Development 103 Supplement, 141-153 (1988)Printed in Great Britain © T h e Company of Biologists Limited 1988

141

The developmental specification of the vertebrate skull

PETER THOROGOOD

Department of Biology, Medical and Biological Sciences Building, Bassett Crescent East, Southampton SO9 3TU, UK

Summary

The initial form of the embryonic bony skull isdetermined in two ways; cranially, by the relativegrowth of the developing brain, and facially, by thechondrocranium. Both are essentially acting as struc-tural templates around which the bony components ofthe skull are assembled. Assuming, therefore, that thespecification of form and pattern in the facial skeletonoccurs at the formation of the chondrocranium, thispaper will focus on precisely how the chondrocraniumforms. Any acceptable explanation of chondrocranialmorphogenesis must satisfy at least two prerequisites.First, given the constancy of chondrocranial form invertebrates, any model proposed should be equallyapplicable to all vertebrates. Second, it should enableus to answer questions of homology concerning theskull and, in particular, provide explanation for thoseinstances where 'homologous' structures have a differ-ent (lineage) composition.

From studies limited to a small number of amphib-ian, avian and mammalian species, it is apparent that

chondrogenesis in the vertebrate skull is largely, if notentirely, elicited by epitheliomesenchymal tissue inter-actions. Analysis of such interactions (and of thosepromoting osteogenesis) reveals that these are matrix-mediated and, recently, the expression of certain'relevant' matrix components has been shown to bedevelopmentally regulated in a fashion that correlateswith the location and timing of these interactions.From these, and related, observations a morphogen-etic model, the so-called 'Flypaper Model', has beenproposed to explain the specification of chondro-cranial form. A number of predictions arising fromthat model are currently being tested experimentallyand the current status of the model is reviewed.Finally, the ability of this model to satisfy the pre-requisites defined above is assessed.

Key words: morphogenesis, skull, chondrogenesis, neuralcrest, epitheliomesenchymal interactions.

Introduction

Almost two hundred years ago, in 1790, whilestrolling in a Venetian cemetery, Johannes WolfgangGoethe and his manservant came across a sheep'sskull. We are told (de Beer, 1937) that the skull waspicked up by the manservant and handed to Goethewho studied it reflectively. Undoubtedly, Goethe hadstudied skulls of various types before but it was thisparticular incident that led him to propose that thevertebrate skull represented a series of modifiedvertebrae, thus making it serially homologous withthe backbone. In fact, Goethe's 'Vertebral theory ofthe skull' (Goethe, 1820) was wrong but the argu-ments for and against this interpretation continuedfor some seventy years before T. H. Huxley deliveredthe final coup de grdce in his Croonian Lecturepresented to the Royal Society in 1858. Ironically,

one legacy of this idea but one not originally envis-aged by Goethe, was that the entire head might besegmented (Goodrich, 1930). Recently the possibilitythat head mesenchyme, or at least mesodermallyderived mesenchyme, might be segmented has beenresurrected in the context of somitomeres (e.g.Meier, 1981) and as a result of the claim thatsomitomeric organization exists transiently in theembryonic vertebrate head (Jacobsen & Meier, 1987)even though recognizable head somites apparentlynever subsequently form. However, this is currentlythe topic of much debate since an alternative view,propounded recently, is that the head is largely anevolutionary novelty, unsegmented and 'added' ontothe anterior end of a primitive segmented ancestor(Gans & Northcutt, 1983; Northcutt & Gans, 1983;and see Alberch & Kollar, this volume).

Nowadays we do not believe the skull to represent

142 P. Thorogood

a series of modified vertebrae or to be a segmentedstructure (other than in the occipital region and in thegill arches of the viscerocranium) but, nevertheless,we must still ask ourselves how this topographicallycomplex form is specified in development. In fact, theinitial form of the vertebrate bony skull is determinedembryonically in two major ways; cranially, by therelative growth of the developing brain and, facially,by the chondrocranium, a solid unsegmented cartila-ginous structure which presages the appearance ofbone in the developing head (strictly speaking, thechondrocranium consists of 'viscerocranial' and'neurocranial' components; although the single term'chondrocranium' will be used throughout this re-view, most of what follows pertains to the neuro-cranium). Both cranial and facial factors are essen-tially operating epigenetically in that they act asstructural templates around which, and on which, thebony components are subsequently formed (mem-brane and endochondral bone, respectively). Thus,specification of primary form of the facial bonyskeleton occurs with the formation of the chondro-cranium (Bosma, 1976). As is characteristically thecase in morphogenesis, 'order builds upon (pre-existing) order' and, therefore, if the shape of thechondrocranial template is altered, then correspond-ing changes in the form of the skull will result.

The extent to which the basic chondrocranialpattern has been conserved during the course ofevolution is striking. This pattern comprises twopaired elements, the trabeculae anteriorly and theparachordals posteriorly, which grow and fusetogether along mutually contacting edges to form aplate-like cartilage around the margin of which threepairs of capsular cartilages form. These cartilaginouscapsules, olfactory, optic and otic, protect andsupport the organs of smell, sight and hearing,respectively (see right half of Fig. 3). Phylogeneticdifferences in chondrocranial form may be seen butarise simply as consequences of allometric growth(especially determined by the expansion rate of theenclosed brain and sense organs, e.g. Hanken, 1983),heterochronic shifts in the rate and timing of develop-mental events (Alberch & Alberch, 1981; Wake,1980), and sometimes functional adaptations to theenvironment and circumstances of embryonic devel-opment (see Gans, this volume). However, allowingfor some limited variation, chrondrocranial pattern isessentially conserved across the vertebrate taxa (deBeer, 1937).

The mesenchyme cells from which the chondro-cranium forms have a dual origin, being derived fromboth mesoderm and from the neural crest ('ecto-mesenchyme'). Whether or not factors specifying aparticular differentiative fate, or eliciting differen-tiation, are identical for cells derived from both of

these two, broadly defined, lineages is not known atthis stage (see later), but undoubtedly both sourcescontribute chondrogenic cells to the chondrocranium.The precise contributions of the two lineages to thebony and cartilaginous elements of the avian skullhave been analysed independently in two laboratories(Le Lievre, 1978; Noden, 1978; and see Le Douarin &Couly, this volume), both using the heterospecificquail/chick grafting system and, broadly speaking,there is reasonable concordance between the two setsof data (minor disparities may relate simply to pro-cedural differences in grafting technique). Individualskeletal elements may be entirely mesodermal inorigin, or entirely ectomesenchymal or even of mixedcomposition with contributions from both mesodermand neural crest. In the last of these three possi-bilities, there is no mixing, at the level of individualcells, and the cells from each source comprise defin-able regions of the element in question. However,there is no detectable border or interface betweensuch regions other than those that are revealed bytechniques that identify cell origin in heterospecificgrafting experiments (Noden, 1983; and see Noden,this volume).

In addition to this 'cataloguing' of lineage compo-sition of the craniofacial skeleton it has also beenpossible, by analysis of the fate of cells in orthotopicgrafts before their differentiation and at progressivelyearlier intervals postgrafting, to trace migrationroutes of neural-crest-derived cells as they movearound within the developing head prior to reachingthe site at which they will differentiate. Thus, in theavian embryo maps of migration route are known forcranial crest cells at all axial levels along the anteriorneural tube (Noden, 1975).

The chondrocranium is, therefore, a topographi-cally complex shape, composed of cartilage differen-tiating from mesenchyme cells of two lineage sources.At this point, it will be evident that elucidatingmechanisms determining skeletal pattern in the headis (perhaps) a unique problem. Where skeletal tissuesform in the trunk or limb, no large-scale rearrange-ments of cells occur as they do in the head. Thus,skeletogenic cells at limb and trunk sites might bethought of as differentiating in situ whereas in thehead a further dimension exists i.e. that of extensiverearrangements of cells (principally neural-crest-derived) both actively, by migration, and passively,by displacement due to growth.

The requirements of a model

I suggest that the question of how skull form isspecified in the embryo might be better restated as'how is chondrocranial form specified...'?'. Further-

more, any causal model proposed to explain morpho-genesis of the chondrocranium must satisfy severalprerequisites. First, given the relative constancy ofchondrocranial pattern in the vertebrates, any modelput forward should be equally applicable to allvertebrates from Agnatha to Man. It should not beunique or specific to any particular experimentalsystem that might have been used to generate theunderlying data. Second, the model should enable usto answer questions concerning homology of parts ofthe skull and, in particular, provide an explanationfor those instances where 'homologous' structures intwo or more species have a different (lineage) compo-sition. Variation in lineage composition of individualelements clearly creates interpretational problemswhen homologizing structures, where embryonic ori-gin rather than function or morphology is the chiefcriterion (Presley & Steele, 1976; Bellairs & Gans,1983; and discussed at length in de Beer, 1937). Asimilar challenge is presented by those situationswhere experimental perturbation, usually created byheterotopic grafting, can shift the lineage compo-sition of those 'mixed' elements to which both NCand mesoderm-derived cells normally contribute. Forexample, grafts of first branchial arch NC to moreposterior myelencephalic levels in the avian embryonot only create ectopic mandibular elements but, ofrelevance to this question, considerably increase thecrest contribution to the otic capsular cartilage whichwas nevertheless described as 'normal in both shapeand size, despite the substitution of crest-derivedmesenchyme for some mesoderm cells' (Noden,1983). Such changes in composition of individualelements, although created experimentally, parallelchanges arising phylogenetically (i.e. during thecourse of evolution). Morphogenetic models mustencompass both classes of phenomena.

A third point, but one which is an ancillaryquestion rather than a prerequisite of any model, is toask if mechanisms initiating and controlling chondro-genesis in the head are in any sense qualitativelydifferent from those mechanisms underlying cartilageformation in the rest of the axial skeleton, i.e. alongthe length of the vertebral column. It has beensuggested elsewhere that 'induction' of cartilage atthree different sites, head, trunk and limb, probablyreflects three different mechanisms (Kratochwil,1983). Yet, in the absence of a clear and comprehen-sive molecular understanding of the events leading tochondrogenesis in any one of these three systems,such an assumption seems unjustified and the possi-bility of a common mechanism should not be dis-missed too readily (discussed in Hall, 1987).

Skull development 143

Skeletogenic differentiation in the head

Although currently we know virtually nothing of thecommitment state, potency or precise lineage of thecells fated to give rise to the chondrocranium orindeed any of the head skeleton, it is clear thatepitheliomesenchymal tissue interactions, operatinglocally, have some fundamental role in causing headmesenchyme to differentiate into cartilage and bone.For instance, isolated premigratory cranial neuralcrest, from amphibian or avian embryos, lacks theability to form cartilage or bone in vitro and yet inassociation with certain epithelia, in vitro or in vivo,cranial crest will differentiate chondrogenically andosteogenically (Drews, Kocher-Becker & Drews,1972; Epperlein & Lehmann, 1975; Tyler & Hall,1977; Bee & Thorogood, 1980; Graveson & Arm-strong, 1987). Thus, factors extrinsic to the cells arecausally implicated but, given our current ignoranceover the commitment state of the cells concerned,these factors may simply constitute a permissiveenvironment or alternatively be providing an instruc-tive signal (Wessells, 1979). It has been arguedelsewhere that consequently such interactions arebest described in more neutral terms as 'chondrogen-esis-promoting' or 'osteogenesis-promoting' until weacquire evidence of something more than just asimple elicitation of differentiation (Thorogood, Bee& von der Mark, 1986; Thorogood, 1987). Variousexperimental strategies exploited over a number ofyears have provided a limited understanding of themechanisms involved. Thus, removal of ectodermfrom the presumptive cranial aspect of the avian heador ablation of underlying neuroepithelium results inan absence of calvarial and/or frontal membranebones (Schowing, 1968; Tyler, 1983). Such epithelio-mesenchymal interactions underlying membranebone formation have been investigated in an exten-sive range of tissue dissociation/recombination ex-periments carried out in vitro by Hall and colleagues,working primarily on avian mandibular development(reviewed Hall, 1982,1984). One of several importantconclusions to emerge from this programme of workwas that osteogenesis-promoting interactions arematrix-mediated, possibly effected by a collagenouscomponent of the mandibular ectodermal basal lam-ina (Hall, van Exan & Brunt, 1983). However,mandibular mesenchyme is capable of mounting anosteogenic response to epithelia from sites other thanin the head (e.g. limb bud and dorsal trunk ecto-derms) so clearly this interaction is not specific in thestrictly instructive sense. Furthermore, although thepresence of an epithelium is necessary for osteogen-esis to ensue, the form or morphology of the develop-ing bony element is clearly a property of themesenchymal partner in the recombination and is not

144 P. Thorogood

determined by the epithelium (Hall, 1981).Parallel studies on the avian embryo, in my own

laboratory, examining the developmentally earlier,chondrogenesis-promoting, tissue interactions (usinga comparable range of transfilter culture techniquesand ultrastructural assessment of the in vivo interfaceat which the interaction takes place), reached similarconclusions; that is, chondrogenesis also is elicited bya matrix-mediated interaction (Smith & Thorogood,1983; Thorogood & Smith, 1984). Immunocyto-chemical analysis of extracellular matrix at theinterface between the two cell populations in onesuch chondrogenesis-promoting interaction (betweenpresumptive pigmented retina and mesencephalic-crest-derived periocular mesenchyme) revealed acompositional change correlated with the duration ofthe interaction. More precisely, type II collagen, thepredominant collagen species of cartilage matrix andtherefore regarded as a specific marker for chondro-genic differentiation, is transiently expressed at theinterface, i.e. at the basal aspect of the epithelial

tissue. Further investigation revealed that type IIcollagen is expressed elsewhere in the head, inaddition to previously reported ocular locationswithin the corneal stroma and vitreous (see Figs 1, 2).Principally these locations are around the basolateralaspects of the diencephalon, mesencephalon andrhombencephalon, around optic and otic vesicles(Thorogood et al. 1986) and around the olfactoryconchi (Croucher & Tickle, personal communi-cation). In fact, distribution of this collagen speciesmaps precisely with the sites of those interactionsknown to generate the constituent parts of the chon-drocranium (see Fig. 3). Not only is the spatialpattern of expression coincident but so too is thetiming of transient type II expression which corre-lates with the duration of the interactions concerned(Thorogood et al. 1986). Thus, there is a strongspatial and temporal correlation between the tran-sient expression of type II collagen and the sites andduration of matrix-mediated interactions which gen-erate the component parts of the chondrocranium.

Fig. 1. Phase-contrast (A,C) andimmunofluorescent (B,D)micrographs prepared usingaffinity-purified, polyclonal rabbitanti-chick type II collagen,followed by incubation in anFITC-conjugated goat anti-rabbit.(A,B) T.S.rhombencephalon ofstage-16 quail embryo. Notestaining only around basal surfaceof neuroepithehum.(C,D) T.S.notochord at cervicallevel of stage-24 quail embryo.Staining, predictably, inperinotochordal sheath(artefactually detached on oneside), plus a very small amount ofstaining within the notochorditself. Bar, 30 ̂ m.

Stage 10 Stage 12 Stage 14 Stage 16

Skull development 145

Stage 18 Stage 20

Fig. 2. Diagram summarizing the distribution of type II collagen around the brain and notochord of the early avianembryo. The upper row illustrates the dorsal aspect, the lower row, the ventral aspect, for a series of developmentalstages. OpV, optic vesicle; Mes, mesencephalon; Rh, rhombencephalon; No, notochord; Met, metencephalon;Mye, myelencephalon; Op, invaginating otic pit; I, invaginating optic vesicle; Ov, otic vesicle; OpC, optic cup; L, lens;Te, telencephalon. Not drawn to scale (taken from Thorogood et al. 1986).

Collagen type II distribution

Ventrolateral surfaces of:

- Diencephalon

- Mesencephalon

- Rhombencephalon

Around optic lobes

Around otic vesicles

Chondrocranial elements

Olfactory capsule*

'Trabecula —

Optic capsule

Parachordal

••Otic capsule

NotochordFirst vertebra

Fig. 3. Diagram relating the distribution of collagen type II (as depicted in Fig. 2) to the component parts of thecartilaginous neurocranium which form as a result of tissue interactions at these sites (the viscerocranial part of thechondrocranium is not shown; the neurocranial part has been idealized to a general vertebrate pattern. Taken fromThorogood et al. 1986). * Although not shown in this diagram there is a prior, transient, expression of type II collagenrelating to the olfactory capsules; it has recently been located around the basal surface of the forming olfactory conchibut at stages somewhat later than the range illustrated in Fig. 2 (Croucher & Tickle, personal communication).

146 P. Thorogood

This result is reminiscent of the notochord-sclero-tome interaction in which sclerotome cells, derivedfrom the medial aspect of the somite, migrate towardsthe midline, undergo a matrix-mediated interactionwith the notochord and differentiate chondrogeni-cally to form the cartilaginous primordia of thevertebral bodies. The perinotochordal matrix or'sheath" contains, in addition to typical basal laminacomponents, two components characteristic of carti-lage, i.e. type II collagen and cartilage-specific pro-teoglycan, synthesized briefly by the notochord itself.In vitro exposure of sclerotome cells to perinotochor-dal sheath, or to either of these two components,elicits a sclerotomal synthesis of the same two mol-ecules, i.e. the collagen and the proteoglycan (Kosher& Church, 1975; Lash & Vasan, 1978), and in afunctionally stable form (Vasan, 1987) as a chondro-cytic phenotype becomes established.

The origin of the transient type II collagen in thepresent system is crucially important if we are tounderstand the causal sequence of events at thecellular level. Is it produced by the mesenchyme cellsat these interfaces, destined to differentiate intocartilage and, if so, why is it only briefly expressed?At some sites, completion of the interaction (i.e. theestablishment of chondrogenic commitment in themesenchyme) and disappearance of the type II arefollowed by a period of up to several days beforeovert differentiation commences within the mesen-chyme. Currently we believe that the type II collagenis epithelially derived and, as might be deduced fromFig. 2, largely derived from neuroepithelium.Although type II is typically a stromal collagen andcharacteristically made by chondrocytes, there arethree precedents for epithelial synthesis - thenotochord (see earlier), the corneal epithelium(Linsenmayer, Smith & Hay, 1977) and lastly, a trueneuroepithelium - the neural retina (Smith, Linsen-mayer & Newsome, 1976). These synthesize thetype 11 collagen found in the perinotochordal sheath,primary corneal stroma and the vitreous, respect-ively. Evidence in support of a neuroepithelial originin the present system comes from two sources. First,the precise ultrastructural location of the type II atthe tissue interfaces concerned has been defined usingTEM immunogold techniques and reveals that, dur-ing the interactions, transiently expressed type II isclosely associated with the basal lamina of the epi-thelium and is not around adjacent mesenchyme cells(see Fig. 4). If synthesized by the mesenchyme, ageneral distribution in the extracellular compartmentmight be anticipated but this is not seen (although apolarized secretion by those mesenchyme cells at theinterface cannot be excluded at this stage). Second,the most unequivocal evidence comes, perhaps pre-dictably, from in situ hybridization experiments using

a cDNA probe specific to chick type II collagenmRNA. Work in progress indicates that type IImessage is detectable in cells of the notochord, oticvesicle, neuroepithelium and neural tube 'long be-fore' message is detectable in the responding mesen-chyme (Masando Hayashi - personal communi-cation). Whereas the previously describedimmunocytochemical evidence simply defines thespatiotemporal distribution of the accumulated geneproduct, in situ hybridization data, when complete,will permit a definition of that window in develop-mental time during which the type II genes aretranscriptionally active. It is likely that such data willbe esssential for a full understanding of the mechan-ism of chondrocranial pattern specification.

How chondrocranial pattern might be specified- the 'Flypaper Model'

The previous section deals with identification of apossible signalling molecule which, it has been specu-lated, is produced by an epithelium and which directs,or at least influences, in some unspecified way, thedifferentiative fate of responding mesenchyme.Given that in craniofacial development we aredealing with a system in which there is extensivemesenchymal cell rearrangement, by both active cellmigration and by passive displacement, then themechanism that ensures precise localization of thosemesenchyme cells capable of mounting a response tosuch a signal is crucially important. Thus, control ofcell migration, particularly as it applies to neuralcrest, becomes fundamental in skull morphogenesisand for two reasons. First, it is the event of cellmigration that creates opportunities for interactionsto take place between epithelia and migrating neuralcrest cells or their progeny. Second, it is the time andposition of arrest of migration, within the developinghead, that determine precisely where ensuing skele-togenic differentiation will be expressed (Thorogood,1981). Through closer study of arrest of migration wehave found that matrices derived from certain regionsof epithelium in the embryonic head can cause arrestof NC cell motility in culture. Thus, mesencephaliccrest cell behaviour is dramatically changed on plasticsurfaces 'conditioned' by pigmented retina extracellu-lar matrix (Fig. 5); the cells cease directional move-ment, become spindle-shaped or round-up and mayeven aggregate in small groups (Thorogood & Smith,1984; Yallup, Smith-Thomas & Thorogood, 1988).This behavioural response closely parallels eventsoccurring in vivo when mesencephalic NC cells,having arrived in the periocular region, cease migrat-ing and condense around the basal aspect of the opticvesicle/optic cup as a prelude to differentiating into(scleral) cartilage several days later.

Skull development 147

B

Fig. 4. Transmission electron micrographs illustrating pre-embedding immunogold staining. Tissue was lightly fixed in4% paraformaldehyde/0-25 % glutaraldehyde, incubated initially in affinity-purified, rabbit polyclonal anti-chick type IIcollagen, and subsequently in Staphylococcal protein A conjugated with 15 nm gold particles, before a standard 2-5 %glutataraldehyde fixation, post-osmication and resin embedding. The sections shown here are of sections left unstained(by standard techniques) in order that the gold particles are clearly visible. (A) Perinotochordal sheath of stage-14 chickembryo; nc, basal cytoplasm of single notochordal cell; arrow, basal surface of cell overlaid by basal lamina. Noteabundance of type II staining in the reticulate lamina; conventional staining of adjacent sections confirms that the goldparticles coincide with fibrous components of the extracellular matrix. (B) Matrix surrounding a partially isolated oticvesicle from a stage-16 chick embryo. Here, even in the unstained specimen, the location of the gold particles on thefibrous components of the matrix is evident; ov, a single epithelial cell of the otic vesicle wall; arrow, basal surface. Bar,200 nm.

These studies on factors which might position NCcells by causing localized arrest of migration, togetherwith the previously described studies on tissue inter-actions affecting differentiative fate, have been usedas the basis for formulating a morphogenetic model -the so-called 'Flypaper Model' (Thorogood, 1987).This is best explained stepwise and sequentially but,bearing in mind that this is, in all probability, acontinuous and stochastic process, the followingapply.

(1) The early developing head is best viewed as aconvoluted bag of folded epithelium (undergoing itsown morphogenesis) enclosing an extensive extra-cellular compartment which is only sparsely popu-lated by mesodermally derived cells.

(2) Into the matrix-filled spaces, NC cells migrateautonomously and opportunistically, migrating wher-ever there is matrix capable of supporting theirlocomotion.

(3) Thus, the migratory behaviour of these cells is,to some extent, controlled by extracellular matrixand, in particular, by epithelially derived matrices.Certain matrices will serve a 'trapping' functioncausing localized arrest of cell migration (as thepigmented retina does in vitro).

(4) At such sites, cells will accumulate andundergo a matrix-mediated interaction with thoseepithelia and differentiate into cartilage. Thus, thepattern of cell trapping will epigenetically determinethe form of the chondrocranium (certain matrixmolecules may serve a dual role, functioning in bothsteps 3 and 4). This possibility emerges from obser-vations that type II collagen, shown to be present atthe tissue interfaces concerned, is very potent inarresting NC cells in culture, whereas fibronectin andlaminin, widely distributed matrix glycoproteins, ap-parently promote migration in vitro (B. Yallup,unpublished observations).

148 P. Thorogood

(5) The spatial pattern of these 'trapping' eventsand the ensuing tissue interactions are a consequenceof epithelial folding (i.e. of epiblast and, later, ofectoderm and neuroepithelium). Thus, topographicaldifferences in invagination and evagination of thevarious epithelia in the head will create differentpatterns of cell trapping. Precisely how epitheliallyassociated matrix in the developing head becomesregionally different in composition is not clear. Theuniform absence of immunocytochemically detect-able type II, prior to stages 9-10, suggests that itarises regionally during epithelial morphogenesis andis not present in the 'unfolded' blastoderm (seeFig. 6).

The model therefore reduces control of chondro-cranial form to simply two parameters. First, themanner in which the various epithelia undergo theirown morphogenesis and, second, the rate and timing

of neural crest migration (passive displacement ofmesenchyme cells is undoubtedly implicated too buthas not been dealt with separately at this stage ofmodel development). It is generally recognized thatminor quantitative changes in particular developmen-tal programmes can have profound morphologicalconsequences (Gould, 1977; Alberch, 1982) and 1propose that minor quantitative shifts in these twoparameters underlie much of the phylogenetic diver-sity of skull form and congenital abnormality of theskull. Different patterns of cell trapping will generatedifferent, and sometimes new, chondrocranial forms.Because the chondrocranium serves as a structuraltemplate for deposition of the bony elements of theskull (at least facially), then one further epigeneticconsequence will be a change in overall skull form.Genetic control of skull form will therefore be me-diated indirectly through gene loci affecting these two

0-2 mm

Fig. 5. Traces of pathways of cell migration, over a 4h period, at the margin of outgrowth from explanted, stage-9,chick mesencephalic neural crest (arrows indicate radial direction away from centre of explant). (A) Control cells grownon plastic. (B) Cells grown on plastic previously conditioned with matrix deposited by stage-16 pigmented retina; theepithelial sheet is stripped off prior to explantation of crest cells. Note the extensive and directional migration of thecontrol cells and the reduced path length and apparent lack of directionality of those cells on matrix. The inset boxesrepresent sample areas at the outgrowth margin from which such traces were made; black, represents the cell sheet;white, the intercellular areas. See text for further details (taken from Yallup el al. 1988).

Skull development 149

folding

Fig. 6. Diagram depicting the possible distributions of atrapping component in matrix (depicted as +; e.g. type IIcollagen) prior to epithelial morphogenesis in the head.(A) Uniform distribution of trapping component;(B) Trapping component uniformally absent. As a resultof folding, the distribution of trapping componentbecomes expressed regionally around areas of high radiusof curvature, either because it disappears locally(subsequent to A) or because it appears locally(subsequent to B). The lower part of the diagram is acomposite to indicate the broad range of possibleepithelial morphogenesis and known distribution oftype II collagen, x, Invagination from surface ectoderm,as in the otic pit/vesicle; y, neural tube; z, evaginationfrom the neural tube as in the optic vesicle. Analysisindicates that B is correct in that type II collagen is notdetectable before stage 10 and its subsequent appearanceis concomitant with, or immediately follows, major eventsof epithelial morphogenesis in the head. However, thetime of first appearance of type II mRNA is not currentlyknown.

parameters, such as regulation of synthesis and turn-over of matrix components, cell surface receptors formatrix components and cytoskeletal organization andfunction. It could be argued that skull form is not somuch 'specified' in a strict genetic sense but deter-mined by local interactions between cells and theirenvironment, operating within what has been calledthe 'epigenetic domain' (Alberch, 1982).

Testing the model

A number of predictions can be made which we are

currently testing experimentally and which will bebriefly reviewed here:

(1) If 'trapping' underlies the arrest of cellmigration, as suggested, then migrating neural crestcells should express a receptor or binding protein ofsome sort to mediate attachment to those matrixcomponents ostensibly responsible for trapping. Cur-rent work reveals that premigratory cells synthesize,and migrating cells express at their cell surface, abinding protein for type II collagen - Anchorin CII(Thorogood, Hoffmann & von der Mark, in prep-aration). Anchorin CII is a 34K membrane proteintypically expressed by chondrocytes and which has ahigh-binding affinity for the abundant type II col-lagen normally found in the chondrocytes' matricialenvironment (von der Mark, Mollenhauer, Pfaffle,van Menxel & Muller, 1986). Immunocytochemicalstudy of cultured mesencephalic crest has shown thatAnchorin CII is expressed by only a subpopulation ofcells (which raises interesting questions about neuralcrest population composition) and it is for these cellsthat we propose Anchorin CII mediates trapping asthey encounter type II during their migration in thedeveloping head (see Fig. 7).

(2) Supplementation of the endogenous trappingmatrix, by introducing exogenous type II collagenectopically somewhere along the migration route,should cause precocious arrest of migration and mayeven result in ectopic cartilages being formed. Exper-iments using implanted microbeads as a supportmedium for a variety of matrix molecules arecurrently in progress.

(3) The parallel between chondrogenesis-promot-ing interactions in the head and in the rest of the axialskeleton was briefly mentioned earlier. From data onthe notochord/somit interaction, it can be predictedthat other matrix components, thought or known tohave a role in this interaction (Vasan, 1987), will showsimilar patterns of distribution to that displayed bycollagen type II during craniofacial development.Cartilage-specific proteoglycan is found not only incartilage but also in perinotochordal sheath alongwith type II. One matrix marker that has been usedfor locating this, is the monoclonal antibody MZ15,which binds keratan sulphate and therefore recog-nizes the keratan sulphate-rich domain on the proteo-glycan in both cartilage (Zanetti, Ratcliffe & Watt,1985) and perinotochordal sheath (Smith & Watt,1985). Mapping of MZ15-staining patterns duringcraniofacial development in the avian embryo revealsa large degree of codistribution with the previouslydescribed patterns for type II collagen. For instance,both molecules are present at the same stages aroundthe otic vesicle (Heath & Thorogood, in prep-aration). However, the lack of codistribution at someother sites (e.g. no MZ15 staining around the optic

150 P. Thorogood

Fig. 7. Phase-contrast (A,C) and immunofluorescent (B,D) micrographs of living mesencephalic neural crest cells inculture, prepared using a polyclonal rabbit anti-chick chondrocyte Anchonn CII, followed by incubation in an FITC-conjugated goat anti-rabbit. (A,B) A cluster of crest cells; arrows indicate cells which do not express Anchorin CII inB. (C,D) An isolated crest cell expressing Anchorin CII (N.B. although initially flattened at the outset of staining, asthe cells are unfixed, they tend to round up during the process). Bar, 10/un.

vesicle) indicates that this will not be a simple story ofcomplete codistribution. Further work using morespecific markers, such as monoclonals against thecore protein of cartilage-specific proteoglycan, mightbe especially valuable in this context.

(4) The results described earlier arise entirely fromwork on avian embryos. It is anticipated that theFlypaper model will be applicable generally to ver-tebrates to describe an evolutionarily conserved de-velopmental mechanism underlying the phylogeneti-cally constant chondrocranial pattern. If that is so,then comparable data should emerge from investi-gation of other vertebrate taxa. We have started sucha comparative study and it has already emerged, fromour current work on mammalian craniofacial devel-opment, that type II collagen is found on the basalaspect of some neuroepithelia and anlagen of thesense organs of the mouse embryo (Wood &Thorogood, in preparation). A pattern of transient

expression is found which is similar although notquite identical to that described for the bird. Equallyimportant will be investigations of type II distributionand associated cell-matrix interactions in the Mower'vertebrate taxa, in particular the agnathans, teleostsand amphibians.

Counter evidence?

The theme of the Flypaper model is that during earlycraniofacial development, local interactions betweenepithelia and mesenchyme will determine not onlydifferentiative fate of the mesenchyme cells but alsospecify skeletal pattern (ectomesenchyme has beendealt with principally but mesodermally derived mes-enchyme can be incorporated into the model). This isunlike our current understanding of events in thedeveloping limb where primary specification of skel-

Skull development 151

etal pattern is normally regarded as residing in themesoderm (Saunders & Reuss, 1974; and see later).However, the model also is in contradiction toNoden's hypothesis of 'regional morphogenetic speci-fication' which postulates that much of the skeletalpattern in the head is 'programmed' in neural crestcells prior to their migration, possibly even in crestprecursors within the neuroepithelium (Noden, 1983,1986). This proposal is based on experiments in whichheterotopic grafts were made, between quail andchick, of first arch crest into sites formerly occupiedby the host's own second and third arch crest (an-terior- and mid-myelencephalic levels, respectively).Graft cells migrated along routes normal for secondor third arch crest cells but proceeded to give rise to asecond, i.e. duplicate, set of first arch skeletal struc-tures. Local interaction, of the sort proposed earlier,would have little, if any, role in pattern specificationof this type. These experiments are clearly importantand the result is probably telling us something funda-mental about skeletal patterning, but is 'regionalmorphogenetic specification' a feasible interpret-ation? More specifically, do these results constitutean unanswerable challenge to the Flypaper model?

It is important to remember that although morpho-genesis is 'driven' at the level of individual cells (bychanges in cell shape, adhesivity, contractility, etc.) itis the collective activity of those cells that generatestissue form and pattern; thus morphogenesis is apopulation phenomenon. It is difficult to think of amolecular or cellular mechanism maintaining patternspecification within a cell population unless the cellsmaintain their spatial relationships to each other.However, maintenance of neighbour-to-neighbourcontact between cells seems unlikely during the large-scale migration that crest cells undergo. One furtherproblem is that not only does first arch crest produceectopic first arch derivatives but so too does morerostral crest. Crest that normally gives rise to fronto-nasal, maxillary and trabecula skeleton will give riseto a duplicate first arch skeleton when grafted tomyelencephalic levels (Noden, 1983). This last obser-vation seems incompatible with the idea of an early(premigratory) specification of pattern appropriatefor each axial level. In fact, the result, arguably,might be used as evidence for a local specification ofpattern occurring during or after migration. Whatmay be important is whether or not crest cellsencounter type II collagen, and similarly activematrix molecules, during their migration and, if so,where and when? (in fact, a correlation has beenmade between lack of chondrogenic commitment atthe end of migration and the absence of type IIcollagen along that particular migration route; seep. 508, Thorogood et al. 1986). Local interactionsoccurring between cranial crest cells, derived from

perhaps any axial level, and an appropriate epithelialenvironment might elicit the formation of an ectopic,duplicate mandibular skeleton if the formation of amandible normally is the result of epigenetic phenom-ena.

Given the extensive rearrangements undergone bycell populations of both mesenchymal lineages duringcraniofacial development, it makes functional sensethat another tissue might provide a frame of pos-itional reference for subsequent differentiation ofmesenchymal cells. The epithelia of the head,although undergoing extensive invaginations andevaginations, display little cell mixing and thereforeprovide just such a planar reference system. It isperhaps significant that similar conclusions have beenreached independently in experimental studies oftooth development (Lumsden, 1987), the patterningof sensory neurites (Lumsden & Davies, 1984) andfrom a paleontological assessment of the evolution ofthe vertebrate dermal skeleton (Thomson, 1987). Theapparent absence of such a role for ectoderm in thelimb may simply reflect an intrinsic differencebetween axial and appendicular skeletons or relate tothe fact that limb mesenchyme does not display cellrearrangements on the scale seen in the developinghead.

Does the model satisfy the prerequisites?

Two prerequisites were stipulated at the outset. Thefirst was that any acceptable model should be gener-ally applicable to all vertebrates. With the Flypapermodel, although it arises from work on avian em-bryos, there is no inherent reason why it should notbe applied to the embryos of other vertebrate taxa.Preliminary observations on mammalian embryossuggest that this is indeed possible. Quantitativevariation of the parameters of the model couldgenerate a variety of chondrocranial patterns and,therefore, of skull form. It also makes functionalsense that the developing brain and sense organsshould, themselves, determine precisely where theirprotective and supportive tissue will be formed. Themodel also lends itself to interpretation, and predic-tion, of congenital abnormality by permitting a defi-nition of 'sensitive periods' during the specification ofskull form.

The second prerequisite was that the model shouldhelp to explain different lineage composition ofotherwise homologous cartilages and, similarly, toexplain the experimentally created alterations inproportional contributions from the two mesenchy-mal lineages, to 'mixed composition' skeletal el-ements which are otherwise normal in shape and size.Although in this review less emphasis has been placed

152 P. Thorogood

on craniofacial mesoderm, there is no evidence thatmesoderm cells cannot respond to matrix signals orcues in a fashion identical to ectomesenchyme cells(and in the sclerotome, they clearly do). In fact, theFlypaper model does not distinguish where mesen-chyme cells come from. The (neuro)epitheliumsimply specifies, to a responsive mesenchyme, whereand when cartilage should form. Viewed in this way,differential lineage composition of 'homologous'skeletal elements, in terms of ectomesenchymal ormesodermal contributions, becomes largely irrel-evant.

Much of the experimental work has been supported bygrants from the Medical Research Council (UK) and theWellcome Trust. Also I am grateful to Klaus von der Markfor allowing me to cite our unpublished results, and toMasando Hayashi for informing me of his current unpub-lished work. Fig. 5 was provided by Bridget Yallup, andFig. 6 was drawn by Helen Wood - thank you.

References

ALBERCH, P. (1982). Developmental constraints inevolutionary processes. In Evolution and Development(ed. J. Bonner), pp. 313-332. Berlin: Springer Verlag.

ALBERCH, P. & ALBERCH, J. (1981). Heterochronicmechanisms of morphological diversification andevolutionary change in the neotropical salamander,Bolitoglossa occidentals (Amphibia: Plethodontidae).J. Morph. 167, 249-264.

BEE, J. & THOROGOOD, P. (1980). The role of tissueinteractions in the skeletogenic differentiation of avianneural crest cells. Devi Biol. 78, 47-66.

BELLAIRS, A. D'A. & GANS, C. (1983). A reinterpretationof the amphisbaenian orbitosphenoid. Nature, Lond.302, 243-244.

BOSMA, J. (1976). Development of the Basicranium(editor). U.S. Department of Health, Education andWelfare publication No. (NIH), 76-989.

DE BEER, G. (1937). Development of the Vertebrate Skull.Oxford: Oxford University Press.

DREWS, U., KOCHER-BECKER, U. & DREWS, U. (1972).Die Induktion von Klemenknorpel ausKopfneuralleistenmaterial durch prasumptivenKiemendarm in der Gewebkultur und dasBewegungsverhalten der Zellen wahrend ihrerEntwicklung zu Knorpel. Wilhelm Roux' Archiv.EntwMech. Org. 171, 17-37.

EPPERLEIN, H. H. & LEHMANN, R. (1975).Ectomesenchymal-endodermal interaction system(EEIS) of Triturus alpestris in tissue culture.2. Observations on the differentiation of visceralcartilage. Differentiation 4, 159-174.

GANS, C. & NORTHCUTT, G. (1983). Neural crest and theorigin of the vertebrates. Science 220, 268-274.

GOETHE, J. W. (1820). Zur Natunvissenschaftenuberhaupt, besonders zur Morphologic Cited in deBeer (1937).

GOODRICH, E. S. (1930). Studies on the Structure andDevelopment of Vertebrates. London: Macmillans.

GRAVESON, A. C. & ARMSTRONG, J. B. (1987).Differentiation of cartilage from cranial neural crest inthe Axolotl (Ambystoma mexicanum). Differentiation35, 16-20.

GOULD, S. J. (1977). Ontogeny and Phytogeny.Cambridge, MA: Harvard University Press.

HALL, B. K. (1981). Specificity in the differentiation andmorphogenesis of neural crest-derived scleral ossiclesand of epithelial scleral papillae in the eye of theembryonic chick. J. Embryol. exp. Morph. 66,175-190.

HALL, B. K. (1982). The role of tissue interactions in thegrowth of bone. In Factors and Mechanisms InfluencingBone Growth (ed. A. S. Dixon & B. G. Sarnat),pp. 205-215. New York: A. R. Liss Inc.

HALL, B. K. (1984). Genetic and epigenetic control ofconnective tissues in the craniofacial structures. BirthDefects: Original Article Series, 20 (no. 3), 1-17. NewYork: A. R. Liss Inc.

HALL, B. K. (1987). Initiation of chondrogenesis fromsomitic, limb and craniofacial mesenchyme: search fora common mechanism. In Somites in DevelopingEmbryos (ed. R. Bellairs, D. A. Ede & J. W. Lash),pp. 247-259. New York: Plenum.

HALL, B. K., VAN EXAN, R. & BRUNT, S. (1983).Retention of epithelial basal lamina allows isolatedmandibular mesenchyme to form bone. J. Cramofac.Genet. Devi Biol. 3, 253-267.

HANKEN, J. (1983). Miniaturization and its effects oncranial morphology in plethodontid salamander, genusThorius (Amphibia: Plethodontidae). II. The fate ofthe brain and sense organs and their role in skullmorphogenesis and evolution. J. Morph. 177, 255-268.

JACOBSEN, A. & MEIER, S. (1987). Somitomeres: theprimordial body segments. In Somites in DevelopingEmbryos (ed. R. Bellairs, D. A. Ede & J. W. Lash),pp. 1-16. New York: Plenum Press.

KOSHER, R. A. & CHURCH, R. L. (1975). Stimulation ofin vitro chondrogenesis by procollagen and collagen.Nature, Lond. 258, 327-330.

KRATOCHWIL, K. (1983). Embryonic induction. In CellInteractions and Development (ed. K. M. Yamada), pp.99-122. New York: John Wiley & Sons.

LASH, J. W. & VASAN, N. S. (1978). Somitechondrogenesis in vitro; stimulation by exogenousextracellular matrix components. Devi Biol. 66,151-171.

LE LIEVRE, C. (1978). Participation of neural crest-derived cells in the genesis of the skull in birds.J. Embryol. exp. Morph. 47, 17-37.

LINSENMAYER, T. F., SMITH, G. N. & HAY, E. D. (1977).Synthesis of two collagen types by chick embryoniccorneal epithelium in vitro. Proc. natn. Acad. Sci.U.S.A. 74, 39-43.

LUMSDEN, A. G. S. (1987). The neural crest contributionto tooth development in the mammalian embryo. InDevelopmental and Evolutionary Aspects of the NeuralCrest (ed. P. F. A. Maderson), pp. 261-300. NewYork: John Wiley & Sons.

Skull development 153

LUMSDEN, A. G. S. & DAVIES, A. M. (1984).Chemotactic guidance of developing sensory neuritesby appropriate target epithelium. /. Embryol. exp.Morph. 82 Supplement, 229.

MEIER, S. (1981). Development of the chick embryo:morphogenesis of the prechordal plate and cranialsegments. Devi Biol. 83, 49-61.

NODEN, D. (1975). An analysis of the migratorybehaviour of avian cephalic neural crest cells. DeviBiol. 42, 106-130.

NODEN, D. M. (1978). The control of avian cephalicneural crest cytodifferentiation. I. Skeletal andconnective tissues. Devi Biol. 67, 296-312.

NODEN, D. M. (1983). The role of the neural crest inpatterning of avian cranial skeletal, connective andmuscle tissue. Devi Biol. 96, 144-165.

NODEN, D. (1986). Origins and patterning of craniofacialmesenchymal tissues. J. Craniofac. Genet, devl Biol. 2,15-31.

NORTHCUTT, G. & GANS, C. (1983). The genesis of neuralcrest and epidermal placodes: a reinterpretation ofvertebrate origins. Q. Rev. Biol. 58, 1-28.

PRESLEY, R. & STEELE, F. L. D. (1976). On the homologyof the alispenoid. J. Anat. 121, 449-459.

SAUNDERS, J. W., JR & REUSS, C. (1974). Inductive andaxial properties of prospective wing-bud mesoderm inthe chick embryo. Devi Biol. 38, 41-50.

SCHOWING, J. (1968). Mise en Evidence du r61e inducteurde l'encephale dans I'oste'ogenese du craneembryonnaire du poulet. /. Embryol. exp. Morph. 75,165-188.

SMITH, G. N., LINSENMAYER, T. F. & NEWSOME, D. A.(1976). Synthesis of type II collagen in vitro byembryonic chick neural retinal tissue. Proc. natn.Acad. Sci. U.S.A. 73, 4420-4423.

SMITH, J. C. & WATT, F. M. (1985). Biochemicalspecificity of Xenopus notochord. Differentiation 29,109-115.

SMITH, L. & THOROGOOD, P. (1983). Transfilter studies onthe mechanism of epithelio-mesenchymal interactionleading to chondrogenic differentiation of neural crestcells. J. Embryol. exp. Morph. 75, 165-188.

THOMSON, K. S. (1987). Speculations concerning the roleof the neural crest in the morphogenesis and evolutionof the vertebrate skeleton. In Developmental andEvolutionary Aspects of the Neural Crest (ed. P. F. A.Maderson), pp. 301-338. New York: John Wiley &

Sons.THOROGOOD, P. (1981). Neural crest cells and

skeletogenesis in vertebrate embryos. Histochem. J. 13,631-642.

THOROGOOD, P. (1987). Mechanisms of morphogeneticspecification in skull development. In Mesenchymal-Epithelial Interactions in Neural Development (ed. J. R.Wolff, J. Sievers & M. Berry), pp. 141-152. Berlin:Springer-Verlag.

THOROGOOD, P., BEE, J. & VON DER MARK, K. (1986).Transient expression of collagen type II atepitheliomesenchymal interfaces during morphogenesisof the cartilaginous neurocranium. Devi Biol. 116,497-509.

THOROGOOD, P. & SMITH, L. (1984). Neural crest cells:the role of extracellular matrix in their differentiationand migration. In Matrices and Cell Differentiation (ed.R. B. Kemp & J. R. Hinchliffe), pp. 171-185. NewYork: A. R. Liss Inc.

TYLER, M. (1983). Development of the frontal bone andcranial meninges in the embryonic chick: anexperimental study of tissue interactions. Anat. Rec.206, 61-70.

TYLER, M. S. & HALL, B. K. (1977). Epithelial influenceson skeletogenesis in the mandible of the embryonicchick. Anat. Rec. 188, 229-240.

VASAN, N. S. (1987). Somite chondrogenesis: the role ofthe microenvironment. Cell Diff. 21, 147-159.

VON DER MARK, K., MOLLENAUER, J., PFAFFLE, M., VAN

MENXEL, M. & MULLER, P. K. (1986). Role ofAnchorin CII in the interaction of chondrocytes withextracellular collagen. In Articular CartilageBiochemistry (ed. K. Kuettner), pp. 125-141. NewYork: Raven Press.

WAKE, D. B. (1980). Evidence of heterochronicevolution; a nasal bone in the Olympic salamander,Rhyacotriton olympicus. J. Herpet. 14, 292-295.

WESSELS, N. K. (1979). Tissue Interactions andDevelopment. Menlo Park, CA: W. A. Benjamin Inc.

YALLUP, B., SMITH, L. & THOROGOOD, P. (1988).

Epithelially derived extracellular matrix modulatesneural crest cell motility. Development.

ZANETTI, M., RATCLIFFE, A. & WATT, F. (1985). Twosubpopulations of differentiated chondrocytes identifiedwith a monoclonal antibody to keratan sulfate. /. CellBiol. 101, 53-59.