A: Left side of a white larva at stage 39, 4 days after injection of DiI into an anterior location of the lefttrunk neural fold. Irregular labeling of the dorsal (d) and middle (m) lateral line nerve with DiI. B: Enlarged inset from A showing neuromasts (arrows)dorsal to the middle lateral line nerve; irregular DiI-labeling along the nerve stains possibly neuromast primordia and glial cells. C: Immunostaining ofa transverse vibratome section through the midtrunk of the larva in A with a primary anti-tubulin antibody and a FITC-conjugated secondary antibodyreveals two green fluorescent subepidermal spots on either side of the dorsal trunk that label the dorsal and midde lateral line nerve. Green spots onthe left side colocalize with DiI-labeled glial cells. D: Immunostaining of transverse microwave-treated paraffin sections through the trunk of a larva(stage 41) with a primary anti-GFAP-antibody, a secondary biotinylated antibody, and a tertiary avidin-peroxidase complex (Vectastain ABC-Kit). Glia(arrow; blue color) around the middle lateral line “nerve” and in the neural tube (nt) reacted positive and (E) colocalized with the middle lateral line nerve(arrow, brown color) dorsal to a blood vessel (v) in a similar transverse microwave-treated paraffin section (stage 41 larva) that reacted positive witha primary anti-tubulin antibody, a secondary biotinylated antibody, and a tertiary avidin-peroxidase complex (Vectastain ABC-Kit).

TRUNK NEURAL CREST CELLS IN THE AXOLOTL EMBRYO 395

tion appears to be compromised com-pared with that in wild-type counter-parts.

To examine late stages of neuralcrest migration, wild type embryoswere injected as above, but at stage 35

(Fig. 8A) and fixed 4 days post-injec-tion (stage 40). As late as stage 40, DiIlabeling resulted in many labeled mi-gratory neural crest cells (Fig. 8B,C).The majority of labeled cells movedalong the ventromedial pathway or

into the dorsal fin with some migra-tion observed along the lateral path-way.

It is difficult to perform DiI labelingof the trunk neural tube much laterthan stage 35 due to the impenetrablenature of the epidermis. Based on lar-vae injected at stage 35 (Fig. 8A) andinvestigated 4 days later (stage 40), itis likely that neural crest cells are stillemigrating from the neural tube afterstage 35 (Fig. 8B,C; Table 1).

Pigment cells (melanophores, xan-thophores) start to visibly differenti-ate under the epidermis by stage 35.By stages 38–40 neurons and glialcells are condensing into dorsal rootganglia at the ventrolateral margin ofthe neural tube as revealed in trans-verse sections stained with anti-tubu-lin and anti-GFAP (not shown), re-spectively. In 14-mm-long larvae,tyrosine hydroxylase (TH)-positivechromaffin cells start to appear, preced-ing TH# sympathetic neurons that dif-ferentiate in larvae $24 mm (notshown; see also Vogel and Model, 1977).

Fig. 6. A–F: Dorsal mid trunk injections (stage 25, premigratory stage). Three separate focalinjections of DiI were applied along the mid trunk neural tube/premigratory neural crest of wildtype(D/D;6A) and white mutant (d/d) embryos (6D) at stage 25. Neural crest cells in the wild typemigrate first on the lateral route where they are more numerous than on the medial pathway. In thewhite mutant, lateral migration remains strongly reduced. The front of cells coincides with thepositon of the middle lateral line nerve. Medial migration dominates and resembles that in the wildtype (compare F with C).

Fig. 7. A–F: Dorsal mid trunk injections (stage 33, early migration stage). Three separate focalinjections of DiI were carried out along the mid trunk neural tube/early migratory neural crest of wildtype (D/D; A) and white mutant (d/d; D) embryos at stage 33. In wild type embryos one day after theinjections, many more labeled cells were observed on the lateral than on the medial pathway (B).By 2 days (C), numbers were approximately equal. In white embryos one day after the injection, fewlabeled cells were found on both routes (E). By 2 days, many more labeled cells were found on theventromedial route but their number stagnated laterally (F).

Fig. 8. A–C: Dorsal mid trunk injections (stage35, late migration stage). Wild-type (D/D) em-bryos (two groups, B and C, with 5 individuals ineach group) were injected with DiI into the dor-sal neural tube/neural crest at late migration(stage 35) and fixed 4 days post-injection (stage40). At stage 40, ventromedial migration andmigration into the dorsal fin dominate.

396 EPPERLEIN ET AL.

Some Neural Crest CellsMigrate Contralaterally

In birds and frogs, some neural crestcells cross the midline to migrate onthe contralateral side (Hall and Horst-adius, 1988; Couly et al., 1996; Le-Douarin and Kalcheim, 1999). In axo-lotl, this possibility was firstinvestigated by Horstadius and Sell-man (1946) using vital dyes. To extendthis work with higher resolution label-ing methods, we performed unilateralgrafts of left neural folds that hadbeen injected with fluorescent dex-trans at the two-cell stage, thus label-ing all neural fold cells in the graft.

Prior to the onset of trunk neuralcrest migration (stage 30/31), fluores-cent cells derived from a rhodaminedextran–labeled neural fold were con-fined to the epidermis and neural tubeon the grafted side and to a premigra-tory neural crest on top of the neuraltube, which also contained unlabeledcells presumably from the contralat-eral neural fold (Fig. 9A). The neuralcrest was surrounded by basementmembranes as revealed by anti-fi-bronectin staining (Fig. 9A). At earlymigratory stages (stages 32–34), fluo-rescent neural crest cells from a left,FITC-labeled neural fold entered theventromedial pathways on both theipsilateral and contralateral sides(Figs. 9B,C). When the left trunk neu-ral fold contains a FITC- and the righta rhodamine dextran graft (Fig. 9D),contralateral migration of neuralcrest cells from each fold was ob-served. We show rhodamine dextran–labeled neural crest cells on the lat-eral and ventromedial pathway of thecontralateral side (Fig. 9D).

In horizontal sections (Fig. 9F)through the dorsal trunk of an embryoat stage 35 bearing a FITC-dextrangraft on its left side (Fig. 9E), singlelabeled cells were clearly identified onthe dorsolateral and ventromedialpathway on both sides of the embryo.Interestingly, those cells moving ven-tromedially were observed within thenarrow sclerotome (Fig. 9B), althoughno labeled neural crest cells were ob-served migrating through the myo-tome as proposed by Vogel and Model(1977). These results show that neuralcrest cells can migrate to the con-

tralateral side of the embryo and thenfollow analogous migratory pathways.

Both Neural Crest andRohon-Beard NeuronsOriginate From Neural FoldsRohon-Beard cells are transient sen-sory neurons in the dorsal trunk neu-ral tube of fish and amphibian em-bryos (Hall and Horstadius, 1988; LeDouarin and Kalcheim, 1999). Theyform a primary sensory system thatdegenerates when dorsal root gangliadevelop (Hughes, 1957; Roberts andPatton, 1985) and persist in axolotl atleast through larval stages (DuShane,1938).

Rohon-Beard cells were identifiedusing a combination of anti-tubulinstaining and DiI injections in sectionsthrough the trunk of embryonic (4mm) to larval (17 mm) stage axolotls(Fig. 10A). In larvae, Rohon-Beardcells appeared as large neurons at thedorsal-most aspect of the spinal cord(Fig. 10B), in contrast to interneuronsthat are localized more laterally (Rob-erts, 2000). Both cell types were about2–3 times larger than neighboringcells of the spinal cord. In youngerlarvae (9 mm; stage 39/40), severallarge cells with a rectangular orrounded shape were observed; thesecan be distinguished from Rohon-Beard cells because they reach fromthe ependymal layer to the outer sur-face of the tube and have nuclei atdifferent levels from the center (Fig.10C,D; see Sauer, 1935). Thus, theyare likely to be mitotic neuronal pre-cursors.

To examine the origin of Rohon-Beard cells in the axolotl, DiI wasinjected into the left trunk neuralfold of neurulae (stage 16), similar tothose injections performed to exam-ine neural crest migratory patterns.Embryos were subsequently fixed atembryonic stage 33–34. In horizon-tal sections through the dorsal neu-ral tube, several giant cell bodies atthe left margin of the neural tubewere labeled with DiI and appearedseveral times larger than adjacentneurons. Furthermore, they ex-tended nerve fibers laterally into theintersomitic clefts (Fig. 10E,F), char-acteristic of Rohon-Beard neurons(for details see Hughes, 1957; Taylorand Roberts, 1986; Roberts, 2000).

The same injections also gave rise toneural crest derivatives and epider-mis. This result suggests that theRohon-Beard cells, epidermis, andneural crest cells share a commonorigin from the dorsal neural folds(Fig. 10E,F).

DISCUSSION

Neural Crest Cells FromDifferent Sites Within theNeural Fold HaveEquivalent Developmentaland Migratory Potential

In amphibian embryos, it has beenproposed that different neural crestderivatives may arise from differentlateral or medial positions within theneural folds. Testing this proposal inaxolotl, Brun (1985) had shown thatmelanophores differentiate from themedial cranial neural fold (“L.N.F.” inBrun, 1985) but not from lateral epi-dermis adjacent to lateral cranial neu-ral fold. Furthermore Northcutt et al.(1996) had demonstrated in axolotlthat the medial wall of (cranial) neu-ral folds forms neural crest and tubederivatives, whereas the lateral walldifferentiates into dorsal and dorso-lateral ectoderm including neurogenicplacodes. The latter authors empha-sized that “the neural folds . . . .mustbe qualified by division into medialand lateral fields which each give riseto multiple lineages” (Northcutt et al.,1996). It also has been reported thatneural plate adjacent to neural fold inaxolotl gives rise to melanophoreswhereas epidermis adjacent to foldcontributes to spinal and cranial gan-glia (Moury and Jacobson, 1990).

Because previous studies did not de-finitively address which neural crestpopulations differentiate from posi-tions within medial or lateral neuralfold positions, we re-examined this is-sue in the axolotl trunk. However, wefailed to find any difference in thepathways chosen, distances traversed,or derivatives produced by neuralcrest cells derived from either medialor lateral trunk neural folds. Thus,there was no evidence for obvious me-diolateral differences or of sublin-eages within the premigratory neuralcrest in the trunk neural folds.

TRUNK NEURAL CREST CELLS IN THE AXOLOTL EMBRYO 397

Fig. 9.

Fig. 10.

Fig. 9. Migration of trunk neural crest to thecontralateral side. The left trunk neural fold ofwild-type (D/-) hosts (stages 15–17) was re-placed orthotopically by a neural fold from darkdonors (same age) labeled with FITC or rhoda-mine dextran. A: Dorsal part of transverse sec-tion through the trunk of a stage 30/31 embryoto demonstrate the distribution of labeled tis-sues that were derived from the left, graftedrhodamine labeled neural fold before the onsetof neural crest migration. Labeled cells from theleft neural fold graft were observed in the leftepidermis (epi) and left neural tube (nt). Theneural crest (nc, arrow) cannot be divided intoleft or right portions; it is partially labeled andcontains also unlabeled cells from the right fold.B–D: Transverse sections through different em-bryonic trunks at early migratory stages of theneural crest (stages 30/31–34). Embryos con-tain either one FITC-dextran labeled neural foldon their left side (B,C) or one FITC-dextran- onthe left and one rhodamine-dextran labeled foldon their right side (D). Sections were immuno-stained with a primary polyclonal anti-fibronec-tin antibody visualized by a FITC-conjugated (A)or a Cy3-conjugated secondary antibody (B,C).Fluorescent neural crest cells entered ventro-medial (arrowheads) and lateral pathways (ar-rows) on the ipsi- and contralateral side. In B,fluorescent neural crest cells on the left medialpathway pass through the sclerotome. E: Leftside of a larva (stage 35) bearing a FITC-dex-tran-labeled trunk neural fold graft on the leftside. F: Horizontal section through larva shownin E at a middle neural tube level. Single labeledcells are visible on the medial (arrowheads) andlateral pathway (arrows) of both the ipsi- andcontralateral side. B,C,F: counterstaining withdapi.

Fig. 10. Origin and distribution of Rohon-Beardneurons. Reinvestigation of Rohon-Beard cellson sections through the axolotl trunk from em-bryos (stage 33) to larvae (17 mm length).A: Schematic showing planes of sectioning inan axolotl embryo at stage 34. B: In transversesections through the trunk of larvae (12 mm),large neurons (arrowheads) were observed atthe dorsal-most aspect of the spinal cord fol-lowing immunostaining with anti-tubulin, whichfulfills the criteria of Rohon-Beard cells.C,D: Horizontal section through the dorsal neu-ral tube of a younger larva (stage 39/40). C:Immunostaining with primary anti-tubulin andsecondary Cy3-conjugated antibody followedby dapi staining. D: Same section uncolorized.Large cells with a rectangular (arrow) orrounded shape (arrowhead) are mitotic precur-sors and not Rohon-Beard cells because theyare continuous from the ependymal layer to theouter surface of the neural tube and have theirnuclei at different distances from the centrallumen. E,F: Horizontal section (E, DiI, and dapi;F, laser scanning micrograph) through the dor-sal neural tube of an embryo (stage 33–34),which had received DiI injections into the lefttrunk neural fold. Three giant labeled cell bodiesat the left margin of the neural tube extendnerve fibres laterally into the intersomitic clefts.They are likely Rohon Beard cells. nt, neuraltube; som, somite; nf, nerve fibre.

398 EPPERLEIN ET AL.

Anterior Trunk Neural FoldGives Rise to More LaterallyMigrating Cells ThanPosterior FoldIn dorsal trunks of white mutant em-bryos injected with DiI at stage 25, aquantitative investigation showedthat labeled cells were distributed onthe lateral migratory route at stage36–37 and stopped in an area of themiddle lateral line nerve (Fig. 6F).These DiI-labeled cells (14 per 5 indi-viduals) formed only a fraction (about10%) of all lateral melanophorescounted in white controls (!30/indi-vidual); however, it was unclearwhether their arrest at the lateral linenerve was coincidental or reflected acontribution to derivatives of the lat-eral line system. To distinguish be-tween these possibilities, we injectedDiI focally into either anterior or pos-terior positions of the trunk neuralfold (stage 16). The number of labeledcells on the lateral route was fourfoldhigher for anterior than posterior in-jections. While anterior injections con-tributed labeled cells to the lateralline primordium (Northcutt et al.,1994), posterior injections failed to doso. Thus, labeled cells migrating froman injected anterior neural fold maycontain neuronal (ganglionic) andglial precursors for the lateral line inaddition to few pigment cells.

Further evidence for this assump-tion was corroborated by a qualitativecharacterization of the labeled cellsdistributed, either after injecting DiIinto anterior or posterior trunk neuralfold as before (Figs. 4A, 5A) or bygrafting GFP transgenic neural foldsinto white hosts (Fig. 4C–H). The lat-ter experiments, in particular, left nodoubt that midtrunk neural crest doesnot supply glial and neuronal cells forthe lateral line. The results confirmedthe view that these cells appear toarise from anterior trunk neural foldlevels behind the otic placode (North-cutt et al., 1994), which migrate pos-teriorly together with or in the vicinityof lateral line primordia/lateral linenerves.

Neural Crest Migrates FirstLaterally and Then MediallyDiI labeling of premigratory stages(stage 25) and early migratory stages

(stage 33) of the trunk neural crest inaxolotl revealed that neural crest cellsin wild type embryos migrate first dor-solaterally and then ventromediallyand are more numerous on the formerroute. In contrast, dorsolateral migra-tion is restricted in the white mutantwhereas ventromedial migration (Fig.6F) resembles that in the wild type(Figs. 6C, 7C). At late-migratorystages (stage 35), there was still mod-erate migration of neural crest cellsalong the dorsolateral pathway, butmost labeled cells moved along theventromedial pathway or into the dor-sal fin.

Our results contrast with previousfindings by Vogel and Model (1977),who, based on (3H)thymidine-labelingof neural folds, suggested that ventro-medial migration of presumptive neu-rons and glial cells at stage 34 hadadvanced similar to pigment cell mi-gration along the dorsolateral path-way (staging according to Schrecken-berg and Jacobson, 1975; stage 34 isequivalent to stage 32 in the table ofBordzilovskaya et al., 1989; see Ep-perlein and Junginger, 1982, for com-parison). Similarly, we found no evi-dence for migration of neural crestcells through the somite (stage 30) asVogel and Model (1977).

The early migration of axolotl neu-ral crest along the dorsolateral path-way contrasts with that in severalother amniotes/anamniotes. Dorsolat-erally, neural crest migration is about300–400 "m advanced compared tomedial migration. Dorsolateral cellsneed about 1 day to cover such a dis-tance (Fig. 6B,C). One possible expla-nation for this earlier lateral migra-tion is that presumptive pigment cellsmay be committed to their fate prior toinitiation of migration (Epperlein andLofberg, 1984) and that this prespeci-fication is essential for migrating dor-solaterally. Thus, pigment cells in ax-olotl might follow the “phenotype-directed model” for neural crestmigration as proposed for pigment cellprecursors in the chick (Erickson andReedy, 1998). However, results byWakamatsu et al. (1998) argueagainst such a general prespecifica-tion of the neural crest. Chicken neu-ral crest cells migrating dorsolaterallyappear initially to include neuronalcells in addition to melanocytes andthese are later eliminated by apopto-

sis. Similarly, we observed glial cellsand neurons migrating dorsolaterallyat least from anterior trunk neuralfolds in the axolotl.

The migration pathways followedby neural crest cells in axolotl are sur-prisingly similar to those previouslydescribed in the chick. For example,axolotl neural crest cells migrate lat-erally between the epidermis and anarrow dermomyotome (referred to as“dermatome” in Sobkow et al., 2006;Epperlein et al., unpublished data),comparable to the dorsolateral migra-tion between the epidermis and der-momyotome observed in chick (Erick-son et al., 1992). Also as in chick(Bronner-Fraser, 1986), medial neuralcrest cells in axolotl migrate through asclerotome (Fig. 9B), albeit very nar-row.

It is uncertain whether an earlierlateral migration of pigment cells isan evolutionary advantage for axolotlembryos. Pigment cells might providethe larvae with disguise. On the otherhand, Rohon-Beard cells constitute aprimary sensory system in axolotl,perhaps making formation of deriva-tives along the ventromedial pathwayless crucial at early times in develop-ment.

Dual Neural Crest andSomite Origin for theMesenchyme of the DorsalFinClassically, the mesenchyme of thedorsal fin in lower vertebrates hasbeen assumed to be neural crest de-rived. The dorsal fin epidermis is in-duced by the neural crest and removalof the neural folds leads to the absenceof the dorsal fin (Raven, 1931; DuSh-ane, 1935; Bodenstein, 1952). Follow-ing injections of DiI into dorsal midtrunks of axolotl tailbuds at stage 25(Fig. 6A–F) and 33 (Fig. 7A–F), weobserved only a few labeled cells in thedeveloping dorsal fin after 2 days(stages 36–37). When DiI was injectedinto larvae at stage 35 (Fig. 8A–C), thedorsal fin mesenchyme was abun-dantly labeled after 4 days (stage 40).Although DiI injections hit only a frac-tion of all neural crest cells present ata certain stage, we found the numberof labeled mesenchymal cells 2 daysafter an injection (stages 36–37) sur-prisingly small.

TRUNK NEURAL CREST CELLS IN THE AXOLOTL EMBRYO 399

By grafting single GFP# labeledsomites into unlabeled hosts, we re-cently found that the dermomyotomeparticipates in early stages of mesen-chyme formation of the dorsal fin(Sobkow et al., 2006). Thus, in axolotlneural crest appears to make only alimited contribution to the mesen-chyme of the developing dorsal fin atearly larval stages; later its contribu-tion appears to increase as shown byDiI labeling of dorsal trunks at stage35 (Fig. 8B,C).

EXPERIMENTALPROCEDURES

Embryos

Wild-type (dark, D/-) and white mu-tant (dd) embryos of the Mexican ax-olotl (Ambystoma mexicanum) wereobtained from the axolotl colony inBloomington, Indiana, or from ourown colonies in Dresden. GFP em-bryos were spawned from a %-actinpromoter-driven GFP germline trans-genic animal that had been producedby plasmid injection (Sobkow et al.,2006). Embryos were kept in tap wa-ter at room temperature or at 7–8°Cand were staged according to the nor-mal table of Bordzilovskaya et al.(1989). Before being used for grafting,embryos were washed thoroughlywith tap water and sterile Steinbergsolution (Steinberg, 1957) containingantibiotics (Antibiotic-Antimycotic,Gibco) and then decapsulated me-chanically.

DiI-Injections

CellTracker CM-DiI (C-7000; Molecu-lar Probes, Eugene, OR) was dissolvedin absolute ethanol to a concentrationof 1 mg/ml and further diluted in 4 or9 parts of 10% sucrose in water justbefore injection. For injection, eitherglass micropipettes were used thatwere backfilled with DiI solution andattached to a Parker Hannifin Corpo-ration Picospritzer II assembly orglass capillaries that were connectedto a IM 300 injector (Narshige). DiIwas injected into embryos as indicatedbelow and monitored with differentstereo and epifluorescence micro-scopes.

Trunk neural fold/neural tube.

DiI injections were made into threedifferent rostrocaudal locations of theleft lateral or medial trunk neural foldof dark neurulae (stages 15–19) to de-termine whether regional differencesexist in migratory potential and cellfate at different mediolateral posi-tions within the neural folds (Figs. 1,2A–F; Table 1). Injected embryos werefixed 2 or 4 days post-injection andvibratomesectioned (transverse sec-tions, see below).

DiI was injected into one middle an-terior or posterior location of the lefttrunk neural fold of dark and whiteneurulae (stage 16) to examinewhether and in which way trunk neu-ral crest in these sites contributes tolateral line nerves (Figs. 1, 3A–C, 4, 5;Table 1). The anterior injection sitewas localized at the level of the pro-spective 2nd somite, the posterior onein the most posterior positon of the leftfold at the level of still unsegmentedmesoderm. White injected embryoswere fixed 2, 4, and 5 days post-injec-tion, dark ones after 4 and 5 days andmostly investigated as whole mounts.Some embryos fixed 4 days after theinjection were vibratomesectioned(transverse sections, see below). If po-sitions or identity of labeled cells wereunclear, sections were counterstainedwith anti-fibronectin to visualize tis-sue borders.

DiI injections were made into threedifferent rostrocaudal locations of thedorsal midtrunk neural tube of darkand white embryos at stages 25, 33,and 35 to examine whether there weredifferent migratory potentials oftrunk neural crest cells (Figs. 6A–F,7A–F, 8A–C; Table 1). Embryos werefixed at different times post-injectionand vibratomesectioned (transversesections, see below).

Quantification of DiI-labelingand controls.

Quantification. For a quantitativeevaluation of the distribution of DiI-labeled neural crest cells, approxi-mately 20 transverse sections (100 "mthick) were cut through the trunk ofDiI-labeled embryos until stage 34and 30 until stage 39/40. Five speci-mens were sectioned per group. Thelabeled cells in sections through allspecimens of a group were counted

and mirror-imaged to the left side of asummary section that represents thegroup (Figs. 2,3,6–8, Table 1).

Controls. The distribution of mela-nophores in whole white larvae wasdetermined between gills and cloacaat stages 36 and 42 in order to obtaina measure that would allow compari-son with the migration of Di-labeleldcells in sectioned white embryos. Atstage 36, six larvae were used andsubjected to Dopa treatment (Epper-lein and Lofberg, 1984) in order to en-hance the visibility of melanin pig-ment in melanophores. At stage 42when melanophores are clearly visi-ble, we used 10 larvae. In both stages,we found a mean value of !30 mela-nophores in the dorsolateral trunk.

FITC- and RhodamineDextran InjectionsDark axolotl embryos at the two- tofour-cell stage were kept at room tem-perature in a deepening of an agardish filled with 5% ficoll in 1/10 Stein-berg solution containing antibiotics.One blastomere was injected withFITC- or rhodamine dextran (Molecu-lar Probes; 50 mg/ml Steinberg solu-tion; 50 nl/injection) through the eggmembrane using an IM 300 injector(Narshige). After one day, the em-bryos were transferred into normalstrength saline. At the neurula stageparts of fluorescent trunk neural foldswere used for grafting.

Axolotl Embryos ExpressingGreen Fluorescent Protein(GFP)White axolotl embryos transgenic forGFP were produced by mating a maleGFP transgenic with a normal whitefemale (Sobkow et al., 2006). %-actinpromoter-driven GFP expression firstbecame clearly visible in the neurulastage. Trunk neural fold material wasgrafted into white donors in order toinvestigate a possible contribution ofneural crest derivatives to the lateralline system (see below).

Neural Fold GraftingExperimentsHead and trunk neural fold materialpre- or postlabeled in different ways

400 EPPERLEIN ET AL.

(see below) from sterile neurulae(stages 15–17) was grafted isochroni-cally into orthotopic locations of darkor white hosts. Before the operations,host neurulae were placed into a deep-ening of an agar dish filled with coldSteinberg saline. Grafting was per-formed under sterile conditions usingtungsten needles.

Whole left trunk neural folds weretransplanted orthotopically from darkFITC- dextran (n&5) or rhodaminedextran (n&2) labeled neurulae intodark unlabeled hosts or from whiteGFP# transgenic neurulae (n&6) intowhite unlabeled hosts. In additionsmall left midtrunk (n&5) or posteriortrunk neural fold fragments (n&5)were grafted orthotopically fromwhite GFP# donors into white hosts.In some cases, both trunk neural foldsof dark embryos were removed andthe left fold replaced with a FITC- andthe right one with a rhodamine dex-tran–labeleld neural fold fragment.Finally, both trunk neural folds ofwhite hosts were removed and onlythe right fold replaced with a GFP#fold (n&10). Donor and hosts were atstages 15–17. After 2–4 days, migra-tion of labeled trunk neural crest cellsthrough somites (FITC-label) or con-tribution to the lateral line (GFP#transgenic) was analyzed.

Fixation, Embedding, andSectioningEmbryos were fixed in 4% paraformal-dehyde (PFA) in PBS overnight andkept either in fixative for a later use inhistology or were transferred directlyinto 100% methanol and kept there at'20°C before use in in situ hybridiza-tion. Some individuals were examinedin an epifluorescence microscope forthe distribution of DiI- or GFP-labeledcells. For histology, specimens werewashed in PBS for 1 h, sectioned witha vibratome (vibratome series 1000sectioning system, Ted Pella, Inc.) intransverse or horizontal planes at 100"m thickness, stained as indicated be-low, and examined with an epifluores-cence microscope. Some PFA-fixedembryos were embedded into paraffinfor microwave-treatment of trans-verse sections (5 "m), which wereused for immunohistochemical dem-onstration of lateral line nerves andglia.

Histochemistry/Histology/ImmunostainingWhite larvae at stage 35/36 were sub-jected to the Dopa reaction (Epperleinand Lofberg, 1984) in order to enhancethe visibilitiy of melanophores and fa-cilitate their exact counting. The num-bers of melanophores were needed forassessing the numbers of DiI labeledcells in white embryos.

Some vibratome sections (100 "m)were cut through PFA-fixed DiI-,FITC- or rhodamine dextran-, or GFP-labeled embryos and counterstainedwith an anti-fibronectin antibody tovisualize tissue borders. Sections werethen incubated with a polyclonal anti-fibronectin antibody (Dako, Hamburg,Germany) followed by a FITC- or Cy3-conjugated goat-anti-rabbit secondaryantibody (Dianova, Hamburg, Ger-many). Some sections were stainedwith dapi (0.1–1 "g/ml in PBS) tomark cell nuclei. Vibratome sectionsthrough larvae that had received onefocal DiI injection into the left anteriortrunk neural fold were also stained foranti-tubulin or anti-GFAP to reveallateral line nerves or glia, respec-tively. For anti-tubulin staining, a pri-mary monoclonal antibody (W. Half-ter, Pittsburgh) was used followed bya FITC-conjugated goat-anti-mousesecondary antibody (Dianova). Stain-ing of glia was carried out with aprimary polyclonal GFAP-antibody(Dako) followed by a FITC-conjugatedgoat-anti-rabbit secondary antibody(Dako). Because staining with fluores-cence techniques was sometimes weak(anti-tubulin) or had high background(anti-GFAP), some unlabeled controllarvae of a similar age as DiI-injectedspecimens were embedded into paraf-fin. Microwaved transverse sections (5"m) were stained with the same pri-mary antibodies but with biotinylatedsecondary antibodies and tertiary avi-din-peroxidase complexes (VectastainABC-Kit).

In Situ HybridizationAxolotl Snail riboprobes were pre-pared as previously described (Epper-lein et al., 2000). In situ hybridizationwas performed on albino and wild-type axolotl embryos as described byHenrique et al. (1995), with the addi-tion of an overnight wash in MAB-T.

In situ–hybridized wild-type embryoswere washed for 2 hours in severalchanges of PBS and bleached for 1–2hr in 1% H202/5% formamide/.5(SSCunder intense illumination to revealhybridization

Laser Scanning MicroscopyHorizontal vibratome sections (100"m) through the dorsal trunk of DiI-injected embryos (stage 33–34) werealso investigated in the laser-scan-ning microscope (Zeiss; (10 objective)to gain a better resolution of RohonBeard cells and their nerve fibres.

Image AnalysisWhole embryos and sections were an-alyzed with an epifluorescence micro-scope and images were captured witha Spot digital camera (Visitron). Us-ing Adobe Photoshop software, thecontrast and brightness of capturedimages were optimized, and brightfield images of whole embryos werecombined with corresponding fluores-cence images. Separate images werecaptured from sectioned material la-beled with DiI, fluorescent-dextransor GFP (neural crest cells), anti-fi-bronectin (tissue borders), anti-tubu-lin (lateral line nerve), anti-GFAP(glial cells), and dapi (cell nuclei), be-fore combinations of two or three ofthem were superimposed into a singleimage.

ACKNOWLEDGMENTSThis research was supported by DFG(EP8/7-1) to H.H.E., by NS051051 toM.B.F., and a James H. Zumberge Re-search and Innovation Fund and aDonald E. and Delia B. Baxter Foun-dation award to M.A.J.S. H.H.E.thanks G. Northcutt (San Diego) andG. Schlosser (Bremen) for valuable in-formation on the lateral line, I. Beck,P. Mirtschink, F. Heidrich, and T.Schwalm for analysis of cell migrationor computer work, and S. Bramke forhistology.

REFERENCES

Bodenstein D. 1952. Studies on the devel-opment of the dorsal fin in amphibians. JExp Zool 120:213–245.

Bordzilovskaya NP, Dettlaff TA, DuhonST, Malacinski GM. 1989. Developmen-tal-stage series of axolotl embryos. In:

TRUNK NEURAL CREST CELLS IN THE AXOLOTL EMBRYO 401

Armstrong JB, Malacinski GM, editors.The developmental biology of the axolotl.New York: Oxford University Press. p201–219.

Bronner-Fraser M. 1986. Analysis of theearly stages of trunk neueral crest mi-graion in avian embryos using nmono-clonal antibody HNK-1 Dev Biol 115:44–55.

Brun RB. 1985. Neural fold and neuralcrest movement in the Mexicansalamander Ambystoma mexicanum. JExp Zool 234:57–61.

Chibon P. 1966. Analyse experimentale dela regionalisation et des capacites mor-phogenetiques de la crete neurale chezl‘Amphibien Urodele Pleurodeles waltlii.Michah Mem Soc Fr Zool 36:1–107.

Collazo A, Bronner-Fraser M, Fraser SE.1993. Vital dye labelling of Xenopus lae-vis trunk neural crest reveals multipo-tency and novel pathways of migration.Development 118:363–376.

Collazo A, Fraser SE, Mabee PM. 1994. Adual embryonic origin for vertebratemechanoreceptors. Science 264:426–430.

Couly G, Grapin-Botton A, Coltey P, LeDouarin NM. 1996. The regeneration ofthe cephalic neural crest, a problem re-visited: the regenerating cells originatefrom the contralateral or from the ante-rior and posterior neural fold. Develop-ment 122:3393–3407.

Dalton HC. 1953. Relations between devel-oping melanophores and embryonic tis-sues in the Mexican axolotl. Gordon M(ed) Pigment cell growth. New York: Ac-ademic Press. P 17–27.

DuShane GP. 1935. An experimental studyof the origin of pigment cells in am-phibia. J Exp Zool 72:1–31.

DuShane GP. 1938. Neural fold derivativesin the amphibia: pigment cells, spinalganglia and Rohon-Beard cells. J ExpZool 82:193–215.

Elena de Bellard M, Bronner-Fraser M.2005. Neural crest migration methods inthe chicken embryo. Methods Mol Biol294:247–267.

Epperlein HH, Junginger M. 1982. Thenormal development of the newt, Tritu-rus alpestris (Daudin). Amphibia Rep-tilia 2:295–308.

Epperlein HH, Lofberg J. 1984. Xan-thophores in chromatophore groups ofthe premigratory neural crest initiatethe pigment pattern of the axolotl larva.Roux Arch Dev Biol 193:357–369.

Epperlein HH, Lofberg J. 1990. The devel-opment of the larval pigment patterns inTriturus alpestris and Ambystoma mexi-canum. Adv Anat Embryol Cell Biol 118:1–99.

Epperlein HH, Meulemans D, Bronner-Fraser M, Steinbeisser H, Selleck MA.2000. Analysis of cranial neural crest mi-gratory pathways in axolotl using cellmarkers and transplantation. Develop-ment 127:2751–2761.

Erickson CA, Reedy MV. 1998. Neuralcrest development: the interplay be-tween morphogenesis and cell differenti-ation. Curr Top Dev Biol 40:177–209.

Erickson CA, Duong TD, Tosney KW. 1992.Descriptive and experimental analysis ofthe dispersion of neural crest cells alongthe dorsolateral path and their entryinto ectoderm in the chick embryo. DevBiol 151:251–272.

Gilmour, DT, Maischein, HM, Nusslein-Volhard, 2002. Migration and function ofa glial subtype in the vertebrate periph-eral nervous system. Neuron 34:577–588

Guillory G, Bronner-Fraser M. 1986. An invitro assay for neural crest cell migra-tion through the somites. J Embryol ExpMorphol 98:85–97.

Hall BK, Horstadius S. 1988. The neuralcrest. London: Oxford University Press.

Henrique D, Adam J, Myat A, Chitnis A,Lewis J, Ish-Horowicz D. 1995. Expres-sion of a Delta homologue in prospectiveneurons in the chick. Nature 375:787–790.

Horstadius S, Sellman S. 1946. Experi-mentelle untersuchungen uber die deter-mination des knorpeligen kopfskelettesbei Urodelen. Nova Acta Regiae Soc SciUps Ser 13:1–170.

Hughes A. 1957. The development of theprimary sensory system in Xenopus lae-vis (Daudin). J Anat 91:323–338.

Keller RE, Lofberg J, Spieth J. 1982. Neu-ral crest cell behavior in white and darkembryos of Ambystoma mexicanum: epi-dermal inhibition of pigment cell migra-tion in the white axolotl. Dev Biol 89:179–195.

Knecht AK, Bronner-Fraser M. 2002. In-duction of the neural crest: a multigeneprocess. Nat Rev Genet 3:453–461.

Krotoski DM, Fraser SE, Bronner-FraserM. 1988. Mapping of neural crest path-ways in Xenopus laevis using inter- andintra-specific cell markers. Dev Biol 127:119–132.

Le Douarin NM. 1969. Particularites dunoyau interphasique chez la Caillejaponaise (Coturnix coturnix japonica).Utilisation de ces particularites comme’marquage biologique’ dans les recher-ches sur les interactions tissulaires et lesmigrations cellulaires au cour de l‘onto-genese. Bull Biol Fr Belg 103:435–452.

Le Douarin NM, Kalcheim C. 1999. Theneural crest. Cambridge: CambridgeUniversity Press.

Le Douarin NM, Teillet MA. 1974. Experi-mental analysis of the migration and dif-ferentiation of neuroblasts of the auto-nomic nervous system and ofneurectodermal mesenchymal deriva-tives, using a biological cell markingtechnique. Dev Biol 41:162–184.

Lofberg J, Ahlfors K, Fallstrom C. 1980.Neural crest cell migration in relation toextracellular matrix organization in theembryonicaxolotltrunk.DevBiol75:148–167.

Lofberg J, Nynas-McCoy A, Olsson C, Jon-sson L, Perris R. 1985. Stimulation ofinitial neural crest cell migration in theaxolotl embryo by tissue grafts and ex-tracellular matrix transplanted on mi-crocarriers. Dev Biol 107:442–459.

Luo R, Gao J, Wehrle-Haller B, Henion PD.2003. Molecular identification of distinct

neurogenic and melanogenic neural crestsublineages. Development 130:321–330.

Macmillan GJ. 1976. Melanoblast-tissueinteractions and the development of pig-ment pattern in Xenopus larvae. J Em-bryol Exp Morphol 35:463–484.

Moury JD, Jacobson AG. 1989. Neural foldformation at newly created boundariesbetween neural plate and epidermis inthe axolotl. Dev Biol 133:44–57.

Moury JD, Jacobson AG. 1990. The originsof neural crest cells in the axolotl. DevBiol 141:243–253.

Northcutt RG, Catania KC, Criley BB.1994. Development of lateral line organsin the axolotl. J Comp Neurol 340:480–514.

Northcutt RG, Barlow LA, Catania KC,Braun CB. 1996. Developmental fate ofthe lateral and medial walls of the neu-ral folds in axolotls. Am Zool 36:74A.

Raven CP. 1931. Zur Entwicklung derGanglienleiste. I. Die Kinematik derGanglienleistenentwicklung bei denUrodelen. Wilhelm Roux Arch Entwick-lungsmech Org 125:210–293.

Reedy MV, Faraco CD, Erickson CA. 1998.The delayed entry of thoracic neuralcrest cells into the dorsolateral path is aconsequence of the late emigration ofmelanogenic neural crest cells from theneural tube. Dev Biol 200:234–246.

Roberts A. 2000. Early functional organi-zation of spinal neurons in developinglower vertebrates. Brain Res Bull 53:585–593.

Roberts A, Patton DT. 1985. Growth conesand the formation of central and periph-eral neurites by sensory neurones in am-phibian embryos. J Neurosci Res 13:23–38.

Sauer FC. 1935. Mitosis in the neural tube.J Comp Neurol 62: 377-405.

Schlosser G. 2002. Development and evo-lution of lateral line placodes in amphib-ians I. Development. Zoology (Jena) 105:119–146.

Schreckenberg GM, Jacobson AG. 1975.Normal stages of development of the ax-olotl Ambystoma mexicanum. Dev Biol42:391–400.

Selleck MA, Bronner-Fraser M. 1995. Ori-gins of the avian neural crest: the role ofneural plate-epidermal interactions. De-velopment 121:525–538.

Selleck MA, Bronner-Fraser M. 1996. Thegenesis of avian neural crest cells: a clas-sic embryonic induction. Proc Natl AcadSci USA 93:9352–9357.

Serbedzija GN, Bronner-Fraser M, FraserSE. 1989. A vital dye analysis of the tim-ing and pathways of avian trunk neuralcrest cell migration. Development 106:809–816.

Serbedzija GN, Fraser SE, Bronner-FraserM. 1990. Pathways of trunk neural crestcell migration in the mouse embryo asrevealed by vital dye labelling. Develop-ment 108:605–612.

Smith SC, Lannoo MJ, Armstrong JB.1990. Development of the mechanorecep-tive lateral-line system in the axolotl:placode specification, guidance of migra-

402 EPPERLEIN ET AL.

tion, and the origin of neuromast polar-ity. Anat Embryol (Berl) 182:171–180.

Sobkow L, Epperlein HH, Herklotz S,Straube WL, Tanaka EM. 2006. A germ-line GFP transgenic axolotl and its use totrack cell fate: dual origin of the fin mes-enchyme during development and thefate of blood cells during regeneration.Dev Biol 290:386–397.

Steinberg MS. 1957. A nonnutrient culturemedium for amphibian embryonic tissue.Carnegie Inst Washington Year b56:347–348.

Taylor JS, Roberts A. 1983. The early de-velopment of the primary sensory neu-rones in an amphibian embryo: a scan-ning electron microscope study. JEmbryol Exp Morphol 75:49–66.

Vogel DL, Model PG. 1977. Development ofthe sympathetic system in the Mexicanaxolotl, Ambystoma mexicanum. DevBiol 56:76–96.

Wakamatsu Y, Mochii M, Vogel KS,Weston JA. 1998. Avian neural crest-de-rived neurogenic precursors undergo ap-optosis on the lateral migration path-way. Development 125:4205–4213.

TRUNK NEURAL CREST CELLS IN THE AXOLOTL EMBRYO 403