Specification and Morphogenesis of the Zebrafish Larval

Kimmel et al., page 1

Specification and Morphogenesis of the Zebrafish Larval HeadSkeleton

Charles B. Kimmel1,3, Craig T. Miller 1, and Cecilia B. Moens21Institute of Neuroscience, 1254 University of Oregon, Eugene, Oregon 97403-1254, and2Division of Basic Science and Program in Developmental Biology, Fred Hutchinson CancerResearch Center, 1100 Fairview Avenue North, Seattle, Washington 98109.

(in press, Developmental Biology)

3To whom editorial correspondence should be addressed. Fax (541) 346-4548. E-mail:[email protected].

Forward genetic analyses can reveal important developmental regulatory genes, and howthey function to pattern morphology. This is because a mutated gene can produce a novel,sometimes beautiful phenotype that, like the normal phenotype, immediately seems worthunderstanding. Generally the loss-of-function mutant phenotype is simplified from thewild-type one, and often the nature of the pattern simplification allows one to deduce howthe wild-type gene contributes to patterning the normal, more complex, morphology. Thistruism seems no less valid for the vertebrate head skeleton than for other and simpler casesof patterning in multicellular plants and animals. To show this, we review selectedzebrafish craniofacial mutants. ‘Midline group’ mutations, in genes functioning in one ofat least three signal transduction pathways, lead to neurocranial pattern truncations thatare primarily along the mediolateral axis. Mutation of lazarus/pbx4, encoding a hox-genepartner, and mutation of valentino/kreisler, a hox-gene regulator, produce anterior-posterior axis disruptions of pharyngeal cartilages. Dorsoventral axis patterning of thesame cartilages is disrupted in sucker/endothelin-1 mutants. We infer that different signaltransduction pathways pattern cartilage development along these three separate axes.Patterning of at least the anterior-posterior and dorsoventral axes have been broadlyconserved, e.g., reduced Endothelin-1 signaling similarly perturbs cartilage specification inchick, mouse, and zebrafish. We hypothesize that Endothelin-1 also is an upstreamorganizer of the patterns of cellular interactions during cartilage morphogenesis.

mailto:[email protected]


Key words: craniofacial patterning, skeleton, head, specification, morphogenesis, Hoxgenes, Endothelin-1, zebrafish

SPECIFICATIONS AND MORPHOGENESIS GO HAND-IN-HAND

Development in multicellular animals includes assignments of fates to naive embryonic cells,and the arrangement of these cells into higher-level functional assemblages. How cascades ofdevelopmental regulatory genes function to mediate these two kinds of processes, specificationand morphogenesis, is currently under intense study in a variety of experimental systems. A likelygeneralization is that controls of specification and morphogenesis are intimately interconnected.It would seem sensible that, at the very least, fate specification must include instructions aboutmorphogenesis along with instructions towards how the cell should undergo functionalspecialization. Hence, both differentiation and morphogenesis may ultimately be controlled by thesame upstream developmental regulatory genes. A clear example in zebrafish is the T-boxtranscription factor Spadetail (Griffin et al., 1998). Loss-of-function mutation of the spadetailgene results in prominent cell-autonomous defects in somitogenesis (a morphogenetic process)and myogenesis (muscle differentiation). spadetail functions very early in development,expression is present and maintained specifically in presomitic mesoderm during gastrulation.Genetic targets under its positive regulation include myoD, a myogenic gene (Weinberg et al.,1996; Amacher and Kimmel, 1998), and a protocadherin (paraxial protocadherin) a putativemorphoregulatory gene (Yamamoto et al., 1998). Loss of these separate downstream functionscorrelates with, and possibly explains, the two classes of phenotypic disturbances observed inspadetail mutants.

A SIMPLE SYSTEM APPROACH TO HEAD SKELETAL PATTERNING

Below we explore development of the early head skeleton in zebrafish as a system forunderstanding hierarchical genetic control of organogenesis. Skeletal organs, bones andcartilages, have several attractive attributes for such analysis. In the skeletal system, perhapsmore than any other, organ morphology is intimately connected with its function. Because of thisfact, the shapes and sizes of skeletal elements can be presumed to be under stringent selection,and the development of organ form exquisitely regulated. Indeed, because mineralized bonefossilizes readily, we understand much of the history of how skeletal elements have evolved alongvarious vertebrate lineages. Adding developmental genetic analyses to this understanding iscurrently revealing how skeletal development itself might have evolved.

Facilitating a mechanistic analysis, the skeletal system has relatively few types of specializedcells. In young zebrafish the system is further simplified: A functional larval skeleton developsover the course of only a few days. The early skeletal elements are very small and mostly made ofcartilage, on which we focus here. Few cells of only two types, chondrocytes and perichondrialcells, comprise these cartilages, with simple monolayered arrangements discussed further below.Most of the cartilages are located not far beneath the organism's surface, where they can be easilyand directly visualized in the intact preparation as they form. As in other vertebrates thecartilaginous head skeleton has two prominent subdivisions, the neurocranium protecting the brainand sensory organs, and the pharyngeal skeleton supporting the feeding and gill-breathingstructures (Fig. 1). Mutational analyses to be discussed in the sections below (see also Schilling,1997) include both regions, and we also describe a hypothesis of cartilage morphogenesis, the


'joints build stacks'hypothesis, motivatedfrom descriptive studies ofchondrogenesis in wild-type and mutant embryos.

MIDLINE GROUPMUTATIONSPERTURBDEVELOPMENT OFTHE BASICRANIALCARTILAGES

In his monumental1894 study of monstersand the lessons we canlearn from them aboutdevelopmental patterning,Bateson included the"bulldog trout" (Fig. 2).In this deformed fish theanterior head (but not thelower jaw) is dramaticallycompressed. In contrast,the form and size of thelower jaw is normal.Because of the disparitythe lower jaw protrudeswell forward of thecranium, just as can be thecase in a genuine bulldog:upper and lower facial

parts are not well matched. The example reveals some local autonomy in the development ofthese two facial regions.

Bateson does not go on to examine the skeletal deformities of the bulldog trout directly, nordoes the bulldog trout, found by chance in a Scottish lake, tell us anything about how or whydevelopment went wrong. Just over a hundred years later, "bulldog zebrafish", i.e. developmentalmutants, such as the silberblick (slb) mutant shown in Fig. 3, provide the beginnings of anexplanation. The bulldog phenotype appears at the early larval stage (i.e. shortly after the embryohatches, roughly at 3 days postfertilization, and begins postembryonic development). Similarphenotypes are observed in larval fish homozygous for loss-of-function alleles at any of severalseparate chromosomal loci. Interestingly, the mutants were all first identified by defects arisingearlier, during embryogenesis, and were all placed into a common embryonic phenotypic class, the'midline' group (Brand et al., 1996). The mutants were classified this way because, irrespective of

FIG. 1 The young larval zebrafish (A, 5 days postfertilization, left side view) and thelayout of its cartilaginous head skeleton. For further description see Schilling andKimmel (1997) and Cubbage and Mabee (1996). Anterior is to left in this and otherfigures in the paper except where indicated. B shows a ventral view of an Alcian bluelabeled, whole-mounted preparation. Cartilages are indicated in the first ormandibular arch (pq, m), and in the second or hyoid arch (hs, ch). C-E showdrawings of such preparations. C. The neurocranial (or basicranial) cartilages andnotochord from the dorsal aspect. The eyes fit into the shallow grooves along thesides of the ethmoid plate and trabeculae. The otic vesicles fit into the prominentcavities to either side of the notochord and parachordal cartilages . The brain'sposterior pituitary fits into the prominent midline cavity ahead of the notochord, thehypophysial fenestra. D. The neurocranium (diagrammatically elevated dorsalwardsfor the sake of clarity) and the pharyngeal skeleton in side view. E. The pharyngealskeleton in ventral view. Abbreviations A: anterior. bb: basibranchial. bh: basihyal.cb: ceratobranchial. ch: ceratohyal. D: dorsal. ep: ethmoid plate. hb: hypobranchial.hs: hyosymplectic. ih: interhyal. L: lateral. M: medial. m: Meckel's. not: notochord.P: posterior. pch: parachordal. pq: palatoquadrate. tr: trabecula. V: ventral.


FIG. 2. The "bulldog-headed trout", acraniofacial monster found in nature. FromBateson (1894).

FIG. 3. The "bulldog-headed zebrafish",comparing the wild-type (WT, A) and slbmutant (B) embryo at 2 days of development(unpublished photographs, courtesy of Dr.Corinne Houart). C, D. Basicranialcartilages labeled with Alcian blue,dissected and laid out as a flat mount,viewed from the dorsal aspect (unpublished,B. Ullmann & C. B. K.). pc: polar cartilageregion, a distinctive region of joiningbetween the parachordals and trabeculae.The polar cartilages form as separateelements in some organisms (e.g. see DeBeer, 1937; Goodrich, 1930), but probablynot in zebrafish. Abbreviations as in Fig. 1.

the genetic pathway affected (see below), they all shared embryonic phenotypes known to resultfrom perturbed development at the midline. Forexample, mutant embryos in the midline group allexhibit cyclopia: the eyes are closer to the midline thannormal, sometimes fused together at the midline. Thestrong inference, recognized by Brand et al. (1996), isthat defective midline signaling during embryonicdevelopment underlies the later craniofacial bulldogphenotype as well as the earlier, embryonicphenotypes.

Although the bulldog facial phenotype might notitself hint atinvolvement of

the midline in patterning, the skeletal phenotypeimmediately does so. Staining the larval cartilages withAlcian blue reveals a profound neurocranial deficiencythat underlies the bulldog face in slb mutants (Piotrowskiet al., 1996). The defects show most clearly in a dissectedout and flat-mounted preparation (Fig. 3C, D). In thewild-type early larva, the basicranium anterior to theregion known as the polar cartilages consists of a bilateralpair of rods, the trabeculae, that are separated in themidline by a prominent hypophyseal fenestra. Thetrabeculae come together more anteriorly and they joinone another, fusing together in the midline in a regiontermed the trabeculae communis. More anterior still thispaired midline cartilage flares outwards laterally as theethmoid plate. In slb mutants this pattern is simplified.Ahead of the polar cartilage region there is only a singlecartilaginous rod present just at the midline, and justabout the diameter of a single trabecula.

In cellular terms, a complex and precisely shaped two-dimensional array of cartilage is converted into a muchsimpler one-dimensional array when a functional geneproduct (a Wnt protein; see below) is absent. The changein the basicranium appears to be very indirect, dueultimately to an early mesodermal signaling defect thatperturbs formation of the prechordal plate, i.e., themidline mesoderm of the head rudiment (see below). Thathead skeletal defects might be due to defectivedevelopment of the prechordal plate would not be news to embryologists experimenting withsalamander embryos nearly 70 years ago. They showed that defects of the same nature, inparticular a fusion of the pair of trabeculae within the midline, resulted from treatment of earlyembryos with teratogenic chemicals (e.g. LiCl) that perturbed early midline development and alsoproduced cyclopia (review: De Beer, 1937, p. 447).


FIG. 4. A phenotypic series of asubset of midline group mutants,chameleon (con), detour (dtr),schmalspur (sur), you-too (yot),and iguana (igu), arranged inapproximate order of severity.Dorsal views, as in Fig. 1C. Twoexamples of con mutants areincluded, the other mutations areat separate genetic loci.Abbreviations as in Fig. 1(modified from Brand et al.,1996, withsur added).

Considering the midline group of mutants collectively, one observes a range of severity of thephenotypic defects. Figure 4 shows this range for a subset of themidline group mutants, representing the genes schmalspur, detour,chamelion, you-too, and iguana. The presentation follows Brandet al. (1996) who arranged the mutants into a phenotypic series byordering them according to the severity of skeletal reduction. Thisarrangement revealed quite a curious sequence with respect to thepattern change as well (Fig. 4). Along the series, except forschmalspur, there seems to be progressive loss of cartilage fromboth anterior and lateral positions. Furthermore, the loss seemsnot to respect the names assigned to the cartilage regions. That is,the ethmoid plate shrinks both in its anterior-posterior andmediolateral aspects, and in the same mutants the trabeculaeshorten along the anterior-posterior axis as well, and areprogressively displaced from lateral to a median position.

The somewhat coordinated anterior-posterior and medial-lateral pattern truncation seems worth further exploration. Not allmutants might obey the same rule. Brand et al. (1996) report thatthe schmalspur mutant phenotype is distinctive in that moreposterior defects (midline cartilage fusion in the region of the polarcartilages) without corresponding anterior changes, i.e. theethmoid plate appears normal.

According to Brand et al., the defects in the midline group ofmutants show up along the entire length of the ventralneurocranium. However, we are impressed that significant patternchanges they show are all anterior to the parachordals: thetransition from unaffected to affected cartilages occurs ratherabruptly at the polar cartilage region (Fig. 3C). In even their mostsevere example, iguana (igu, Fig. 4) the cartilages posterior to thistransition look approximately the same as in the wild type. Yetonly unpatterned islands of cartilage are present anterior to thepolar cartilage region in iguana mutants. We conclude that themutants in this midline group are revealing genes involved inpattern regulation of particularly the anterior basicranium.

If the skeletal disturbances to the basicranium are indeedlimited in anterior-posterior extent, as we just argued, how can weunderstand why this might be so? Following Brand et al. (1996),

suppose that the cartilage defects follow from defects in the embryonic midline. Hence thedifference between anterior and posterior basicranium could be due to an anterior-posteriordifference in the midline, due to a difference in the cells responding to the midline signal, or due toa difference in both signaling and responding cell types together. Indeed there is evidence forchanges in both types: The putative developmental boundary region, the polar cartilage region(Fig. 3), develops just ahead of a prominent and famous transitional zone in the embryonic midline


mesoderm, a known signaling center. Here the notochord ends and the prechordal plate begins.Further, the developmental origin of the cells forming the neurocranial cartilages might change inthe same location. According to fate mapping studies in avians (Le Lièvre, 1978) and amphibians(Chibon, 1967), and extirpation experiments in lamprey (Languille and Hall, 1986; 1988), cranialneural crest forms the trabeculae but not the parachordals, which seemingly come from paraxialmesoderm. Hence, the polar cartilage region may mark an important transitional zone in both themidline signaling cells and the responding presumptive cartilage cells.

MIDLINE GROUP GENES FUNCTION IN THREE DIFFERENT SIGNALTRANSDUCTION PATHWAYS

Subsequent to their discoveries in mutagenesis screens, a number of the genes in the midlinegroup have been molecularly identified. silberblick encodes a Wnt-11 ortholog (Heisenberg et al.,2000). cyclops encodes a Nodal ortholog (TGFbeta superfamily, Rebagliati et al., 1998; Sampathet al., 1998; review Schier and Shen, 2000). one-eyed pinhead encodes a Crypto ortholog (EGF-CFC gene family; Zhang et al., 1998; review: Shen and Shier, 2000) that functions downstream toNodal signaling. schmalspur encodes a FoxH1/FAST1 homolog (winged-helix transcriptionfactor; Pogoda et al., 2000; Sirotkin et al., 2000), also in the Nodal signal-transduction pathway.syu is an ortholog of sonic hedgehog (Schauerte et al., 1998), and yot encodes a Gli-2 ortholog(Karlstrom et al., 1999), a transcription factor regulated by Hh signaling.

Hence, the mutations, where known, identify genes that function in cell-cell signalingpathways, and at least three signal transduction pathways are involved. From expression analysisit seems likely that two, possibly all three, of these signals do not act directly on the cartilage-forming cells. If such is the case, this is an important lesson for understanding craniofacial mutantphenotypes: the genes responsible for severe craniofacial phenotypes might be remote to cartilagedevelopment itself.

Both cyc/nodal and syu/shh are expressed by the anterior midline mesoderm. but slb/wnt11has a complex expression pattern, including expression in head neural crest that might eventuallyform the anterior basicranium. However, the perhaps more important expression domain is withinearly paraxial mesoderm, before mesoderm has migrated to reach where the head will form(Heisenberg et al., 2000). Here it acts to turn on a "noncanonical" Wnt pathway involving

intracellular Ca2+ release and a G-protein-dependent activation of kinases (review; Kühl et al.,2000) to mediate cell polarity and polarized cell movement. These movements underlie the earlyextension of the embryonic axis, and the initial formation of the midline head mesoderm in the lategastrula (Heisenberg and Nüsslein-Volhard, 1997; Heisenberg et al., 2000). Such movements gowrong in slb/wnt11 mutants, and the prechordal plate is severely defective. Such a function, at ca.8 hours postfertilization (h) is remote indeed to head cartilage formation, chondrogenesisoccurring near the end of embryogenesis, two days later.

Similarly, the role of cyc/nodal signaling must be indirect on the cartilage-forming cells. Hereexpression is present in prechordal plate, i.e. just at the region adjacent to where the anteriorbasicranium forms, but the time of expression seems wrong. The cyc/nodal gene is expressed asthe presumptive mesodermal cells move anteriorwards along the axis during gastrula, but shortlyafter this migration is completed expression is downregulated. This occurs several hours beforeneural crest migration begins, such that it seems unlikely that the crest cells ever encounter the


Cyc/Nodal signal when they arrive at their destination beneath the forebrain. This timing mightmean that cyc/nodal signaling serves to set up a subsequent midline signal that in turn affectsneural crest migration and/or cartilage specification at the midline. Supporting this scenario, amarker for premigratory neural crest, crestin, is downregulated in cyc mutants (Rubinstein et al.,2000).

Syu/shh expression begins in mesoderm in the early-midgastrula, and like cyc/nodalexpression, expression of syu/shh is downregulated after the prechordal plate mesoderm hasmigrated to underlie forebrain neurectoderm. Hence it also seems unlikely that a Syu/Shh signalfrom prechordal plate acts on cartilage-forming neural crest. However, we propose that in factdefective Syu/Shh signaling is a part of the proximate cause of the cartilage dysmorphogenesis inall of these mutants. In response to induction by the prechordal plate (and that clearly depends oncyc/nodal functioning in the prechordal plate; Hatta et al., 1994) the responding neurectodermbegins a course of development that includes its own upregulation of syu/shh. This expressionpersists through the time (pharyngula stages) when potential cartilage-forming cells move into theneighborhood of the ventral neurectoderm and are potentially ready to respond. Hence, in thispossible scenario the Shh signal that acts on postmigratory cartilage-forming neural crest comesfrom the ventral ectoderm of the primordial forebrain, not the prechordal plate. The distinctivesur mutant phenotype mentioned above is partly explained by this model: In most of the midlinegroup mutants expression of syu/shh is perturbed throughout the anterior ventral forebrain. In surmutants specifically, syu/shh expression is missing just in a gap that corresponds in anterior-posterior level to the basicranial defect (Brand et al., 1996).

That signaling from neurepithelium can promote neurocranial cartilage formation, andparticipate in its patterning is well known from transplantation studies (review; Thorogood,1983). Furthermore, in limb and in pharyngeal cartilage development there has beendemonstrated an essential role of Shh in chondrogenesis, probably acting directly on the cartilage-forming cells (Hu and Helms, 1999). The model predicts that one might be able to rescue thebasicranial defects of any of the mutants by locally supplying a source of Shh to the postmigratorycranial crest.

The phenotypes that we have been discussing are largely limited to the neurocranium. Inparticular, they do not include the pharyngeal cartilages, and in some respects this is surprising.For example, an extended process of the dorsal cartilage in the first pharyngeal arch (thepalatoquadrate's pterygoid process) articulates with the anterior-lateral region of the ethmoid, aregion missing in slb/wnt11 mutants. Hence we expected to see a corresponding deformation ofthe palatoquadrate in these mutants, but their study did not reveal any change (B. Ullmann and C.B. K., unpublished findings). Perhaps only the severest of the midline group mutants havepharyngeal cartilage phenotypes: cyc/nodal mutants have extensive deletion of the anteriorbasicranial cartilage (like the igu phenotype shown in Fig. 4), and here there are extensivecartilage fusions in the midline of the mandibular arch as well (C. T. M., unpublished findings).

REGULATION OF hox GENE FUNCTION AND AP PATTERNING

The segmentally organized pharyngeal cartilages of the larval zebrafish have a primitiveanterior-posterior (AP) organization, with paired dorsal and ventral elements in the first and


FIG. 5. Cartilage phenotypes and dlx2 expression in WT and lzrmutants, from Pöpperl et al., 2000). A, B: Alcian blue stained, flat-mounted cartilages of the first and second pharyngeal segmentsprepared from day 4 embryos. Left side views, dorsal to the top andanterior to the left. In the WT (A) the elements are separated by joints.Abbreviations as in Fig. 1. In the mutant (B) the dorsal elements aswell as the ventral elements in the two segments are fused to oneanother. DV fusions within each segment occur as well; the DV jointsare missing as is the interhyal cartilage (ih) in the second segment. C,D: Dorsal views (anterior to the left) of 20 h embryos labeled by RNAin situ hybridization for expression of dlx2. The three separate bilateralpatches of expression in the WT (C) correspond to the three streams ofpostmigratory neural crest that have populated the pharyngeal archprimordia. The streams are fused together in the lzr mutant.

second arches, and paired ventral elements in the five more posterior gill arches (Fig. 1). Theelements in the first and second arches are morphologically unique, and thus can be used asdifferentiated markers of AP identity within the head periphery.

Anterior-posterior patterning in vertebrate embryos, as in Drosophila, depends on the Hoxgenes, whose overlapping expression domains demarcate regions of positional identity (reviewedin Lumsden and Krumlauf, 1996). Hox gene expression is regulated in a complex and inter-dependent manner (reviewed in Nonchev et al., 1997; Studer et al., 1998). At the top of thehierarchy lies an essential responsiveness, on the part of particular Hox genes, to retinoic acid(RA; reviewed in Marshall et al., 1996; Gould et al., 1998). Although a gradient of RA along theAP axis has been difficult to demonstrate, the phenotypes of embryos in which RA synthesis or

responsiveness is blocked, eithergenetically or pharmacologically,strongly suggests a requirement forRA in vertebrate AP patterning(Niederreither et al., 2000; White etal., 2000; Gale et al., 1999; van derWees et al., 1998).

Fusions and duplications ofpharyngeal cartilages of quitedifferent sorts to those describedabove are observed in mutants inwhich Hox gene function is altered,either due to mutations in Hoxpartners or in upstream regulators.The cartilage fusions in lazarus (lzr)mutants are among the morespectacular (Fig. 5A, B; Pöpperl etal., 2000). In wild types thecartilages in the adjacent pharyngealsegments (adjacent arches) areseparate elements from one another,although as shown in Fig. 5A, thedorsal cartilages in the hyoid andmandibular arches are closeneighbors. The dorsal and ventralcartilages in each segment are

separate as well, and they articulate at a prominent joint (this joint region including the smallinterhyal cartilage in the hyoid arch). In lzr mutants the hyoid and mandibular cartilages are quitethoroughly fused to one another, dorsal with dorsal, ventral with ventral and dorsal with ventral.Clearly the processes that underlie formation of isolated chondrogenic islands in wild types aredefective in lzr mutants, implying that the function of the wild-type gene is to pattern cartilageislanding. The islanding mechanism is not lost altogether: there are none of the fusions at themidline in lzr mutants, such as are present in mutants of the midline group.


Fig. 6. Cartilage phenotypes and hox gene expression inWT and val mutants.A-E Ventral views, anterior to the top, of flat-mountedcartilages in 7-day old larvae, from Moens et al., 1998. A.The WT pharyngeal segments 2-5. The hyoid segment(second pharyngeal segment) uniquely contains a smallinterhyal (ih) cartilage. At this stage, there is no separatehypohyal cartilage, the serial homolog of the hypobranchials.The ceratohyal (ch) is distinctively larger in size from itssegmental homologs, the ceratobranchials (cb1-5).Ceratobranchials are tapering elements, as shown in B forcb1 (third pharyngeal segment) in another WT preparationat higher magnification. C-E. The cartilages in the same(third) segment in three individual val mutants. Theprincipal element, normally a ceratobranchial is oftentruncated and thickened, now more like a ceratohyal (C).Separate hypobranchials may be missing (arrow in D), andsmall interhyal-like elements are sometimes present

What is this islanding function (orfunctions)? lazarus encodes a protein of thePbx homeodomain family (Pbx4; Pöpperl etal., 2000), expressed ubiquitously. Suchproteins are known to interact directly withHox proteins (reviewed in Mann and Chan,1996). Hox and Pbx proteins form adimeric DNA-bound complex in theregulation of transcriptional targets (Passneret al., 1999; Piper et al., 1999). Hence onemight expect pbx genes to function in thesame general manner as hox genes.However, we caution that some of pbxgene-mediated functions in Drosophila donot depend on hox gene interaction (Casaresand Mann, 1998; Abu-Shaar and Mann,1998; González-Crespo, S. et al., 1998). Infact we do not know, but only suppose, thatin patterning cartilage development the wild-type Lzr/Pbx4 protein works with a Hoxprotein partner.

The dorsal to dorsal and ventral toventral cartilage fusions occur along theanterior-posterior axis, and they can berelated to defects in segmental patterningand in anterior-posterior specification ofidentity in a straightforward manner. Inwild-type zebrafish, as in other vertebrates,at least some of the neural crest that willlater develop as cartilage migrates from thedorsal lateral margin of the neural plate inthree streams present along the AP axis.These streams can be visualized byexpression of the neural crest marker dlx2(Fig. 5C, D), and are also apparent in timelapse recordings (C. B. K. & R. Keynes,unpublished observations). In lzr/pbx4mutants, at least as assayed by dlx2expression, the streams are fused together;

the crest appears to migrate as "a single, uninterrupted sheet, ... with neural crest cells populatingthe normally crest-free zones lateral to the otic vesicle and to r3" (Pöpperl et al., 2000).

This fusion of the streams of migrating crest that will form the pharyngeal cartilages mightdirectly underlie the cartilage fusions that occur in the AP axis. We suppose that the anterior twostreams in wild types give rise, in a restricted manner, to the postmigratory arch neural crest (theectomesenchyme) of the corresponding anterior two pharyngeal segments, the mandibular and


hyoid. Evidence for this supposition in zebrafish comes from clonal analysis: single premigratoryneural crest cells contribute progeny to one or the other of these pharyngeal segments but not toboth together. The segments, by this analysis, are cell lineage compartments (Schilling andKimmel, 1994). Disruption of compartmentation, as would seem to be the case in lzr/pbx4mutants, might be at least partly responsible for the AP fusions of cartilage elements in the firstand second arches. Time lapse and cell lineage analyses in the mutant are required to ascertainwhether individual neural crest cells contribute to both first and second arch cartilages, as wouldbe predicted by this model.

A further issue is whether the fusion of crest streams, and therefore, as we propose, ofcartilage elements, is itself due to an intrinsic defect in the specification of AP identity inpremigratory neural crest populations, or to the absence of lzr-dependent signals that wouldotherwise serve to separate the crest into streams. Hox genes, and by inference, lzr/pbx4, areclearly implicated in the former mechanism, such that an AP “hox code” established within thehindbrain is carried into the periphery by the neural crest (Hunt et al., 1991). It is easy to imaginehow the downstream effectors of Hox/Pbx4 complexes could result in the mutual repulsion ofcrest streams based on their AP identity. However there is also support for a model by whichcranial crest is separated into streams by extrinsic influences. A number of groups havedemonstrated the existence of non-cell-autonomous signals, acting either within the hindbrain orwithin the mesoderm of the head periphery, that influence the organization of neural crest intostreams (Farlie et al., 1999; Graham et al., 1993; Sechrist et al., 1993). Recent chimeric analysisof the ErbB4 mutant mouse demonstrated that the mis-migration of a late population of neuralcrest cells in that mutant results from the absence of signals that normally function to keep thesestreams separate (Golding et al., 2000). The primary role of ErbB4, a receptor expressed inhindbrain rhombomeres 3 and 5, may therefore be to establish a crest-inhibitory domain adjacentto those segments. Although lzr/pbx4 mutants have a related phenotype, the lzr/pbx4 expressionpattern offers little in the way of an indication of where it might primarily function to control crestmigration. Whether lzr/pbx4 functions autonomously within the neural crest or non-autonomouslyin its environment remains to be determined by genetic mosaic analysis. In either case, dysfunctionof hox genes is implicated by the lzr/pbx4 mutant phenotype (Pöpperl et al., 2000).

While lzr/pbx4 mutants exhibit a simplified cartilage pattern that may result from loss of hoxfunction, valentino (val) mutants undergo a cartilage duplication that correlates with gain of hoxfunction. val encodes a bZIP transcription factor, homologous to the mouse kreisler (kr) gene,and a regulator of hox gene expression within hindbrain rhombomeres 5 and 6 (Moens et al.,1998, Cordes et al., 1994; Manzanares et al., 1997, 1999a, b; McKay et al., 1994). In valmutants, an ectopic cartilage resembling the small interhyal cartilage characteristic of the hyoidarch is associated with the dorsal-most aspect of the ceratobranchial cartilage in the thirdpharyngeal arch (Fig. 6A-E; Moens et al., 1998). Normally the third and more posterior archesare simplified versions of the first two arches, containing only ventral elements homologous toMeckel’s and the ceratohyal. The ectopic putative interhyal in the third arch is reminiscent ofectopic cartilage elements on the hyoid bone of kreisler mutant mice (Frohman et al., 1993).

The cartilage duplication in val/kr mutants correlates closely with changes in hox expression.In both the fish and the mouse, loss of val/kr function results in the ectopic expression in the thirdarch of hox genes normally in the second arch, and the loss of expression of hox genes normally inthe third arch. Thus hoxb2, which is normally expressed in the second arch at 19 h, is ectopically


expressed in val/kr mutants in a population of crest cells migrating into the third arch, whilehoxb3 expression in the region of the hindbrain that gives rise to third arch crest is reduced (Fig.6H, I; Prince et al., 1998). The loss of hoxb3 expression is a direct effect of loss of val/krfunction, as kr has been shown to be a direct regulator of hoxb3 expression in the mouse(Manzanares et al., 1997). The ectopic hoxb2 expression may result from the loss, in val/krmutants, of almost all of r5. As a result, r4-derived crest, which normally migrates anterior to theotic vesicle and into the 2nd arch, migrates caudal to the otic vesicle and into the 3rd arch as well,carrying with it hoxb2 expression (Fig. 6F, G) and, by inference, the wherewithal to specify aninterhyal cartilage. Alternatively, the r6-derived third crest stream could be homeoticallytransformed into second arch identity.

Mutations in hox genes themselves have not been described so far in zebrafish. Targetedmutagenesis of Hoxa2 in the mouse shows a crucial role for this gene in patterning along theanterior-posterior axis (Gendron-Maguire et al., 1993; Riijli et al., 1993). The gene is expressedin the hyoid but not the mandibular pharyngeal segment. Hoxa2 mutants have a homeotic,anterior transformation phenotype: skeletal derivatives of the mandibular segment are duplicatedand derivatives of the hyoid segment are deleted, revealing its role as a selector gene specifyingsegmental identity of these anterior pharyngeal segments. Gain-of-function analyses have beencarried for Hoxa2 in the chick and Xenopus (Grammatopoulos et al, 2000; Pasqualetti et al.,2000), and the results of these studies are entirely consistent with it functioning as a homeoticselector gene, for the phenotypes are essentially the opposite to those of the loss-of-functionmutants. Thus, when transgenic techniques are used to express Hox2 in the mandibular segment,the cartilages are homeotically transformed to a hyoid phenotype, i.e. there is posteriortransformation.

We cannot so easily reconcile dorsal-ventral patterning defects such as we observe in lzrmutants with hox gene patterning. Recently Hoxa2 and Hoxb2 have been implicated indorsoventral patterning in the mouse hindbrain (Davenne et al., 1999), but it is not known if theyhave a corresponding role in the pharyngeal arches. As we come to next however, it is clear thatdorsal and ventral cartilages have separate identities.


Fig. 7. Phenotype and gene expression in WT and sucmutants (from Miller et al., 2000). WT are to the left andsuc mutants to the right. A, B: The craniofacial appearancein left side view of WT and a suc mutant at day 4. C, D.The cartilage phenotype at day 4. Alcian blue-labeled, flat-mounted elements of the mandibular and hyoid arches(pharyngeal segments 1 and 2). Abbreviations as in Fig. 1.The asterisks in D show regions of ventral cartilage that maycorrespond to m and ch in the first and second segmentsrespectively. Scale bar 100 µm. E. Whole-mount RNA insitu preparation (36 h, left side view) showing segmentalexpression of suc/et-1 in the ventral pharyngeal arches. Thefirst four segments are indicated (1-4), and labeling is alsopresent in endodermal pouches 2 and 3 (p2, p3). F-K.Targets of suc signaling, as revealed by comparing the RNAexpression patterns in WT and suc mutants. Left side viewsof whole mounted embryos. F, G: goosecoid (gsc), 30 h. H,I: dHAND, 28 h. J, K, dlx2 (in red) and EphA3 (in blue), 32h. Scale bar (K) 50 µm.

A CONSERVED ENDOTHELIN-1 SIGNAL SPECIFIES VENTRAL CARTILAGEIDENTITY IN THE PHARYNGEAL ARCHES

One hundred and nine mutations were isolated in a large-scale genetic screen in Tübingen bytheir prominent craniofacial phenotypes (Schillinget al., 1996; Piotrowski et al. 1996; e.g. Fig. 7A,B). Staining the mutants with Alcian bluerevealed the underlying defects in the pharyngealcartilages (e.g. Fig. 7C, D, see also below, Fig. 9).Among them, six mutations representing fourgenes named sucker (suc), sturgeon (stu),schmerle (she), and hoover (hoo) were placedtogether into a single phenotypic "anterior arch"class because they primarily affected cartilages inthe mandibular and hyoid arches (Piotrowski et al.,1996).

Anterior arch mutants share three relateddefects in both the mandibular and hyoid arches.Ventral cartilages are variably reduced in size,changed in orientation, and fused to the dorsalones (Piotrowski et al., 1996; Kimmel et al.,1998), i. e., the joint normally present between thedorsal and ventral cartilages is missing. Mutationof suc causes the most severe loss of first andsecond arch ventral cartilages. This phenotypesuggested an Endothelin-1 (Et-1) gene as acandidate for harboring the suc mutation, because

in mouse embryos homozygous for targeted disruption of Et-1, Meckel's cartilage in themandibular arch and the ventral cartilage in the hyoid arch are both severely reduced (Kurihara etal., 1994). That is, the craniofacial phenotypes of homozygous mouse Et-1 mutants and zebrafish


homozygous suc mutants are essentially identical. Genetically mapping of suc revealed conservedsynteny of the relevant genomic regions between zebrafish and humans, and mapping, seqencing,and rescue analyses demonstrated that suc encodes a zebrafish Et-1 homolog (Miller et al, 2000).

Endothelin-1 is a secreted peptide (Yanagisawa et al., 1988), and marker analyses suggeststhat in the pharyngeal arches it is secreted by cells that closely neighbor postmigratory neural crestcells in the ventral pharyngeal arches (Miller et al., 2000). Here the peptide could serve as apositional signal to specify the crest towards development of ventral cartilages (review: Francis-West et al., 1998). Expression of suc/et-1 begins during late segmentation stages (ca. 16 h),several hours after crest migration begins and many hours before cartilage formation(chondrogenesis) begins. suc/et-1 expression is localized to at least three pharyngeal arch tissues-- the paraxial mesodermal arch cores, the surface ectoderm, and pharyngeal endoderm. Neuralcrest itself does not appear to express the gene. In agreement with this finding, mosaic analysissuggests Suc/Et-1 signaling function is required in the environment of the neural crest cells, notautonomously in the crest (Miller et al., 2000).

At the time (pharyngula stages; ca. 22-28 h) when signaling and response seems to occur, thecrest cells that will form the cartilages shown in Fig. 7C have completed (or nearly completed)their migration. Observed gene expression defects in suc/et-1 mutants begin at this later,postmigratory stage and ET-1 can rescue ventral cartilage formation in suc/et-1 mutants as late as28 h, well after most of the cranial neural crest has completed migration (Miller et al., 2000).Hence the signal seems unimportant for crest migration, rather it may act in specification at apostmigratory stage. The same conclusion was reached in studies of the Et-1 receptor, EndrA,function in mice (Clouthier et al., 2000).

Correlating with the later defects being restricted to more ventral cartilages, suc/et-1 isexpressed only ventrally in the embryonic pharyngeal arches (Fig. 7E). There, as we propose, itacts to specify the postmigratory neural crest cells to their ventral identities. By this model a highlevel of the Suc/Et-1 signal is a necessary requirement in the specification of ventral cartilageidentity. Whether this signal is sufficient for ventral specification could be tested in gain offunction experiments, but so far no such experiments have been reported.

MOLECULAR TARGETS AND DEVELOPMENTAL ROLES OF ENDOTHELIN-1SIGNALING

How is the Suc/Et-1 signal transduced into a 'ventral identity' response? The Et-1 signaltransduction pathway includes a membrane-bound metalloprotease Endothelin-Converting-Enzyme-1 (ECE-1) (Xu et al., 1994). This enzyme, present in both cells that make Et-1 and cellsthat receive Et-1, cleaves a longer precursor form of big Et-1 into its secreted mature 21-aminoacid form (Xu et al., 1994; Yanagisawa et al., 1998). A seven-transmembrane, G-protein-coupledreceptor, endothelin type A receptor (EndrA), is expressed by the responding cells. Targeteddisruption in mice of either ECE-1 (Yanagisawa et al., 1998) or of EndrA (Clouthier et al., 1998)produces a craniofacial phenotype indistinguishable to that of Et-1 mutants. Furthermore,pharmacological inhibition of ENDRA in avian embryos inhibits ventral arch cartilage formation,suggesting a role for Et-1/ENDRA signaling has been highly conserved in gnathostome evolution(Kempf et al., 1998). As yet, genes homologous to Ece-1 and EndrA have not been described in


zebrafish; they are, of course, candidates for other, as yet molecularly unknown, anterior archgenes identified in the mutant screens (see below).

The signal transduction pathway following EndrA receptor activation by Et-1 has been wellcharacterized in physiological contexts (see Huggins and Pelton, 1997; Pollock and Highsmith,1998 for review). However, signal transduction downstream of EndrA during embryonicdevelopment is largely unknown. Regardless of how the pathway works, a prominent downstreamresponse is well understood, a transcriptional activation of target genes. Like the Suc/Et-1 signalitself, this response to the signal by neural crest-derived cartilage progenitors has beenevolutionarily conserved to a large extent between mouse and zebrafish. Targets in both speciesinclude homeobox genes of Dlx and Msx families, the homeobox gene goosecoid (gsc), and thebHLH gene dHAND (Fig. 7; Clouthier et al., 1998; Thomas et al., 1998; Clouthier et al., 2000,Miller et al., 2000).

In zebrafish, there clearly are several distinct styles of target regulation. Some genes,including dHAND, EphA3, msxE, and dlx3 begin expression specifically in the ventral archmesenchyme of the pharyngula (Fig. 7H, J). The domains of expression are all similar andcorrelate both temporally and spatially with suc/et-1 expression itself. Further, the expression ofthese genes is severely reduced in suc/et-1 mutants (Fig. 7I, K). Hence, as we suppose, thesegenes are activated in these ventral arch domains specifically by the Suc/Et-1 signal. In contrast,regulation of the gene dlx2, which in combination with dlx1 is critical for arch patterning (Qui etal., 1995; 1997), seems more complex. Expression of dlx2 begins before suc/et-1 expression isdetected, and may be present in all migrating crest, including that destined for the dorsal domainas well as the ventral domain. Later, the ventral but not dorsal expression of dlx2 comes under thepositive control of suc/et-1, as revealed by down-regulation of ventral but not dorsal dlx2expression in suc/et-1 mutants. A third variation on the theme is of gsc. This gene comes on wellafter the Suc/Et-1 signal (at about 28-30 h), first in the ventral domain and then in the dorsaldomain (Fig. 7F). Ventral but not dorsal expression depends on the signal, as revealed by themissing ventral expression domain in suc/et-1 mutants (Fig. 7G).

Pharyngeal arch expression of mammalian orthologs of many of these genes also requires Et-1/EndrA signaling. In Et-1 and EndrA mutant mice, ventral (distal) arch dHAND expression isdramatically affected (Thomas et al., 1998; Clouthier et al., 2000). EndrA mutant mice also lackGsc and Dlx3 expression (Clouthier et al., 1998; Clouthier et al., 2000). Several differences existhowever; for example Et-1 mutant mice have normal Msx1 expression (Thomas et al., 1998)whereas an msx gene in fish seems to more directly require Et-1 signaling (Miller et al., 2000).

A wonderfully interesting issue for future studies is to learn how the upregulation of thesetranscription factors by suc/et-1 relates to patterning the later cartilage morphogenesis that themutant phenotype shows us is controlled by the signal. Mutagenesis of some of these targetgenes in the mouse (e.g. Gsc, Dlx2, dHAND; Yamada et al., 1995; Rivera-Perez et al., 1995; Qiuet al, 1995; Thomas et al., 1998) and fish (dHAND, C.T.M., D. Yelon, D. Stainier, and C.B.K,unpublished results) reveals that their functions are also essential for properly patternedcraniofacial chondrogenesis. What is missing at this point for better understanding is a carefulcomparative analysis of the phenotypes of embryos bearing mutations in the upstream anddownstream genes. In zebrafish, an interesting beginning of such pathway analysis can be madeby differences in the phenotypes of suc/et-1 mutants and other mutants in the anterior arch class.


The latter are generally less severe than suc/et-1- with respect to the loss of ventral cartilage(Kimmel et al., 1998; see below). Where the suc/et-1 targets have been examined in these othermutants (e.g. in schmerle, C. T. M., unpublished work) they are also observed to be

downregulated, but not to such a great extent as in suc/et-1-. Hence these studies support theproposals that the genes all function in the same pathway (Piotrowski et al., 1996; Kimmel et al.,1998) and further that transcriptional regulation of target genes in the cells responding to theSuc/Et-1 signal is quantitatively correlated with ventral cartilage development.

Another suc/et-1 target in zebrafish that immediately connects to morphogenesis is the geneEphA3, expressed like dHAND in the ventral mesenchyme specifically, and downregulated insuc/et-1 mutants (Fig. 7J, K). Eph genes encode receptor tyrosine kinases. They are activated bycell-bound ligands, ephrins, and generally Eph activation results in an avoidance response of theEph-expressing cell towards the ephrin-expressing neighbor. Such avoidances have been shownto be important for establishing polarity, e.g. in the polarized spread of Eph-expressing retinalaxons across an ephrin-expressing midbrain tectum, the synaptic target tissue of these axons(review: Holder and Klein, 1999). Eph-ephrin mediated repulsion can also function indevelopmental boundary formation and segmental restriction (Robinson et al., 1997). The bestknown case is in the hindbrain. This tissue is made from a segmental series of rhombomeres(review; Guthrie, 1995), and mixing of cells of adjacent rhombomeres is restricted (Fraser et al.,1990; Birgbauer and Fraser, 1994). Correlated with the restriction, cells on one side of adeveloping rhombomere boundary express an Eph and cells on the other side the correspondingephrin (Xu et al., 1999). When this situation is perturbed, as in overexpression studies (Xu et al.,1995; Mellitzer et al., 1999), the restrictions towards mixing are lifted. Furthermore, in thevalentino mutant discussed earlier, loss of rhombomere boundaries correlates with the loss ofnormal ephrin-Eph boundaries in the r4-r7 region of the hindbrain (Cooke et al., 2001).

We can readily suppose that ephrin-Eph mediation of cellular patterning, including avoidanceinteractions at developmental boundaries, is also a feature of cartilage development, and is undercontrol of Et-1 signaling. As noted above, a prominent phenotype of all of the anterior archmutants is the loss of the joint region that normally separates the dorsal and ventral cartilages ineach segment. Imagine that the joint is the site of a developmental boundary present in wild-typeembryos but missing in mutants. The position where the joint appears corresponds with theborder of the EphA3 expression domain. We know that the dorsal and ventral cartilages in asingle segment arise as two separate sites of chondrification within a single precartilagecondensation (Fig. 8; Schilling and Kimmel, 1997). Further, the joint develops between thesesites as a special region of cells that do not chondrify: rather the dorsal and ventral zones remainseparate as chondrification spreads through the condensation (Fig. 8). Hence the joint regionbehaves as a morphogenetic dorsal-ventral boundary within the precartilage condensation.Presence of EphA3 on the cells on the ventral side of this putative developmental boundarypredicts an appropriate ligand (e.g. ephrinA2 or ephrinA3) on the dorsal side, if indeed the Eph-ephrin system is included in the molecular machinery that patterns the joint.


Fig. 8. Chondrification of the dorsal (pq) and ventral (m) cartilageswithin a single precartilage condensation in the mandibular segment.Left-side views of Alcian blue-labeled whole mounted embryos,photographed with Nomarski optics (from Schilling and Kimmel (1997).A. 53 h. The eye is to the top of the field and the rudiment of theadductor mandibular muscle (am, mediating movement between pq andm) is visible. Other abbreviations as in Fig. 1. The dorsal cartilage, thepalatoquadrate (pq) has begun to chondrify (become Alcian-positive)within the same mandibular condensation that will also form Meckel’scartilage (m). An unlabeled bridge of condensed mesenchyme, presentjust beneath the muscle rudiment, connects these two labeled regionsand shows there is but a single condensation. See also Fig. 10, for theappearance of the condendensation in a live embryo. This bridgingregion will eventually develop as the joint between pq and m. B.Another embryo at about the same stage in which m is also lightlylabeled, and the joint-forming bridge remains unlabled. The arrowheadindicates the hyomandibular pouch, separating the mandibular and hyoidarches. C. 60 h. Both cartilages, but not the joint region between them,are strongly labeled. Scale bar 50 µm.

Finally, downregulation of EphA3 could also at least partly underlie what we propose is apolarity change in the anterior arch mutants. A prominent aspect of the suc/et-1 mutantphenotype is the orientation of the remnant of persisting ventral cartilage (Fig. 9; Kimmel et al.,1998). As we indicate by the direction of the green arrows in Fig. 9, the regions of cartilage we

interpret as a reduced Meckel's cartilage in the mandibular archand ceratohyal in the hyoid arch both point in a posteriordirection rather than an anterior direction as in as in the wild type.Correlating with the less severe loss of ventral cartilage in she andstu mutants the orientations of the same cartilage regions in these

mutants are more normal, stu- more so than she-. Hence, in away that matches severity of the phenotype measured by amountof ventral cartilage present in the mutants there is a progressivelymore severe polarity change of the ventral cartilage in the series

stu- < she- < suc/et-1-. As explained next, the polarity changecould be revealing a reversal in the direction of outgrowth of theventral cartilage during morphogenesis. The reversal could wellbe a consequence of a missing repulsive interaction between cellsthat normally have distinct ventral and dorsal identities.

MORPHOGENESIS: JOINTS MAKE STACKS

In the suc/et-1-dependent regulation of EphA3 we envisage aparticular mode of cellular behavior that produces specific

cartilage form. Namely, repulsivecellular interactions occur at theincipient dorsal-ventral joint regionand are ultimately responsible, in partanyway, for morphogenetic behaviorscarving out separate dorsal andventral elements. As we propose tobe the case in suc/et-1 mutants, whenthe behaviors fail so too doesmorphogenesis. Repulsion betweencells may be part of the suite ofinteractions that occur during thestages of “chondrogenesis”, whenspecified but still functionallyimmature (or "undifferentiated")mesenchymal cells in so-calledprecartilage condensations form

specifically-shaped elements of now specialized chondrocytes and cartilage matrix. Past authorshave conceptually separated cartilage cytodifferentiation (histogenesis) and cartilage shaping(morphogenesis; e.g. Thorogood, 1983; Noden et al., 1999). Features of both can occur togetherin the development of the zebrafish head skeleton. The embryonic cartilage matrix begins to belabeled with Alcian blue just as the embryo enters the third day following fertilization (Fig. 8).


Fig. 9. A phenotypic series of anterior arch mutants,arranged in approximate order of severity (unpublishedexamples, C. T. M; see also Kimmel et al., 1998). A.Alcian-stained flat mounts of mandibular and hyoidcartilages. Abbreviations and orientation as in Fig. 1D. The3 mutations are at 3 separate loci; sucker (suc) encodesEndothelin-1, sturgeon (stu) and schmerle (she) have notbeen identified molecularly. The mutants are arranged byseverity of the phenotypes, as judged by the amount ofcartilage present in the mutants; stu- is the least severe andsuc- the most severe. B. Drawings illustrating ourinterpretation of the mutant phenotypes. Red: dorsal

til i th t t G t l til

Nearly simultaneously, we also see early signs of cells taking on regular lined-up arrangementsthat could underlie cartilage shaping.

Patterning can be complex from theoutset, and exquisitely precise. For example,we have shown, studying Alcian-labeledmaterial, that in the mandibular and hyoidsegments on each side of the pharynx separatechondrogenic sites develop stereotypically outof a single precartilage condensation(Schilling and Kimmel, 1997). The patternfor the mandibular segment was discussedabove, where we show two separatechondrogenic sites present in thecondensation represent the future dorsalpalatoquadrate and ventral Meckel’s cartilage(Fig. 8). In the hyoid segment there are threesites, two of them corresponding differentregions of the dorsal hyosymplectic cartilage,and the third site representing the ventralcartilage in this segment (the ceratohyal). Innewer, unpublished studies we have recentlyconfirmed this pattern by using confocalmicroscopy to image the developingcondensations in live embryos stained vitallywith BODIPY ceramide. Figure 10 shows anexample of this imaging at two timesseparated by about six hours. So far we havestudied one of these chondrogenic sites in themost detail, a region of the dorsal cartilagetermed the symplectic (Kimmel et al., 1998).It has the simplest structure, for thesymplectic develop as a single file of cartilage

cells (as shown for a stage about two days later in Fig. 11A, sy). We first recognize thesymplectic rudiment when it has as few as two or three cells that emerge out of the precartilagecondensation, about 5 are present in Fig. 10A. These cells appear at a specific location of thecondensation, just where the dorsal-ventral joint will eventually develop, and more cells add tothe symplectic row as development continues (Fig. 10B). Using lineage tracer dye to labelindividual cells in the developing symplectic (Kimmel et al., 1998), we found that lengthening ofthe cell row involves progressive addition of cells, not randomly along its length, but preferentiallyat or near the joint region, from which the symplectic thus appears to grow outwards. Thedirection of outgrowth is indicated by the arrow in Fig. 11A.


Fig. 10. Morphogenesis (chondrogenesis) in the anterior pharyngeal wall,as revealed by BODIPY ceramide labeling in live WT preparations (leftside view) A. 52 h, B. 58 h (unpublished confocal micrographs, M. Jonesand C. B. K.). Negative images are shown, allowing the various regions ofdeveloping cartilages to be pseudocolored for clarity of presentation. Themature cartilages are shown in the same orientation in Fig. 1D. Here thetwo distinctive regions of the dorsal hyoid cartilage (the hyosymplectic,Fig. 1) are indicated separately, the strut-like symplectic region (sy), andthe plate-like hyomandibular region (hm). The more dorsal part(hyomandibular region; for more detail see Kimmel et al., 1998) is leftuncolored. Both of these cartilage-forming regions, as well as the ventralcartilage in this segment, the ceratohyal (ch), are developing out of acommon primordium, the hyoid condensation. In contrast, thepalatoquadrate (pq), and Meckel’s cartilage (only partly included here)can be seen to be developing out of another shared primordium, themandibular condensation, as described previously from analysis of fixedmaterial (Schilling and Kimmel, 1997; see also Fig. 8).

This pattern can be understoodby proposing that the specificneighborhood of the condensationthat later forms the dorsal-ventraljoint acts, at the earlier stage, as anorganizing center forchondrogenesis. Cells in the jointneighborhood are recruited to joininto cell rows, or 'stacks', perhapsby intercalating with one another,suggested from the lineage studies(Kimmel et al., 1998). By thishypothesis the symplectic developsas a single stack elongating in aventral direction out of the jointneighborhood. This 'joints makestacks' hypothesis can readilyaccomodate the polarity changediscussed above in suc/et-1mutants: We imagine the polarizedstacking of cells making the ventralcartilage remnants is disrupted.Stacks now emerge from themutant joint region with reversed

orientation (Fig. 11B). Confocal time-lapse analyses of chondrogenesis in wild-type and mutantembryos could directly test the hypothesis.

By the joints makes stacks hypothesis the patterning of cartilage shape is proximatelycontrolled, at least in part, within the precartilage condnsation itself, rather than by thesurrounding environment. If so, environmental signals would act to organize the primordium atan earlier stage. For the Suc/Et-1 signal, such timing nicely fits the early expression observed ofsuc/et-1 -- expression is present more than a day before chondrogenesis is initiated. We as yethave no evidence for later primordium-intrinsic versus extrinsic control of patterning, but olderexperiments in other species support intrinsic regulation. As studied particularly in avianembryos, precartilage condensations put into organ cultures develop into cartilages ofrecognizably different shapes according to the specific rudiment isolated (Jacobson and Fell, 1941;Weiss and Amprino, 1940; reviewed in Thorogood, 1983).

The mode of cartilage development just described, wherein separate cartilages arise within asingle condensation, also occurs in the mandibular arch and during cranial and pharyngealdevelopment of other species of fish (Bertmar, 1959). However, a different pattern is present inlimb development of a variety of tetrapods. Shubin and Alberch (1986) found that separatecartilages within the limb each arise from separate Alcian blue-positive condensations. Thecondensations themselves arise by a characteristic pattern of branching and budding from anearlier single Alcian-positive condensation. So the relative timing of matrix deposition versuselement formation seems quite different in the two modes. However, we think that the difference


Fig. 11. Abnormally curving chondrocyte stacks in the ventralregion of the hyoid cartilage of a suc mutant embryo, as compared tothe pattern in a WT (to the left). Arrows indicate proposedpolarities (see text), as in Fig. 9. We propose that the abnormalpolarities of the ventral cartilages when Suc/Et-1 signaling isdefective is underlain by abnormal orientation of stacking of thechondrocytes (modified from Kimmel et al., 1998). hm, sy:hyomandibular and symplectic regions of the hyosymplecticcartilage Other abbreviations as in Fig 1 Scale bar: 100 µm

is not really a fundamental one, for wesee an example of what looks like thelimb bud pattern in the hyoid arch. Theinterhyal cartilage, positioned betweenthe hs and the ch, develops about a daylater than its neighbors, and it developsby budding out of the young hs. In factit comes from the same dorsal-ventraljoint region we have been considering,but only after this region has alsodeveloped an Alcian-positive matrix.

PROSPECTUS

For each of the kinds of mutants weconsidered, the phenotypes are pattern

deformations, simplified from the wild-type pattern in particular ways rather than just beingrandom losses of pattern elements. The mutations are all loss-of-function alleles. Connectingthese two observations suggests directly how the wild-type functions of the particular genes arecontributing to pattern complexity.

The midline group mutants (represented by slb/wnt11- for example) deform a bilateral two-dimensional sheet of complex morphology into a one-dimensional array just along the midline.The implied wild-type function of the genes of this group is to position the initial normallybilateral sites of chondrogenesis of the anterior basicranium, and perhaps by specifying wheremigrating neural crest will settle down, as further studies could reveal. Further, molecular studiesreveal that these genes function in intracellular signaling pathways crucial for embryonic midlinedevelopment, and suggest a particular scenario in which Shh from the ventral neural tube is alikely proximate influence patterning cartilage development along the mediolateral axis.

Cartilage fusions in lzr/pbx4 mutants correlate with earlier fusions of segmental streams ofmigrating neural crest that will form these cartilages. Hence we suggest that this gene functionsin establishment of developmental boundaries along the anterior-posterior axis that restricts neuralcrest migration, and that restricted migration in part underlies the formation of separate cartilagesin adjacent pharyngeal segments. Because Pbx and Hox proteins interact, hox gene function isimplicated. The same is the case for val/kr function; we can interpret the cartilage phenotype inval/kr mutants as being due to either a mis-migration of neural crest along the AP axis or to achange in AP identity; either interpretation is consistent with defective hox gene-dependentpatterning.

The anterior arch mutant (e.g. suc/et-1) phenotypes simplify pattern by reducing developmentof ventral pharyngeal cartilages and fusing them into the dorsal ones. Along with the molecularwork establishing where and when the Suc/Et-1 signal must act, the studies show a crucialsignaling that occurs relatively late in the embryo: The local environment acts upon a primordiumof postmigratory neural crest. Genes that could directly affect cartilage morphogenesis (EphA3)are controlled by this signal.


The mutational analyses have revealed genes that function in several different geneticpathways. Further, by the phenotypic changes we see that each genetic pathway connects topatterning along only one of the three coordinate axes of the embryo -- medial-lateral (Wnt,Nodal, Hedgehog), anterior-posterior (Pbx, Val, Hox and hence perhaps involving retinoic acid),and dorsal-ventral (Et-1). These findings motivate the hypothesis that head cartilage morphologyarises through largely separate control of pattern formation along these three separate axes.Characterizing new mutants is a way to learn whether this proposal has any validity.

All of the genes identified so far act in pattern specification, apparently rather distantlyupstream of cartilage morphogenesis itself. Hence the mutants now available may not be veryuseful for examining details of control of morphogenesis: e.g. what underlies the assembly of cellrows postulated by the joints make stacks hypothesis? Upstream functions might be expected forsome of the mutants, like ones reviewed above with pleiotropic phenotypes that include disruptionof embryogenesis before cartilage progenitors are even present. Hence, a way to find later actinggenes is to screen later, and screen more specifically for very discrete cartilage phenotypes.Screens by Alcian blue labeling of larval cartilages are in progress, and it is clear from initialresults that mutants with very subtle phenotypes can be identified and recovered, and that theprevious screens are far from saturation. Hence the prospect for continued progress is brightindeed.

ACKNOWLEDGMENTS

We are very grateful to Bonnie Ullmann and Martha Jones, who contributed to new studiesdescribed here. Unpublished work was supported by NIH grants DE13834, NS17963 andHD22486.

REFERENCES

Abu-Shaar, M., and Mann, R. (1998). Generation of multiple antagonistic domains along the proximodistal axis duringDrosophila leg development. Development 125, 3821-3830.

Amacher, S. L., and Kimmel, C. B. (1998). Promoting notochord fate and repressing muscle development in zebrafish axialmesoderm. Development 125, 1397-1406.

Bateson, W. (1894). "Materials for the Study of Variation." Macmillan, London.Bertmar, G. (1959). On the ontogeny of chondral skull in Characidae, with a discussion on the chondroncranial base and the

visceral chondrocranium in fishes. Acta Zool. 40, 203-364.Birgbaur, E., and Fraser, S. E. (1994). Violation of cell lineage restriction compartments in the chick hindbrain. Development

120, 1347-1356.Brand, M., Heisenberg, C. P., Warga, R. M., Pelegri, F., Karlstrom, R. O., Beuchle, D., Picker, A., Jiang, Y. J., Furutani-Seiki,

M., Van Eeden, F. J. M., Granato, M., Haffter, P., Hammerschmidt, M., Kane, D. A., Kelsh, R. N., Mullins, M. C.,Odenthal, J., and Nuesslein-Volhard, C. (1996). Mutations affecting development of the midline and general body shapeduring zebrafish embryogenesis. Development 123, 129-142.

Casares, F. and Mann, R. (1998). Control of antennal versus leg development in Drosophila. Nature 392, 723-726.Chibon, P. (1967). Marquage nucléaire par la thymidine tritiée des dérivés de la crête neurale chez l’amphibien urodèle

Pluerodeles waltlii Macihah. J. Embryo. Exp. Morph. 18, 343-358.Clouthier, D. E., Hosoda, K., Richardson, J. A., Williams, S. C., Yanagisawa, H., Kuwaki, T., Kumada, M., Hammer, R. E.,

and Yanagisawa, M. (1998). Cranial and cardiac neural crest defects in endothelin-A receptor-deficient mice. Development125, 813-824.

Clouthier, D. E., Williams, S. C., Yanagisawa, H., Wieduwilt, M., Richardson, J. A., and Yanagisawa, M. (2000). Signalingpathways crucial for craniofacial development revealed by Endothelin-A receptor-deficient mice. Dev. Biol. 217, 10-24.

Eph signalling functions downstream of Val to regulate cell sorting and boundary formation in the caudal hindbrain


Cooke, J. E., Moens, C. B., Roth, L. W. A., Durbin, L., Shiomi, K., Brennan, C., Kimmel, C. B., Wilson, S. W. and Holder, N.(2001) Eph signalling functions downstream of Val to regulate cell sorting and boundary formation in the caudal hindbrain.Development, in press.

Cordes, S. P., and Barsh, G.S. (1994). The mouse segmentation gene kr encodes a novel basic domain-leucine zippertranscription factor. Cell 79, 1025-34.

Cubbage, C. C., and Mabee, P.M. (1996). Development of the cranium and paired fins in the zebrafish Dano rerio(Ostariophysi, Cyprinidae). J. Morphol. 229, 121-160.

Davenne, M., Maconochie, M. K., Neun, R., Pattyn, A., Chambon, P. Krumlauf, R., ad Rijli, F. M. (1999). Hoxa2 and Hoxb2control dorsoventral patterns of neuronal development in the rostral hindbrain. Neuron 22, 677-691.

De Beer. G. R. (1937). The Development of the Vertebrate Skull. Oxford: Oxford University Press. Reprinted 1985, Chicago:Chicago University Press.

Farlie, P. G., Kerr, R., Thomas, P., Symes, T., Minichiello, J., Hearn, C. J., and Newgreen, D. (1999). A paraxial exclusionzone creates patterned cranial neural crest cell outgrowth adjacent to rhombomeres 3 and 5. Dev. Biol. 213, 70-84.

Francis-West, P., Ladher, R., Barlow, A., Graveson, A. (1998). Signalling interactions during facial development. Mech. Dev.75, 3-28.

Fraser, S., Keynes, R., and Lumsden, A. (1990). Segmentation in the chick embryo hindbrain is defined by cell lineagerestrictions. Nature 344, 431-435.

Frohman, M. A, Martin, G. R., Cordes, S. P., Halamek, L. P., and Barsch, G. S. (1993). Altered rhombomere-specific geneexpression and hyoid bone differentiation in the mouse segmentation mutant, kreisler (kr). Development 117, 925-936.

Gale, E., Zile, M., and Maden, M. (1999). Hindbrain respecification in the retinoid-deficient quail. Mech. Dev. 89, 43-54.Gendron-Maguire, M., Mallo, M., Zhang, M., and Gridley, T. (1993). Hoxa-2 mutant mant mice exhibit homeotic

transformation of skeletal elements derived from the cranial neural crest. Cell 75, 1317-1331.Golding, J. P., Trainor, P., Krumlauf, R., and Gassmann, M. (2000). Defects in pathfinding by cranial neural crest cells in mice

lacking the neuregulin receptor ErbB4. Nature Cell Biol. 2, 103-9.Gonzalez-Crespo, S., Abu-Shaar, M., Torres, M., Martinez, A. C., Mann, R. S., Morata, G. (1998). Antagonism between

extradenticle function and Hedgehog signalling in the developing limb. Nature 394,196-200.Goodrich, E. S. (1930). Studies on the Structure and Development of Vertebrates. Macmillan. London.Gould, A., Itasaki, N., and Krumlauf, R. (1998). Initiation of rhombomeric Hoxb4 expression requires induction by somites and

a retinoid pathway. Neuron 21, 39-51.Graham, A., Heyman, I., and Lumsden, A. (1993). Even-numbered rhombomeres control the apoptotic elimination of neural

crest cells from odd-numbered rhombomeres in the chick hindbrain. Development 119, 233-245.Grammatopoulos, G. A., Bell, E., Toole, L., Lumsden, A., and Tucker, A. S. (2000). Hoeotic transformation of branchial arch

identity after Hoxa2 overexpression. Development 127, 5355-5365.Griffin, K. J. P., Amacher, S. L., Kimmel, C. B., and Kimelman, D. (1998). Molecular identification of spadetail: regulation of

zebrafish trunk and tail mesoderm formation by T-box genes. Development 125, 3379-3388.Guthrie, S. (1995). The status of the neural segment. Trends Neurosci. 18, 74-79.Hatta, K., Püschel, A. W., and Kimmel, C. B. (1994). Midline signaling in the primordium of the zebrafish anterior central

nervous system. Proc. Natl. Acad. Sci. USA 91, 2061-2065.Heisenberg, C.-P. and Nüsslein-Volhard, C. (1997). The function of silberblick in the positioning of the eye anlage in the

zebrafish embryo. Dev. Biol. 184, 85-95.Heisenberg, C.-P., Tata, M., Rauch, G.-J., Saude. L., Concha, M. L., Geisler, R., Stemple, D. L., Smith, J. C., and Wilson, S.

W. (2000). Silberblick/Wnt11 mediates convergent extension movements during zebrafish gastrulation. Nature 405, 76-81.Holder, N., and Klein, R. (1999). Eph receptors and ephrins: effector molecules of morphogenesis. Development 126, 2033-

2044.Hu, D., and Helms, J. A. (1999). the role of sonic hedgehog in normal and abnormal craniofacial morphogenesis. Development

126, 4873-4884.Huggins, J. P, and Pelton, J. T. (Eds.) (1997). “Endothelins in Biology and Medicine.” CRC Press. Boca Raton.Hughes, M. K., and Hughes, A. L. (1993). Evolution of duplicate genes in a tetraploid animal, Xenopus laevis. Mol. Biol. Evol.

10, 1360-1369.Hunt, P., Gulisano, M., Cook, M., Sham, M. H., Faiella, A., Wilkinson, D., Boncinelli, E., and Krumlauf, R. (1991). A distinct

Hox code for the branchial region of the vertebrate head. Nature 353, 861-4.Jacobson, W., amd Fell, H. B. (1941). The developmental mechanics and potencies of the undifferentiated mesenchyme of the

mandible. Q. J. Microsc. Sci. 82, 563-586.Karlstrom, R. O., Talbot, W. S., and Shier, A. F. (1999). Comparative synteny cloning of zebrafish you-too: mutations in the

Hedgehog target gli2 affect ventral forebrain patterning. Genes Dev. 13, 388-393.Kempf, H., Linares, C., Corvol, P., and Gasc, J.-M. (1998). Pharmacological inactivation of the endothelin type A receptor in

the early chick embryo: a model of mispatterning of the branchial arch derivatives. Development 125, 4931-4941.Kimmel, C. B., Miller, C. T., Kruse, G., Ullmann, B., BreMiller, R. A., Larison, K. D., and Snyder, H. C. (1998). The shaping

of pharyngeal cartilages during early development of the zebrafish. Dev. Biol. 203, 245-263.

Kühl. M., Sheldahl, L. C., Park, M., Miller, J. R., and Moon, R. T. (2000). The Wnt/ Ca2+ pathway. Trends Genet. 16, 279-283.


Kurihara, Y., Kurihara, H., Suzuki, H., Kodama, T., Maemura, K., Nagai, R., Oda. H., Kuwaki, T., Cao, W.-H., Kamada, N.,Jishage, K., Ouchi, YU., Azuma, S., Toyoda, Y., Ishikawa, T., Kumada, M. and Yazaki, Y. (1994). Elevated blood pressureand craniofacial abnormalities in mice deficient in endothelin-1. Nature 368, 703-710.

Languille, R. M., and Hall, B. K. (1986). Evidence of cranial neural crest contribution to the skeleton of the sea lamprey,Petromyzon marinus. In "Progress in Developmental Biology", Part B, pp. 263-266. Alan R. Liss, Inc.

Langille, R. M., and Hall, B. K. (1988). Role of the neural crest in development of the cartilaginous cranial and visceralskeleton of the medaka, Oryzias latipes (Teleostei). Anat. Embryol. 177, 297-305.

Le Lièvre, C. S. (1978) Participation of neural crest-derived cells in the genesis of the skull in birds. J. Embryo. Exp. Morph.47, 17-37.

Lumsden, A., and Krumlauf, R. (1996). Patterning the vertebrate neuraxis. Science 274, 1109-1115.Mann, R. and Chan, S.-K. (1996) Extra specificity from extradenticle: the partnership between HOX and PBX/EXD

homeodomain proteins. Trends Genet. 13, 772-774.Manzanares. M., Cordes, S., Kwan, C. T., Sham, M. H,, Barsh, G.S., and Krumlauf, R. (1997). Segmental regulation of Hoxb-

3 by kreisler. Nature 387, 191-5.Manzanares, M., Cordes, S., Ariza-McNaughton, L., Sadl, V., Maruthainar, K., Barsh, G., and Krumlauf, R. (1999a).

Conserved and distinct roles of kreisler in regulation of the paralogous Hoxa3 and Hoxb3 genes. Development 126, 759-69.Manzanares, M., Trainor, P.A., Nonchev, S., Ariza-McNaughton, L., Brodie, J., Gould, A., Marshall, H., Morrison, A., Kwan,

C.T., Sham, M.H., Wilkinson, D.G., and Krumlauf, R. (1999b). The role of kreisler in segmentation during hindbraindevelopment. Dev. Biol. 211, 220-37.

Marshall, H., Morrison, A., Studer, M., Popperl, H., and Krumlauf, R. (1996). Retinoids and Hox genes. FASEB J. 10, 969-78.Mellitzer, G., Xu, Q., and Wilkinson, D. G. (1999). Eph receptor and ephrins restrict cell intermingling and communication.

Nature 400, 77-81.McKay, I. J., Muchamore, I., Krumlauf, R., Maden, M., Lumsden, A., and Lewis, J. (1994). The kreisler mouse: a hindbrain

segmentation mutant that lacks two rhombomeres. Development 120, 2199-211.Miller, C. T., Schilling, T. F., Lee, K.-H., Parker, J., and Kimmel, C. B. (2000). sucker encodes a zebrafish Endothelin-1

required for ventral pharyngeal arch development. Development 127, 3815-3828.Moens, C. B., Cordes, S. P., Giorgianni, M. W., Barsch, G. S., and Kimmel, C. B. (1998). Equivalence in the genetic control of

hindbrain segmentation in fish and mouse. Development 125, 381-391.Niederreither K, Vermot J, Schuhbaur B, Chambon P, and Dolle P. (2000) Retinoic acid synthesis and hindbrain patterning in

the mouse embryo. Development 127, 75-85.Noden, D. M., Marcucio, R., Borycki, A.-G., and Emerson, C. H., Jr. (1999). Differentiation of avian craniofacial muscles. I.

Patterns of early regulatory gene expression and myosin heavy chain synthesis. Dev. Dyn. 216, 96-112.Nonchev, S., Maconochie, M., Gould, A., Morrison, A., and Krumlauf, R. (1997) Cross-regulatory interactions between Hox

genes and the control of segmental expression in the vertebrate central nervous system. Cold Spring Harb. Symp. Quant.Biol. 62, 313-23.

Pasqualetti, M., Ori, M., Nardi, I., Rijli, F. M. (2000) Ectopic Hoxa2 induction after neural crest migration results in homeosisof jaw elements in Xenopus. Development 127, 5367-5378.

Passner, J. M., Ryoo, H. D., Shen, L., Mann, R. S., and Aggarwal, A. K. (1999). Structure of a DNA-bound Ultrabithorax-Extradenticle homeodomain complex. Nature 397, 714-719.

Piper, D. E., Batchelor, A. H., Chang, C. P., Cleary, M. L., and Wolberger, C. (1999). Structure of a HoxB1-Pbx1 heterodimerbound to DNA: role of the hexapeptide and a fourth homeodomain helix in complex formation. Cell 96, 587-597.

Piotrowski, T., Schilling, T. F., Brand, M., Jiang, Y. J., Heisenberg, C. P., Beuchle, D., Grandel, H., Van Eeden, F. J. M.,Furutani-Seiki, M., Granato, M., Haffter, P., Hammerschmidt, M., Kane, D. A., Kelsh, R. N., Mullins, M. C., Odenthal, J.,Warga, R. M., and Nuesslein-Volhard, C. (1996). Jaw and branchial arch mutants in zebrafish . II: anterior arches andcartilage differentiation. Development 123, 345-356.

Pollack, D. M., and Highsmith, R. F. (1998). "Endothelin receptors and Signaling Mechanisms." Springer-Verlag and R.G.Landes Co., Berlin, Heidelberg, Germany and Georgetown, TX USA.

Pogoda, H., Solnica-Krezel, L., Driever, W., and Meyer, D. (2000). The zebrafish forkhead transcription factor FoxH1/Fast1 isa modulator of nodal signaling required for organizer formation. Curr. Biol. 10, 1041-1049.

Pöpperl, H., Rikhof, H., Chang, H., Haffter, P., Kimmel, C. B., and Moens, C.B. (2000). lazarus is a novel pbx gene thatglobally mediates hox gene function in zebrafish. Mol. Cell 6, 255-67.

Prince, V. E., Moens, C. B., Kimmel, C. B., and Ho, R. K. (1998). Zebrafish hox genes: expression in the hindbrain region ofwild-type and mutants of the segmentation gene valentino. Development 125, 393-406.

Qiu, M., Bulfone, A., Martinez, S., Meneses, J. J., Shimamura, K., Pedersen, R. A., and Rubenstein, J. L. R. (1995). Nullmutation of Dlx-2 results in abnormal morphogenesis of proximal first and second branchial arch derivatives and abnormaldifferentiation in the forebrain. Genes Dev. 9, 2523-2538.

Qui, M., Bulfone, A., Ghattas, I., Meneses, J. J., Christensen, L., Sharpe, P. T., Presley, R. Pedersen, R. A., and Rubenstein, J.L. R. (1997). Role of the Dlx homeobox genes in proximodistal patterning of the branchial arches: Mutations of Dlx-1, Dlx-2, and Dlx-1 and -2 alter morphogenesis of proximalskeletoal and soft tissue structures dervived from the first and secondarches. Dev. Biol. 185, 165-184.


Rebagliati, M. R., Toyama, R., Fricke, C., Haffter, P., and Dawid, I. B. (1998). cyclops encodes a nodal-related factor involvedin midline signaling. Proc. Roy. Natl. Adad. Sci. USA 95, 9932-9937.

Rijli, F. M., Mark, M., Lakkaraju, S., Dierich, A., Dolle, P., and Chambon, P. (1993). A homeotic transformation is generatedin the rostral branchial region of the head by disruption of Hoxa-2, which acts as a selector gene. Cell 75, 1333-1349.

Rivera-Pérez, J. A., Mallo, M., Gendron-Maguire, M., and Gridley, T. (1995). goosecoid is not an essential component of themouse gastrula organizer but is required for craniofacial and rib development. Development 121, 3005-3012.

Robinson, V., Smith, A., Flenniken, A. M., and Wilkinson, D. G. (1997). Roles of Eph receptors and ephrins in neural crestpathfinding. Cell Tiss. Res. 290, 265-274.

Rubinstein, A. L., Lee, D., Luo, R., Henion, P. D., and Halpern, M. E. (2000). Genes dependent on zebrafish cyclops functionidentified by AFLP differential gene expression screen. Genesis 1, 86-97.

Sampath, K., Rubinstein, A. L., Cheng, A. M. S., Liang, J. O., Fekany, K., Solnica-Krezel, Korzh, V., Halpern, M. E., andWright, C. V. E. (1998). Induction of the zebrafish ventral brain and floorplate requires cyclops/nodal signalling. Nature395, 185-189.

Schauerete, H. E., van Eden, F. J., Fricke, C., Odenthal, J., Strähle, U., and Haffter, P. (1998). Sonic hedgehog is not requiredfor the induction of medial floor plate cells in the zebrafish. Development 125, 2983-2893.

Schier, A. F., and Shen, M. M. (2000). Nodal signalling in vertebrate development. Nature 403, 385-389.Schilling, T. (1997). Genetic analysis of craniofacial development in the vertebrate embryo. Bioessays 19, 459-68.Schilling, T.F., and Kimmel, C.B. (1994). Segment and cell type lineage restrictions during pharyngeal arch development in

the zebrafish embryo. Development 120, 483-494.Schilling, T., and Kimmel, C. (1997). Musculoskeletal patterning in the pharyngeal segments of the zebrafish embryo.

Development 124, 2945-60.Schilling, T. F., Piotrowski, T., Grandel, H., Brand, M., Heisenberg, C. P., Jiang, Y. J., Beuchle, D., Hammerschmidt, M.,

Kane, D. A., Mullins, M. C., Van Eeden, F. J. M., Kelsh, R. N., Furutani-Seiki, M., Granato, M., Haffter, P., Odenthal, J.,Warga, R. M., Trowe, T., and Nuesslein-Volhard, C. (1996). Jaw and branchial arch mutants in zebrafish. I: branchialarches. Development 123, 329-344.

Sechrist, J., Serbedzija, G. N., Scherson, T., Fraser, S. E., and Bronner-Fraser, M. (1993). Segmental migration of thehindbrain neural crest does not arise from its segmental generation. Development 118, 691-703.

Shen, M. M., and Shier, A. F. (2000) The EGF-CFC gene family in vertebrate development. Trends Genet. 16, 303-309.Shin, M. K., Levorse, J. M., Ingram, R. S., and Tilghman, S. M. (1999). The temporal requirement for endothelin receptor-B

signalling during neural crest development. Nature 402, 496-501.Shubin, N. H. and Alberch, P. (1986). A morphogenetic approach to the origin and basic organization of the tetrapod limb.

Evol. Biol. 20, 318-390.Sirotkin, H. I., Gates, M. A., Kelly, P. D., Schier, A. F., and Talbot, W. S. (2000) fast1 is required for the development of

dorsal axial structures in zebrafish. Curr. Biol. 10, 1051-1054.Studer M., Gavalas, A., Marshall, H., Ariza-McNaughton, L., Rijli, F. M., Chambon, P., and Krumlauf, R. (1998). Genetic

interactions between Hoxa1 and Hoxb1 reveal new roles in regulation of early hindbrain patterning. Development125:1025-36.

Thomas, T., Kurihara, H., Yamagishi, H., Kurihara, Y., Yazaki, Y., Olson, E. N., and Srivastava, D. (1998). A signalingcascade involving endothelin-1, dHAND and Msx1 regulates development of neural-crest-derived branchial archmesenchyme. Development 125, 3005-3014.

Thorogood, P. V. (1983). Morphogenesis of cartilage. In Cartilage, vol. 2. Development, Differentiation and Growth. Hall, B.K., ed. Academic Press, New York. pp. 223-254.

van der Wees J., Schilthuis J. G., Koster C. H., Diesveld-Schipper H., Folkers, G.E. , van der Saag, P.T., Dawson, M. I., ShudoK., van der Burg, B., and Durston, A. J. (1998). Inhibition of retinoic acid receptor-mediated signalling alters positionalidentity in the developing hindbrain. Development 125, 545-56.

Weinberg, E. S., Allende, M. L., Kelly, C. S., Abdelhamid, A., Murakami, T. Andermann, P., Doerre, O. G., Grunwald, D. J.,and Riggleman, B. (1996). Developmental regulation of zebrafish MyoD in wild-type, no tail, and spadetail embryos.Development 122, 271-280.

Weiss, P., and Amprino, R. (1940). The effect of mechanical stress on the differentiation of scleral cartilage in vitro and in theembryo. Growth 4, 245-258.

White, J .C., Highland, M., Kaiser, M., and Clagett-Dame, M. (2000). Vitamin A deficiency results in the dose-dependentacquisition of anterior character and shortening of the caudal hindbrain of the rat embryo. Dev. Biol. 220, 263-84.

Xu, D., Emoto, N., Giaid, A., Slaughter, C., Kaw, S., deWit, D., and Yanagisawa, M. (1994). ECE-1: a membrane-boundmetalloprotease that catalyzes the proteolytic activation of big endothelin-1. Cell 78, 473-485.

Xu, Q., Alldus, G., Holder, N., and Wilkinson, D. (1995). Expression of truncated Sek-1 receptor tyrosine kinase disrupts thesegmental restriction of gene expression in the Xenopus and zebrafish hindbrain. Development 121, 4005-4016.

Xu, Q., Mellitzer, Robinseon, V., and Wilkinson, D. G. (1999). In vivo cell sorting in complementary segmental domainsmediated by Eph receptors and ephrins. Nature 399, 267-271.

Yamada, G., Mansouri, A., Torres, M., Stuart, E. T., Blum, M., Schultz, M., De Robertis, E. M., and Gruss, P. (1995).Targeted mutation of the murine goosecoid gene results in craniofacial defects and neonatal death. Development 121, 2917-2922.


Yamamoto, A., Amacher, S. L., Kim, S.-H., Geissert, D., Kimmel, C. B., and De Robertis, E. M. (1998). Zebrafish paraxialprotocadherin is a downstream target of spadetail involved in morphogenesis of gastrula mesoderm. Development 125,3389-3397.

Yanagisawa, M., Kurihara, H., Kimura, S., Tomobe, Y., Kobayashi, M., Mitsui, Y., Yazaki, Y., Goto, K., and Masaki, T.(1988). A novel potent vasoconstrictor peptide produced by vascular endothelial cells. Nature 332, 411-415.

Yanagisawa, H., Yanagisawa, M., Kapur, R. P., Richardson, J. A., Williams, S. C., Clouthier, D. E., de Wit, D., Emoto, N.,and Hammer, R. E. (1998). Dual genetic pathways of endothelin-mediated intercellular signaling revealed by targeteddisruption of endothelin converting enzyme-1 gene. Development 125, 825-836.

Zhang, J., Talbot, W. S., and SAchier, A. F. (1998). Positional cloning identifies zebrafish one-eyed pinhead as a permissiveEGF-related ligand required during gastrulation. Cell 92, 241-251.

Documents

Specification and Morphogenesis of the Zebrafish Larval