A Developmental Model For Evolution of The Vertebrate Exoskeleton and Teeth The Role of Cranial and Trunk Neural Crest

10
A Developmental Model for

Evolution of the Vertebrate
Exoskeleton and Teeth
The Role of Cranial and Trunk Neural Crest

MOYA M. SMITH and BRIAN K. HALL
EXOSKELETAL DEVELOPMENT AND EVOLUTION

An exoskeleton is extensive in the head, trunk, and tail of agnathan and
gnathostome fishes, where it forms a thick, rigid armor in most fossil fishes,
although many only have a covering of separate denticles. This armor consists
of three basic layers of mineralized tissue surrounding soft tissue components.
There is a superficial odontode layer (see pp. 391 and 393 for discussion of
odontodes) with enamel, dentine, attachment bone, and pulp chambers; a
middle layer with trabeculae of cellular or acellular bone surrounding soft
tissue vascular spaces; and a basal laminated layer of compact cellular or
acellular bone, with zones of attachment fibers. Among extant osteichthyan
fishes are those that retain some of these odontode components in their dermal
scales as a mineralized exoskeleton. These are the so-called living fossils, the
MOYA M. SMITH Division of Anatomy and Cell Biology, United Medical and Dental
Schools of Guy's and St Thomas's Hospitals, London, SEI 9RT, England.
BRIAN K.
HALL Department of Biology, Dalhousie University, Halifax, Nova Scotia, B34 HJl, Canada.
Evolutionary Biology, Volume 27, edited by Max K. Hecht et al. Plenum Press, New York,
1993.
387
388
M. M. Smith and B. K. Hall
Polypteridae and the extant coelacanth, as well as catfishes. Odontode components are retained as placoid scales in all chondrichthyans (see Figs. 1 and 2).
In addition to this general body covering of scales, part of the fin skeleton-the terminal mineralized portion of the fin rays, or lepidotrichia-is
considered to be exoskeleton, because these skeletal elements are often associated with enamel and dentine layers and do not form first in cartilage,
but directly as membrane bone. Knowledge about the developmental embryonic origin of these different skeletal dermal components is fragmentary
and sometimes at variance. The skeletogenic cells of the paired and median
fins are said to have different embryonic origins: those forming lepidotrichia
of the paired fins being derived from dermatomal or splanchnic mesenchyme,
those forming lepidotrichia ofthe median fin being derived from neural crestderived mesenchyme (Schaeffer, 1987).
The developmental origin of the exoskeleton is mostly assumed to be
from neural crest (M. M. Smith and Hall, 1990). Conclusions extrapolated
to fishes from much experimental data gathered from mammals, birds, and
amphibians are that both the odontogenic and skeletogenic components of
the exoskeleton are cranial neural crest derived. Lack of developmental evidence for a skeletogenic role of trunk neural crest stands in stark contrast to
the vast amount of data showing cranial neural crest to be both skeletogenic
and odontogenic.
A conclusion that neural crest forms all of the competent skeletogenic
mesenchyme contributing to the dermal skeleton (lepidotrichia and scales)
of fishes requires either a considerable extension of our existing knowledge
on distances and rates of migration of cranial neural crest, or an acceptance
that trunk neural crest must be skeleto-odontogenic in lower vertebrates. A
skeleto-odontogenic trunk neural crest challenges the assumption that cranial
neural crest alone accounts for the initiation and differentiation of all skeletal
cells in the extensive exoskeleton in lower vertebrates.
In this review an explanation is sought for the extensive distribution,
diverse morphology, and histology of the exoskeleton in extant and fossil
fishes, whether that exoskeleton has its developmental and evolutionary origin
in the neural crest, and, if so, whether in cranial or trunk neural crest. This
is an attempt to provide a more logical explanation for the developmental
mechanism that allows an extensive exoskeleton to develop in the trunks and
tails of the majority of extinct fishes and in the trunk and tail skeleton in
some extant fishes. We compile and evaluate information from exoskeletal
evolution, morphology, histology, and experimental and molecular studies
of skeletal development. We propose a model of skeletal origin and diversity
in the exoskeletons of "lower" vertebrates that also explains loss or reduction
of the exoskeleton within most "higher" vertebrates (amphibians, reptiles,
Vertebrate Exoskeleton and Neural Crest
389
birds, and mammals). A central question considered is adequacy of the evidence for lack of skeletogenic potential in trunk neural crest.
Five possible cellular origins of scleroblasts (cells capable of making mineralized connective tissues) to form the trunk exoskeleton are considered:
1. Scleroblasts differentiate from cranial neural crest after caudad migration into the trunk and without involvement of other mesenchymal
populations.
2. Scleroblasts form from trunk neural crest and make the exoskeleton
without involvement of other mesenchymal cell populations.
3. Scleroblasts differentiate from trunk neural crest, but are dependent
for their differentiation on somitic mesoderm (dermomyotome).
4. Scleroblasts differentiate from mesenchyme derived from somatic lateral plate mesoderm.
5. Scleroblasts differentiate from mesenchyme derived from somatic lateral plate mesoderm, but are dependent for their differentiation on
trunk neural crest.
Principles of Skeletal Development and Evolution
Neural Crest versus Mesodermal Cellular Origins

Throughout this review it is assumed that homologous parts of the skeleton in extant and fossil vertebrates share the same embryonic cellular origins:
neural crest for teeth, visceral arch skeleton, some ofthe neurocranium, and
all dermal bones except a few of the dermatocranium (see p. 390; Couly et
al., 1993); mesoderm for cartilage bones of most of the neurocranium, and
all of the axial and appendicular skeleton. Cellular origins are therefore divisible
into ectodermal or mesodermal skeletal lines. Within each are two histogenetic
modes of development, intramembranous and endochondral. For example,
perichondral bone of the endoskeleton is a bone tissue forming in membrane
(the same developmental mechanism seen in dermal bone, but specifically
here in the connective tissue sheath on the surface of the cartilage), but the
majority of the bone in the endoskeleton is formed by endochondral ossification of the cartilaginous models of those bones. Again, visceral cartilage
elements (endoskeleton from cranial neural crest) are replaced by bone endochondrally after the perichondral collar surrounding them has been made
in membrane, as in the mesodermally-derived endoskeleton. Perichondral
bone is developmentally and phylogenetically earlier than endochondral bone
(Patterson, 1977; Maisey, 1988; M. M. Smith and Hall, 1990) [see Fig. 1 of
390
M. M. Smith and Hall (1990) for an illustration of the components of exoand endoskeletons].
As a basis for evaluating the role of developmental mechanisms of constraint or plasticity in evolutionary change, we use previous postulates established by M. M. Smith and Hall (1990), in particular those built upon Patterson
(1977), concerning the separate origins in development and evolution of the
exo- and endoskeleton.
Prior to Patterson's conclusions, many statements had been made about
the presumed interchangeability of dermal and cartilage bones, such as "there
cannot be any fundamental difference between the endoskeleton and the exoskeleton" (Jarvik, 1959). Although Patterson stated that the endoskeleton and
exoskeleton are distinct and have separate phylogenetic and ontogenetic origins, little was said about the embryonic origin of the participating cells. This
was because little was then known, although in a paper in the same volume
Schaeffer (1977) postulated that "ectomesenchyme is present in the dermis
throughout the body and. . . can form dermal bone as well as dental organs."
Schaeffer was aware of the association of enamel and dentine with the dermal
skull, scales, and lepidotrichia in all fossil fishes and some extant ones, such
as Polypterus, and of the dependent link with mesenchymal neural crestderived cells for tooth development. These concepts of Schaeffer are central
to this review.
The information from developmental biology on the origin of cells that
contribute to the dermal bones of the skull in birds has provided a picture in
which most of the cranial vault is mesodermal in origin (frontal, parietal,
supraoccipital, exoccipital, and orbitosphenoid) but the facial skeleton is neural
crest in origin (squamosal, supraorbital, lacrimal, sclerotic, premaxilla, maxilla,
dentary, splenial, parasphenoid, palatine), a distribution illustrated by Le
Douarin (1982, Fig. 3.6). Recent publications by eouly et al. (1992) have
further elaborated on the contribution of the paraxial mesoderm and the
prechordal plate to the head skeleton and produced many interesting additions
to the accepted story. They found that neural crest cells were in all of the
dermis prior to membrane bone formation, with the implication that all dermal bones of the skull come from neural crest (eouly et al., 1992, Fig. 9, and
Table 2) with only the chondrocranial bones, basisphenoid, orbitosphenoid,
and otic capsule arising from paraxial mesoderm. In their subsequent paper
(eouly et al., 1993), they have further refined the information by taking chimeric embryos to the stage where frontal and parietal bones can be identified.
They have found that, as with the facial skeleton, these bones also originate
in neural crest. Of great importance, they have also found that all the 'prechordal' chondrocranium originates from neural crest. This new information
considerably alters our concept of the development of the head skeleton (facial
and skull roof) and the chondrocranium anterior to the notochord. As eouly
391
et al. concluded, "The anterior limit of the notochord then corresponds to

the limit where the mesodermal skeleton ends and the neural crest derived
skeleton begins." This new information from studies on the chick indicates
a more restricted contribution of the paraxial mesoderm to the skull than
previously shown, and may be vital to our understanding of the evolution of
the cranial and trunk dermal skeleton in lower vertebrates. It now seems that
all the dermal bones of the head originate in neural crest except parts of the
occipital bone; these have contributions from the sclerotomal mesenchyme
of the first four somites. One of the postulates of a fundamental difference
between the exoskeleton (cranial and trunk) and the axial and appendicular
endoskeleton, to which this new data has a direct bearing, is that they have
different embryonic cell origins, and that this is evolutionarily conserved.
Postulates and Potential Mechanisms
First, and fundamental to the assumptions discussed in this review, is
the postulate that all dentoskeletal components of the exoskeleton require
neural crest, both for induction and for participation in skeletogenesis, and
that inductive interactions occur with epithelia. We consider it crucial for
understanding evolutionary origins and for comparing exoskeletons with endoskeletons to look for evidence of compatibility or discordance of the initial
signal for specification of the type of skeletal tissue in these epitheliomesenchymal interactions. But as M. M. Smith and Hall (1990) commented, the
scarcity of data on mechanisms of induction of skeletogenic and odontogenic
potential in fishes prevents such a comparison until experimental data are
provided. Data for the identification of the origin of cell types are also not
available for the dermal skeleton in fishes, although some are available for
the neural crest origin of the visceral skeleton in lampreys and teleost fishes
(Langille and Hall, 1988a,b).
Although teeth and skin denticles in extant and fossil forms have been
assumed to be similar structures since the publications of Williamson (1849),
Hertwig (1879), and Goodrich (1907), it was Schaeffer's (1977) review of the
dermal skeleton in fishes that first clearly linked common structure with shared
morphogenetic mechanisms. Incorporated in this hypothesis was the concept
that "the dermal skeleton develops from a single, modifiable morphogenetic
system." He had envisaged regulation of morphogenetic change through this
interactive system as significant for evolution of the dermal skeleton. We now
propose that part of the "morphogenetic system" is the dental primordium
of interactive dental cap ectoderm and dental papillary and follicular ectomesenchyme (Fig. 1). This system forms the developmental basis of the odontode, the primitive exoskeletal building block (see Fig. 2). In addition to the
odontode, a skeletal primordium would be involved, i.e., a subpopulation of
392
suc. prim .
i. en. epith.
o. en. epith.
epith.
dent. cap
dent. foil.
dent. pap.
osteo. condo
boo
c
FIG. 1. Three stages of development of the epithelioectomesenchymal interactions involved
in skeletoodontogenesis, showing the three sets of condensations that are proposed as
subpopulations of the neural crest involved in induction and morphohistogenesis of the
vertebrate dentoskeletal tissues in development of the entire exoskeleton . This is based
on the classic stages of morphogenesis of mammalian tooth germs and skeletal support
tissues of the jaws: (a) bud stage, primordial tissue of the odontogenic epithelium and
ectomesenchyme, with adjacent condensations of osteogenic and chondrogenic ectomesenchyme; (b) bell stage, morphogenesis within the enamel organ and dental papilla,
and histodifferentiation of these cells to produce the tooth shape and tissues, enamel and
dentine; more extensive membrane bone and established cartilage; (c) cap stage, three
393
ectomesenchymal cells dependent upon specific epithelium for induction and

competent to produce dermal bone to support the odontodes (see Fig. 1) and,
in the head, cartilage of the jaws and the pharyngeal arches.
Much earlier than Schaeffer (1977), however, Moss (1960, 1968a,b) had
argued that "the earliest ossified vertebrates possessed the intrinsic capacity
to produce the entire spectrum of vertebrate skeletal tissues," and that this
required the acquisition in evolution of two essential developmental processes,
(1) "inductive interaction between ectoderm and neural crest derivatives"
and (2) "epidermal co-participation," implied to occur between epithelial
cells and neural-crest-derived mesenchymal cells, both in morphogenesis and
in histogenesis. We specifically propose that such intrinsic capacity would
have extended from head to tail in those early vertebrates possessing an extensive bony armor of separate or fused denticles, i.e., in the majority of early
vertebrates. In fact, Moss (1968a) concluded that "Dentine did not evolve
from bone. Dentine could arise as soon as the neural crest had evolved and
with it the topographic relationships and their inductive sequences of tooth
formation." The proposals in the next section accept this dichotomy between
odontogenic and skeletogenic development and evolution and discuss the role
of the odontode as a basis for the exoskeleton.
The Odontode as Dermal Exoskeleton

The unit structure that develops from the interactive morphogenetic system proposed by Moss and elaborated by Schaeffer is the odontode.
The term odontode was first proposed by 0rvig (1967), and later expanded
(0rvig, 1977) to include all hard tissue units of the dermal skeleton, composed
of enamel, dentine, and pulp. Reif( 1982) subsequently expanded this concept
to include oral denticles and teeth. Two assumptions are that teeth and dermal
denticles are homologous, constrained by common developmental patterns,
and that dermal denticles in the oral cavity gave rise to teeth during the
cellular aggregates of the odontode/tooth, epithelial dental cap, ectomesenchymal dental

papilla, and ectomesenchymal dental follicle; with the osteogenic condensation and early
bone, and chondrogenic condensation and early cartilage; the developmental origins of
four differentiated cell types are indicated. Abbreviations: amelo., ameloblasts; bo., bone;
ca., cartilage; chondro., chondrocyte; chond. cond., chondrogenic condensation; chond.
ecto., chondrogenic ectomesenchyme; de., dentine; dent. cap, dental cap; dent. foil.,
dental follicle; dent. pap., dental papila; dent. lam., dental lamina; en., enamel; en. epith.,
enamel epithelium; enam. org., enamel organ; epith., epithelium; odonto., odontoblast;
odonto. ecto., odontogenic ectomesenchyme; odonto. epith., odontogenic epithelium;
osteo., osteocyte; osteo. cond., osteogenic condensation; osteo. ecto., osteogenic ectomesenchyme.
394
evolution of jawed vertebrates (Fig. 2). This axiom may be seriously challenged
by new information on the relationship of conodonts to vertebrates (see p.
416). These postulated chordates are now proposed as the sister group of all
vertebrates. They possess arrays of teeth in an oral/pharyngeal cavity, but
show no evidence of mineralized dermal armor.
The odontode sensu 0rvig did not include the tooth-integrated basal
tissue "bone of attachment" developed after but in topographic relationship
with dentine of the exoskeleton. We include bone of attachment as part of
the fundamental odontogenetic unit. All three tissues [enamel, dentine (pulp),
denticle I tooth
skin denticles
teeth
dental lamina
oral denticles
':' f
.<j< x .,".. }{"" ,"', ! L,
odontogeniC cells
denticle I tooth
odontogenic primordium
a
FIG. 2. Model redrawn from Reif (1982) to show his "odontode regulation theory" and
teeth developing from a dental lamina. It assumes that a zone of inhibition exists around
the developing primordium (extent indicated by the arrows) and further that it is still effective
in the fully formed tooth/denticle; evidence for the molecular/cellular control has not been
presented. All assume that the primitive craniate condition is microsquamous and that
skin denticles gave rise to teeth through the development of a dental lamina as shown in
(b), in which new denticles form in the inhibition-free space developed with growth of the
tissues, and new teeth develop from the dental lamina, at the inhibition-free end. In (a)
three stages show initiation, functional position, and loss by shedding of an odontode.
This provides the possibility that a new, larger denticle can develop in the inhibition-free
space; also the assumption is that shedding of denticles is the primitive craniate condition.
395
and bone] are present and anyone could have evolved earlier than the others
and be present in the fossil record independently of the others. This could
happen in development if part of the odontode differentiation program was
suppressed through a heterochronic shift, or if one cell type had not acquired
competence in the odontode system. Whether dermal bone or dentine or
bone in the odontode evolved first is contentious [see M. M. Smith and Hall
(1990) for a review, and M. M. Smith (1991)] and has been made more so
by the proposal that conodonts are vertebrates, a contention based on recent
histological evaluation of mineralized conodont elements (serving as teeth)
in which dentine is absent but cellular bone present (Sansom et aI., 1992)
(see p. 417) and on new descriptions and a reassessment of the anatomy of
conodonts (Aldridge et aI., 1993).
Reif(1982) based his "odontode regulation theory" of the dermal skeleton
on Schaeffer's proposal, but extended it to propose that regulatory processes
(we suggest, involving heterochrony, and cell or matrix-based specific regulatory molecules) are the means of varying differentiation programs of odontodes. Such variation would produce different unit products in different combinations from a single morphogenetic system, which would be the
developmental basis for evolutionary change of the dermal skeleton.
Reifproposed a model for evolution of the dermal skeleton starting from
a microsquamous condition in which early vertebrates possessed an exoskeleton of nongrowing odontodes covering the whole body uniformly [examples
are thelodont scales, and placoid scales of sharks (M. M. Smith and Hall,
1990, Fig. 5)]. Implicit in this proposal is that odontodes evolved before dermal
bone, and this was absent from these exoskeletons. Each odontode was easily
shed from its functional site by soft tissue resorption, as there was not an
ankylosis to underlying bone. Regrowth was from a new primordium developed in the soft tissue space (Fig. 2). Growth was from new morphogenetic
units developing as soon as inhibition-free space was created between existing
odontodes. This model depends on the concept of alternate fields of induction
and inhibition. Variability within these units could occur by altering such
parameters as spacing, size, shape, tissue proportions, or tissue types.
We examine possible developmental mechanisms that may regulate and
modify this system, and will suggest certain developmental constraints that
apply. The ultimate basis for understanding control mechanisms will be
through molecular control of genetic expression in the phenotype, but, as
suggested by M. M. Smith and Hall (1990), this will operate within the developmental framework of causal cascades of dependent, sequential steps, on
a controlled spatial and temporal basis. There are some very exciting new
developments in the understanding of the regulatory role of hom eo box genes
in an individual's development. That is, those that apply to pattern formation,
as discussed by Hanken and Thorogood (1993) and Thorogood (1993) with
396
regard to the evolution and development of the vertebrate skull, "mechanisms

of pattern formation-the events and processes by which cells are organized
into predictable spatial arrangements (morphologies) oftissues in their proper
locations." We discuss this below (p. 406).
In one early vertebrate with good preservation of histology and separate
odontodes ["the enigmatic third vertebrate" (Denison, 1967)], M. M. Smith
(1991) described enamel, cellular dentine, and cellular bone. There is good
evidence that an exoskeleton, with abundant odontodes, was extensive in the
trunk, tail, and head of the other earliest vertebrates [see M. M. Smith and
Hall (1990) for a description of these], and was much more developed than
the endoskeleton in the same regions. It is not possible to comment on uncalcified cartilage; it may either not have been present or have been present
but not preserved, an issue also discussed by M. M. Smith and Hall (1990).
In summary, odontodes are products ofa morphogenetic unit (the odontode primordium) developing through epitheliomesenchymal cooperative interactions, in constrained, causal, and temporal sequences. Modifications of
the differentiation program result in different phenotypes in the dermal skeleton in fishes, including teeth of the oral cavity.
Lepidotrichia as Dermal Fin Rays

One aspect of the skeletal tissues of the dermal skeleton that should be
addressed in this review is the development and embryonic origin of the
dermallepidotrichia, the distal elements of the fin rays formed in association
with the actinotrichia and which develop without passing through a cartilaginous precursor. Frequently in many fossil fishes, and in some extant fishes
such as Polypterus, dentine with an enamel covering is fused to the underlying
lepidotrichia. Any developmental model for skeletal tissue evolution and
variation should take into account this part of the dermal skeleton, especially
because it is suggested that neural crest mesenchyme is implicated as one
source of cells.
Both Hall (1991) and Thorogood (1991) considered the embryonic source
of skeletal tissues when evaluating the developmental basis of limb evolution.
Hall discussed the origin of the dermal fin skeleton as probably neural crest,
and the loss of these elements in the transition from fin to limb. Thorogood
commented that the neural crest origin of the lepidotrichia remains to be
demonstrated. Prior to this, Schaeffer (1977) had reviewed the origins of cells
that contribute to the induction and skeletogenesis of the median and lateral
paired fins, concluding that induction of median fins is by neural crest-derived
cells, but that the lateral fins are different, being induced by lateral plate
mesoderm. Probably the most significant statement for this review is that by
Schaeffer (1977), who commented that the histological resemblance between
397
lepidotrichia and scales does not indicate that one evolved from the other,
but that "scales and lepidotrichia are. . . manifestations of the same morphogenetic system." This statement captures the basis for our model of development of the dermal component of the fin skeleton.
Both Schaeffer (1977) and Patterson (1977) rejected the delamination
theory for derivation from epitheliodermal morphogenetic units of superficial
and deeper separate elements ofthe dermal skeleton [conceived by Holmgren
(1940) and applied to the dermal fin rays by Jarvik (1959)] as having any
value as an explanatory principle for development or evolution of the fin
skeleton. But it had, for some time previously, been accepted, leading to the
implication that "there is some sort of interchangeability between the dermal
skeleton and endoskeleton" (Patterson, 1977, p. 110). This is a concept accepted neither by Patterson nor by the authors of this review (see section
below on the experimental embryology of fin rays p. 431).
Odontogenic Dependence on Neural Crest and Extent a/Competence

An implication of the postulate that all dentoskeletal elements of the
exoskeleton require neural crest as a component and are based on the odontode
as the developmental system is that neural crest is necessary for induction,
morphogenesis, and differentiation of odontodes wherever they occur on the
body and in whichever vertebrates they occur. In mammals, reptiles, and
amphibians, odontodes are found in the oral tissues, where they function as
teeth, whereas in extant fishes they occur throughout the exoskeleton, on
scales (Bhatti, 1938; M. M. Smith, 1979), fin spines (Maisey, 1979), and lepidotrichia (Geraudie, 1988). In all documented cases neural crest is required
for induction and development of teeth (p. 428) (M. M. Smith and Hall,
1990). No data are available, however, concerning neural crest involvement
in the development of teeth or denticles of fishes.
If we consider that most of the earliest recorded fossil vertebrates had an
extensive exoskeleton composed of either thick dentoskeletal mineralized tissues or separate denticles inserted via their basal bone into a soft connective
tissue, then, following our postulate of neural crest origin of odontodes, neural
crest would have provided dentoskeletogenic competence over the entire vertebrate body. Presumably this competence was acquired before mesodermallyderived mesenchyme had evolved competence to develop perichondral membrane bone or endochondral bone. We do not know if dermal bone was ever
derived from a mesodermal mesenchymal population (see p. 390).
Gans and Northcutt (1983, 1985) and Northcutt and Gans (1983) proposed that neural crest was a vertebrate innovation, from which the first vertebrate skeletal tissues arose. They did not address the question of whether it
was initially cranial neural crest that was skeletogenic or both cranial and
398
trunk together. If we extend the concept that neural crest and the tissues
derived from this source account for all vertebrate synapomorphies, one of
which is the dentoskeletal tissue, to all fossil fishes, then the following is also
true. In the earliest recorded vertebrates with isolated scales (odontodes) in
their skin, without extensive sheets of dermal bone (M. M. Smith, 1991),
then, at least in some early basal vertebrates, neural crest was probably the
only skeletogenic population in the dermis. 0rvig, T, (1951), Obruchev (1964),
Karatajute-Talimaa (1978), Reif (1982), Maisey (1988), and most recently
Blieck (1992) concluded that micromeric dermal armor (scales from head to
tail, and absence of plates of bone), as in thelodonts, was the primitive vertebrate condition. Thelodonts are therefore one basal vertebrate group in
which neural crest would have been the only skeletogenic population of both
head and trunk, because no sheets of bone developed in their exoskeleton,
and there is no evidence of either calcified or noncalcified cartilage.
Evidence for Skeletogenic/Odontogenic Cranial versus Trunk Neural Crest

In a review of the development and evolutionary origin of skeletal tissues
in vertebrates, M. M. Smith and Hall (1990) distinguished between skeletogenic and odontogenic potentials in cranial neural crest populations and argued
for the existence of subpopulations of neural crest cells morphogenetically
specified and with restricted differentiative potential. These were cells which
after migration and superficial location beneath epithelia would, as an odontogenic subpopulation, differentiate into tooth buds and produce dentine and
bone of attachment. The other skeletogenic subpopulations of cranial neural
crest would differentiate into cartilage, dermal bone, perichondrium and periosteum, and periosteal and endochondral bone, for cartilages in the visceral
skeleton and bone in the facial skeleton.
These suggestions of separate skeletogenic and odontogenic cell populations were expanded further by Atchley and Hall (1991) in a review using
the mouse mandible as a model for development of complex morphological
structures (see p. 402).
Evidence for a substantial role of cranial neural crest in induction and
skeletogenesis of the head exoskeleton and endoskeleton in tetrapods began
to accumulate as data was provided from experimental embryology. By contrast, trunk neural crest has been consistently demonstrated not to share this
skeletogenic potential. In this context, several references to dermal armor of
fossil fishes were made. These were put on one side for later investigation by
de Beer (1947) and explained by Wilde (1955) as dermal bone and a denticle
system developed from head ectomesenchyme with a protective function,
assumed to later extend into the trunk. Gaunt and Miles (1967) realized that
399
there was a fundamental question to be answered in considering the possible

neural crest origin of these exoskeletal tissues, "the apparent common origin
in ontogeny of membrane bone and dentine from neural crest is of great
interest and raises fundamental questions which cannot be answered
decisively."
Although much ingenious and delicate experimental embryology has
been carried out to address the question of trunk neural crest competence,
the vital set of experiments creating temporal and spatial conditions in which
competence (potential) of trunk neural crest to express odontogenic and skeletogenic ability can be expected has only just begun to be undertaken in
amphibian embryos (Graveson, 1993) (see following section) and has yet to
be initiated in fish embryos.
Evolutionary considerations would suggest that dentine and/or boneforming capabilities of trunk neural crest have either been lost or are latent,
but suppressed in amphibians, reptiles, birds, and mammals. (We have excluded consideration of the embryonic origin of osteoderms in amphibians
and reptiles in this review, due to lack of both data and understanding.) Several
reasons can be put forward to explain why this potential is not expressed
when amphibian, avian, or mammalian trunk neural crest is grafted into the
head in place of cranial neural crest, an experimental design frequently used
to test the potential of trunk neural crest. The first reason may be spatial,
reflecting lack of migration of trunk neural crest in the cranial site (Graveson,
1993). The second reason may be the lack of suitable extracellular substrates
(fibronectin, etc.) required to maintain trunk neural crest cells. The third
reason is temporal; trunk neural crest may not confront inductive epithelia
at the appropriate time for required epithelial-mesenchymal interactions
to occur.
One set of experiments has been carried out by Lumsden (1987) in
postconception-aged mice from embryonic day 8 (E8) to day 15 (EI5). Heterochronic explants of odontogenic mandibular epithelium (E 10) combined
with premigratory trunk neural crest from (E8) resulted in the formation of
teeth and adjacent bone, but no cartilage. The segment of neural crest taken
in Lumsden's experiments was from the level of the open posterior neuropore,
or future cervical somites (S5-1O); in this restricted location (caudad to S5),
the most rostral trunk neural crest could form dentine and bone of attachment
in response to mandibular ectoderm [this has been discussed in detail by
M. M. Smith and Hall (1990, p. 333)]. This result alone suggests that postcranial ectomesenchyme can be induced to participate in odontogenesis and
osteogenesis, but not chondrogenesis. Recent work by Graveson (1993) and
A. Graveson, M. M. Smith, and B. K. Hall (in preparation) has also revealed
that the segment of axolotl cranial neural crest immediately caudad to the
400
chondrogenic region will make teeth but not cartilage, confirming in amphibians Lumsden's (1987) conclusions that some mammalian postcranial
ectomesenchyme is odontogenic.
The experimental demonstration of odontogenic capability in mammalian trunk neural crest and axolotl caudad cranial neural crest suggests
that the potential to make odontogenic tissues should be present in lower
vertebrates. As suggested previously, "because postcranial skeletal tissues develop from mesoderm and not from neural crest, developmental biologists
have drawn an absolute dichotomy between a cranial skeletogenic/odontogenic
and a trunk non-skeletogenic/non-odontogenic neural crest." (Smith and Hall,
1990, p. 332). The region oftrunk neural crest with odontoskeletogenic ability
might be expected to be more extensive in those extant fishes with an exoskeleton than so far shown for amphibians.
Dependence of Neural Crest on Epithelium for Acquisition of Morphogenetic

Specificity
Before discussing the role of neural crest in patterning skeletal morphogenesis, we should consider the dependence on, and role of, epithelia in this
interaction. Lumsden (1988) concluded that "tooth initiation involves an
inductive interaction between region specific epithelium and competent but
unspecified mesenchyme cells, normally derived from the cranial neural crest"
and also "a prepatterned distribution of inductive potency may exist in the
epithelium. . . thus controlling the spatial organization of the dentition as
a whole." This model could also provide a patterning mechanism for scales/
denticles in the entire body exoskeleton of fishes, with an inductive signal
from the epithelium, perhaps only effective outside zones of inhibition. Reif
(1980) proposed a model of morphogenesis in the dermal skeleton of sharks
to explain regulation of new scale development using the inhibitory field model
for teeth of Osborn (1978) (see Fig. 2).
Lumsden (1987) had previously cautioned about conclusions that ectomesenchyme produces the initial signal for differentiation of teeth, although
it may initiate tooth morphogenesis. Many experiments addressing tooth
morphogenesis involve ectomesenchyme that has been primed to make teeth,
before histological recognition of tooth sites-"the patterning process involved
in tooth initiation, controlling its timing and positioning, are critical events
that must precede the morphological appearance of the tooth germ." In a
precisely chosen series of times and sites, Lumsden (1987) demonstrated that
tooth development in mammals is initiated via a permissive signal of specific
oral ectoderm on competent ectomesenchyme.
Graveson (1993, and personal communication) has found in isochronic
and heterochronic recombinations of neural crest with ecto/endoderm as ex-
401
plants from amphibian neurulae that ectoderm alone will not induce teeth
from neural crest, but endoderm will. In mammals endoderm also could be
a prerequisite for tooth initiation, because the precise boundaries of ectoderm
and endoderm in the oral cavity are not known (A. Lumsden, personal
communication).
It may be significant that there appears to be a difference between initiation of tooth and bone morphogenesis, although both share dependence on
epithelium for the inductive signal. Hall (1983) showed that foreign epithelia
can permit mandibular ectomesenchyme to differentiate into membrane bone
in the chick (i.e., the tissue generating the permissive signal is not site specific),
but that the shape of the bone is specified by the ectomesenchyme. Experiments
by Noden (1983) also led to the suggestion that some populations of cranial
neural crest, those that are destined to make the visceral cartilages, are morphogenetically specified prior to emigration. Such divergent mechanisms
would effectively uncouple the patterning mechanism of the two components
of the exoskeleton, "odontodes and membrane bone," and those of the viscerocranium. The potential patterning mechanisms in the vertebrate skull
have been recently reviewed by Hanken and Thorogood (1993), who suggest
that one fundamental topic that remains to be adequately addressed is "the
relation of initial embryonic patterning to adult skull morphology, and especially to the dermatocranium." These issues are further discussed in the
following section.
Exoskeletal Diversity and a Developmental Model

The greatest diversity of skeletal tissues is apparent among the exoskeletons of lower vertebrates. Variation occurs in tissue type (enamel, enameloid;
mesodentine, semidentine, orthodentine); pulp (open, closed, or infilled with
dentine or bone); bone (cellular or acellular); arrangement of bone tissues
(spongy or compact); bone composition (from surface parallel lamellae, or
lamellae concentric with vessels); whether bone tissue has inserted attachment
fiber bundles; whether tubercles (odontodes) on the dermal armor are resorbed,
or nonresorbed and superimposed with new growth; whether skin denticles
are free (odontodes), or ankylosed to bone; whether dermal bone occurs as
extensive, continuous sheets, or as small plates called jointed tesserae. A model
offering a precise understanding of development, in terms of spatial and temporal patterning, as well as specific molecular data on control systems, would
be powerful aids to explaining the evolution and variation of such skeletal
types and their arrangement (see p. 406).
A developmental model from which skeletal evolution and diversity in
the dermal armor can be explained is based on the concept of both a single
402
T ABLE I.
The Cranial Neural Crest-Derived Populations of Cells Associated

with the Dentary of the Mammalian Mandible
Cell population
Cell types derived from the population
Chondroblastic/first arch
Skeletogenic/ramus
Chondroblasts of Meckel's cartilage

Osteoblasts of ramal bone of dentary, chondroblasts of
callus cartilage in fractured mandibles, and
secondary chondroblasts of the mandibular
symphysis
Osteoblasts of bone of the condylar, coronoid, and
angular processes; chondroblasts of secondary
cartilage of the bony processes
Odontoblasts of the teeth, fibroblasts of the dental pulp
Osteoblasts of alveolar bone, fibroblasts of the
periodontal ligament, cementoblasts of cementum
Skeletogenic/processes
Odontogenic/papilla
Skeletogenic/follicle
morphohistogenetic unit, the odontode (utilizing a sequence of epitheliomesenchymal interactions), and a bone morphohistogenetic unit (blastema or
condensation), both with intrinsic capability of inducing cell differentiation,
establishing and controlling shape (morphogenesis), and producing variation
in tissue type and tissue arrangement. Development of the three tissues of
the odontode (enamel, dentine, and bone of attachment) is linked in a causal
sequence, independent of, although coordinated with, the development of
deeper basal bone.
Cell Populations of the Mammalian Mandible

A model for coordination of odontode patterning and association with
a dermal bone in the exoskeleton may be developed from one produced for
a restricted part of the dermal skeleton in mammals, the mandible and teeth.
Atchley and Hall (1991) proposed that the population of migratory neural
crest cells that reach the maxillary and mandibular arches to contribute to
the teeth, cartilage, and the basal dermal bone are restricted in their potential
and establish three subpopulations, one only osteogenic, one only chondrogenic, and one only odontogenic. A further restriction of the odontogenic
population into two subpopulations is discussed below, dental papilla (dentine
and pulp-forming) and dental follicle (cementum, periodontal ligament, and
lining alveolar bone-forming, the latter assumed to be homologous with bone
of attachment in lower vertebrates). Evidence for such restrictions is provided
by Palmer and Lumsden (1987) and Vaahtokari et al. (1991) and discussed
by M. M. Smith and Hall (1990, p. 320). Atchley and Hall (1991) proposed
403
a classification ofthese cells which we now modify and expand. (Tables I and
II). The following cell populations can be identified.
1. A chondrogenic population differentiating into Meckel's cartilage, the

cartilaginous precursor of the lower jaw, and the skeletal element directing growth of the mandible.
2. A skeletogenic population forming bone of the body (ramus) of the
dentary, chondroblasts of the symphyseal cartilages, and chondroblasts
in repair of fractures of the body of the mandible. Symphyseal cartilage
arises after bone formation; they are both secondary cartilages. The
skeletogenic population therefore produces osteoblasts of ramal bone,
callus cartilage in fractures, and secondary cartilage of the symphysis
(Table II).
3. A skeletogenic population forming the bone and secondary cartilage
of the three bony processes of the dentary, the condylar, coronoid,
and angular processes.
4. Populations associated with the teeth, called odontogenic by Atchley
and Hall (1991) to signify that they produce odontoblasts and deposit
dentine at a particular stage in their development. Initially, these are
the cells of the condensed dental mesenchyme associated with the
developing enamel organ (Fig. la). This population of cells subdivides
into two, one population becoming the dental papilla and the other
the dental follicle (Fig. 1b). Cells in the dental papilla differentiate
into odontoblasts and fibroblasts of the pulp. These cells are not osteogenic. They are the population labeled odontogenic/papilla in Table
II. Cells in the dental follicle differentiate into osteoblasts of the alveolar
bone, fibroblasts of the periodontal ligament, and cementoblasts of
TABLE II.
A Further Summary of Cell Types Formed from the Cell

Populations Summarized in Table I
Cell typea
Cell population
CB
Chondroblastic/Meckel's
Skeletogenic/ramus
Skeletogenic/processes
Odontogenic/papilla
Skeletogenic/follicie
OB
CC
SC
+
+
+
+
OD
PFb
PDLFb
CB, chondroblasts; OB, osteoblasts; CC, callus cartilage chondroblasts; SC, secondary chondroblasts of secondary cartilage; OD, odontoblasts; PFb, pulpal fibroblasts; PDLFb, periodontal
ligament fibroblasts; C, cementoblasts.
404
the cementum, three tissues anchoring the tooth into the ramal bone
of the mandible. These are the cells labeled skeletogenic/follicle in
Table II.
The tissue types formed by these different cell populations, all of which
arise from the cranial neural crest, are further summarized in Table II.
Models for morphogenesis of such a complex structure as the exoskeleton
will require knowledge of such subpopulations of ectomesenchymal cells.
Evidence that neural crest may have such population heterogeneity comes
from in vitro studies using quail and mouse premigratory neural crest (Heath
et al., 1992). These showed at least four separate populations within individual
crest cultures, each with a unique antibody reactivity pattern. These results
using both mesencephalic and trunk premigratory neural crest revealed different epitope expression 15 hr after explantation of premigratory neural crest
at a stage when all cells were still morphologically identical. Essentially these
in vitro studies are compatible with previous cell lineage studies using dye
injection to mark emigrating neural crest cells (Fraser and Bronner-Fraser,
1991). These data demonstrated that restriction of neural crest cell fate must
occur relatively late in migration, and that migrating trunk neural crest cells
can be multipotent.
In a review of neural crest lineages, Marusich and Weston (1991) discussed
how developmentally restricted intermediate subpopulations might arise sequentially by differential responses to developmental cues on their migration
pathways. The molecular cues to determine phenotypic expression are part
of the control mechanisms that select odontogenic and skeletogenic populations from intermediate populations of partially committed neural crest cells.
An explanation of skeletal diversity in the exoskeleton of vertebrates,
using developmental and molecular data on the epitheliomesenchymal interactions of teeth and supporting dermal bone, will be proposed from data
now available from the development of mammalian teeth (see next subsection).
We propose that the same set of sequential restrictions would operate to produce subpopulations of neural crest cells in the trunk and head dermal skeleton.
Odontogenic and Skeletogenic Subpopulations of the Cranial Neural Crest

Odontogenic and skeletogenic populations of cranial neural crest cells
are clearly a fundamental part of mammalian tooth development. The range
of cell types arising from these populations localized, or at least recognizable,
in the dental papilla and dental follicle is enormous. They produce odontoblasts of the teeth, osteoblasts of alveolar bone, cementoblasts of cementum,
and fibroblasts of both the pulp and the periodontal ligament.
Clearly, the odontogenic lineage of the dental papilla separates out from
the skeletogenic lineage of the dental follicle at an early stage in morphodif-
405
ferentiation ofthe tooth germ. Exactly when is not clear. This is an important
subdivision, separating as it does an odontoblast-fibroblast lineage from a
cementoblast-osteoblast-fibroblast lineage. Presumably a common progenitor
capable of forming all these cell types exists at the pre-dental papilla/follicle
stage (as shown in Fig. la, odontogenic ectomesenchyme). Experiments of
Palmer and Lumsden (1987) show a choice in differentiation dependent upon
location of the ectomesenchyme. Contact with inner epithelium produces
odontoblasts, whereas contact with outer epithelium produces cern en to blasts
(see Fig. 1c).
Experimental evidence for the existence of such separate odontogenic
and skeletogenic lineages comes from experiments in which isolated dental
follicles have been shown capable of forming these tissues (Ten Cate and
Mills, 1972; Ten Cate, 1975; Osborn, 1984; Osborn and Price, 1988; Hall,
1992a,b) and from the physiological individuality of alveolar from ramal
bone. Alveolar bone has a much higher turnover rate than does ramal bone
and is dependent on biomechanical factors for its continued existence (Stutzman and Petrovic, 1989; Petrovic et aI., 1990; King et al., 1992). Alveolar
bone is rapidly resorbed when teeth are extracted; ramal bone is not. Ramal
bone develops in relative independence from any extrinsic influences from
either the teeth or from muscles inserting on the dentary.
Cells of the skeletogenic lineage of the dermal bone can form cartilagethe secondary chondroblasts found in alveolar bone. Cells of the odontogenic
lineage are not chondrogenic. Secondary cartilage only forms after osteogenesis
has been initiated and arises as a localized response to functional need. The
ability of skeletogenic cells to form secondary cartilage is a property not shared
with cells of the follicle, but is confined to osteogenic cells of the ramus (which
form symphyseal cartilage) and with osteogenic cells of the bony processes
(which form secondary condylar, coronoid, and angular cartilages).
Several independent criteria therefore can be used to distinguish between
and among skeletogenic and odontogenic cell populations. These include the
forming primary tissues, the metabolic nature of the forming bone, dependence
of the bone on mechanical and extrinsic factors such as muscle or tooth
action, and whether and what type of cartilage forms.
Murine cranial neural crest contains both odontogenic and skeletogenic
lineages. Rostral murine trunk neural crest contains only the odontogenic
lineage, including its osteogenic subpopulation, but probably not the skeletogenic lineage (Lumsden, 1985, 1987, 1988). Amphibian cranial neural crest
contains both odontogenic and skeletogenic lineages. Rostral amphibian trunk
neural crest contains odontogenic lineages [see the studies of Graveson (1993)].
Thus, studies on murine and amphibian neural crest indicate odontogenic
potential existing outside the cranial neural crest. Potential to form teeth
extends beyond the cell populations normally forming teeth. Odontogenic
406
and skeletogenic cell populations can be distinguished and separated from

one another along the neural axis from rostral (cranial) to caudal (trunk),
odontogenic populations extending further caudal than skeletogenic
populations.
Given these studies on mouse and amphibian embryos and given that a
dermal exoskeleton of dentine and bone exists in many fishes, we would
expect trunk neural crest of such fishes to possess odontogenic cell lineages.
Indeed, in two teleost fishes, the goldfish and the Nibe croaker (Nibea mitsukurii), dentine and dermal bone can be obtained from cell lines derived from
tumors of neural crest origin (Matsumoto et aI., 1983).
On the other hand, we would expect that vertebrates, such as birds, having
lost teeth during their evolution, would have lost the odontogenic lineages
from both cranial and trunk neural crest cell populations. Cross-species tissue
recombination experiments of Kollar and Fisher (1980) indicate this to be
the case, at least for mesenchyme derived from cranial neural crest (they did
not test trunk neural crest). Teeth develop when murine tooth mesenchyme
is combined with chick mandibular oral epithelium, a finding indicating chick
epithelium to have retained the ability to induce mouse dental mesenchyme
to differentiate into odontoblasts and deposit dentine, and to be induced by
murine mesenchyme to form enamel in these chimaeric teeth. Teeth failed
to form, however, when chick mandibular mesenchyme was recombined with
mouse dental epithelium. This again emphasizes the primacy of the epithelium
in initiation of the sequence of inductive interactions that typifies this epitheliomesenchymal model. Chick neural crest-derived mesenchyme has lost
dentine-forming capabilities, even though oral epithelium retains enamelforming potential, albeit unrealized other than under experimental conditions.
Molecular Control Mechanisms

In the first two parts of this section we sought to elaborate a cellular basis
for exoskeletal diversity using, as model systems, cell populations in the mammalian mandible and the discordance between skeletogenic and odontogenic
cell populations within and between cranial and trunk neural crest. Eventually,
we will require an explanation of exoskeletal diversity that includes molecular
control mechanisms. Some signposts on the road to that explanation are provided in this section.
Molecular control mechanisms will have to be sought for at least the
following developmental processes or events: (1) the origin of neural crest
and mesodermally-derived scleroblast-forming cells; (2) the developmental
origins of the neural crest; (3) segregation of the neural crest into cranial and
trunk domains; (4) segregation of subpopulations of cells within neural crest
and mesoderm; (5) onset and timing of differentiation of skeletogenic and
407
odontogenic cell types; (6) patterning of cell populations into skeletogenic

and odontogenic tissues and organs; (7) regulation of turnover, replacement
and growth of skeletogenic and odontogenic tissues and organs; (8) coupling
or uncoupling of formation, rates, and timing of development of tissues such
as dentine and bone within morphogenetic units such as odontodes and their
accompanying osteogenic condensations. Needless to say, we are a long way
from understanding the entirety, subtlety, or complexity of even one of these
processes or events.
Analysis of molecular control mechanisms will require an understanding
of whether control resides within individual cell populations, i.e., is intrinsic,
or whether it resides, in whole or in part, within adjacent cell populations,
i.e., is extrinsic or epigenetic (Hall, 1992a,b). In reality, exoskeletal and dental
development is a subtle mix of intrinsic cellular properties acquired during
ontogeny (and sometimes lost later in ontogeny) that are only realized following interactions with adjacent cell populations and/or their extracellular
products, also subject to their own temporal and spatial regulation during
ontogeny.
Furthermore, differentiation, morphogenesis, and growth of individual
cell populations, tissues, or organs may be controlled by different mixes of
intrinsic and epigenetic controls, mixes that can change throughout ontogeny.
In this way, differentiation, morphogenesis and growth may be coupled or
uncoupled, co- or independently regulated, or cause or effect. Such flexibility
provides a powerful basis for developmental/ontogenetic, and evolutionary/
phylogenetic modulation of the exoskeleton and teeth (M. M. Smith and Hall,
1990; Atchley and Hall, 1991; Hall, 1992a,b).
Patterning of the Neural Crest. Given the important role assigned to the
neural crest in the development and evolution of exoskeletal and dental tissues,
we begin our analysis by assessing the types of molecular control mechanisms
responsible for patterning the neural crest. Such mechanisms are likely to
have been how cranial and trunk neural crests were delineated and skeletogenic
and odontogenic neural crest cells established. Langille and Hall (1993) have
recently analyzed patterning of the neural crest, indeed of the entire head, in
some detail. As the issue of whether the head or its component elements is
segmented is discussed in that review and in some depth in three recent volumes devoted to the vertebrate skull (Hanken and Hall, 1993) segmentation
per se will not be considered here. Hanken and Thorogood (1993), in their
analysis of the role of pattern formation in the vertebrate skull, compare two
different models, one in which pattern is specified intrinsically and the other
in which an extrinsic, molecular prepattern is expressed by the neuroepithelium, to which skeletogenic mesenchyme conforms. It may be very significant
that the first model applies generally to the viscerocranium (the morphogenetic
specificity of neural crest referred to earlier in Noden's experiments); the sec-
408
ond, applies primarily to the neurocranium where gross spatial patterning is

established by the matrix molecule, type II collagen, expressed transiently by
the neuroepithelium and sensory epithelium as a prepattern, to which the
mesenchyme responds. Experimental research is offering a molecular basis
for intrinsic differences in the pattern-forming ability of neural crest cells,
based on homeobox genes, as discussed later (see also Hanken and Thorogood,
1993). These gene products bind to other genes and regulate their expression,
and in this way are candidates for determining the development of visceral
arch-specific morphological patterns.
An important problem when considering patterning of the neural crest
is the establishment of boundaries. What mechanisms determine where brain
stops and spinal cord begins, where forebrain ends and midbrain starts, where
cranial neural crest ends and trunk neural crest begins, where the posterior
boundary of the skull lies? Since the pioneering scenario of the neural crest
origin of the vertebrate head provided by Northcutt and Gans a decade ago,
the skull has been seen as a composite of a neural crest anterior (rostral)
portion fused onto a more ancient posterior (caudal) mesodermal portion.
As the neural crest arises in the developing neural tube, it is to the latter
that we turn for molecular clues of patterns that might be imposed onto the
neural crest while it is still within the neural ectoderm. The most obvious
pattern, long suspected but only recently directly demonstrated, is organization
of the developing brain into segmentally arranged rhombomeres (Lumsden
and Keynes, 1989; Lumsden, 1990; Fraser et al., 1990). This neuromeric
organization and the boundaries that it establishes are reflected in the patterning of the migration of neural crest cells away from the brain (Hunt et
al., 1991; Hunt and Krumlauf, 1991), i.e., primary patterning of the brain is
imposed onto neural crest-derived mesenchyme.
The molecular basis of this pattern has been shown to reside in the pattern
of expression of homeobox genes, which are either restricted to individual
neuromeres or whose rostral or caudal limits of expression coincide with
neuromeric boundaries (Holland, 1988; Wilkinson et aI., 1989; Kessel and
Gruss, 1990; Lonai and Urtreger, 1990; McGinnis and Krumlauf, 1992). Hox
genes continue to be expressed in spatially regulated patterns in mesenchyme
derived from neural crest cells, as is discussed in the following section.
Direct evidence for the role of homeotic and other patterning genes in
establishing boundaries comes from such recent studies as that by Krauss et
al. (1992) on the role of the gene pax[ b] in determining the midbrain-hindbrain
boundary in the zebrafish, Brachidanio rerio, pax[b] is expressed in the midbrain and future midbrain region of the neural tube from gastrulation onward.
Injection of antibodies against pax[b] into fertilized eggs resulted in specification of abnormalities at the midbrain-hindbrain boundary. These boundary
anomalies were associated with failure of expression of endogenous pax[b] in
409
the posterior midbrain, and the alteration of such other patterning genes as
wnt-l and en-2. Thus, we see that individual genes and their downstream
effects play specific roles in establishment of boundaries within individual
organ systems.
Evidence for a patterning role of Hox genes comes from experiments in
which animals (usually mice) are made transgenic by insertion of a homeobox
gene such that patterns of spatial expression of the genes are altered, the gene
being expressed either more rostrally or more caudally than normal. Such
ectopic expression of Hox genes repatterns mesenchymal cell populations and
skeletal elements.
Lufkin et al. (1991) disrupted Hox-I.6 in embryonic stem cells using
homologous recombination and then introduced the transformed cells into
the germline of mice. The resulting embryos exhibited defective hindbrain
development, lacked some cranial nerves and ganglia, and had abnormal
inner ears and malformed skull bones. Hox-I.6 has a rostral boundary of
expression; its disruption altered structures associated with the rostral
hindbrain.
Chisaka and Capecchi (1991) used gene targeting to disrupt Hox-I.5.
The resulting homozygous embryos displayed a syndrome-absence of thymus
and parathyroid glands, diminished thyroids, heart defects, and craniofacial
malformations-strikingly reminiscent of DiGeorge syndrome in humans.
DiGeorge syndrome is thought to result from defective neural crest development (Hall and Horstadius, 1988). Hox-I.5 is one of the earliest Hox genes
expressed in the mouse embryo. Disruption of Hox-I.5 therefore produces
defects in many tissues.
Other Hox genes have more specific actions. Insertion of Hox-l.l into
mouse embryos produces an extra vertebra, a proatlas (Kessel et al., 1990).
Perturbation of Hox genes by retinoic acid causes anterior shifts in Hox gene
expression and posterior transformations of vertebral type (Kessel and
Gruss, 1991).
Hox-4.2 has a pattern of expression in the mouse such that the most
rostral (anterior) boundary of expression in mesoderm is at the level of the
somites forming the first cervical vertebra. If the pattern of expression of H ox4.2 is experimentally manipulated so that the gene is expressed more rostrally,
the occipital bones are transformed homeotically, into the next most posterior
skeletal element, cervical vertebrae (Lufkin et aI., 1992). These authors discuss
their results in the context of evolutionary transformations, with variation in
expression boundaries of Hox genes as a genetic mechanism responsible for
evolutionary modification of the skeleton.
Because homeobox genes provide a positional code utilized by both neural
crest and mesodermally-derived skeletal elements, Hox gene transformations
are particularly relevant to our analysis of exoskeletal and tooth patterning.
410
Patterning of the tissue producing exoskeletal scleroblasts (the neural crest)

by homeobox genes provides a molecular mechanism for patterning the exoskeletal or tooth derivatives of the neural crest along the body axis. Example
of tooth patterning involving Hox-7.1 and Hox-8 are discussed in the next
section.
Whether the embryonic ectoderm is also patterned is more difficult to
say. Couly and LeDouarin (1990) demonstrated an organization of anterior
embryonic ectoderm into regions they called ectomeres, but it is not yet accepted that they represent true cell-restriction areas. Ectomeric boundaries
do not correspond to neuromeric boundaries; indeed, they cut across neuromeric boundaries, so that a simple one-to-one correspondence between ectomeres and neuromeres does not exist. Nevertheless, if this ectodermal regionalization demonstrated in avian embryos exists in toothed vertebrates
such as fishes and mammals, it would provide a mechanism for regionalizing
ectoderm that will subsequently participate in formation of exoskeleton and/
or teeth. The molecular basis of ectomeric organization does not exist.
Odontodes and Condensations. In this section we assess molecular control
mechanisms by evaluating current knowledge concerning the two fundamental
units of the exoskeleton: the odontode, responsible for forming the odontogenic
tissues (p. 393) and the osteogenic condensation, responsible for forming bone
of attachment or alveolar bone (p. 402). There are sufficient parameters in
common between basic mechanisms that molecular control of odontodes and
condensations will be considered together. The cellular parameters of these
two fundamental units were discussed earlier.
Few studies on exoskeletal or dental development in fishes are available,
Nakajima's studies on pharyngeal teeth in larval and adult Japanese fishes
(Nakajima, 1984, 1987, 1990; Nakajima and Vue, 1989) and Huysseune and
Sire's (1992a) study on mandibular teeth being recent notable exceptions.
Given that we lack molecular information on odontodes and condensations
in fishes but do have information on molecular aspects of tooth and alveolar
bone formation in mammals (primarily derived from studies on mice), the
developing mammalian tooth will serve as our guide to molecular control of
odontode formation. We must bear in mind, of course, that mammalian teeth
are derived structures whose evolutionary history is removed from odontodes
of fishes. Nevertheless, given the parsimony of developmental mechanisms
that pervades vertebrate embryos in general (Langille and Hall, 1989; Hall,
1992a,b) and the common developmental bases of differentiation and morphogenesis of teeth and skeletal tissues (M. M. Smith and Hall, 1990) we feel
justified in taking molecular control of mammalian teeth as a realistic signpost
pointing the way to the types of information required to understand exoskeletal
and tooth development/evolution in fishes. A similar argument can be made
for utilizing knowledge of preosteogenic condensations acquired from avian
411
or mammalian embryos as the basis for exploring exoskeletal development

and evolution in fishes.
All the available literature on molecular control of mammalian tooth
development cannot be exhaustively reviewed in this chapter; recent overviews
of molecular control of tooth development are Mina et al. (1990), Slavkin
(1990, 1991), and Thesleff (1991). Rather, and to indicate the general categories
of molecular control involved, we provide a summary of the type of molecular
information that has accumulated concerning mammalian tooth development.
Vertebrate teeth are composite organs formed by two mineralized extracellular matrices: dentine, a mesenchymal (neural crest) product, and enamel,
an epithelial (ectodermal) product. Consequently, the history of tooth development is the history of interactions between epithelial cells, from which
enamel-depositing ameloblasts develop, and neural crest-derived mesenchymal
cells, from which dentine-depositing odontoblasts develop. Epithelialmesenchymal interactions therefore epitomize, characterize, and constrain
tooth development (Kollar, 1986; Slavkin, 1990, 1991; Thesleff, 1991). Intrinsic potential in each of these two cell layers is only realized following
activation by the other cell layer.
Mammalian odontodes, as they proceed from bud, to cap, to bell stages,
are classic representatives of the interplay between intrinsic and extrinsic control. It is at the cap stage that condensed tooth mesenchyme can first be
subdivided into odontoblast-forming cells of the dental papilla on one hand,
and the bone, cementum, and periodontal ligament-forming cells of the dental
follicle on the other (see p. 393). Tooth morphogenesis is established at the
bell stage; secretion of dentine and enamel quickly follows (see Fig. 1).
Essentially, two data sets pertaining to molecular control of tooth development are available: one is based on localization or visualization of molecules at particular stages, and the other derives from studies of the perturbation of molecules.
Syndecan, a proteoglycan on cell surfaces, and tenascin, a component
of extracellular matrices that binds to syndecan (Salmivirta et ai., 1991), are
present throughout tooth mesenchyme, initially at the bud, but in abundance
at the cap stages (Thesleff et ai., 1987, 1988). Vainio et al. (1989) demonstrated
that both syndecan and tenascin are induced because of interaction between
epithelial and mesenchymal components of developing teeth. Syndecan, tenascin, and probably fibronectin are required for initiation of the important
condensation stage of tooth development (Vainio et ai., 1992; Vainio and
Thesleff, 1992).
Although fewer data are available, a similar scenario is emerging for
control of preosteogenic condensations (Hall and Miyake, 1992). The initiating
action of an epithelial-mesenchymal interaction and subsequent interplay
between extracellular molecules such as fibronectin, cell-surface and cell-
412
adhesion molecules, and growth factors such as TGF-iJ (see below) facilitate
the formation of condensations and the selective gene activation that accompanies condensation formation.
Coregulation of these molecules therefore facilitates formation of the
fundamental anlage for teeth and for bone of attachment (alveolar bone).
The importance of the condensation at the onset of selective gene activation
during tooth or skeletal development cannot be overemphasized; see Hall
and Miyake (1992) for a recent review.
Once cells have accumulated in the odontogenic or preosteogenic condensation, growth factors increasingly regulate tooth and bone development.
Growth factors are small peptides. Originally thought only to act on transformed cells, growth factors have now been shown to be key players in regulating embryonic development; see Hall and Ekanayake (1991) for an analysis
of the role of growth factors in regulating development of the neural crest and
of neural crest derivatives. Growth factors have been visualized in developing
murine tooth primordia and experimental manipulation of individual growth
factors has begun.
. Once tooth development has been initiated, transforming growth factor
(TQF)-iJl can be visualized in the enamel organ, in the stellate reticulum and
in tieveloping tooth mesenchyme at both bud and cap stages of both mice
and rats (Cam et aI., 1990; D'Souza et at., 1990). TGF-iJl, 2, and 3 have been
vi!jualized at the epithelial-mesenchymal junction at these stages (Pelton et
at, 1990; Millan et al., 1991).
The most convincing demonstration for an association between TGFiJl and epithelial-mesenchymal interactions during tooth development comes
from the elegant study ofVaahtokari et al. (1991), who documented the temporal sequence of expression of TGF-iJl mRNA in dental epithelium and
demonstrated its regulation by dental mesenchyme. TGF-iJ mRNA was shown
to be present in dental epithelium at the bud stage and to increase rapidly as
tooth primordia progressed from bud to cap stages, at which time mRNA
was intense in the dental epithelium and initiated in dental mesenchyme.
Message was not seen in dental mesenchyme until odontoblasts began to
differentiate at 15 days of gestation. At the bell stage when the inner enamel
epithelium begins to form (18 days of gestation in the mouse), TGF-iJ mRNA
can no longer be visualized, but expression reappears in ameloblasts at 1 day
posthatching.
Significantly, Vaahtokari and colleagues demonstrated regulation by
dental mesenchyme of TGF-iJ mRNA synthesis in dental epithelium. This
demonstration was achieved by separating and recombining epithelial and
mesenchymal components of developing teeth. No TGF-iJ mRNA was expressed in dental epithelium combined with nondental mesenchyme. When
dental epithelium was recombined with dental mesenchyme, however, TGF-
413
{3 mRNA was expressed in the epithelium. It is clear that dental mesenchyme

regulates the synthesis ofTGF-{3 mRNA in dental epithelium. These workers
noted that TGF-{3 synthesis was controlled by the product of the Hox-7.1
gene, providing a potential link between epithelial-mesenchymal interactions,
homeobox genes, and growth factor expression; see page 414 for homeobox genes.
Nerve growth factor (NGF) and NGF receptor (NGFR) have been
localized in preameloblasts and odontoblasts and temporal regulation of
NGFR suggested; NGFR is only transiently expressed in inner dental epithelium and in odontoblasts (Mitsiadis et aI., 1992). The temporally and
spatially complex patterns of expression displayed by NGF and NGFR are
consistent with changing roles in regulation of both differentiation and
morphogenesis.
Epidermal growth factor (EGF) receptors are present in the enamel
organ at the outset of tooth development, but are lost from both inner and
outer enamel epithelia as they differentiate from the enamel organ (Cam
et aI., 1990) [see Partanen (1990) for a review]. EGF is found in both
epithelial and mesenchymal components of developing murine teeth
(Slavkin et aI., 1990).
EGF within odontodes is temporally regulated (Partanen and Thesleff,
1987; Topham et aI., 1987). At the bud stage, EGF is found in the dental
epithelium but not in the dental mesenchyme, a pattern that is reversed at
the cap stage. EGF persists in the dental follicle, the precursor of alveolar
bone, cementum, and periodontal ligament. EGF regulates gene expression
in both epithelial and mesenchymal components of developing teeth; EGF
inhibits synthesis of type I collagen by odontoblasts and synthesis of enamel
protein by ameloblasts when tooth buds are cultured in the presence of EGF
(Hata et aI., 1990).
A role for EGF has been forcefully demonstrated by Kronmiller et al.
(1991), who showed that an antisense oligodeoxynucleotide to EGF blocked
odontogenesis in tooth germs from E9 embryos explanted in vitro, apparently
by blocking initiation of odontoblast differentiation. EGF itself is transcriptionally regulated by retinol (vitamin A). Mandibles from 9-day-old embryos
exposed to 1-5 JLg retinol/ml produce supernumerary teeth associated with
enhanced epithelial proliferation and increased expression of EGF mRNA
(Kronmiller et aI., 1992). Therefore tooth patterning, including specification
of the number of teeth primordia, can be modulated by agents such as retinoids
which act on growth factors such as EGP. As discussed above when considering
patterning of the neural crest, retinoids do act as pattern-generating molecules
by modifying the patterns of expression of hom eo box genes. Retinoids have
now been shown to specify patterns in a number of embryonic organ systems
414
(Maden et al., 1992; Murphy et al., 1992). As discussed below, a link between
retinoids, growth factors, and homeobox genes has also been established.
Homeobox genes, increasingly being recognized as important components of axis, polarity, and segmentation specification in invertebrates and
vertebrates, have already been described as key molecular players in specification of the neural crest and its regionalization into cranial and trunk
neural crest. They also appear to play a role in specification of tooth
morphogenesis.
Mackenzie et al. (1991) found Hox-7.1 expression to be maximal in
dental mesenchyme at the cap stage when dental mesenchyme condenses and
tooth morphogenesis is initiated. It may be of morphogenetic significance
that H ox-7.1 can be activated by retinoids, and that retinoic acid receptor
and retinoic acid-binding proteins have both been localized in developing
teeth. Theslef et al. (1990) represented Hox-7.1 as active early in the cascade
of molecular changes leading to expression of syndecan, tenascin, growth
factors, and ultimately to differentiation of dental mesenchymal cells as odontoblasts and deposition of dentine. Epithelial-mesenchymal signaling regulates
this molecular cascade.
Another homeobox gene, Hox-B, has also been shown by Mackenzie et
al. (1992) to be localized at sites and times oftooth morphogenesis, including
localization in a proximodistal gradient in dental mesenchyme at the sites of
future tooth formation. Hox-B is found in the oral epithelium at the dental
placode stage, a spatial and temporal localization consistent with a potential
role in specification of tooth position.
A direct link between epitheliomesenchymal interactions, Hox gene
expression, and osteogenesis of neural crest-derived mesenchyme has been
demonstrated by Takahashi et al. (1991). Quox-7 (the quail homologue of
mouse Hox-B, now msx 2) is only expressed in mandibular mesenchyme of
the embryonic quail when epithelium is present. Osteogenesis of mandibular
mesenchyme depends upon interaction with mandibular epithelium (Hall,
1983). It is obviously tempting to link these two observations causally and to
speculate that the mandibular epitheliomesenchymal interaction initiates osteogenesis after activation of Quox-7 (msx 2): msx 1 (Hox-7) in the chick is
known to regulate muscle differentiation. Also in the chick msx 1 is expressed
in the mesenchyme of all the branchial arches (A. Graham, personal
communication).
So we see that cell surface molecules (syndecan), components of extracellular matrices (tenascin, fibronectin), growth factors (EGF, TGF-,B), homeobox genes (Hox-7.1, B.1), and retinoids represent five classes of interdependent molecules involved in molecular control of odontode formation.
The potential for developmental and evolutionary modulation of such pentapartite control is enormous.
415
Origins of Skeletal Tissues and an Evolutionary Model

Since the review of evolutionary origins of vertebrate skeletogenic and
odontogenic tissues by M. M. Smith and Hall (1990), a new interpretation
of the presence of osteocytic bone for one of the Harding Sandstone early
vertebrates has been published (M. M. Smith, 1991), confirming previous
suggestions that cellular tissue was present (Denison, 1967); also, a new report
appeared on the histology of the phosphatic mineralized elements known for
over a century as conodonts (Sansom et aI., 1992) in which a new interpretation
is made of the inner core "white matter" in which the minute spaces are
interpreted as evidence of osteocytic bone with integral cell spaces and interconnecting cannaliculi. This cellular tissue and the basal tissue, which is compared with calcified cartilage of Eriptychius, one of the two named vertebrates
of the Harding Sandstone, are used as synapomorphic vertebrate characters
for conodonts (Sansom et ai., 1992; Briggs, 1992; Aldridge et ai., 1993), because
cellular bone is certainly present in dermal denticles of the third vertebrate
of the Harding Sandstone, and calcified cartilage is represented as a mineralized
tissue in close association with the acellular bone of Eriptychius.
Previous work on conodont histology (Gross, 1954) had looked for a
resemblance to acellular bone (aspidin) because this was the tissue character
ofheterostracans, the primitive agnathan group. This phylogeny predicts that
aspidin would be present in a more primitive outgroup (acellular bone assumed
to precede cellular bone). The new interpretations that odontode-related cellular bone is present in conodonts and primitive vertebrates have major implications for the evolution of vertebrate skeletal tissues, namely, that cellular
bone is more primitive than acellular bone, and that it may be more primitive
than orthodentine (see Fig. 5). This would suggest that the latter type of vertebrate tissue, with only the cell processes in spaces without the inclusion of
the cell body, is a later evolutionary development, and also that conodonts
are the primitive sister group of the higher craniates, excluding myxinoids
(Briggs, 1992; Aldridge et ai., 1993) (see cladogram in Fig. 4b).
In this new phylogeny conodont phosphatic tissues present many problems for discussion of the evolution of odontoskeletogenic tissues in the exoskeleton in vertebrates, and we propose that these should be considered in
this review. Among these are (I) the presence of oral mineralized tissues before
any in the dermal skeleton, (2) the apparent absence of dentine, and (3) the
presence of cellular bone in conodonts and osteostracans, but its absence in
heterostracans. As Briggs (1992) concluded from the new evidence on conodont histology, "The absence of dentine and the presence of cellular bone
in a highly adapted feeding apparatus in the earliest vertebrates leave current
hypotheses of the early evolution of their skeletal tissues in some disarray."
416
We therefore evaluate conodonts in our evolutionary model, and reassess

the conclusions drawn previously (M. M. Smith and Hall, 1990) from the
available fossil evidence on the sequence of evolution of vertebrate skeletal
tissues, with particular reference to the evolution of cranial neural crest as a
vertebrate innovation. Several major questions are raised by accepting conodonts as vertebrates: (1) Are conodont elements (teeth or tooth plates) independently evolved from those of gnathostomes, or are they the earliest
examples of teeth and homologous with those in gnathostomes? Also, did
these conodont teeth arise before a dermal skeleton? A consequence of an
affirmative answer to the latter question would be that loss of teeth in all
other agnathans was secondary, and that teeth had been reacquired in gnathostomes. (2) Does the presence of enamel, tooth-related bone, and tooth-related
cartilage suggest a cranial neural crest odontoskeletogenic origin of the visceral
skeleton, before neural crest made dermal bone as a body armor? (3) The
absence of dermal skeletal denticles in conodonts must question the proposed
primitiveness of such denticles in the vertebrate exoskeleton. The reverse
view-"odontodes ofthe dermal skeleton being already present in the dermal
skeleton before any skeletal structures in the oral cavity" (M. M. Smith and
Hall, 1990)-has been long held.
No previously postulated evolutionary scenario envisages mineralized
pharyngeal teeth before a body armor. Gans and Northcutt (1983), however,
argued that a cartilage-supported muscular pharynx was an early prerequisite
of vertebrates before gas exchange through the skin could be sacrificed for a
protective, sensory dental skeletal armor. Acceptance of this argument
prompted Hall and Horstadius (1988) to comment that "early association of
neural-crest-derived skeletal tissues (visceral cartilages) with the pharynx establishes the dichotomy between a skeletogenic cranial and non-skeletogenic
trunk neural crest." A calcified cartilage support and tooth elements at the
rostral end of a pharynx are now known for conodonts (Sansom et al., 1992;
Aldridge et al., 1993), but no cartilaginous supporting elements remain in the
fossils of the whole animal.
Can we therefore use the evolutionary record of conodont tissues in an
evolutionary model to explain the order in which skeletal tissue components
arose? Did these tissues arise because of evolving developmental competence
of various populations of neural crest or mesodermal mesenchyme to make
the whole variety of skeletal tissues? Such questions are central to this review.
Conodonts as Vertebrates
The debate on conodont affinities (a separate phylum, nemerteans, mollusks, chaetognaths, cephalochordates, or vertebrates) has been readdressed
in a paper by Aldridge et al. (1993) in which six new specimens of the conodont
417
animal are described, others reevaluated, and details given of newly discovered
element assemblages. These new anatomical characters, plus the histology of
the conodont elements described as being vertebrate tissues (Sansom et aI.,
1992), together with details of how these worked as a feeding apparatus (Purnell
and von Bitter, 1992; Purnell, 1993), make acceptance of these phosphatic
mineralized mouth parts as vertebrate teeth much more compelling than previously thought.
Conventionally, conodonts had been studied as one type of phosphatic
microfossil, with any of the affinities listed above. Sansom et al. (1992), in
an initial report of euconodont histology, assigned them to vertebrates on the
basis of both their tissue composition (enamel homologues, cellular bone,
and calcified cartilage) and their tissue growth, as if in apposition from a
dermal-epidermal interface in a toothlike structure (conodont elements). As
discussed by Sansom et al. (1992), other workers had previously claimed the
presence of most examples of extant and fossil vertebrate hard tissues, as well
as similarities with the un mineralized keratinous teeth of myxinoids. There
has been little agreement, however, on definitive evidence of a vertebrate
affinity for conodonts.
One remarkable feature is that conodonts are found over a very wide
stratigraphic range, from long before the previously accepted earliest vertebrates are recorded (at least 30 million years prior to this), until all major
vertebrate groups, except the mammals, had evolved (350 million years ago),
a range from the Cambrian to the Triassic.
Although isolated conodont elements have been known as fossils for 130
years, it was not until relatively recently that soft-body impressions of the
whole animal were discovered with these elements in situ (Briggs et al., 1983).
Sansom (1992) has documented the earliest notes on conodonts and their
monographic description in 1856 by Pander. From their location in the cephalic region as a spatially preserved assemblage, conodont elements were
interpreted as a feeding apparatus within an oral, or at least a pharyngeal,
cavity. There is no evidence of any paired branchial cartilages as functional
skeletal supports for a moveable pharyngeal apparatus, but their presence is
not precluded; the soft tissues are extremely poorly preserved (M. P. Smith,
personal communication).
Subsequent to the whole-animal descriptions, a more reliable basis for
restoring the in vivo arrangement of "conodont elements" was discovered.
This was a lower Silurian conodont animal, Panderodus, with a bedding plane
assemblage of elements preserved as arranged in the living animal (Mikulic
et al., 1985). M. P. Smith et al. (1987) reconstructed the three-dimensional
arrangement of this assemblage with the bases of the paired elements arranged
on an arch and the tips in an overlapping occlusion. From this reconstruction
they discussed the probability of a contiguous cartilaginous basal support for
418
each half set of the elements, noting that basal filling material of several Ordovician taxa resembled globular calcified cartilage of the earliest vertebrates.
They concluded that the 14 bilaterally opposed conodont elements were a
feeding apparatus. Sansom (1992) has reevaluated this specimen and reconstructed the apparatus as two arched bilateral blades of eight pairs, and an
additional, 17th, symmetrical element along the midline. Using an analysis
based on comparison between two functional hypotheses, Purnell (1993) has
concluded that conodonts were not suspension feeders, but used the elements
of the feeding apparatus in toothlike function. His reconstruction of this apparatus is reproduced in Fig. 3 to show their position relative to the whole
animal, and also how the posterior P-elements might slice against each other,
with overlapping occlusion.
Two sets of coniform elements set in rows parallel to the trunk axis
suggested a similarity of function with the feeding apparatus of myxinoids,
and led to the proposal that the jaw musculature may be homologous, but
associated with differently derived elements. In fact, the latest report on the
histology of conodont elements (Sansom et aI., 1992) shows many differences
between these and keratinous teeth of myxinoids. Indeed, as M. P. Smith
(1990) pointed out, the elements of all three agnathan groups (myxinoids,
petromyzontids, and heterostracans) are histologically different. M. M. Smith
and Hall (1990) also did not support homology of hagfish teeth with those of
other vertebrates. In the light of new histological data, however, we do have
to reconsider the status of conodont elements as potentially representing the
earliest form of craniate skeletal tissues.
Evidence from conodont whole bodies and intact assemblages reinforced
the hypothesis that conodonts are more closely related to jawless craniates
than to cephalochordates (Aldridge et al., 1986; M. P. Smith, 1990). The
cladogram then proposed, however, contained an unresolved dichotomy between myxinoids and heterostracans (Fig. 4). Blieck (1992) identified the central problem of conodont relationships as the diversity of opinion on the
histological types present, and concluded that features used to determine
chordate affinity were not convincing. He presented a cladogram with conodonts in an unresolved position between cephalochordates and myxinoids,
as reproduced in Fig. 4c. The relationships of conodonts within the vertebrates
will remain a source of debate until the histological characters are further
refined. This is a view endorsed by Forey and Janvier (1993) in their review
of agnathan relationships and early vertebrate history, who "await more research on a wide variety of conodonts before histological structure may be
used as evidence of vertebrate affinities." On the basis of current information,
however, they can be placed closer to the heterostracans than the myxinoids,
as in the cladogram (Fig. 4b) suggested by Briggs (1992). Sansom (1992) and
Aldridge et al. (1993) have reassessed the phylogenetic position of conodonts
419
,,'
.;.'/
.' .'
.. :,. ....
.,'
.,..'/
"'"
'~
-'
I
I
b
FIG. 3. The conodont elements arranged in !$itu as a projected reconstruction from their
anteroventral position in the conodont animal, based on a partially disarticulated bedding
plane assemblage of a single apparatus of Vogelgnathus campbelli from the lower Carboniferous, as illustrated by Purnell and von Bitter (1992, Figs. 1 and 3). They concluded
that "the anterior elements grasped the food, which was then cut and ground by the Pb
and Pa elements." The whole apparatus functioned as an integrated feeding structure,
grasping, cutting and grinding to process the food. Part (b) shows the cutting function in
operation with the Pa elements interacting as in "serrated scissors." [By permission of
the authors and Nature. Copyright 1992, Macmillan Magazines Ltd.]
420
...-----A.f:nrifJOiclea
...----CO""~~
-<IOfJta
FIG. 4. Different cladograms taken from the publications of (a) Gans (1989), (b) Briggs
(1992), and (c) Blieck (1992), illustrating the different views on the relationships of conodonts to chordates and to the agnathan vertebrates, the Heterostraci.
within the vertebrates and produced a cladogram for pregnathostome vertebrates based on the consensus of that of Janvier (1981). They have used the
presence of cellular dermal bone, enamel homologues, and calcified cartilage
to place conodonts crownward of myxinoids and a sister group of higher
vertebrates.
Chordate/Vertebrate Origins and the Primitive Skeletal Tissues

Conodonts may represent an initial approach to skeletization of the feeding apparatus and an early experiment of neural crest in making a set of
mineralized tissues. This interactive developmental program, initiated by ectodermal cells, was probably of a type that was later modified and extended
in evolution to make odontodes and teeth with recognizable dentine that later
covered the whole body. In a recent paper M. M. Smith (1992) has argued
for the primitiveness of enamel in vertebrate odontodes/teeth and that heterochrony can explain the early or late appearance of this ectodermal secretion
product in teeth of different types, an explanation that would allow conodonts
421
to develop enamel perhaps before dentine. Observation that conodont elements and some associated elements belonging to a species of Pseudooneotodus
possessed a distinctly recognized enamel layer and possibly also dentine (Sansom, 1992, and personal communication) emphasizes the primacy of ectoderm
in the odontogenic morphogenic system and its ability to induce or produce
phosphatic mineralized tissues.
In our view, enamel and dentine of vertebrate teeth are always associated
with cellular bone, or alternative attachment tissue, and cellular dentine may
be more primitive than dentine without cell bodies, but with cell processes,
as evidenced by the early vertebrate tissues (M. M. Smith, 1991). There is a
problem with the allocation of the euconodont tissues to bone or an early
type of dentine based solely on the size and arrangement of the cell and cell
process spaces, and this will need to be resolved by examination of many
examples from further species and by using other parameters. Primitive dentines are generally described as belonging to one of four patterns as shown in
Fig. 5, based on those suggested by 0rvig (1967) and figured by Baume (1980,
Fig. 4.1), and Reif (1982, Fig. 2). The implication has been that this represents
an evolutionary series with a progression from primitive mesodentine to the
a.
Mesodentine
b. Semidentine
c.
Metadentine
d. Orthodentine
FIG. 5. Four types of dentine all occurring in early agnathan vertebrates, proposed by
0rvig (1967) as sequential stages in an evolutionary series in which progressive changes
in development of the primitive (a) mesodentine and (b) semidentine released the odontoblast body from the mineralized matrix, leaving only a branching network of cell process
spaces leading from a central polarized cell process as in (c) metadentine and (d) the more
regular orthodentine. The number of cell bodies enclosed in mesodentine relative to cell
process spaces varies considerably among the fossil agnathan osteostracans.
422
most advanced orthodentine as found in human teeth, but also present in

some heterostracan dermal armor of the lower Silurian forms [examples of
these different types of dentine have been illustrated in M. M. Smith and Hall
(1990, Figs. 12,13, and 15). As stated previously, "The question of the relative
primitiveness of dentine and bone is one that has aroused the most debate"
(M. M. Smith and Hall, 1990), and the assumption that bone arose earlier
than dentine was challenged with the proposition that "the earliest skeletal
tissues were dental." The tissue beneath the enamel-like layer in conodont
elements could represent the primitive papillary product, a fine-gauge syncitium of cell process and cell body spaces, without any polarity to them,
more bonelike than dentinelike, but with the potential to be transformed into
the types recognized as mesodentine, semidentine, and metadentine in the
later evolving vertebrates. In this sense the conodont tissues present in elements
that function as teeth could be the most primitive type of dentine and be
present before any tissue developed in a bonelike way, as in fetal or dermal
membrane bone.
Suggestions that cellular bone is more primitive than acellular were discussed by M. M. Smith (1991) with a new evaluation of the earliest vertebrates
with well-preserved histology. The presence of cellular bone in conodonts
polarizes this character for vertebrate evolution. In both it is bone of attachment that forms in association with the tooth and fixes the tooth/odontode
to a support system, either soft connective tissue, calcified cartilage, or bone.
The decision that conodont elements function as teeth or dental plates,
is now more universally accepted [see Jeppsson (1979) and Conway Morris
(1980, 1989) for discussion of all previous views]. We would suggest that the
skeletogenic ability of neural crest evolved first in the head, and that dermal
denticles or a body armor of sheets of dentine and bone did not precede oral
teeth or dental plates in the fossil record. It is interesting that experimental
embryology in amphibians indicates that development of teeth is frequently
found to be spatially and temporally associated with cartilage of the visceral
skeleton (A. Graveson, M. M. Smith, and B. K. Hall, in preparation). In
cichlid fishes, tooth formation and attachment is intimately associated with
(and possibly inductively coupled to?) resorption of cartilage (Huysseune and
Sire, 1992a,b).
If conodont elements were the first craniate experimentation with mineralized tissues in the mouth/pharynx, the following reconstruction can be
proposed as being in accord with some ofthe elements of the scenario discussed
by Gans (1987). Predatory, muscularized conodonts with paired sensory organs
(derived from embryonic placodal systems) for sight and balance (M. P. Smith,
personal communication; Aldridge et al., 1993; Aldridge and Theron, 1993),
serving as distance receptors, were armed with two laterally opposed sets of
piercing teeth for food trapping, supported by calcified cartilage ofthe visceral
423
skeleton (derived from cranial neural crest). The success ofthis reconstruction
depends on the evolutionary novelty of the cranial neural crest providing
migratory cells, as suggested by Gans and Northcutt (1983, 1985), with the
potential to induce and differentiate into odontoskeletogenic cells that produce
mineralized tooth tissues and cartilage alongside the pharyngeal muscles.
At this stage the trunk neural crest would not have evolved its skeletogenic
role. Prepatterning of ectoderm alongside the cranial region of the neural
crest, in addition to the placodal tissue, remaining superficial to ectomesenchyme of the branchial region, would provide the initial induction signal.
This is suggested by experimental work in the chick that shows regions of
superficial ectoderm that correspond with the spatial distribution of neural
crest cells (Couly and Le Douarin, 1990). None of this will preclude the possibility that an electroreceptive sensory system would have existed in the soft
tissues of the skin as the first true distance-receptive system, as proposed by
Gans (1987). As Gans implied, enhancement ofthis system would have been
secondary by positional stabilization of the signal receptors/sources, with either
dentine or bone deposition associated with this system. This would be acquired
when skeletogenic ability of extensively migratory neural crest cells evolved
in the dermis. These patterns generated in the phylogeny of vertebrates would
be conserved and operate as constraints in a developmental model.
Teeth are not present in the mouths of the other fossil agnathan groups;
this poses a far more fundamental question of homology. Again, if not homologous, then teeth were acquired independently several times in vertebrate
evolution and the involvement of neural crest in conodont elements cannot
be assumed. The search for teeth in other agnathan groups becomes even
more imperative (see p. 438). Also, the significance of enamelinlike proteins
in hagfish teeth must be reassessed. Although internal pharyngeal teeth are
not known, dermal structures at the margins of the mouth in heterostracans,
referred to as dental plates by Halstead (1973) and Janvier (1981), are reported
to be used in food gathering, but assumed not to be homologous with teeth.
Recent discovery of an articulated Silurian agnathan fish representing the
oldest and most primitive heterostracan adds further information on the arrangement of oral plates surrounding an anteroventral mouth (Wilson and
Soehn, 1990; Soehn and Wilson, 1990). Soehn and Wilson report that "We
see no evidence of anterior denticles, sharp ends, or slicing surfaces on the
oral plates of Athenaegis, although there is a finely ornamented area on the
rostral brim." They further suggest that there was "unified opening and closing
movements of a scoop-shaped mouth" and that "they most likely fed upon
plankton or detritus." Their histology and mode of growth are unknown, so
that new data of this type would almost certainly lead to a new interpretation
of the relationship of these structures to other feeding apparatuses. This would
424
be crucial to theories on the relationship between teeth, tooth elements in

conodonts, and dermal armor in other established early vertebrates.
With respect to the great variety of dentine and bone in the early fossil
record, Gans (1987), noting the complex armor of heterostracan and osteostracan agnathans, suggested that "each represents a derived condition relative
to the animals showing the 'earliest' ossification. The fossils and their 1:UIDour
give us a record of an evolutionary experiment that is analogous to the agnathan experiment with paired appendages." It would seem that there is now
a possibility of the earliest skeletization being part of a toothed apparatus in
the head of a craniate animal.
In the next section we review and compile what some of these developmental patterns might be, as deduced from experimental embryology.
EXPERIMENT AL EMBRYOLOGY OF SKELETOGENESIS IN

LOVVER VERTEBRATES
The Role of the Neural Crest

Although a few studies rely on natural markers, most knowledge about
the neural crest is from experimentally marking or manipulating embryonic
tissues, admitedly from a very small number of species.
Most studies of skeletogenesis, apart from those on mammals, have been
in the chick or in amphibians, both anurans and urodeles. Most information
available on neural crest in fishes concerns nonskeletogenic tissues. Migration
pathways of trunk neural crest cells were traced using tritiated thymidine in
a teleost, Barbus conchonius (Lamers et al., 1981). Distribution and migration
of HNK-l immunoreactive neural crest in both the head and the trunk of
three species of teleost (Xiphorous helleri, X. maculatus, and Orizias latipes)
were reported by Sadaghiani and Vielkind (1990). In neither study were embryos studied long enough for any comments to be made on skeletogenesis.
The basis for reviewing details of the involvement of neural crest in
skeletogenesis in fishes is the accepted view that all vertebrates develop a
neural crest tissue with an underlying common pattern of morphogenesis and
migration, which, at least in the cranial region, is skeletogenic. As remarked
by Hall and Horstadius (1988), neural crest is "characterized by similarity in
site of embryological origin, by similar regionalization within the neural tube,
by similar migratory behaviour and by production of a similar range of cell
and tissue types." It is axiomatic that this pattern is conserved throughout
the vertebrates and is synapomorphic. For instance, to date, no publications
425
demonstrate an origin of odontogenic tissue other than from neural crestderived mesenchyme. Comments by Kemp (1990) from extirpation studies
of the neural crest in the Australian lungfish Neoceratodus that "cells of the
neural crest are of limited importance in this animal" and also "that lungfishes
are so far from the main stream of vertebrate evolution that their neural crest
cells are atypical (in not producing dental or skeletal tissues)" (Kemp, 1993)
are hard to reconcile with all other published accounts, in particular with the
underlying concept of conserved developmental strategies, which allows the
statement that "the neural crest is a quintessential vertebrate characteristic
. . . producing a myriad of craniate tissues" (Hall and Horstadius, 1988, p.
20), a major group of which is skeletogenic, i.e., cartilage, bone, and dentine.
The question of similarities between patterns in development of the skeleton from cranial neural crest has been reviewed by Hall, in Hall and Horstadius (1988), in which he acknowledges that little is known about neural
crest in fish. He cited the work of Damas (1944, 1951), an early example of
experimental work on neural crest skeletal structures in the lamprey, in which
defective chondrogenesis in the head cartilage resulted from exposure of developing larvae to light. Interestingly, the conclusions were that these defects
resulted from a failure of the interactions between epithelium, neural crest,
and pharyngeal endoderm, developmental mechanisms assumed to operate
in lampreys as they do in amphibians. The possibility that defective chondrogenesis was due to lack of neural crest cell migration was not discussed.
Extirpation and transplantation experiments ofNewth (1950, 1951, 1956)
on Lampetra planeri and L. jluviatalis were very thoroughly discussed by
Hall (1987) and Hall and Horstadius (1988) with an assessment of their significance. Despite initial failure by Newth to show that cartilage was produced
by cranial neural crest, he continued with further experiments (Newth, 1956)
which indicated that although lamprey neural crest was capable of forming
cartilage in flank grafts, it no longer received the required inductive influence,
whereas in branchial grafts induction was provided and cartilage developed
from donor neural crest. Damas (1951) had also speculated that contact with
branchial arch epithelium was necessary for chondrogenesis to occur in the
lamprey, Lampetra jluviatis. As Hall and Horstadius (1988) summarized,
"both the cellular origin of these cartilages and the requirement for activation
via inductive interaction with embryonic epithelia corresponded well with
previous results in amphibia." The questions raised here are, has the epithelium
lost or did it never acquire its inductive ability in the trunk during the evolution
of "higher" vertebrates? Also, is migration of neural crest necessary to determine when the epithelial inductive signal will be produced?
A recent study of the sea lamprey, Petromyzon marin us, by Langille
(1987) and Langille and Hall (1988a) used very precise extirpations of regions
of 250-mm craniocaudad lengths to analyze deletions of the cartilaginous
426
skeleton. These extirpation studies showed good correlation between the region
extirpated and the cartilages that failed to develop and relatively few abnormalities of structure, certainly not massive ones. From these studies, the
chondrogenic portion ofthe lamprey neural crest was mapped. It shows good
correspondence with maps developed for other vertebrate embryos, absence
of chondrogenic potential in the anterior prosencephalon and caudad to the
sixth pair of somites. Absence of cartilages following extirpation showed that
regulation was not a significant mechanism, and reinforced data that specific
regions of cranial neural crest alone contribute to the branchial arches in
fishes (see below).
Regional information for neural crest cells is also important. The more
posterior branchial arches are derived from the more posterior region of cranial
neural crest. Some cartilages (i.e., parachordals, otic capsules) developed normally following deletion of the neural crest, the conclusion being that these
are not derived entirely from neural crest mesenchyme. Couley et al. (1993)
have now provided very precise evidence in the chick for the cellular origin
of the sphenoid (a cartilage bone from the parachordals) from cephalic paraxial
mesoderm, and the basipresphenoid from cranial neural crest (see p. 390).
Hall and Horstadius (1988) concluded that this mapping of the chondrogenic
part of the cranial neural crest in lampreys represents a pattern developed
early in vertebrate evolution and conserved among divergent groups.
Langille and Hall (1988b) were also the first to report an experimental
study mapping the cranial neural crest of teleost fishes (the Japanese medaka,
Orizias latipes) preceded by a baseline study of normal development of the
head skeleton (Langille and Hall, 1987). In the experimental study sections
of neural crest were removed from stage 18-19 neurula embryos. From an
analysis of changes to the head skeleton it was possible to conclude that most
of the anterior neurocranium and the entire viscerocranium received neural
crest contributions during development. But of even greater interest was the
mapping of neural crest cells contributing to individual elements from particular rostrocaudal regions along the neural axis. This pattern was found to
be very similar to the regional contributions from the neural crest to the head
skeleton in lampreys, urodeles, and birds (Langille and Hall, 1988a; Fig. 7).
These patterns are evolutionarily conserved.
An elegant set of experiments in teleost fish (Lamers et al., 1981) uses
transplantation of previously labeled grafts to reveal migration pathways of
trunk neural crest cells. They used donor neural crest from an embryo prelabeled with [H3]-thymidine and injected labeled neural crest into rhombencephalic and trunk regions of 1- to 16-somite host embryos. A migratory
pattern similar to that observed in birds was seen. There was, however, one
small difference; the ventral stream did not penetrate mesoderm, but migrated
within spaces between the neural tube/notochord and somites. This difference
427
was explained by Lamers and colleagues as possibly due to a relatively much

earlier period in fish development of "explosive growth and differentiation of
the somites, to become functional within a few hours after their formation."
Sadaghiani and Vielkind (1990) used HNK-l antibody visualization to
mark and follow premigratory and migratory neural crest cells in developmental series of three species of teleost. Their results were similar to those
reported in birds and mammals, which they concluded "reflected the close
similarity of embryogenesis amongst the vertebrates." In an earlier study of
normal development of the neural crest in Xiphophorus teleost fishes (Sadaghiani and Vielkind, 1989) and in their HNK-l labeling study, Sadaghiani
and Vielkind commented that one marked difference between these fishes
and other vertebrates was segregation of neural crest cells from the dorsolateral
aspect of a neural keel, rather than from the dorsal midline as in other vertebrates. They described ridges ofHNK-l-positive cells appearing topographically lateral to, and temporally after, neural keel formation, and identified
these as populations of neural crest cells. One route for migration of neural
crest cells in the trunk is below the epithelium. These superficially-migrating
cells are presumed to give rise to pigment cells. This issue will be addressed
in the following section on the potential of trunk neural crest, as will the issue
of the dorsal location of many neural crest-derived cells.
An early report (Lopashov, 1944) of extirpation and grafting in several
teleost fishes (Misgurnis jossilis, Nemacheilus barbatulus, Perca f/uviatilis)
showed that cartilage developed in neural crest cells grafted into yolk sacs,
supporting a neural crest origin of cranial cartilage in these fishes. It seems
one must assume either that yolk sac epithelium is inductive on neural crest
cells for chondrogenesis, or that these neural crest cells were specified prior
to transplantation.
The labeling study by Lamers et al. (1981) was not directed at skeletal
tissue development. Nevertheless, they showed an accumulation of trunk
neural crest-derived cells under sensory organs of the neuromasts. Such a
localization can be linked to, and provide indirect evidence with, observations
that the first preosteogenic cells aggregate in association with neuromasts.
Whether the early aggregating cells observed by Lamers and colleagues would
have become neural or connective (skeletal) tissue is unclear.
All the experimental studies just discussed reveal different aspects of the
neural crest as a potent source of cells in embryogenesis, giving rise to a great
diversity of cells and tissues with a remarkable precision of distribution. In a
summary of work on the neural crest and its derivatives since 1950, Weston
(1970) emphasized the understanding of developmental problems provided
by each different experimental method, and evaluated those conclusions that
describe processes and those that allow causal analyses. Experiments involving
428
the neural crest were grouped into three types as presented above, ablation,
explantation, and markers.
The full range of the prospective potential of the neural crest, independent
of any environmental developmental cues, is evidenced in explantation studies.
The normal fate of crest cells is best determined by labeling and orthotopic
grafting, provided that, as Weston pointed out, the region excised from the
donor for grafting is very precisely determined and host age controlled to
avoid precocious migration of crest cells. Permanent and cell-specific markers
allow patterns of early migration and localization to be followed, provided
that the chosen markers do not transfer to other cell populations and that
they are not toxic to the cells in normal development.
Cell-marking experiments are probably where the most advances are
being made today, with the development of very precise methods for injecting
specific, spatially and temporally located sites in the embryo, even into single
cells. Such experiments allow real cell lineage studies to be performed, and
as discussed by Lumsden (1989), begin to allow the distinguishing of subsets
of cells and "provide for the first time direct evidence that some neural crest
cells are multipotential at the time of their initial migration." As Lumsden
(1989) summarized, the elegant studies of Bronner-Fraser and Fraser (1988)
have allowed "direct observation of cellular diversification from multipotent
progenitors in the emergent crest and paves the way for detailed examination
of the process offate restriction by marking cells . . . at different points along
their paths of migration." This type of investigation will be of special value
in the trunk neural crest, and is in progress (M. M. Smith, A. G. S. Lumsden,
and P. V. Thorogood) in relation to skeletogenesis in fishes, where somatic
lateral plate mesenchymal cells make a major contribution to the skeleton,
as will be reported in the subsection following the next one.
Teeth from Cranial Neural Crest

The role of cranial neural crest in inducing teeth and contributing cells
to make dentine has been firmly established since the work of Adams (1924)
on urodele amphibians. All subsequent work has been expertly reviewed by
Gaunt and Miles (1967), Lumsden (1987), and Hall and Horstadius (1988).
It is known with some precision from the work of Chibon (1966) on
amphibians that odontogenic potential is very precisely localized in cranial
neural folds. These locations are craniorostral regions expressed as a series of
30 segments, starting from the medial axis and progressing through 180.
These segments covering 30-100 correlate with tooth formation. Cells from
the 30-70 regions produce teeth of the upper jaws, cells of the 70-100
regions teeth of the lower jaws.
429
Direct evidence that the role of neural crest in tooth development is not
just evocative, but includes the contribution of cells differentiating as tooth
germs, was provided by the experiments of Chibon (1966, 1967) in which
tritiated thymidine-labeled donor cranial neural crest cells appeared in odontoblasts and pulp cells of larvae with orthotopic grafts.
The significance of tissue interactions in odontogenesis was emphasized
by Lumsden (1987), who cited the dual origin of the enamel organ from
ectoderm or endoderm as the reason for de Beer (1947) suggesting that ectomesenchyme rather than epithelium provided the initial role in tooth induction. Later work, however, in particular the studies of Wagner (1949,
1955), Henzen (1957), and Lumsden (1987), suggested that stomodeal ectoderm and endoderm provided the first inductive signal for the initiation of
tooth development, and that although the cranial neural crest has odontogenic
competence in anurans, the signal is lacking from or suppressed in the ectoderm until metamorphosis. An important aspect of metamorphosis relates
to the mechanisms controlling when signals are switched on and how such
signals are modified in amphibians that have lost the tadpole stage, i.e., those
demonstrating direct development (Hanken, 1986).
Some of the most interesting experiments are those of Wagner (1949,
1955) and Henzen (1957), who used reciprocal orthotopic grafts to produce
chimaeric tooth germs in anuran and urodele larvae. Wagner (1955) found
that urodele ectoderm would allow urodele-type teeth to form from anuran
neural crest, the dental papilla forming from frog neural crest-derived cells.
He concluded that stomatodeal ectoderm lacks the inductive signal in larval
anurans, acquiring it after metamorphosis. In the reciprocal experiment, where
urodele cranial neural crest was grafted orthotopically into anuran hosts, no
teeth formed, but a urodele-type visceral skeleton developed. As well as supporting the conclusion that anuran cranial neural crest is competent to form
teeth but does not receive inductive signals because ectoderm is deficient, this
experiment also suggests that the permissive signal is generated by ectoderm,
and that cartilage formation does not depend on the same inductive signal
as does tooth formation.
Remarkably, it was not until much later that direct evidence for the
neural crest contribution to teeth in mammals was provided by Lumsden's
(1987) explant experiments. This study also demonstrated that oral epithelium
provided the initial signal for tooth development. Because the boundary between ectoderm and endoderm in the oral cavity of mammals has not been
precisely established, whether tooth initiation depends on endoderm or ectoderm cannot be established.
There are no data for the involvement of neural crest in odontogenesis
in fishes. Langille and Hall (1988b), in their extirpation study of cranial neural
crest in the Japanese medaka, demonstrated neural crest involvement in the
430
cartilaginous skeleton of the branchial region, but their study did not allow
comments to be made on the origins of dermal bone or teeth. Sadaghiani and
Vielkind (1990) commented on a positive reaction with HNK-l antibodies
in the pharyngeal teeth of xiphophorids, but quite inexplicably they found a
strong reaction in the epithelium of the dental organ and apparently not in
the dental papilla.
The experiments on amphibian tissues using extirpation and transplantation, or recombinations as explants, provide information on the range of
potential of cranial neural crest cells. As Lumsden (1987) commented, the
use of explants addresses the following questions:
1.
2.
3.
4.
Is odontogenic ectomesenchyme committed at mid-neurula stage?

Or is it committed at a later time during migration?
Or after contact with pharyngeal endoderm?
Or after contact with stomodeal ectoderm?
Lumsden (1987) concluded that Avery's (1954) experiment in which

cranial neural crest of Ambystoma mexicanum formed teeth in ectopic sites
indicated that "a degree of committment had been acquired by the crest
before migration," but he suggested that "this committment is not complete";
there is a requirement for interaction with epithelium before odontogenesis
is effected. Such a requirement would be a permissive signal from epithelium
to competent ectomesenchyme.
Sellman (1946) first explored the potential of cranial neural crest as heterotopic grafts to the trunk region of Ambystoma larvae. He found that teeth
are not formed unless three tissues are present-odontogenic cranial neural
crest, stomatodeal ectoderm, and pharyngeal endoderm. But the transplantation method leaves open the possibility of contact with mesodermal mesenchyme, either at the future site or with the explant. Wilde (1955) confirmed
this tripartite dependence in an explant system using Ambystoma mexicanum,
where he could be sure that mesodermally-derived mesenchyme was excluded.
As with Sellman's study, teeth were only obtained if all three tissues were
present in the explant.
Neither Sellman's nor Wilde's experiment answers the third and fourth
questions posed by Lumsden (1987), viz. whether contact with pharyngeal
endoderm or stomodeal ectoderm is required to commit ectomesenchyme
for odontogenesis. This is because cranial neural crest cells cannot be traced
sequentially in their interactions with pharangeal endoderm and stomatodeal
ectoderm. Cassin and Capuron (1979) appear to have provided a solution to
this problem, although their results differ from Sellman's and Wilde's. As well
as obtaining teeth with all three tissues in both explant and blasocoel transplants, they obtained teeth with cranial neural crest and endoderm, but not
with cranial neural crest and ectoderm. This would seem to implicate phar-
431
angeal endoderm rather than stomodeal ectoderm as the epithelium providing

the initial signal for commitment of cranial neural crest to odontogenesis.
Cassin and Capuron (1979) concluded that endoderm promotes the differentiation of cranial neural crest into both odontoblasts and chondroblasts,
and that both can only differentiate from cranial neural crest after contact
with pharangeal endoderm. Despite this dependence of cartilage and teeth on
pharyngeal endoderm, cartilage and teeth develop independently of one another; development of bone required cartilage to be present.
Results from work in progress (A. Graveson, M. M. Smith, and B. K.
Hall, in preparation) testing all combinations of tissues in explants of Ambystoma mexicanum without serum (which was used in all previous studies)
confirm that teeth and cartilage will form when cranial neural crest (CNC),
endoderm, and ectoderm are present in the explants; when endoderm and
CNC are cocultured, but not when ectoderm alone is cocultured with CNC.
These workers also confirmed that teeth and cartilage form independently of
one another, although there is a close topographic relationship between some
cartilages and teeth. Remarkably, spatial positioning seen in vivo was maintained in explants. A symmetry developed whereby rows of teeth opposed
each other across an epithelially-lined space, with the oldest teeth on the
outside of the explants (labiobuccal) and tooth buds on the inside (lingual).
Bone of attachment formed after teeth had formed, but never before. Bone
of attachment was in continuity with the dentine cone, linking teeth to one
another and to the perichondrium of the cartilage. Membrane bone of the
dentary or palate had not formed at this stage.
Trunk Neural Crest or Mesodermal Origins of Trunk

Odontogenic and Skeletogenic Tissues
We have argued in previous sections that major questions in the development of the postcranial skeleton in fishes are still to be answered:
1. What is the embryonic origin of odontogenic and skeletogenic tissues
in trunk and tail exoskeleton of all major groups of fossil fishes, all
chondrichthyan fishes, and several species of extant osteichthyan fishes,
including the coelacanth and armored catfishes?
2. Can it be demonstrated from experimental studies on lower vertebrates, including extant fishes, that the cell lineage of scleroblasts
producing denticles (tubercles and dermal bone) is from trunk
neural crest?
The data produced so far to answer these questions are negligible. There
are some fascinating, general, nonexperimental, observations on dermal skel-
432
etons in fishes requiring explanation. Hall (1991) stated, "there is lack of

critical data for key developmental events" and "the embryonic origin of
mesenchymal elements offish fins would advance our knowledge of the evolution of connective tissue in higher vertebrates."
Until very recently there was no obvious reason to seek any skeletogenic
capability of trunk neural crest cells. Classic studies on the neural crest origin
ofskeletogenic and odontogenic tissues, performed largely, if not exclusively,
on amphibian and avian embryos, established cranial but not trunk neural
crest as skeletogenic in both avian and amphibian embryos [see above, and
Le Douarin (1982), Hall and Horstadius (1988), and M. M. Smith and Hall
(1990) for reviews]. Trunk neural crest has therefore long been regarded as
neither skeletogenic nor odontogenic. Reasons for such dichotomous thinking
concerning capabilities of cranial and trunk neural crest are many.
Most fate mapping studies demonstrate cranial neural crest to be skeletogenic (and odontogenic in toothed vertebrates), not that the trunk neural
crest is not. As reviewed by Le Douarin (1982) and Hall and Horstadius
(1988), many experimental studies in amphibians have shown that trunk
neural crest transplanted into the head does not make cranial skeletal bone,
cartilage, or teeth. Similarly, cranial neural crest transplanted into the trunk
does not express its skeletogenic potential.
Evidence, therefore, for inability of trunk neural crest to form skeletal
or dental tissues is negative and by exclusion, rather than positive and from
direct demonstration. This remains so, even though several early studies specifically sought to address whether trunk neural crest was skeletogenic or
odontogenic. They did so by grafting trunk neural crest into the cranial neural
tube and failing to observe chondrogenesis, osteogenesis, or odontogenesis
(Horstadius and Sellman, 1946; Sellman, 1946; Chibon, 1966). As has been
demonstrated by Graveson (1993), however, failure to observe development
of skeletal or dental tissues in such transplantation experiments need not
necessarily mean trunk neural crest lacks these potentials. Heterotopic transplantation experiments are performed assuming grafted trunk neural crest
cells undergo normal patterns of migration and interactions with cranial inducers. Graveson (1993) has shown this need not, and indeed often is not,
the case; see p. 431.
None of the postcranial skeleton has been shown to be of neural crest
origin in any vertebrates. An "exception" to this statement is cartilage nodules
in the hearts of domestic fowl, shown to arise in mesenchyme derived from
the cardiac neural crest. Cardiac neural crest cells migrate into the heartforming region to form some heart mesenchyme, conotruncal region, and
cartilage. It is arguable whether this cartilage is part of the skeleton or ectopic.
In either case, it is a skeletogenic tissue in the trunk of neural crest, albeit
cranial neural crest, origin.
433
The postcranial endoskeleton arises from somitic or lateral plate mesoderm. Within the postcranial endoskeleton, axial and rib skeletons arise from
somitic mesoderm, appendicular and girdle skeletons from lateral plate mesoderm (Hall, 1986, 1991; Hall and Horstadius, 1988). As documented by
M. M. Smith and Hall (1990) and Hall (1991), there is little if any likelihood
of any trunk endoskeleton being of neural crest origin. Whether trunk exoskeleton in vertebrates such as fishes is of neural crest origin is the issue.
In our earlier review (M. M. Smith and Hall, 1990), we did not discuss
the developmental origin of the exoskeleton of either medial or paired fins in
fishes because little is known, and what is stated is partly contradictory. Schaeffer (1977), although accepting that there is no experimental evidence for involvement of neural crest in the development of body scales and fin lepidotrichia, proposed, on the basis of similarity of tissue composition, that
"ectomesenchyme is involved in the formation of all dermal calcifications."
"Similarity of tissue compositions" may not be a good criterion, unless specific
combinations of tissues are used, e.g., dentine and bone.
Although trunk endoskeleton is not of neural crest origin, connective
tissue of the median fin folds in the amphibians Ambystoma mexicanum and
Pleurodeles waltii is derived from trunk neural crest (Raven, 1931, 1936; Du
Shane, 1935; Holtfreter, 1935; Detwiler, 1937; Chibon, 1966). Furthermore,
median, unpaired fins are induced to form under an inductive influence arising
from trunk neural crest (Du Shane, 1935). This lone study needs to be reinforced by further analyses.
Paired fins, on the other hand, arise following inductive influences either
from trunk neural crest or lateral plate mesoderm, but do not contain neural
crest-derived connective tissues (Balinsky, 1975) [see reviews in Schaeffer
(1977), Thomson (1987), Hall (1991), and Table III]. The few studies in these
areas also need to be augmented by additional experimental studies on additional species.
Both in his 1977 review on the dermal skeleton in fishes and in his 1987
review comparing early craniate development with other deuterostomes,
Schaeffer (1977, 1987) suggested that different cell populations provide the
initial inductive signal and form scleroblastic mesenchyme in paired and unpaired median fins. Somatopleural mesenchyme provides the cells in paired
TABLE III. The Relationship between Trunk Neural Crest and Median
(Unpaired) and Lateral (Paired) Fins in Amphibians and Fishes
Fin type
Neural crest induced
Contains trunk neural crest connective tissue
Median unpaired
Lateral paired
++++
++++
434
fins, while neural crest-derived mesenchyme provides cells in unpaired, median

fins. Schaeffer (1987) stated that "there is substantial evidence that the epithelial unpaired fin-fold is invaded by mesenchyme from the trunk neural
crest in fishes," citing Terentiev (1941). The latter author had shown, in urodele
amphibians, that trunk neural crest could induce dorsal fin folds to form from
both trunk and head ectoderm, and concluded that both positional and morphogenetic information came from trunk neural crest.
Differences in inductive cell populations between median and paired fins
are unexpected. All fins are thought to have been derived evolutionarily from
continuous fin folds, and to develop using the same sequence of morphogenetic-histogenetic events (Hall, 1991; Thorogood, 1991). In sequence these
would be, first, development of an apical epidermal ridge; second, formation
of actinotrichia from ectoderm with which they retain contact; third, formation
of lepidotrichia from mesenchyme as the first elements of the dermal skeleton,
aligned by the actinotrichia (Wood, 1982; Schaeffer, 1987; Thorogood, 1991).
Schaeffer (1987) further believed that "the initial inductive signal for the
paired appendages comes from somatopleural mesoderm" and that "the somatopleural mesenchyme interacts with the presumed ectomesenchymal actinotrichial substratum," both activities being necessary for the development
of pectoral fins. Evidence for this comes from Wood (1982), who suggested,
on the basis of an ultrastructural study, that the cells that migrate along actinotrichia (assumed to form from the ectoderm) in developing teleost pectoral
fins arise from adjacent somatopleural mesenchyme. The only experimental
work bearing on this is a paper by Lopashov (1950), who showed that lateral
plate mesoderm when transplanted to the abdominal wall of teleost fishes
induced ectoderm to form fin folds.
There is no evidence that ectomesenchyme makes a contribution to either
actinotrichia or lepidotrichia during fin development. Experimental work in
early stages ofa project by P. V. Thorogood, M. M. Smith, D. Amanzee, and
A. G. S. Lumsden to determine the lineage of scleroblasts in caudal and
pectoral fins ofthe zebrafish appears to show that some early-migrating cells,
aligned along actinotrichia in the caudal fin, originate from neural crest, but
their differentiative fate has yet to be demonstrated.
In a study of the fine structure of the pectoral fin dermoskeleton in two
quite primitive osteichthyan fishes, Polypterus and Calamoichthyes, Geraudie
(1988) described ganoine-covered lepidotrichia, which she concluded (despite
lack of dentine) implied a role for epithelium in producing an enamel-like
part to this cover. She further commented that absence of dentine in pectoral
fins but its presence in caudal fins "could be related to the disappearance of
neural crest cells with skeletogenic fate within the fish dermis during evolution." This comment was based on the presumption that only mesodermal
cells formed osteoblasts of lepidotrichia, no data being available for an ec-
435
tomesenchymal origin. Geraudie (1988) concluded that although dermoskeleton, scales, and lepidotrichia are the result of "dermoepithelial interaction," the component covered by ganoine was either mesodermal or
ectomesenchymal in origin.
Le Douarin (1982), in her review of the source of mesenchymal cells
from neural crest, commented that, "Morphogenesis of the dorsal fin (in
amphibians and fishes) is the result of tissue interactions between ectomesenchyme (trunk origin) and dorsal ectoderm." Citing the work of Twitty and
Bodenstein (1941) and Bodenstein (1952), she concluded that both mediodorsal and flank ectoderm in amphibians was competent to participate in fin
formation, while admitting that "in fishes, there has been no systematic study
of the migration and differentiation of neural crest cells."
Most early transplant work also showed that interaction with pharyngeal
endoderm was necessary for skeletal differentiation to occur. This has been
confirmed by recent explant studies in which actual cellular contact between
ectomesenchyme and pharyngeal endoderm was a prerequisite for transmission of inductive signals (Epperlein and Lehmann, 1975).
Since a study by Lamers et al. (1981) of the migration routes of trunk
neural crest cells in fishes, recent papers have documented both timing and
pathways of this migration in the chick (Serbedzija et al., 1989) and in teleost
fishes (Sadaghiani and Vielkind, 1990). In vital dye studies of these premigratory and migratory trunk neural crest cells in the chick, Bronner-Fraser
and Fraser (1988, 1991) and Fraser and Bronner-Fraser (1991) concluded
that "trunk neural crest cells are not restricted in developmental potential,"
but "have the capacity to develop into a wide variety of both neuronal and
non-neuronal phenotypes." Their work is encapsulated by their interpretation
that "the majority of premigratory and migrating neural crest cells appear to
be multipotent," but "there may be minority populations of predetermined
cells." There is very little information on when these cells become committed
or of the factors influencing phenotypic selection.
Scarcity of data on migration, commitment, and selection of phenotype
of trunk neural crest cells leaves the question of their skeletogenic potential
in fish entirely open, with the corollary that research should be directed toward
migration and differentiation of neural crest cells in fishes, where it could be
predicted that induction of, and commitment to, exoskeletal skeletogenesis
would have been evolutionarily preserved. Such work is in progress.
SUMMARY OF POSTULATED MECHANISMS OF

SKElETOGENESIS
We provide two summaries in this section, a summary of the data and
postulates presented and an overview of the developmental model for the
436
evolution of exoskeleton and teeth. The data discussed can be summarized

under 12 headings.
1. The exoskeleton includes in its entirety denticles, scales, armor (dorsal
and ventral plates, branchial plates, etc.), fin lepidotrichia, and teeth, in agnathan and gnathostome fishes.
2. The dermal exoskeleton is made up of alternative combinations of
three basic layers: enamel, dentine, bone of attachment; overlying cellular or
acellular spongy bone; overlying cellular or acellular compact bone.
3. The exoskeleton is a neural crest derivative wherever it forms along
the body.
4. Evidence for a skeletogenic and odontogenic cranial neural crest, well
established from experimental analyses, is reviewed. Trunk neural crest has
in the past been taken as both nonskeletogenic and non odontogenic. This
presumption is challenged in this review.
S. The neural crest versus mesodermal origins ofskeletogenic and odontogenic mesenchyme from which exoskeletal and endoskeletal tissues arise is
discussed.
6. Activation of skeletogenic and odontogenic mesenchyme by epithelialmesenchymal inductive tissue interactions is discussed, as is the source of the
patterning.
7. The odontode is defined as the "single morphogenetic modifiable system" envisaged by Schaeffer and as a fundamental differentiative and morphogenetic unit of the exoskeleton by Reif. Within the odontode develop
enamel, dentine, bone of attachment, and pulp.
8. Osteogenic condensations are identified as the second fundamental
differentiative and morphogenetic unit of the exoskeleton. From such condensations comes the basal bone to which exoskeletal scales, denticles, or
teeth are attached, and from which come the lepidotrichia.
9. Individual cell populations can be recognized within odontogenic
condensations and osteogenic populations. These are discussed, using the
mammalian mandible as the model of a neural crest-derived dermal bone.
Segregation of odontogenic from skeletogenic cell populations is discussed and related to the apparent dichotomy between cranial and trunk
neural crest.
10. Molecular control mechanisms responsible for initiation of differentiation and patterning of exoskeletal elements are identified and discussed.
The important role played by homeotic genes is emphasised; evidence that
homeotic genes pattern skeletal elements along the anterior-posterior (rostrocaudal) body axis is discussed. Extracellular matrix molecules such as tenascin, cell surface molecules such as syndecan, and growth factors are iden-
437
tified and discussed as three classes of molecules regulating differentiation of

cells within odontodes and condensations.
11. The evolutionary origins of exoskeletal tissues are evaluated especially
in the light of recent studies on odontogenic and skeletogenic products present
within conodonts and the placement of conodonts within the vertebrate
lineage.
12. Experimental embryological evidence on the role and patterning of
the neural crest in skeletogenesis and odontogenesis in fishes is evaluated and
discussed in relation to mesodermal production of mesenchyme, including
mesenchymal derivatives in fins.
The developmental model that emerges from this analysis is hierarchical
and combinatorial. Exoskeletal tissues arise in populations of neural crest
cells that are patterned through nonuniform expression of home otic and patterning genes. Ectoderm and mesoderm are also regionally patterned. Anteroposterior (rostrocaudal) patterning acquired in the neural tube is retained
as neural crest-derived cells migrate throughout the body. Further positional
information is imposed on neural crest cell populations as they interact with
epithelia, which are responsible for initiation of cytodifferentiation. Cell surface, extracellular matrical, and growth factor molecules provide a combinatorial temporally and spatially regulated code utilized by neural crest-derived
skeletogenic and odontogenic cells to segregate into subpopulations, differentiate, and initiate patterned morphogenesis. Potential for modulation of
this code, especially through heterochrony, is discussed.
Boundary positions are established by homeotic and patterning genes.
That these boundaries can be experimentally modified by manipulation of
domains of expression of Hox and Pax genes suggests that establishment of
boundaries by combinations of such genes is the mechanism that established
exoskeletal and tooth patterning during vertebrate evolution.
The most parsimonious interpretation of the fossil skeletal and experimental embryological evidence is that initially in some basal vertebrate group,
both cranial and trunk neural crest cells were skeletogenic and odontogenic.
With the loss of exoskeletal denticles/scales, this capability has been retained
as an odontogenic and skeletogenic cranial neural crest, but lost from the
caudal trunk neural crest. Toothless vertebrates have lost the cranial odontogenic capability, but retained the skeletal potential. The separation of odontogenic and skeletogenic potential is further emphasized by the experimental
observations on recombined explants that in mice and amphibians the most
rostral trunk segment is odontogenic, but not chondrogenic. It is anticipated
that fishes with trunk exoskeletal elements, such as dermal denticles, will
possess an odontogenic trunk neural crest that is more extensively rostrally
438
and/or an odontogenic cranial neural crest that is more extensive caudally.

In animals such as armored catfishes and all cartilaginous fishes, the entire
rostrocaudallength of the trunk neural crest would be expected to be odontogenic as a retention of the primitive vertebrate condition. More fundamentally in cartilaginous fishes, the inability to make exoskeletal bone but the
widespread ability to make teeth (denticles) would further reinforce the separation of osteogenic (missing from cranial and trunk neural crest) and odontogenic potential of neural crest.
The evolutionary interpretation of this difference in distribution of potential skeletogenic and odontogenic developmental mechanisms can be either
that "odontodes only," is the primitive condition for vertebrates, accepting
conodonts without an exoskeleton but with tooth elements as a stem-group
vertebrate, or that it is derived, because heterostracans represent the basal
vertebrate group with an extensive bony armor.
NOTE ADDED IN PROOF. Van der Bruggen and Janvier (1993) have just
published some findings on teeth in agnathans. They report preliminary information on the finding of denticles in the pharyngeal cavity of a thelodont
agnathan, Loganellia, and say that they resemble pharyngeal dermal elements
of jawed vertebrates and are the first record of these in any jawless vertebrate.
These observations provide data to support our suggestion that pharyngeal
denticles evolved before marginal teeth on the jaws, and, in conodonts, existed
without any dermal denticles covering the body. The dependent link between
odontogenic cranial neural crest and endoderm in tooth development in amphibians and mammals may have been an early developmental mechanism
in the evolution of teeth, where the pharynx with cartilage bars as support
would have been lined by endoderm.
ACKNOWLEDGMENTS
We wish to thank Jim Hanken for critical reading of an early draft, for
which we are much indebted; and Peter Thorogood for comments on the
final version. M.M.S. thanks in particular the Nuffield Foundation for a Science Travel Grant, NERC for a project grant (GR3/8543) on biomineralization
in conodonts and early vertebrates, and the Royal Society for a Travel Grant;
for discussions, Andrew Lumsden, Anthony Graham, Paul Smith, Ivan Sansom, Richard Cloutier; and for editorial work with the references, Annabelle
Hickman. B.K.H. thanks the I. W. Killam Trust, NSERC of Canada, and,
for discussions, Bill Atchley; both B.K.H. and M.M.S. thank Ann Graveson,
439
Tom Miyake, and Steve Smith for discussions. M.M.S. is most grateful to
Mark Purnell for copies of his drawings and for permission to publish them,
and to him, Dick Aldridge, and Paul Smith for access to papers in press and
permission to quote from them. We thank Carl Gans, whose comments on
the "conodont story" were available to us prior to their publication. We have
not discussed them in this review and await their eventual publication.
REFERENCES
Adams, A. E., 1924, An experimental model of the study of the mouth in the amphibian embryo,
J. Exp. Zool. 40:311-380.
Aldridge, R. J., and Briggs, D. E. G., 1989, The soft body of evidence, Nat. Rist. 5:6-11.
Aldridge, R. J., and Theron, J. N., 1993, Conodonts with preserved soft tissue from a new Ordovician, Konservat-Lagerstiitte, J. Micropalaeontol., in press.
Aldridge, R. J., Briggs, D. E. G., Clarkson, E. N. K., and Smith, M. P., 1986, The affinities of
conodonts-New evidence from the Carboniferous of Edinburgh, Scotland, Lethaia 19:279291.
Aldridge, R. J., Briggs, D. E. G., Smith, M. P., Clarkson, E. N. K., and Clark, N. D. L., 1993,
The anatomy of conodonts, Phil. Trans. R. Soc. B., 338:405-421.
Atchley, W. R., and Hall, B. K., 1991, A model for development and evolution of complex
morphological structures. Bioi. Rev. 66:101-157.
Avery, J. K., 1954, Primary inductions of tooth formation (abstract), J. Dent. Res. 33:702.
Balinsky, B. I., 1975, An Introduction to Embryology, Saunders, Philadelphia.
Baume, L. J., 1980, The biology of pulp and dentine, a historic, terminologic-taxonomic, histologicbiochemical, embryonic and clinical survey, in: Monographs in Oral Science (H. M. Myers,
ed.), Vol. 8, pp. 1-246, S. Karger, Basel.
Bhatti, H. K., 1938, The integument and dermal skeleton of Siluroidea, Trans. Zool. Soc. 24:182.
Blieck, A., 1992, At the origin of chordates, Geobios 25: 10 1-/13.
Bodenstein, D., 1952, Studies on the development of the dorsal fin in amphibians, J. Exp. Zool.
120:213-245.
Briggs, D. E. G., 1992, Conodonts: A major extinct group added to the vertebrates, Science 256:
1285-1286.
Briggs, D. E. G., Clarkson, E. N. K., and Aldridge, R. J., 1983, The conodont animal, Lethaia
16:1-14.
Bronner-Fraser, M., and Fraser, S. E., 1988, Cell lineage analysis reveals multi potency of some
neural crest in avian embryos, Nature 335:161-164.
Bronner-Fraser, M., and Fraser, S. E., 1991, Cell lineage analysis of the avian neural crest, Development Suppl. 2:17-22.
Cam, Y., Neumann, M. R., and Ruch, J.-V., 1990, Immunolocalization of transforming growth
factor fJl and epidermal growth factor receptor epitopes in mouse incisors and molars with
a demonstration of in vitro production of transforming activity, Arch. Oral Bioi. 35:813822.
Cassin, c., and Capuron, A., 1979, Buccal organogenesis in Pleurodeles waltii Michah (urodele
amphibian), Study by intrablastocelic transplantation and in vitro culture, J. Bioi. Buccale
7:61-76.
440
Chibon, P., 1966, Analyse eXpCrimentale de la regionalisation et des capacites morphogenetiques

de la crete neurale chez l'amphibien urodele Pleurodeles waltii Michah, Mem. Soc. Zool. Fr.
36:1-107.
Chibon, P., 1967, Marquage nuc1eaire par la thymidine tritee des derives de la crete neurale chez
l'amphibien urodele Pleurodeles waltii Michah, J. Embryol. Exp. Morphol. 18:343-358.
Chisaka, 0., and Capecchi, M. R., 1991, Regionally restricted developmental defects resulting
from targeted disruption of the mouse homeobox gene hox-l.5, Nature 350:473-479.
Conway Morris, S., 1980, Conodont function: Fallacies of the tooth model, Lethia 13:107-108.
Conway Morris, S., 1989, Conodont paleobiology: Recent progress and unsolved problems, Terra
Nova 1:135-150.
Couly, G., and Le Douarin, N. M., 1990, Head morphogenesis in embryonic avian chimeras:
Evidence for a segmental pattern in the ectoderm corresponding to the neuromeres, Development 108:543-558.
Couly, G. F., Coltey, P. M., and Le Douarin, N. M., 1992, The developmental fate of the cephalic
mesoderm in quail-chick chimeras, Development 114:1-15.
Couly, G. F., Coitey, P. M., and Le Douarin, N. M., 1993, The triple origin of the skull in higher
vertebrates: a study in quail-chick chimeras, Development 117:409-429.
Damas, H., 1944, Recherches sur Ie developpement de Lampetra jluviatilis L. Contribution a
l'etude de 1a cephalogenese des Vertebres, Arch. BioI. (Paris) 55: 1-284.
Damas, H., 1951, Observations sur Ie developpement des ganglions cdiniens chez Lampetra
jluviatilis (L), Arch. BioI. 62:55-95.
De Beer, G. R., 1947, The differentiation of neural crest cells into visceral cartilages and odontoblasts
in Amblystoma, and a re-examination of the germ-layer theory, Proc. R. Soc. B 134:377398.
Denison, R. H., 1967, Ordovician vertebrates from western United States, Fieldiana Geol. 16:
131-192.
Detwiler, S. R., 1937, Observations upon the migration of neural crest cells and upon the development of the spinal ganglia and vertebral arches in Amblystoma, Am. J. Anat. 61:63-94.
D'Souza, R. N., Happonen, R.-P., flanders, K. C, and Butler, W. T., 1990, Histochemical 10calization of transforming growth factor-In in developing rat molars using antibodies to
different epitopes, J. BioI. Buccale 18:299-306.
Du Shane, G. P., 1935, An experimental study of the origin of pigment cells in Amphibia, J.
Exp. Zool. 72:1-31.
Epperlein, H. H., and Lehmann, R., 1975, The ectomesenchymal-ectodermal interaction system
(EElS) of Triturus alpestris in tissue culture. 2. Observations on the differentiation of visceral
cartilage, Differentiation 4:159-174.
Forey, P., and Janvier, P., 1993, Agnathans and the origin of vertebrates, Nature 361:129-134.
Fraser, S. E., and Bronner-Fraser, M., 1991, Migrating neural crest celIs in the trunk of the avian
embryo are multipotent, Development 112:913-920.
Fraser, S., Keynes, R., and Lumsden, A., 1990, Segmentation in the chick embryo hindbrain is
defined by cell lineage restrictions, Nature 344:431-435.
Gans, C., 1987, The neural crest: A spectacular invention, in: Developmental and Evolutionary
Aspects o/the Neural Crest (P. F. A. Maderson ed.), pp. 361-379, Wiley, New York.
Gans, C., 1989, Stages in the origin of vertebrates: analysis by means of scenarios, BioI. Rev. 64:
221-265.
Gans, C., and Northcutt, R. G., 1983, Neural crest and the origin of vertebrates: A new head,
Science 220:268-274.
Gans, C, and Northcutt, R. G., 1985, Neural crest: The implications for comparative anatomy,
Fortschr. Zool.3O:507-514.
Gaunt, W. A., and Miles, A. E. W., 1967, Fundamental aspects of tooth morphogenesis, in:
441
Structural and Chemical Organization of Teeth (A. E. W. Miles, ed.), Vol. 1, pp. 151-197,
Academic Press, London.
Geraudie, J., 1988, Fine structural peculiarities of the pectoral fin dermoskeleton of two Brachiopterygii, Polypterus senegalus and Calamoichthys calabaricus (Pisces, Osteichthyes), Anat.
Rec.221:455-468.
Goodrich, E. S., 1907, On the scales of fishes, living and extinct, and their importance in classification, Proc. Zool. Soc. Lond. 1907:751-774.
Graveson, A., 1993, Neural crest contributions to the development of the vertebrate head, Am.
Zool., in press.
Gross, W., 1954, Zur Conodonten-Frage, Senckenb. Lethaia 35:73-85.
Hall, B. K., 1983, Epithelial-mesenchymal interactions in cartilage and bone development, in:
Epithelial-Mesenchymal Interactions in Development (R. H. Sawyer and J. F. Fallon, eds.),
pp. 189-214, Praeger, New York.
Hall, B. K., 1986, Initiation of chondrogenesis from somitic, limb and craniofacial mesenchyme:
Search for a common mechanism, in: Somites in Developing Embryos (R. Bellairs, D. A.
Ede, and J. W. Lash, eds.), pp. 247-260, Plenum Press, New York.
Hall, B. K., 1987, Tissue interactions in the development and evolution of the vertebrate head,
in: Developmental and Evolutionary Aspects of the Neural Crest (P. F. A. Maderson, ed.),
pp. 215-259, Wiley, New York.
Hall, B. K., 1991, Evolution of connective and skeletal tissues, in: Developmental Patterning of
the Vertebrate Limb (J. R. Hinchliffe, J. M. Hurle, and D. Summerbell, eds.), pp. 303-312,
Plenum Press, New York.
Hall, B. K., 1992a, Evolutionary Developmental Biology, Chapman and Hall, London.
Hall, B. K., I 992b, The origin and evolution of vertebrate skeletal tissues, in: Chemistry and
Biology of Mineralized Tissues (H. Slavkin and P. Price, eds.), pp. 103-111, Elsevier, Amsterdam.
Hall, B. K., and Ekanayake, S., 1991, Effects of growth factors on the differentiation of neural
crest cells and neural crest cell-derivatives, Int. 1. Dev. Bioi. 35:367-387.
Hall, B. K., and Horstadius, S., 1988, The Neural Crest, Oxford University Press, Oxford.
Hall, B. K., and Miyake, T., 1992, The membranous skeleton: The role of cell condensations in
vertebrate skeletogenesis, Anat. Embryol. 186:107-124.
Halstead, L. B., 1973, The heterostracan fishes, Bioi. Rev. 48:279-332.
Hanken, J., 1986, Developmental evidence for amphibian origins, in: Evolutionary Biology, Vol.
20 (M. Hecht, B. Wallace, and I. Prance, eds.), pp. 389-416, Plenum Press, New York.
Hanken, J., and Hall, B. K., 1993, The Vertebrate Skull, Vols. 1-3, University of Chicago Press,
Chicago.
Hanken, J., and Thorogood, P., 1993, Evolution and development of the vertebrate skull: The
role of pattern formation, Trends Ecol. Evol. 8:9-15.
Hata, R., Bessem, c., Bringas, P., Jr., Hsu, M.-Y., and Slavkin, H. c., 1990, Epidermal growth
factor regulates gene expression of both epithelial and mesenchymal cells in mouse molar
tooth organs in culture, Cell Bioi. Int. Rep. 14:509-519.
Heath, L., Wild, A., and Thorogood, P., 1992, Monoclonal antibodies raised against pre-migratory
neural crest reveal population heterogeneity during crest development, Differentiation 49:
151-165.
Henzen, W., 1957, Transplantationen zur enwicklungs-physiologischen Analyse der larvalen
Mundorgane bei Bombinator und Triton, Arch. Entwicklungsmech. Org. 149:387-442.
Hertwig,O., 1879, Uber das Hautskelett der Fische, Morphol. lahrb. 5:1-21.
Holland, P. W. H., 1988, Homeobox genes and the vertebrate head, Development Suppl. 103:
17-24.
442
Holmgren, N., 1940, Studies on the head in fishes, Part I, Development of the skull in sharks
and rays, Acta Zool. Stockh. 21:51-267.
Holtfreter, J., 1935, Uber das Verhalten von Anurenektoderm in Urodelenkeimen, Wilhelm
Rouxs Arch. Entwicklungsmech. 133:427-494.
Horstadius, S., and Sellman, S., 1946, Experimentelle untersuchungen iiber die Determination
des Knorpeligen Kopfskelettes bei Urodelen, Nova Acta Regie Soc. Sci. Upsaliensis Ser. 4
13:1-170.
Hunt, P., and Krumlauf, R., 1991, Deciphering the Hox code: Clues to patterning branchial
regions of the head, Cell 66:1075-1078.
Hunt, P., Gulisano, M., Cook, M., Sham, M., Faiella, A., Wilkinson, D., Boncinelli, E., and
Krumlauf, R., 1991, A distinct Hox code for the branchial region of the head, Nature 353:
861-864.
Huysseune, A., and Sire, J.-Y., 1992a, Bone and cartilage resorption in relation to tooth development in the anterior part of the mandible in cichlid fishes: A light and TEM study, Anat.
Rec.234:1-14.
Huysseune, A., and Sire, J.-Y., 1992b, Development of cartilage and bone tissues of the anterior
part of the mandible in cichlid fishes: A light and TEM study, Anat. Rec. 233:357-375.
Janvier, P., 1981, The phylogeny of the craniata, with particular reference to the significance of
fossil agnathans, J. Vertebr. Paleontol. 1:121-159.
Jarvik, E., 1959, Dermal fin-rays and Holmgren's principle of delamination, Kungliga Svenska
Vetenskapsakad. Handlingar Uppsalla, (Stockholm) 6:1-51.
Jeppsson, L., 1979, Conodont element function, Lethaia 12:153-171.
Karatajute-Talimaa, V., 1978, Silurian and Devonian Thelodonts of U.S.S.R. and Spitsbergen,
Lithuanian Scientific Research Geological Survey Institute, Vilnius, Lithuania [in Russian].
Kemp, A., 1990, Involvement of the neural crest in development of the Australian longfish,
Neoceratodusforsteri (Krell 1870), Mem. Queensland Mus. 28:101-102.
Kemp, A., 1993, On the neural crests of lower vertebrates, in: 7th International Symposium on
Studies of Early Vertebrates, 1991, in press, Miguasha, Quebec.
Kessel, M., and Gruss, P., 1990, Murine developmental control genes, Science 249:374-379.
Kessel, M., and Gruss, P., 1991, Homeotic transformations of murine vertebrae and concomitant
alteration of Hox codes induced by retinoic acid, Cell 67:89-104.
Kessel, M., Balling, R., and Gruss, P., 1990, Variation of cervical vertebrae after expression of a
Hox-l.l transgene in mice, Cell 61:301-308.
King, G. J., Keeling, S. D., and Wronski, T. J., 1992, Histomorphologic and chemical study of
alveolar bone turnover in response to orthodontic tipping, in: Bone Biodynamics in Orthodontic and Orthopedic Treatment (D. S. Carlson and S. A. Goldstein, eds.), pp. 281-298,
University of Michigan, Ann Arbor, Michigan.
Kollar, E. J., 1986, Tissue interactions in development of teeth and related ectodermal derivatives,
in: Developmental Biology, A Comprehensive Synthesis, Vol. 4, Manipulation ofMammalian
Development (L. Browder, ed.), pp. 197-314, Plenum Press, New York.
Kollar, E. J., and Fisher, c., 1980, Tooth induction in chick epithelium: Expression of quiescent
genes for enamel synthesis, Science 207:993-995.
Krauss, S., Maden, M., Holder, N., and Wilson, S. W., 1992, Zebrafish pax[b] is involved in the
formation of the midbrain-hindbrain boundary, Nature 360:87-89.
Kronmiller, J. E., Upholt, W. B., and Kollar, E. J., 1991, EGF antisense oligodeoxynucleotides
block murine odontogenesis in vitro, Dev. Bioi. 147:485-488.
Kronmiller, J. E., Upholt, W. B., and Kollar, E. J., 1992, Alteration of murine odontogenic
patterning and prolongation of expression of epidermal growth factor mRNA by retinol in
vivo, Arch. Oral Bioi. 37:129-138.
Lamers, C. H. J., Rombout, J. W. H. M., and Timmermans, L. P. M., 1981, An experimental
443
study on neural crest migration in Barbus conchonius (Cyprinidae, Teleosti), with special
reference to the origin of the enteroendocrine cells, J. Embryol. Exp. Morpho!. 62:309-323.
Langille, R. M., 1987, The neural crest and the development of the skeleton in lower vertebrates,
Ph.D. thesis, Dalhousie University, Halifax, Nova Scotia, Canada.
Langille, R. M., and Hall, B. K., 1987, Development of the head skeleton of the Japanese medaka,
Oryzias latipes (Teleostei), J. Morphol. 193:135-158.
Langille, R. M., and Hall, B. K., 1988a, Role of the neural crest in development of the trabeculae
and branchial arches in embryonic sea lamprey, Petromyzon marinus (L.), Development 102:
301-310.
Langille, R. M., and Hall, B. K., 1988b, Role of the neural crest in development of the cartilaginous
cranial and visceral skeleton of the medaka, Oryzius latipes (Teleostei), Anat. Embryol. 177:
297-305.
Langille, R. M., and Hall, B. K., 1989, Developmental processes, developmental sequences and
early vertebrate phylogeny, Bioi. Rev. 64:73-91.
Langille, R. M., and Hall, B. K., 1993, Pattern formation and the neural crest, in: The Vertebrate
Skul/, Vol. 1, Development (J. Hanken and B. K. Hall, eds.), in press, University of Chicago
Press, Chicago.
Le Douarin, N. M., 1982, The neural crest, source of mesenchymal cells, in: The Neural Crest,
pp. 55-89, Cambridge University Press, Cambridge.
Lonai, P., and Urtreger, A. 0., 1990, Homeogenes in mammalian development and the evolution
of the cranium and central nervous system, FASEB J. 4:1436-1443.
Lopashov, G. V., 1944, Origins of pigment cells and visceral cartilage in teleosts, C. R. Acad. Sci.
USSR 44: 169-172.
Lopashov, G. V., \950, Experimental investigation of the sources of cellular material and conditions
offormation of the pectoral fins in teleost fishes, C. R. Acad. Sci. USSR 70:137-140.
Lufkin, T., Dierich, A., LeMeur, M., Mark, M., and Chambon, P., 1991, Disruption ofthe Hox1.6 homeobox gene results in defects in a region corresponding to its rostral domain of
expression, Ce1l66:1105-11 19.
Lufkin, T., Mark, M., Hart, C. P., Dolle, P., LeMeur, M., and Chambon, P., 1992, Homeotic
transformation of the occipital bones of the skull by ectopic expression of a homeobox gene,
Nature 359:835-841.
Lumsden, A. G. S., 1985, Tooth morphogenesis: Contributions of the cranial neural crest in
mammals, in: Tooth Morphogenesis and Differentiation (A. Belcour, and J.-V. Ruch, eds.),
pp. 29-40, INSERM.
Lumsden, A. G. S., 1987, The neural crest contribution to tooth development in the mammalian
embryo, in: Developmental and Evolutionary Aspects ofthe Neural Crest (P. F. A. Maderson,
ed.), pp. 261-300, Wiley, New York.
Lumsden, A. G. S., 1988, Spatial organisation of the epithelium and the role of neural crest cells
in the initiation of the mammalian tooth germ, Development Suppl. 103:155-169.
Lumsden, A. G. S., 1989, Multipotent cells in the avian neural crest, Trends Neurosci. 12:8183.
Lumsden, A., 1990, The cellular basis of segmentation in the developing hindbrain, Trends Neurosci. 13:329-335.
Lumsden, A. G. S., and Keynes, R., 1989, Segmental patterns of neuronal development in the
chick hindbrain, Nature 337:426-428.
Mackenzie, A., Lemming, G. L., Jowett, A. K., Ferguson, M. W. J., and Sharpe, P. T., 1991, The
homeobox gene Hox 7.1 has specific regional and temporal expression patterns during early
murine craniofacial embryogenesis, especially tooth development in vivo and in vitro, Development 111:269-285.
Mackenzie, A., Ferguson, M. W. J., and Sharpe, P. T., 1992, Expression patterns of the homeobox
444
gene, Hox-8, in the mouse embryo suggest a role specifying tooth initiation and shape, Development 115:403-420.
Maden, M., Horton, C., Graham, A., Leonard, L., Pizzey, J., Siegenthaler, G., Lumsden, A., and
Eriksson, U., 1992, Domains of cellular retinoic acid-binding protein I (CRABP I) expression
in the hindbrain and neural crest ofthe mouse embryo, Mech. Dev. 37:13-23.
Maisey, J. G., 1979, Finspine morphogenesis in squalid and heterodontid sharks, Zool. J. Linn.
Soc. 66:161-183.
Maisey, J. G., 1988, Phylogeny of early vertebrate skeletal induction and ossification patterns,
Evo!. Bioi. 22:1-36.
Marusich, M. F., and Weston, J. A., 1991, Development of the neural crest, Curro Opin. Genet.
Dev. 1:221-229 .
.Matsumoto, J., Lynch, T. J., Grabowski, S., Richards, C. M., Lo, S. L., Clark, c., Kern, D.,
Taylor, J. D., and Tchen, T. T., 1983, Fish tumor pigment cells: Differentiation and comparison to their normal counterparts, Am. Zool. 23:569-580.
McGinnis, W., and Krumlauf, R., 1992, Homeobox genes and axial patterning, Cell 68:283-302.
Mikulic, D. G., Briggs, D. E. G., and Kluessendorf, J., 1985, A Silurian soft-bodied biota, Science
228:715-717.
Millan, F. A., Denhez, F., Kondaiah, P., and Akhurst, R. J., 1991, Embryonic gene expression
patterns ofTGF!31, 132 and 133 suggest different developmental functions in vivo, Development
111:131-144.
Mina, M., Kollar, E. J., Bishop, J. A., and Rohrbach, D. H., 1990, Interaction between the neural
crest and extracellular matrix proteins in craniofacial skeletogenesis, Crit. Rev. Oral Bioi.
Med. 1:79-87.
Mitsiadis, T. A., Dicou, E., Joffre, A., and Magloire, H., 1992, Immunohistochemical localization
of nerve growth factor (NGF) and NGF receptor (NGF-R) in the developing first molar
tooth of the rat, Differentiation 49:47-61.
Moss, M. L., 1960, Experimental induction of osteogenesis, in: Calcification in Biological Systems,
pp. 323-348, American Association for the Advancement of Science, Washington, D.C.
Moss, M. L., 1968a, Bone, dentin and enamel and the evolution of vertebrates, in: Biology o/the
Mouth (P. Person, ed.), pp. 37-65, American Association for the Advancement of Science,
Washington, D.C.
Moss, M. L., 1968b, Comparative anatomy of vertebrate dermal bone and teeth, I, The epidermal
co-participation hypothesis, Acta Anatom. 71:178-208.
Murphy, P., Davidson, D. R., Hill, R. E., and Morriss-Kay, G. M., 1992, Retinoid-induced
alterations of segmental organization and gene expression in the mouse hindbrain, in: Retinoids
in Normal Development and Teratogenesis (G. M. Morriss-Kay, ed.), pp. 229-240, Oxford
University Press, Oxford.
Nakajima, T., 1984, Larval vs adult pharyngeal dentition in some Japanese cyprinid fishes, J.
Dent. Res. 63:1140-1146.
Nakajima, T., 1987, Development of pharyngeal dentition in the Cobitid fishes, Misgurnus anguillicaudatus and Cobitis biwae with a consideration of evolution ofCyprininiform dentitions,
Copeia 1:208-213.
Nakajima, T., 1990, Morphogenesis of the pharyngeal teeth in the Japanese dace, Tribolodon
hakonensis (Pisces: Cyprinidae), J. Morphol. 205:155-163.
Nakajima, T., and Yue, P., 1989, Development of the pharyngeal teeth in the Big Head, Aristichthys
nobilis (Cyprinidae), Jpn. J. lchthyol. 36:42-47.
Newth, D. R., 1950, Fate of the neural crest in lampreys, Nature 165:284.
Newth, D. R., 1951, Experiments on the neural crest of the lamprey embryo, J. Exp. Bioi. 28:
247-260.
Newth, D. R., 1956, On the neural crest oflamprey embryos, J. Embryol. Exp. Morphol. 4:358375.
445
Noden, D. M., 1983, The role of the neural crest in patterning of avian cranial skeletal, connective,
and muscle tissues, Dev. Bioi. 96:144-165.
Northcutt, R. G., and Gans, C, 1983, The genesis of neural crest and epidermal placodes: A
reinterpretation of vertebrate origins, Q. Rev. Bioi. 58:1-28.
Obruchev, D. V., 1964, Besceljustnye, ryby, in: Osnovy paleontologii [English trans!., 1967, Agnathans, fishes, in: Fundamentals ofPaleontology (J. A. Orlov, ed.), Israel Program for Scientific
Translation, Jerusalem].
0rvig, T., 1951, Histologic studies of placoderms and fossil elasmobranchs. I. The endoskeleton,
with remarks on the hard tissues oflower vertebrates in general, Ark. Zool. 2:321-454.
0rvig, T., 1967, Phylogeny of tooth tissues: Evolution of some calcified tissues in early vertebrates,
in: Structural and Chemical Organization of Teeth (A. E. W. Miles, ed.), Vo!. 1, pp. 45-110,
Academic Press, London.
0rvig, T., 1977, A survey of odontodes (dermal teeth) from developmental, structural, functional,
and phyletic points of view, in: Problems in Vertebrate Evolution (S. M. Andrews, R. S.
Miles, and A. D. Walker, eds.), pp. 53-75, Academic Press, London.
Osborn, J. W., 1978, Morphogenic gradients: Fields versus clones, in: Development, Function and
Evolution of Teeth (P. M. Butler, and K. A. Joysey, eds.), pp. 171-202, Academic Press,
London.
Osborn, J. W., 1984, From reptile to mammal: Evolutionary considerations of the dentition with
emphasis on tooth attachment, Symp. Zool. Soc. Land. 52:549-574.
Osborn, J. W., and Price, D. G., 1988, An autoradiographic study of periodontal ligament development in the mouse, J. Dent. Res. 67:455-461.
Palmer, R. M., and Lumsden, A. G. S., 1987, Development of periodontal ligament and alveolar
bone in homografted recombinations of enamel organs and papillary, pulpal and follicular
mesenchyme in the mouse, Arch. Oral Bioi. 32:281-289.
Partanen, A.-M., 1990, Epidermal growth factor and transforming growth factor-alpha in the
development of epithelial-mesenchymal organs of the mouse, CU". Top. Dev. Bioi. 24:3155.
Partanen, A.-M., and Thesleff, I., 1987, Localization and quantitation of 125I-epidermal growth
factor binding in mouse embryonic tooth and other embryonic tissues at different developmental stages, Dev. Bioi. 120:186-197.
Patterson, C, 1977, Cartilage bones, dermal bones and membrane bones, or the exoskeleton
versus the endoskeleton, in: Problems in Vertebrate Evolution (S. M. Andrews, R. S. Miles,
and A. D. Walker, eds.), pp. 77-122, Academic Press, London.
Pelton, R. W., Dickinson, M. E., Moses, H. L., and Hogan, B. L. M., 1990, In situ hybridization
analysis of TGFi33 RNA expression during mouse development: Comparative studies with
TGFi3l and 132, Development 110:609-620.
Petrovic, A. G., Stutzmann, J. J., and Lavergne, J. M., 1990, Mechanisms of craniofacial growth
and modus operandi of functional appliances: A cell-level and cybernetic approach to orthodontic decision making, in: Craniofacial Growth Theory and Orthodontic Treatment
(D. S. Carlson, ed), pp. 13-74, University of Michigan, Ann Arbor, Michigan.
Purnell, M. A., 1993, Feeding mechanisms in conodonts and the function of the earliest vertebrate
hard tissues, Geology, 21:375-377.
Purnell, M. A., and von Bitter, P. H., 1992, Blade-shaped conodont elements functioned as
cutting teeth, Nature 359:629-631.
Raven, C. P., 1931, Zur Entwicklung der Ganglienleiste. I. Die Kinematik der Ganglienleistenentwicklung, Wilhelm Raux' Arch. Entwicklungsmech. 125:210-292.
Raven, C. P., 1936, Zur Entwicklung der Ganglienleiste. V. lrber die Differenzierung des Rumpfganglienleistenmaterials, Wilhelm Raux' Arch. Entwicklungsmech. 134:122-145.
Reif, W.-E., 1980, A model of morphogenetic processes in the dermal skeleton of elasmobranchs,
Neues Jahrb. Geol. Palaeontol. Abh. 159:339-359.
446
Reif, W.-E., 1982, Evolution of dermal skeleton and dentition in vertebrates: The odontoderegulation theory, Evol. BioI. 15:287-368.
Sadaghiani, B., and Vielkind, J. R, 1989, Neural crest development in Xiphophorns fishes: Scanning
electron and light microscopic studies, Development 105:487-504.
Sadaghiani, B., and Vielkind, J. R, 1990, Distribution and migration pathways of HNK-1immunoreactive neural crest cells in teleost fish embryos, Development 110:197-209.
Salmivirta, M., Elenius, K., Vainio, S., Hofer, U., Chiquet-Ehrismann, R., Thesleff, I., and Jalkanen,
M., 1991, Syndecan from embryonic tooth mesenchyme binds tenascin, J. Bioi. Chern. 266:
7733-7739.
Sansom, I. J., 1992, The palaeobiology of the panderodontacea and selected other euconodonts,
Doctorate thesis, University of Durham, Durham, England.
Sansom, I. J., Smith, M. P., Armstrong, H. A., and Smith, M. M., 1992, Presence of the earliest
vertebrate hard tissues in conodonts, Science 256: 1308-1311.
Schaeffer, B., 1977, The dermal skeleton in fishes, in: Problems in Vertebrate Evolution (S. M.
Andrews, R S. Miles, and A. D. Walker, eds.), pp. 25-52, Academic Press, London.
Schaeffer, B., 1987, Deuterostome monophyly and phlogeny, in: Evolutionary Biology, Vol. 21
(M. K. Hecht, B. Wallace, and G. T. Prance, eds.), pp. 179-235, Plenum Press, New York.
Sellman, S., 1946, Some experiments on the determination of the larval tooth in Amblystorna
rnexicanurn, Odontol. Tidskrift 54: 1-128.
Serbedzija, G. N., Bronner-Fraser, M., and Fraser, S. E., 1989, A vital dye analysis of the timing
and pathways of avian trunk neural crest cell migration, Development 106:809-816.
Slavkin, H. C, 1990, Molecular determinants of tooth development: A review, Crit. Rev. Oral
BioI. Med. 1:1-16.
Slavkin, H. C, 1991, Molecular determinants during dental morphogenesis and cytodifferentiation:
A review, J. Cranio/ac. Genet. Dev. Bioi. 11:338-349.
Slavkin, H. C, Sasano, Y., Kikunaga, S., Bessem, c., Bringas, P., Jr., Mayo, M., Luo, W., Mak,
G., Rall, L., and Snead, M. L., 1990, Cartilage, bone and tooth induction during early
embryonic mouse mandibular morphogenesis using serumless, chemically-defined medium,
Connect. Tiss. Res. 24:41-52.
Smith, M. M., 1979, Scanning electron microscopy of odontodes in the scales of a coelacanth
embryo, Latimeria chalumnae Smith, Arch. Oral BioI. 24: 179-183.
Smith, M. M., 1991, Putative skeletal neural crest cells in early Late Ordovician vertebrates from
Colorado, Science 251:301-303.
Smith, M. M., 1992, Microstructure and evolution of enamel amongst osteichthyan and early
tetrapods, in: Structure, Function and Evolution o/Teeth (P. Smith, 00.), pp. 73-101, Proceedings 8th International Symposium on Dental Morphology, Jerusalem (1989).
Smith, M. M., and Hall, B. K., 1990, Developmental and evolutionary origins of vertebrate
skeletogenic and odontogenic tissues, BioI. Rev. 65:277-374.
Smith, M. P., 1990, The Conodonta-Palaeobiology and evolutionary history of a major Palaeozoic
chordate group, Geol. Mag. 127(4):365-369.
Smith, M. P., Briggs, D. E. G., and Aldridge, R. D., 1987, A conodont animal from the lower
Silurian of Wisconsin, USA, and the apparatus architecture of panderodontid conodonts,
in: Palaeobiology o/Conodonts (R. J. Aldridge, 00.), pp. 91-104, Ellis Horwood, Chichester.
Soehn, K. L., and Wilson, M. V. H., 1990, A complete, articulated heterostracan from Wenlockian
(Silurian) beds of the Delorme Group, Mackenzie Mountains, Northwest Territories, Canada,
J. Vertebr. Paleontol. 10:405-419.
Stutzman, J. J., and Petrovic, A. G., 1989, Responsiveness of alveolar bone to orthodontic treatment
in adult patients, in: Orthodontics in an Aging Society (D. S. Carlson, ed.), pp. 181-199,
University of Michigan Press, Ann Arbor, Michigan.
Takahashi, Y., Bontoux, M., and Le Douarin, N. M., 1991, Epitheliomesenchymal interactions
447
are critical for Quox 7 expression and membrane bone differentiation in the neural crest
derived mandibular mesenchyme, EMBO J. 10:2387-2393.
Ten Cate, A. R., 1975, Formation of supporting bone in association with periodontal ligament
organization in the mouse, Arch. Oral Bioi. 20:137-138.
Ten Cate, A. R., and Mills, S. C, 1972, Development of periodontium-Origin of alveolar bone,
Anat. Rec. 173:69-78.
Terentiev, I. B., 1941, On the role played by the neural crest in the development of the dorsal fin
in Urodela, C. R. Acad. Sci. USSR 31:91-94.
Thesleff, I., 1991, Tooth development, Dent. Update 1991 (November):382-387.
Thesleff, 1., Mackie, E., Vainio, S., and Chiquet-Ehrismann, R., 1987, Changes in the distribution
oftenascin during tooth development, Development 101:289-296.
Thesleff, 1., Jalkanen, M., Vainio, S., and Bemfield, M., 1988, Cell surface proteoglycan expression
correlates with epithelial-mesenchymal interactions during tooth morphogenesis, Dev. BioI.
129:565-572.
Thesleff, 1., Vaahtokari, S., and Vainio, S., 1990, Molecular changes during determination and
differentiation of the dental mesenchymal cell lineage, J. BioI. Buccale 18: 179-188.
Thomson, K. S., 1987, The neural crest and the morphogenesis and evolution of the dermal
skeleton in vertebrates, in: Developmental and Evolutionary Aspects of the Neural Crest
(P. F. A. Maderson, ed.), pp. 301-338, Wiley, New York.
Thorogood, P. V., 1993, Differentiation and morphogenesis of cranial skeletal tissues, in: The
Vertebrate Skul/, Vol. I, Development (J. Hanken and B. K. Hall, eds.), pp. 112-152, University of Chicago Press, Chicago.
Thorogood, P. V., 1991, The development of the teleost fin and implications for our understanding
of tetrapod limb evolution, in: Developmental Patterning of the Vertebrate Limb (J. R.
Hinchliffe, J. M. Hurle, and D. Summerbell, eds.), pp. 347-354, Plenum Press, New York.
Topham, R. T., Chiego, D. J., Jr., Gattone, V. H., II, Hinton, D. A., and Klein, R. M., 1987,
The effect of epidermal growth factor on neonatal incisor differentiation in the mouse, Dev.
BioI. 124:532-543.
Twitty, V. C, and Bodenstein, D., 1941, Experiments on the determination problem, I: The roles
of ectoderm and neural crest in the development of the dorsal fin in Amphibia, II: Changes
in ciliary polarity associated with the induction of fin epidermis, J. Exp. Zool. 86:343-380.
Vaahtokari, A., Vainio, S., and Thesleff, I., 1991, Associations between transforming growth
factor,61 RNA expression and epithelial-mesenchymal interactions during tooth morphogenesis, Development 113:985-994.
Vainio, S., and Thesleff, I., 1992, Sequential induction of syndecan, tenascin and cell proliferation
associated with mesenchymal cell condensation during early tooth development, Differentiation 50:97-105.
Vainio, S., Jalkanen, M., and Thesleff, I., 1989, Syndecan and tenascin expression induced by epithelialmesenchymal interactions in embryonic tooth mesenchyme, J Cell Bioi. 108:1945-1954.
Vainio, S., Jalkanen, M., Vaahtokari, A., Sahlberg, C, Mail, M., Benrfield, M., and Thesleff, I.,
1992, Expression of syndecan gene is induced early, is transient, and correlates with changes
in mesenchymal cell proliferation during tooth organogenesis, Dev. BioI. 147:322-333.
Van der Bruggen, W., and Janvier, P., 1993, Denticles in thelodonts, Nature, 364:107.
Wagner, G., 1949, Die Bedeutung der Neuralleiste fur die Kopfgestaltung der Amphibienlarven,
Untersuchungen an Chimaeren von Triton und Bombinator, Rev. Suisse Zool. 56:519-620.
Wagner, G., 1955, Chimaerische Zahnanlagen aus Triton-Schmelzorgan und Bombinator-Papille.
Mit Beobachtungen uber die Entwicklung von Kiemenzahnchen und Mundsinnesknospen
in den Triton-Larven, J. Embryol. Exp. Morphol. 3:160-188.
Weston, J. A., 1970, The migration and differentiation of neural crest cells, in: Advances in
448
Morphogenesis. Vol. 8 (M. Abercrombie, J. Brachet, and T. J. King, eds.), pp. 41-114,
Academic Press, New York.
Wilde, C. E., 1955, The urodele neuroepithelium, The differentiation in vitro of the cranial neural
crest, J. Exp. Zool. 130:573-591.
Wilkinson, D. G., Bhatt, S., Cook, M., Bonicelli, E., and Krumlauf, R., 1989, Segmental expression
of Hox-2 homeobox-containing genes in the developing mouse hindbrain, Nature 341:405-409.
Williamson, W. c., 1849, On the microscopic structure of the scales and dermal teeth of some
ganoid and placoid fish, Phil. Trans. R. Soc. Land. 139:435-475.
Wilson, M. V. H., and Soehn, K. L., 1990, Discovery of complete Silurian fish, Naturwissenschaften
77:328-330.
Wood, A., 1982, Early pectoral fin development and morphogenesis of the apical ectodermal
ridge in the killifish, Aphyosemion scheeli. Anat. Rec. 204:349-356.

A Developmental Model For Evolution of The Vertebrate Exoskeleton and Teeth The Role of Cranial and Trunk Neural Crest

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

A Developmental Model For Evolution of The Vertebrate Exoskeleton and Teeth The Role of Cranial and Trunk Neural Crest

Încărcat de

Drepturi de autor:

Formate disponibile

10

A Developmental Model for

The Role of Cranial and Trunk Neural Crest

EXOSKELETAL DEVELOPMENT AND EVOLUTION

M. M. Smith and B. K. Hall

Vertebrate Exoskeleton and Neural Crest

Principles of Skeletal Development and Evolution

Neural Crest versus Mesodermal Cellular Origins

M. M. Smith and B. K. Hall

Vertebrate Exoskeleton and Neural Crest

et al. concluded, "The anterior limit of the notochord then corresponds to

M. M. Smith and B. K. Hall

Vertebrate Exoskeleton and Neural Crest

ectomesenchymal cells dependent upon specific epithelium for induction and

The Odontode as Dermal Exoskeleton

cellular aggregates of the odontode/tooth, epithelial dental cap, ectomesenchymal dental

M. M. Smith and B. K. Hall

.<j< x .,".. }{"" ,"', ! L,

Vertebrate Exoskeleton and Neural Crest

M. M. Smith and B. K. Hall

regard to the evolution and development of the vertebrate skull, "mechanisms

Lepidotrichia as Dermal Fin Rays

Vertebrate Exoskeleton and Neural Crest

Odontogenic Dependence on Neural Crest and Extent a/Competence

M. M. Smith and B. K. Hall

Evidence for Skeletogenic/Odontogenic Cranial versus Trunk Neural Crest

Vertebrate Exoskeleton and Neural Crest

there was a fundamental question to be answered in considering the possible

M. M. Smith and B. K. Hall

Dependence of Neural Crest on Epithelium for Acquisition of Morphogenetic

Vertebrate Exoskeleton and Neural Crest

Exoskeletal Diversity and a Developmental Model

M. M. Smith and B. K. Hall

The Cranial Neural Crest-Derived Populations of Cells Associated

Cell types derived from the population

Chondroblasts of Meckel's cartilage

Cell Populations of the Mammalian Mandible

Vertebrate Exoskeleton and Neural Crest

1. A chondrogenic population differentiating into Meckel's cartilage, the

A Further Summary of Cell Types Formed from the Cell

M. M. Smith and B. K. Hall

Odontogenic and Skeletogenic Subpopulations of the Cranial Neural Crest

Vertebrate Exoskeleton and Neural Crest

M. M. Smith and B. K. Hall

and skeletogenic cell populations can be distinguished and separated from

Molecular Control Mechanisms

Vertebrate Exoskeleton and Neural Crest

odontogenic cell types; (6) patterning of cell populations into skeletogenic

M. M. Smith and B. K. Hall

ond, applies primarily to the neurocranium where gross spatial patterning is

Vertebrate Exoskeleton and Neural Crest

M. M. Smith and B. K. Hall

Patterning of the tissue producing exoskeletal scleroblasts (the neural crest)

Vertebrate Exoskeleton and Neural Crest

or mammalian embryos as the basis for exploring exoskeletal development

M. M. Smith and B. K. Hall

Vertebrate Exoskeleton and Neural Crest

{3 mRNA was expressed in the epithelium. It is clear that dental mesenchyme

M. M. Smith and B. K. Hall

Vertebrate Exoskeleton and Neural Crest

Origins of Skeletal Tissues and an Evolutionary Model

M. M. Smith and B. K. Hall

We therefore evaluate conodonts in our evolutionary model, and reassess