Anat Embryol (1998) 198:289–305

© Springer-Verlag 1998

O R I G I NA L A RT I C L E

&roles:A. Huysseune · C. Van der heyden · J.-Y. Sire

Early development of the zebrafish (Danio rerio) pharyngeal dentition (Teleostei, Cyprinidae)

&misc:Accepted: 17 April 1998

&p.1:Abstract In order to build a reference system to assess ongoing in vitro and in situ hybridisation experiments on epithelial-mesenchymal interactions governing odontogenesis in the zebrafish, we describe here the generation of the pharyngeal dentition, and the histological development of teeth up to fourteen days post-fertilization, using serial semithin sections, handmade and computer-assisted reconstructions and transmission electron microscopy. The tooth pattern in larval zebrafish is generated in a predictable, and bilaterally symmetrical manner from shortly before hatching onwards. Characteristics related to tooth development and structure differ considerably from those seen in juvenile specimens and those described for other bony fishes. Particular features related to the cyprinid condition include the complex epithelial connectivity and the mode of attachment of the teeth. &kwd:Key words Zebrafish · Danio · Dentition · Tooth development · TEM&bdy:

Introduction Teeth are present in most classes of vertebrates and are the developmental outcome of reciprocal interactions between two fundamental tissues in the embryo: epithelium and mesenchyme (Thesleff et al. 1995; Huysseune and Sire 1998). To study the interactions responsible for tooth development, scientists, for obvious reasons, have chosen the mammalian tooth as their model. Yet, teeth in mammals present a number of specializations with regard to ancestral teeth and have many derived characters. These relate especially to the number of replacement cyA. Huysseune (✉) · C. Van der heyden Instituut voor Dierkunde, Universiteit Gent, Ledeganckstraat 35, B-9000 Gent, Belgium Tel.: 32.9.2645229; Fax 32.9.2645344; e-mail [email protected],&/fn-block: J.-Y. Sire Université Paris 7-Denis Diderot, CNRS, URA 1137, Case 7077, 2 Place Jussieu, F-75251 Paris, Cédex 05, France

cles (usually only two, i.e. diphyodonty, instead of many, i.e. polyphyodonty), the shape of the teeth (heterodonty derived from homodonty, multicusp pattern of molars) and their number (Butler 1995; Huysseune and Sire 1998). The study of a vertebrate that has numerous teeth, of relatively simple shape, and that are replaced throughout life, unquestionably presents a number of advantages. In particular, a polyphyodont condition, implying repeated initiation of tooth germs, offers interesting possibilities of experimentation. Many actinopterygian fishes present such a condition in their dentition. For several years our laboratories have engaged in the study of odontogenesis in actinopterygian fishes, especially in more highly evolved teleosts such as cichlids (Huysseune 1983, 1989a, b, 1990; Huysseune et al. 1989; Huysseune and Sire 1997b). However, in order to advance our understanding of the mechanisms that govern the epithelial-mesenchymal interactions underlying tooth development, it would be useful to study these in a model for which an important array of genetic and molecular tools is available. The zebrafish (Danio rerio), a small cyprinid fish, has become such a model over the past years (Ekker and Akimenko 1991; Driever et al. 1994; Lele and Krone 1996). In addition, a wide range of mutants of this species is available (Driever et al. 1996). Most studies related to zebrafish development have concentrated so far on the very precocious embryonic development. Surprisingly few studies have focussed on the development of features that largely appear in post-embryonic life (the notable exceptions being the study of Cubbage and Mabee 1996 on the skull and girdles, and of Sire et al. 1997 on the scales). To our knowledge, not one paper has studied the emergence of the dentition in detail. This paper aims at filling this gap and will serve as a reference for ongoing in vitro and in situ hybridisation experiments focusing on the mechanism of tooth initiation in this model. For this purpose, we have studied the emergence of the tooth pattern from the very first appearance of tooth germs, up to fourteen days post-fertilization (i.e. twelve days post-hatching). This period covers the experimental period used in our in vitro studies

290

and largely exceeds the period during which larval-lethal recessive mutants affecting posterior branchial arch development can survive (Schilling et al. 1996). In addition, we have investigated the histological features of first-generation teeth to use them as baseline information when assessing the results of the in vitro experiments.

Materials and methods Materials All developmental stages of the zebrafish used in this study come from laboratory-reared specimens and are expressed as number of days post-fertilization (d PF). Fertilized eggs of zebrafish were allowed to develop and to hatch in tanks of approximately 80 l at 28° C and with a light period of 12 h/day. Under these circumstances, they hatched at 48 h post-fertilization. The larvae were fed Paramecium and commercial baby fish food up to 10d PF, and powdered Tetramin and Artemia nauplii from 10d PF onwards. We selected specimens from 36 h and 48 h post-fertilization and then at daily intervals up to 7d PF, as well as 10, 12 and 14d PF for processing into serial semithin sections. Methods All animals were killed by means of an overdose of the anaesthetic MS-222 (Sandoz, Basel). Specimens were fixed for 2 h, at room temperature, in a mixture containing 1.5% glutaraldehyde and 1.5% paraformaldehyde in 0.1 M cacodylate buffer (pH 7.4), rinsed in the same buffer to which 10% sucrose was added, postfixed in 1% osmium tetroxide in cacodylate buffer containing 8% sucrose and rinsed again. They were next dehydrated using a graded series of ethanol and embedded in epon. The specimens were cut with glass or diamond knives into uninterrupted series of transverse semithin (1- or 2-µm-thick) sections, which were stained with toluidine blue. Other series were interrupted for ultrathin sectioning. Some thin sections were partially demineralized whilst floating on a water bath. Thin sections were contrasted with uranyl acetate and lead citrate and viewed in a Philips 201 transmission electron microscope operating at 80 kV. Handmade as well as computer-assisted reconstructions were prepared starting from drawings of serial sections. Computer-assisted reconstructions were generated using the Jandell 3D software package. A number of juvenile specimens was cleared and stained according to a slightly modified version of the method of Hanken and Wassersug (1981).

Results General features of the dentition The zebrafish, like other cyprinids, lacks teeth on the oral jaws but carries a well-developed pharyngeal dentition. The pharyngeal teeth are implanted on the paired fifth ceratobranchials (ventral elements of the fifth branchial arch s.s.), the only tooth-bearing elements in the entire buccal and pharyngeal region (Fig. 1). These elements start their development as paired cartilaginous anlagen around 2d PF, at which time there is evidence of condensing chondroblasts. They further differentiate as paired rod-like elements, slightly diverging from each other in their anterior half, and directed laterodorsad in

their posterior part. At 3d PF already, they become enveloped with perichondral bone, and membranous apolamellae develop from this bone collar from about 4d PF onwards to form the major part of the juvenile and adult ceratobranchial. In our laboratory conditions, the cartilage starts to resorb (without any endochondral ossification) shortly after 10d PF; in the adult it has disappeared altogether. Throughout ontogeny, each of the paired fifth ceratobranchials carries a number of pharyngeal teeth and tooth germs in various phases of development (up to a total of about 20 on each pharnygeal jaw in juveniles of 4 weeks of age, and to about 10 in 1-year-old adults). The functional teeth abut against a keratinized epithelial pad supported by the basioccipital bone of the skull floor. Emergence of the tooth pattern in early post-embryonic stages The descriptions below are based on the study of serial sections (and reconstructions derived from them) of at least three specimens per stage. The number of days is given as a general guideline and corresponds to the situation found in the laboratory conditions described above (Table 1). Figure 2 represents a number of these stages. The dentigerous area on average spans about 40 µm at 3d PF to 120 µm at 14d PF in an antero-posterior direction. It should be noted that the first appearance of new germs is always difficult to observe because of the elaborate infoldings of the pharyngeal epithelium (see below). Although not apparent at first sight, the initiation of tooth germs follows a regular pattern during the first fourteen days PF (Fig. 3). Teeth arise in transverse rows, new rows being added posterior to old ones. Teeth belonging to a single transverse row are given the same roman numeral. During the first fourteen days PF, one germ breaks this rule by developing in front of the first (I) row; this potential row is labeled i. Dorsal (i.e. more lateral) germs within a row arise earlier than ventral (and more medial) ones and are therefore numbered from dorsal (lateral) to ventral (medial) loci by ascending arabic numerals. At 36 h PF, no tooth germs are visibly present in the pharyngeal region. The first germs are observed at hatching (48 h, or 2d PF) and lie immediately posterior and medial to the condensing fifth ceratobranchial cartilages. In the next few days, they will come to lie in between the diverging posterior portions of the cartilages, which by then are fully differentiated and surrounded by perichondral bone. At hatching, there are two germs on each side, situated at the same transverse level, one medially and one, further developed, more laterally. The lateral (and more dorsal) germ (tooth I1) has already produced some matrix at its tip and is therefore probably the first germ to develop in the dentition. The tooth has fully differentiated at 3d PF, at which time it has already become ankylosed to the perichondral bone through a delicate por-

291

e ba

B

pg

A Fig. 1 A Drawing of the outline of the head of a juvenile (8.8 mm SL) zebrafish (Danio rerio), as observed from a cleared and stained specimen, showing the position of the pharyngeal jaws – fifth ceratobranchials (arrow). B, C Enlarged dorsal (B), ventral (C) view of the pharyngeal jaws and their teeth. The pulp cavity is visible in the teeth (anterior in all three drawings is to the top, ba branchial arches, e eye, pg pectoral girdle). Bars A 1 mm; B, C 0.25 mm&ig.c:/f

-tion of attachment bone. Its tip pierces the epithelium at 4d PF. The medial (and more ventral) germ (tooth I2), at hatching, is still in a stage of morphogenesis and has not yet deposited any matrix. It will start to do so shortly thereafter, since well-formed dentine is present at 3d PF, and it will become ankylosed at 5d PF. During its growth, the tooth base is deflected laterally, so that at the moment of attachment, it has come to lie beneath the base of I1 and the attachment bone of I2 has fused dorsally to the ventral side of the attachment bone of I1. Teeth I1 and I2 thus form a pair of nearly synchronously initiated and closely connected teeth. However, although their tooth bases have fused to one another, the orientation of the two germs is different: the older (and more dorsal) I1 has its long axis more or less in the transverse plane, whereas the younger (and more ventral) I2 is oriented rather antero-posteriorly.

C Caudolateral to I1, an invagination that is barely visible at hatching announces the development of a third germ (II1). Tooth matrix has already been formed at 3d PF and the tooth is well differentiated at 4d PF. At 5d PF it has become partially ankylosed, either to the perichondral bone, or to apolamellae emanating from it. In all specimens of 6d PF, it is attached to the posterior part of the attachment bone of I1. A new germ (II2) develops posterior to I1 in advanced 3d PF specimens. At that stage, the germ is still in the initial phase, its epithelial organ not yet having grown out. It starts to differentiate at about 4d PF but it does not ankylose until 10d PF, and in advanced specimens only. Together with II1 to which base it attaches ventrally, it forms a second pair of teeth, itself attached to the first pair. A fifth germ (III1) appears in well-developed specimens of 5d PF, lateral and posterior to II1 and has acquired a bell-shape by 7d PF. Dental matrix starts to appear as late as 10d PF, and in some specimens only. The tooth has become ankylosed by 14d PF. Ventral and medial to I2 a germ (I3) appears to have been initiated in advanced 5d PF specimens. The tooth further develops ventral to I2. Cytodifferentiation starts at 7d PF at the earliest, and is still not well advanced at 10d PF. Tooth I3 has become ankylosed to I2 by 14 days.

292

Fig. 2A–E Reconstructions from serial transverse sections (left) and schematical representations (right) showing the pharyngeal jaws and the larval teeth in zebrafish. Anterior is below. Numbers refer to the tooth numbers as described in the text. A 2d PF, B 4d PF, C 6d PF, D 10d PF, E 14d PF. In A–D the pharyngeal jaws are represented by the cartilaginous fifth ceratobranchials (CB V). To simplify, neither perichondral bone nor bone apolamellae have been drawn. In E, the jaws are largely bony but include remnants of the cartilage. Teeth are represented by dots on the schemes; black dots represent attached teeth. Attachment bone and tooth proper have not been distinguished in any of the drawings. Bars A–D 50 µm; E 100 µm&ig.c:/f

In well-developed 10d PF specimens, a new germ (III2) has appeared ventral and posterior to II2, but has not yet acquired the bell-shape. It has become ankylosed to III1 at 14d PF, with which it forms a pair of closely connected teeth. Two parallel crest-like invaginations of the pharyngeal epithelium are present, from 4d PF onwards, medial to the anterior teeth, one on each side of the mediosagittal plane. These give rise to a discernible germ (i1) at

293 Table 1 Landmarks in the development of first-generation teeth of the zebrafish (Danio rerio) from fertilization up to 14d PF. Number of hours, days – or interval – represents average time under the lab conditions described previously; variation exists depending on more or less advanced state of development of individuals ( >indicates later than; 1 earliest visible invagination of the epithelium; 2 distinct epithelial downgrowth; 3 attachment bone mineralized throughout)&/tbl.c:&

Tooth

Initiation/start of morphogenesis1

Ongoing morphogenesis2

Start of matrix deposition

Attachment 3

I1 I2 I3 II1 II2 II3 III1 III2 IV1 IV2 i1

36–48 h 36–48 h 5–6 days 2–3 days 3 days 11–12 days 5–6 days 10 days 11–12 days 13–14 days 8–10 days

36–48 h 36–48 h 6–7 days 2–3 days 3–4 days 11–12 days 6–7 days 10–12 days 12–14 days >14 days 10 days

36–48 h 2–3 days 7 days 3 days 4–5 days 13–14 days 10 days 12 days 14 days >14 days 13–14 days

3 days 5 days 14 days 5–6 days 10 days >14 days 14 days 14 days >14 days >14 days >14 days

&/tbl.:

about 10d PF and the germ is well differentiated, but not yet attached, at 14d PF. At 14d PF two more tooth germs have developed on the posterior side of the dentition. One germ (IV1) develops posterior to III1, the other germ (IV2) posterior to III2 and medial to IV1. Tooth IV1 is present at 12d PF already and has clearly differentiated a dentine cone by 14d PF, but the medial germ (IV2) develops only beyond 12d PF and has not produced any matrix at 14d PF. Both these teeth become ankylosed beyond 14d PF. Finally, a germ (II3) develops in the second tooth row posterior to I3; it is present at 12d PF, has differentiated matrix by 14d PF, but attaches only beyond 14d PF. From 12d PF onwards, teeth, usually I1 and II1, are missing in the dentitions examined. In the 14d PF specimen illustrated in Fig. 2, e.g. I1 and II1 are missing on one side, whereas I1 and I2 are missing on the other side. In early post-hatching stages, the pattern observed in every single specimen is very alike. The pattern starts to show some slight variation only from 5d PF onwards.

Fig. 3A–G Schematic representation of the order of tooth appearance in the pharyngeal jaws of the zebrafish, obtained by projecting the loci onto a horizontal plane (dorsal tooth germs become lateral ones in this scheme; ventral tooth germs become medial ones). Anterior is below, lateral to the left; only the right pharyngeal jaw is figured (white dots unattached tooth germs or teeth, hatched dots teeth having been shed, arrows between the dots indicate sequential order of appearance). A 2d PF, B early 3d PF, C 4d PF, D 5d PF, E 10d PF, F 12d PF, G 14d PF&ig.c:/f

However, this variation can always be reduced to the overall level of development of the dentition, e.g. less advanced 5-day-old larvae resembling more advanced earlier stages. Up to 10d PF, left and right side are completely symmetrical. Only at 12d PF does variation become more prominent, and do left and right sides differ in the specimens examined. Development of a first-generation pharyngeal tooth In the following description, three stages of tooth development have been distinguished. The first stage covers initiation and morphogenesis, prior to the deposition of any matrix. Next comes a stage of cytodifferentiation in which the enameloid and dentine matrix are deposited, and finally a stage of attachment. Once ankylosed, the tooth has completed its development and enters a functional phase. Stages of teleost tooth development previously distinguished in other teleosts and termed “a” to “e” (Huysseune 1983; Huysseune and Sire 1997b) cannot be used here because several characters differ from this standard scheme, as explained below. Tooth numbers refer to the loci described in the previous section. Phase of initiation and morphogenesis This phase starts with the commitment of the odontogenic tissues and ends just prior to the deposition of matrix

294

in the newly formed germ. The pharyngeal epithelium prior to the onset of tooth formation is composed of at least three cell layers, the basal layer (facing the basal lamina) consisting of cylindrical cells, the intermediate layer having more cuboidal cells and the superficial layer consisting of flattened cells. At the onset of epithelial invagination, cells of the basal layer become more polarized, with elongated nuclei perpendicular to the basal membrane (Fig. 4). At hatching and shortly thereafter (2 and 3d PF), the mesenchyme immediately underlying the epithelium is highly cellular so that an aggregation of mesenchymal cells, if at all present, cannot be detected opposite a beginning invagination. Afterwards, the mesenchyme becomes more loose, but even so, a clear aggregation of mesenchymal cells that could be interpreted as a presumptive dental papilla is never obvious at the time an epithelial invagination starts to form. The early invagination resembles a disc of approximately four or five cylindrical epithelial cells, overlain by four or five rather cuboidal cells (Figs. 5, 6). This organisation produces an essentially bilayered structure, facing a few mesenchymal cells. In the epithelial population the cylindrical cells are polarized towards the epithelial-mesenchymal junction. They contain a moderate amount of cytoplasm, in contrast to the cells covering them, and to the mesenchymal cells, which have a high N/C ratio. The epithelial-mesenchymal interface presents an undulated aspect due to imbrication of epithelial and mesenchymal cells. The medial side of the disc will subsequently grow deeper into the mesenchyme to produce a crescent-shaped epithelial mass of which the arms lie parallel to the pharyngeal epithelium. A ventral and a dorsal side of the germ can thus be distinguished (Fig. 7 inset). This epithelial invagination, or dental organ, surrounds one or at most a few mesenchymal cells (Fig. 7). The two cell layers of the dental organ, from now on called inner (ide) and outer (ode) dental epithelium, are clearly distinguishable. Cells of the ide are still polarized although the future apical cells are cuboidal. Cells of the ode are somewhat more flattened and their cytoplasm has developed a little. Throughout the dental organ however, cells are poor in organelles. The mesenchymal cells that are now enveloped by the dental organ are plump and show a slightly increased amount of cytoplasm. They constitute the early dental papilla. The orientation of the dental organ differs according to the tooth locus considered. Tooth I1 is oriented more or less transversally, and in transverse sections such as used here, the tooth is therefore cut in a longitudinal plane (Fig. 8). Tooth I2 however, although adjacent to I1, has a predominant antero-posterior orientation and is therefore cut transversely. Prior to the deposition of any matrix, the membrane of the ide cells facing the dental papilla is fairly smooth (Fig. 9). Their cytoplasm, which has become slightly more important, shows numerous free ribosomes and a few small vesicles. The RER is hardly developed, few mitochondria are present. In contrast, the cells of the dental papilla show short finger-like projections that come into contact with the basal lamina (Fig. 9). The cytoplasm of

these cells appears to be more developed than that of the ide cells as it contains a fairly well-developed RER, numerous free ribosomes and several mitochondria. Phase of cytodifferentiation Deposition of the first matrix takes place simultaneously along nearly the entire epithelial-mesenchymal interface with only a moderate gradient from tip to base (Figs. 8, 10). In addition, the process is slightly more advanced along the dorsal side of the germ, that is, the side closest to the pharyngeal epithelium (Fig. 10). The ide cells, or ameloblasts, are well differentiated; their cytoplasm consists mainly of free ribosomes, numerous mitochondria and a few RER cisternae (Fig. 10). The part of the dental papilla surrounded by the dental organ is composed of a cone of approximately five cells on one section, all of which can be considered odontoblasts (Fig. 10). These plump cells closely adjoin each other and, in contrast to the ameloblasts, are characterized by an extremely welldeveloped RER, with numerous parallel cisternae surrounding the centrally located nucleus. Despite many observations on numerous specimens, we have been unsuccessful in identifying a matrix at the tooth tip corresponding to the unmineralized precursor of the enameloid cap of erupted, functional, first-generation teeth. The successive steps of dentinogenesis on the other hand are illustrated from the mid-region of the tooth Fig. 4 One-µm-thick transverse section through the pharyngeal region of a 2d PF zebrafish. The aperture in the pharyngeal epithelium (pe) starts to be formed. Note the polarized epithelial cells in the region of future germ formation (asterisk; m mesenchyme). Bar 10 µm&ig.c:/f Fig. 5 One-µm-thick transverse section through an early invagination-stage tooth germ (arrow) at 2d PF. Bar 10 µm&ig.c:/f Fig. 6 TEM micrograph of a 6d PF zebrafish showing an early stage of invagination of the pharyngeal epithelium (pe). Polarized epithelial cells (ep) face a few mesenchymal cells (m; arrowheads points of invagination). Bar 5 µm&ig.c:/f Fig. 7 TEM micrograph of a germ, indicated by an arrow in the inset, in a 3d PF zebrafish. No matrix has been deposited so far. Note the polarized cells of the basal layer of the pharyngeal epithelium (pe). The dental organ is composed of the inner (ide) and outer (ode) dental epithelium (dp dental papilla). Bar 5 µm. Inset general view of the germ shown in magnification (arrow), as observed from a 1-µm-thick section. The pharyngeal cavity (pc) is now widely open (cb ceratobranchial cartilage). Bar 25 µm&ig.c:/f Fig. 8 One-µm-thick section slightly posterior to that shown in Fig. 7, inset. Note the different orientation of the two germs (I1 and I2) and the mitosis in the outer dental epithelium of I2 (arrowhead). Bar 25 µm&ig.c:/f Fig. 9 High magnification of the interface between ide cells (ide) and cells of the dental papilla (dp) in a germ of a 2d PF zebrafish prior to the onset of matrix deposition. An arrow points to fingerlike processes of the dental papilla cells. Bar 1 µm&ig.c:/f Fig. 10 TEM micrograph of the lateral germ (I1) indicated in Fig. 8, showing the organisation of the dental organ (do) and of the dental papilla (dp). Arrowhead indicates patches of matrix in the prolongation of the tooth base. The dorsal side of the germ is towards the top of the figure (am ameloblast, od odontoblast). Bar 2 µm&ig.c:/f

295

296

because this area is representative for the major part of the developing shaft (Figs. 11–13). Dentine is deposited first as predentine, consisting of a fairly dense meshwork of collagen fibrils, of about 30 to 40 nm in diameter in an electron-lucent background. The fibrils are oriented roughly parallel to the long axis of the tooth. The outer, first-deposited matrix, soon becomes filled with a finely granular electron-dense background substance (Fig. 11). The lamina densa of the basement membrane has disappeared from zones where the background substance is present. The odontoblasts engaged in synthesis of the predentine show an extremely well-developed network of RER cisternae parallel to the surface of the cell. There are no odontoblast processes penetrating into the predentine matrix. The cytoplasmic content of the ameloblasts does not show important changes compared to the situation immediately prior to matrix deposition. In a slightly more advanced stage, mineralization is initiated in the superficial region of the predentine, turning it into dentine (Fig. 12). Opposite regions of mineralization the ameloblast cytoplasm is distinctly poor in organelles. The mineralization front is irregular and some patches of mineral crystals are seen at some distance from the mineralized region proper. Within the mineralized areas, fine needle-like crystals appear to be aligned along the collagen fibrils. Background substance is now present throughout nearly the entire thickness of the predentine layer. The odontoblasts still show a well-developed RER and numerous mitochondria. In a well-advanced stage of dentinogenesis such as found in the more apically situated part of the tooth described above, the meshwork of fibrils is no longer recognizable in the mineralized part (Fig. 13), suggesting a maturation of the dentine. At the ameloblast-dentine interface a highly electron-dense amorphous substance forming a distinct membrane, approximately 100 nm thick, has been deposited. The extent of this membrane does not appear to be related to the dentine maturation below. As before, the ameloblasts are poor in organelles in their distal cytoplasm, especially where they face this membrane. The predentine is reduced due to progression of the mineralization front and extracellular spaces appear at its interface with the odontoblasts. The latter show a reduced amount of RER and some cytoplasmic prolongations at their surface. However, none of them is seen to penetrate into the predentine. Throughout all stages of dentinogenesis, fibril diameter and orientation remain the same. Therefore the dentine matrix in first-generation teeth of zebrafish is homogeneously composed of longitudinally oriented fibrils, 30–40 nm in diameter. Moreover the tooth pulpa consists of a few, closely packed odontoblasts only. No capillaries nor nerve endings have been observed inside the pulpa. Attachment, functional stage and resorption The first teeth to attach in the dentition (i.e. I1, I2, II1, II2) are approximately 12 µm across at their base and about

50 to 70 µm tall prior to attachment. In these four teeth attachment is first realized along the dorsal side of the tooth base. Depending on the tooth considered, the bone support can be the perichondral bone collar along the medial side of the cartilage, membranous apolamellae, the attachment bone of an adjacent tooth germ, or a combination of these. The attachment bone is defined as the matrix that is deposited in the prolongation of the tooth base from the level of the cervical loop down to the surface of the supporting bone (Fig. 14). Attachment bone matrix deposition does not await completion of the tooth base since patches of collagenous matrix can already be observed in Fig. 11 Ultrastructural features of the predentine (pd) in the midregion of a tooth in a 6d PF zebrafish. A layer of predentine, embedded in an electron-dense background substance is separated from the odontoblasts (od) by a layer of more immature predentine (asterisk) with similar fibril composition but lacking the background substance. Bar 0.5 µm&ig.c:/f Fig. 12 Start of predentine (pd) mineralization in a 6d PF zebrafish (de dentine). Bar 0.5 µm&ig.c:/f Fig. 13 Advanced stage of dentinogenesis in a 6d PF zebrafish showing a thick mineralized dentine (de) layer and little predentine (pd). A superficial electron-dense membrane (arrowhead) covers the apical region of the tooth. Bar 0.5 µm&ig.c:/f Fig. 14 Low-power electron micrograph of a tooth base in a 6d PF zebrafish, shortly before ankylosis of the perichondral bone (pb). The dentine (de) is heavily mineralized but not covered by the dense membrane described in Fig. 13. Predentine (pd) persists at the tooth base and as a thin layer against the inner dentinal wall. The limit of the electron-dense background substance coincides with the position of the cervical loop tip (arrowheads). In this particular case one mesenchymal cell (asterisk) adjoins the predentine, bone of attachment (boa) and perichondral bone simultaneously. Bar 2 µm&ig.c:/f Fig. 15 High magnification of the attachment bone (boa) shown in Fig. 14. Small patches of fine needle-like crystals (arrowheads) are dispersed especially in areas where an electron-dense background substance is present, and preferentially against the perichondral bone (pb). Bar 0.5 µm&ig.c:/f Fig. 16 Detail of the base of an ankylosed tooth in a 10d PF zebrafish; slightly demineralized. Note differences in dentine (de) and attachment bone (boa) matrix organisation with respect to the location of the cervical loop tip (arrowhead). Bar 1 µm&ig.c:/f Fig. 17 One-µm-thick section through the pharyngeal region of a 6d PF zebrafish. Tooth I1 is completed and ankylosed via its attachment bone (arrowhead) to the perichondral bone surrounding the ceratobranchial cartilage (cb). Bar 25 µm&ig.c:/f Fig. 18 One-µm-thick section slightly posterior to the region shown in Fig. 17. The tip of tooth I 1 has erupted and lies in the pharyngeal cavity (pc). The enameloid cap (arrowhead) stains weakly, the dentine is darkly stained. Note in this, and in the previous figure, the difference in pulpal content between tooth I 1 and I2. Bar 25 µm&ig.c:/f Fig. 19 TEM micrograph of the tooth tip shown in Fig. 18. The enameloid cap (en) is overlain by an electron-dense membrane (arrowheads) diminishing in thickness towards the tooth base. In the enameloid, only traces of organic matrix are left. The dentine (de) is composed of longitudinally oriented collagen fibrils. Bar 2 µm. Inset Detail of the dentine (de) matrix at a more proximal level of the same tooth. The dentine is possibly overlain by a thin seam of enameloid (arrowhead; ide remnants of the inner dental epithelium). Bar 0.5 µm&ig.c:/f

297

298

the prolongation of the tooth base as soon as dentinogenesis has started (Fig. 10). However, its mineralization starts only when mineralization of the dentine has nearly reached the tooth base. Facing the cervical loop tip, some predentine persists that merges without a clear separation with the developing attachment bone matrix. At the perichondral bone surface too, the attachment bone matrix imperceptibly merges with osteoid matrix (Figs. 14, 15). Like the predentine the attachment bone matrix is composed of a fairly dense meshwork of approximately 30-nm-thick collagen fibrils, but their orientation is no longer predominantly parallel to the long axis of the tooth. Patches of electron-dense background substance are dispersed in this matrix. Mineral nucleation appears to take place preferentially on these patches, although some crystals are seen dispersed in the matrix (Fig. 15). In accordance with the slightly retarded stage of odontogenesis ventrally versus dorsally in the tooth germ, we observe a general delay in the functional characteristics of the different cell populations participating in tooth formation and attachment. This is particularly obvious for the osteoblast-like mesenchymal cells that line the attachment bone along its outer surface (Fig. 14). Along the ventral part of the attachment, these cells are plump and show features indicative of active synthesis (low N/C ratio, well-developed RER, many mitochondria), whereas dorsally cells in the same position have a high N/C ratio and almost no organelles visible in the scarce cytoplasm. Along the pulpal side, the odontoblasts are continued beyond the cervical loop tip (i.e. facing the attachment bone) by cells that resemble them in being large, closely juxtaposed, and possessing an equally well-developed RER network of parallel cisternae. In contrast, their surface facing the attachment bone matrix shows numerous small finger-like projections penetrating a little into the osteoid. Strikingly, a single cell can face the dentine, the attachment bone and the supporting bone simultaneously and apparently is able to participate in the synthesis of all three matrices (Figs. 14, 15). Once the attachment bone matrix is well mineralized, it is no longer distinguishable either from the dentine on one side, or from the supporting bone on the other side and its position can only be traced by the position of the cervical loop tip. Slight demineralization however makes it possible to distinguish both matrices on the basis of different fibril orientation, the attachment bone being more woven-fibred (Fig. 16). Nevertheless, some longitudinally arranged fibril bundles appear to penetrate into the attachment bone. Conspicuous extracellular spaces have appeared in between the odontoblasts proper and the odontoblast-like cells facing the attachment bone, making the pulp look more loosely organized. Eruption appears to be slightly delayed with regard to attachment. Indeed, concomitant with attachment the tip still lies surrounded by the enamel organ. Elaborate epithelial folds exposing the tip become visible shortly thereafter (Figs. 17, 18). Comparative measurements have shown that a tooth does not grow in length any further once mineralization of the attachment bone has

reached the bone support, be it even in a limited portion below the tooth base. A distinct enameloid cap becomes visible in fully ankylosed functional teeth (Figs. 18, 19). It consists of a well-mineralized tissue, containing traces of a fibrillar organic material. The enameloid cap is covered by the electron-dense membrane described previously (see Fig. 13). Once attached, teeth also present obvious differences with respect to the aspect of their pulpal content, even at the LM level. Teeth either show a pulp filled with stellate cells and large intercellular spaces or a pulp filled with closely adjoined cells, with no intercellular spaces (Figs. 17, 18). These characteristics possibly represent two functional stages of a process. Indeed, ultrastructural evidence suggests that, shortly after the tooth has erupted, the inner lining of the dentine wall is subjected to a series of successive changes. In a first step, the dentine surface along the pulpal side no longer presents a smooth surface but has a denticulate aspect with the electron-dense cell membrane of the odontoblasts exactly matching the surface without any intervening space or matrix (Fig. 20). These odontoblasts show an extensive, electron-lucent cytoplasm, poor in organelles. Next, new woven-fibred matrix intermingled with granular material is deposited on the former dentinal surface. This secondary deposition often presents the aspect of bulges with successive electron-dense lines suggesting successive waves of deposition (Fig. 21). The odontoblasts facing this newly deposited matrix show a more electron-dense cytoplasm than described in the previous stage, with numerous mitochondria, a moderate amount of RER and some vesicles. The secondary deposition results in a dentinal wall composed of two distinct layers (Figs. 22, 23): an outer layer (in average 1-µm-wide) with longitudinally arranged fibrils and an irregular, more electron-dense inner layer (0.5-µm-wide in averFig. 20 TEM micrograph of part of a functional tooth (partially demineralized) showing dentine (de), and portions of adjacent ameloblasts (am) and odontoblasts (od). Note the denticulate aspect of the inner dentinal boundary (arrowheads). Bar 0.5 µm&ig.c:/f Fig. 21 TEM micrograph of a stage succeeding the one presented in Fig. 20. Patches of new matrix (arrowheads) have been deposited against the inner dentinal wall (de). Bar 0.5 µm&ig.c:/f Fig. 22 TEM micrograph of part of a functional tooth near the apex of its pulpal cavity, partially demineralized. The secondary deposition (asterisks) is clearly distinguishable from the dentine (de) that was originally deposited. Bar 1 µm&ig.c:/f Fig. 23 High magnification of an area adjacent to that shown in Fig. 22. Note the difference in structure and organization of the dentine (de) and the secondary deposition (asterisk), and the translucent aspect of the odontoblasts (od). Bar 0.5 µm&ig.c:/f Fig. 24 TEM micrograph of the base of a functional tooth (I2) and of a tooth germ (I3) in a 10d PF zebrafish. The attachment bone (boa) of I2 experiences resorption (arrowhead) adjacent to the developing tooth germ (od odontoblasts of I3). Bar 5 µm&ig.c:/f Fig. 25 High magnification of a small, possibly mononuclear, cell (arrowhead) involved in the type of resorption shown in Fig. 24. This part of the cell probably represents a clear zone (boa bone of attachment). Bar 0.5 µm&ig.c:/f

299

300

Fig. 26 Schematic representation of the development of a firstgeneration tooth in the zebrafish, as deduced from serial-1-µm sections of numerous samples. A–E Phase of initiation and morphogenesis, F cytodifferentiation, G attachment, H eruption (stippled epithelium, crosses mesenchyme, cross-hatched supporting bone – perichondral bone or apolamellae, white enameloid, black dentine, densely stippled mineralized attachment bone, asterisk cavitation in the epithelium). Despite a different representation of dentine and attachment bone, the two are not distinguishable at the light microscopical level&ig.c:/f

age). At the apex of the pulp cavity the latter matrix presents the aspect of coarse bundles (Fig. 22) whereas along the shaft it forms a continuous layer, although of uneven thickness (Fig. 23). Where the base of a tooth germ meets the base of a functional tooth, the tissues often become very tightly adjoined. Cells lining the attachment bone of the functional tooth are very flattened and lined immediately by the dental papilla cells of the adjacent germ (Fig. 24). In this area, cells likely to represent clastic cells have been found to line the attachment bone of the functional tooth. Portions of such cells, very poor in organelles, have been observed to tightly encrust the attachment bone matrix (Fig. 25). Epithelial connectivity A special point of interest concerns the epithelial connectivity of the germs, i.e. the developmental relationship between a germ and the epithelium from which it derives. We have been able to establish these connections with certainty through TEM observations only. Like for eruption, observations on epithelial connectivity are obscured by the elaborate folding of the pharyngeal epithelium.

Observations have shown that I1 and I2, II1 and II2, III1 and III2, IV1 and IV2 develop independently from each other directly from the pharyngeal epithelium. However, some connection appears to exist between the same loci in subsequent rows (e.g. I1, II1, III1 and IV1). Tooth I3, although medially connected to the pharyngeal epithelium, appears to branch off partially from the dental organ of I2. Tooth i1 is the only one to develop from a pre-existing crest-like longitudinal epithelial invagination. However, at its posterior margin this tooth germ appears to be connected, through this crest-like structure, to the epithelial invagination of I1. A schematized representation of the development of a first-generation tooth in the zebrafish is given in Fig. 26.

Discussion The appearance of tooth germs on the fifth ceratobranchials in larval zebrafish (Danio rerio) has briefly been reported in some recent papers (Cubbage and Mabee 1996; Schilling et al. 1996), yet these data present no more than a note in passing. To our knowledge, this is the first study that describes, in detail, the emergence of the pharyngeal dentition in early post-embryonic zebrafish. In addition, details are presented related to the development and fine structure of the first-generation teeth. Pattern generation The first tooth in the zebrafish pharyngeal dentition differentiates prior to ossification of the fifth ceratobranchial, in accordance with observations made by Cubbage and Mabee (1996) on cleared and stained specimens. Schilling et al. (1996) note the presence of pharyngeal

301

teeth, ‘three on each side’ on P7 (an erroneous term for the fifth arch). From our results, it is clear that the pattern appears in a predictable and bilaterally symmetrical way at least during the first ten days PF. Any deviation observed during this period relates to the entire pattern and can be attributed to a slight delay in overall development of the specimen considered. At 12d PF, left-right asymmetries become apparent, and the pattern becomes evermore variable afterwards, both between similar-aged specimens and between left and right side in a single specimen (C. Van der heyden and A. Huysseune, unpublished observations). The emergence of the tooth pattern in larvae has been described for cichlids (Huysseune 1983, 1989b, 1990), and, of more relevance to the situation in zebrafish, in a number of cyprinids including carps and Danioninae – but not Danio rerio (Nakajima 1979, 1984, 1987, 1990; Nakajima and Yue 1989). It has not been possible to fit our observations on larval tooth patterns into the cyprinid (or even danioninoid) schemes as erected by Nakajima (1984), in which the larval teeth appear at even and odd positions in waves that sweep alternately. Clearly, in the zebrafish, teeth up to 10d PF are added either medial (and ventral) or posterior to existing teeth or tooth germs, and thus, subsequent rows develop from anterior to posterior. This situation resembles the pattern of tooth initiation in the upper pharyngeal jaw of larval cichlids, be it in the opposite way – in cichlids new germs arise lateral and anterior to existing ones (Huysseune 1983). At about 10d PF, one germ develops anterior to the first row and during further development, more teeth will arise in an anterior direction (C. Van der heyden and A. Huysseune, unpublished observations). There may be several reasons why larval zebrafish do not appear to fit into the schemes erected by Nakajima (1984): (1) Danio rerio is a very small species compared to those described by Nakajima (1984), notably Zacco platypus and Opsariichthys uncirostris; the sequence described in the present study is derived from specimens the largest of which hardly exceeds 5 mm standard length; (2) the sequence of initiations is more compressed and the delay between the first appearance of a tooth and its attachment in zebrafish is much shorter than in the other species described by Nakajima (1984); this rapid development may in part be related to the fact that Danio rerio is a tropical species; (3) because of the size of the dentigerous area (spanning 40–120 µm in an antero-posterior direction) the observations here are necessarily based on the study of serial, 1-µm sections; Nakajima (1984) worked on cleared and stained specimens only and therefore saw teeth only from a stage when mineralized matrix is present. Thus, different techniques may have entailed differences in the patterns observed; (4) finally, important intraspecific variation may exist. Nakajima himself was not able to assess the type of larval dentition in the Danioninae he studied because of too much variation in the appearance of the anterior teeth (Nakajima 1984). Cubbage and Mabee (1996) also revealed the existence of intraspecific variation in the ossification sequence of many cranial bones. Both genetic

and environmental factors were thought to possibly play a role in this variation and the different strains that are used throughout the world may certainly be of significance, not only in the order of ossification, but possibly also in the order of tooth appearance. The fact that, in our material, variation in the pattern does occur beyond 10 days PF only, may be caused at least partially by differential growth rates among individuals. However, the possibility that a change in diet may have an effect must be considered as well. Sire et al. (1997) showed that, in zebrafish, the timing of appearance of the scales, elements most likely phylogenetically related to teeth (Huysseune and Sire 1998), is a function of age as well as size. Teeth I1, II1 and III1 probably correspond to positions ce-0, po-1 and po-2 in Nakajima’s terminology. Furthermore I2, II2 and III2 may represent replacement teeth for I1, II1 and III1 next to which they become attached, yet without shedding of the latter. The teeth coexist for a while; then the medial teeth of the pair are usually shed. According to Nakajima (1984) such teeth belong to different replacement waves, in which case the term ‘firstgeneration tooth’ may not be applied, strictly speaking. However, since they arise virtually simultaneously, the two series are considered here as ‘first-generation teeth’. Retention of the functional tooth after ankylosis of the replacement tooth is responsible for the high number of teeth in larval and early juvenile dentitions of zebrafish as compared to their adult counterparts (C. Van der heyden and A. Huysseune, unpublished observations). This appears to be general for cyprinids (Nakajima 1984). A striking feature is the difference in speed of development of the tooth at different positions. The first tooth to appear is also the first one to become attached, and this process is completed in less than 2 days. In contrast, e.g. I3, although present as a germ at 6d PF, does not become attached before 14d PF and therefore takes more than one week to develop into an ankylosed, functional tooth. Because there is no obvious size difference between these teeth, there is at present no explanation for this differential growth rate. There is nevertheless a slight tendency of ventral teeth to have longer developmental times, possibly due to the larger size that they must acquire in order to become attached. It would appear that the nearly synchronous, and juxtaposed appearance of I1 and I2, II1 and II2, etc. is at variance with assumed inhibitory fields around individual germs (cf. Osborn 1971, 1978). Elucidation of the tooth pattern throughout ontogeny and the variation that exists are the subject of ongoing studies in our laboratories and will be particularly useful in questioning Osborn’s (1971, 1978) inhibition model of tooth initiation. Development and fine structure of the teeth Recently we have described the development of tooth germs, from initiation to attachment, in two other teleost families, the callichthyids (Huysseune and Sire 1997a) and the cichlids (Huysseune and Sire 1997b), covering

302

both light microscopical and ultrastructural aspects. The callichthyids, despite being ostariophysans and thus relatively closely related to the zebrafish, have an atypical dentition in that teeth disappear soon and are not replaced. We therefore refer to the paper dealing with tooth development in the cichlid Hemichromis bimaculatus as a general guideline for tooth development in teleosts. Although there are general resemblances, the development of zebrafish teeth nevertheless differs in a number of notable features from the above-mentioned standard description and these features will be discussed below. In addition, it must be stressed that a number of features described here apply to larval teeth only, and differ in the larger teeth of juvenile and adult zebrafish (C. Van der heyden and A. Huysseune, unpublished observations). Such is also the case in cichlid teeth (Huysseune and Sire 1997b). A first point of difference concerns the number of stages that have been distinguished (five in cichlids, three in zebrafish). More detailed ultrastructural studies on larval teeth may reveal that each of the three phases distinguished in zebrafish can be further subdivided, and this may certainly apply to the replacement teeth in juveniles. The distinction between the cyprinid Danio and the cichlid Hemichromis is due to the fact that in the latter a distinct epithelial invagination is formed prior to reinvagination to form a bell-shaped tooth germ, whereas in Danio the bell-shape is acquired right from the beginning of epithelial ingrowth. In other words, the cervical loop appears to be formed in two distinct ways in the two species. As a result, the first stage in Danio corresponds to stages “a” and “b” in cichlids. This is also true for adult zebrafish: a cervical loop is formed from the beginning of epithelial downgrowth, whereas in cichlids, a long epithelial strand is formed prior to reinvagination (C. Van der heyden and A. Huysseune, unpublished observations). This difference may, however, be due at least partly to the cryptic development of replacement teeth in cichlids versus their superficial development in zebrafish. An unmineralized precursor of the enameloid cap could not be distinguished in zebrafish first-generation teeth. This, along with the fact that dentine is deposited simultaneously over most of the epithelial-mesenchymal interface made it impossible to define a clear stage “c”. Therefore, the stage of cytodifferentiation in zebrafish corresponds to stages “c” and “d” in cichlids. Further differences are described below.

growth. Preodontoblast contacts with the basal lamina similar to those observed here were reported earlier in callichthyids (Huysseune and Sire 1997a) and cichlids (Huysseune and Sire 1997b) and appear to be a common feature in early odontogenesis (Kallenbach and Piesco 1978). It has been suggested that these contacts mediate early epithelial-mesenchymal interactions (Thesleff and Hurmerinta 1981). At present there is no morphological evidence indicative for the factors responsible for initiation of the tooth germ. Recent papers contradict each other as to whether initiation events occur in the epithelium or the mesenchyme and what factors are possibly involved. We refer to Huysseune and Sire (1997b) for a discussion of the literature. Through our ongoing in vitro experiments on larval zebrafish, we hope to contribute positively to this discussion. Related to the issue of tooth initiation is the problem of the epithelial connectivity of the germs, i.e. the localization of emergence from a particular epithelium (either the pharyngeal epithelium proper, or the reduced enamel epithelium of a predecessor). Epithelial connectivity indicates possible developmental relationship between germs, and therefore also gives a clue as to identification of the tooth family to which it belongs (see the discussion by Reif 1982). The identification of the epithelium from which a new germ emerges is hampered in larval as well as in juvenile and adult zebrafish by the fact that the pharyngeal epithelium from early post-embryonic life onwards assumes an elaborately folded aspect, whereby dental organs form the continuation of the crypts between these deep folds, which themselves often surround the tip of a functional tooth. In such cases it is hard to tell whether the germ derives from the pharyngeal epithelium proper or from the dental organ of an older tooth. One should also consider the possibility that a germ may shift in position concomitant with eruptive movements of an adjacent tooth. Such a process has been described in piranhas (Berkovitz and Shellis 1978) where there is a relative movement between the epithelial cord and the tooth germ from which it originates, probably due to eruption of the latter. A shift in the connection of the tooth germ-dental lamina has also been observed in apodans (Casey and Lawson 1981). So far, our observations relative to the first two weeks PF, appear to indicate that all germs derive directly from the pharyngeal epithelium, with the possible exception of I3. This is not always the case for the replacement teeth (C. Van der heyden and A. Huysseune, unpublished observations).

Initiation In contrast to the situation in cichlids, slight polarization of the cells of the basal layer of the presumptive dental epithelium announces imminent tooth germ formation. Presumptive odontogenic mesenchyme cannot, however, be distinguished from non-odontogenic mesenchyme. At e.g., 2d PF, the entire mesenchyme in the pharyngeal region is literally crowded with cells. In later stages, only a few mesenchymal cells face the initial epithelial down-

Cytodifferentiation Zebrafish teeth, from the first generation on, are covered with a hypermineralized substance, presumably enameloid. The unmineralized precursor of enameloid is apparently indistinguishable from early dentine since in no case have we been able to identify a distinctive matrix prior to attachment and eruption. In fact, the enameloid

303

stands out clearly only once it is mineralized. This contrasts with later tooth generations in the zebrafish where unmineralized enameloid can be identified at the tip of early differentiating tooth germs (C. Van der heyden and A. Huysseune, unpublished observations). The dentine of first-generation teeth lacks dentinal tubules. Such a feature has repeatedly been described in teleosts, where adult teeth do contain tubules (e.g. callichthyids: Huysseune and Sire 1997a; cichlids: Huysseune and Sire 1997b) and this condition has been related to the minute thickness of the dentinal walls. We have considered it as an ontogenetic precursor of orthodentine (in the sense of Ørvig 1951, 1967; Peyer 1968; Carlson 1990). This feature deserves further attention, e.g. whether there is a relation to the number of odontoblasts in the pulp cavity and whether the outer layer of first deposited dentine in larger replacement teeth is also devoid of dentinal tubules. In a similar way, the lack of capillaries and nerves in the pulp cavity should be examined in view of the size of the latter. Major changes have been observed in the pulp around the time of completion of the tooth. Teeth that have just realized their attachment have a highly cellular content whereas the pulp of well-established, functional teeth presents a loose aspect with stellate-shaped cells. Most likely these differences reflect the changes that appear to take place along the pulpal wall after attachment, and which are suggestive of a remodeling process. The observations suggest that the dentine wall appears to go through a phase of resorption, followed by a phase of new deposition. Whether or not odontoblasts modulate through cycles of resorptive and secretory activity, or whether cells are recruited from outside the pulp to achieve at least some phase in the process, is unknown at present. To our knowledge, this kind of remodeling has not been described before in any polyphyodont dentition. In mammals, physiological tooth resorption is a common process in deciduous teeth prior to sheding and is carried out by mononuclear (Domon et al. 1994) and multinucleated osteoclasts – also called odontoclasts (Sasaki et al. 1989; Sahara et al. 1992, 1994, 1996) presumably deriving from circulating progenitor cells (Sahara et al. 1996). Resorbed dentine has been reported to be repaired by a cementum-like deposition (Sahara et al. 1992). On the other hand, primary dentinogenesis in mammals (which proceeds until the completion of root formation) is followed by a secondary deposition, not unlike primary dentine (Linde and Goldberg 1993; Smith et al. 1995). Tertiary dentine (either reactive or reparative, depending on the source of odontoblasts participating in the process) has been described in mammals as occurring at specific loci of the pulp-dentine interface in response to environmental stimuli (Smith et al. 1995). In order to define the later deposit in zebrafish either as secondary (and thus part of the normal tooth formation process) or as tertiary, requires a profound knowledge of the cellular events in the pulp and the extent of the process throughout first- and later-generation teeth.

Attachment and resorption Teeth commence their attachment bone formation very soon, in contrast to cichlid teeth, which are well formed (mineralized dentine, mineralized enameloid) before a circular collar of attachment bone is deposited. This is probably in congruence with the fairly synchronous overall progression of differentiation of first-generation teeth in the zebrafish, with little or no proximo-distal gradient. In contrast, odontogenesis in the cichlid firstgeneration dentition proceeds according to a strict proximo-distal gradient (Huysseune and Sire 1997b). Also, the attachment bone in the zebrafish larval teeth is continuous, from the start on, with the dentine, without any separation between both. Only the position of the cervical loop and slight differences in fibril orientation make it possible to distinguish the dentine base from the attachment bone. This is no doubt related to the fact that one cell population, and even a single cell, appears to participate in the formation of all three matrices, dentine, attachment bone and supporting bone. Even though this phenomenon may be limited to first-generation teeth, the observation merits further study, especially in the light of a recent hypothesis concerning the separate evolution of teeth and the bones that bear them (Smith and Coates, 1998a, b). Continuity between dentine, bone of attachment and supporting bone has been repeatedly observed in teleosts (e.g. in callichthyids: Huysseune and Sire 1997a, in cichlids: Huysseune and Sire 1997b). In addition, a single tooth apparently can attach simultaneously to either combination of perichondral bone, membrane bone, and adjacent tooth base. According to Fink and Fink (1981) the ankylosed dentition of Cypriniformes is secondarily derived within ostariophysans although fully ankylosed teeth are primitive for actinopterygians (Fink 1981). Eruption of the tooth tip does not occur unless attachment has been realized. However, in contrast to first-generation pharyngeal teeth in cichlids (Huysseune 1983) eruption in the zebrafish usually lags behind attachment. Evidence suggests that eruption is possibly the result of processes occurring in the pharyngeal epithelium rather than of further changes at the level of the tooth itself. Indeed, the size of the tooth, including its attachment bone, does not differ prior to and after eruption, and the bone support is assumed not to change position. The changes in the epithelium are furthermore reminiscent of the formation of epithelial troughs around the developing tooth plates of the permanent dentition in the recent dipnoan Neoceratodus forsteri (Kemp 1995). This does not exclude that, e.g., rotation of the tooth might assist in the eruption process. Superimposition of reconstructions indeed suggests that, prior to complete ankylosis, teeth that initially lie horizontally become slightly erected thereafter. Zebrafish teeth are never associated with resorption of the branchial cartilage in a way such as described for cichlid teeth (Huysseune 1983; Huysseune and Sire 1992).

304

Conclusions The data presented in this paper offer a broad frame in which further, more detailed morphological studies pertaining to specific aspects of odontogenesis can and must be made. In particular, the present description will serve to calibrate in situ hybridisation experiments and to assess the results of ongoing in vitro experiments dealing with odontogenesis in this vertebrate model. &p.2:Acknowledgements Mrs. G. De Wever and Mrs. F. Allizard are gratefully acknowledged for expert technical assistance in preparing the sections. TEM work and preparation of the photographic prints have been carried out at the ‘Centre Interuniversitaire de Microscopie Electronique’ – CIME Jussieu, Paris 6 and 7. Research performed by C. Van der heyden is financed through a specialisation grant of the ‘Vlaams Instituut voor de Bevordering van het Wetenschappelijk-Technologisch Onderzoek in de Industrie (IWT)’. This work has benefitted from a grant of the ‘Bijzonder Onderzoeksfonds’ of the University of Ghent and has furthermore been supported by an exchange program between the ‘Ministerie van de Vlaamse Gemeenschap’ (Belgium) and the ‘Centre National de Recherche Scientifique’ (France) (C94.005). The paper has also benefitted from discussions in the frame of the COST-action B8 on Odontogenesis.

References Berkovitz BKB, Shellis RP (1978) A longitudinal study of tooth succession in piranhas (Pisces: Characidae), with an analysis of the tooth replacement cycle. J Zool 184:545–561 Butler PM (1995) Ontogenetic aspects of dental evolution. Int J Dev Biol 39:25–34 Carlson SJ (1990) Vertebrate dental structures. In: Carter JG (ed) Skeletal biomineralization: patterns, processes and evolutionary trends, vol 1. Van Nostrand, New York, pp 531–556 Casey J, Lawson R (1981) A histological and scanning electron microscope study of the teeth of caecilian amphibians. Arch Oral Biol 26:49–58 Cubbage CC, Mabee PM (1996) Development of the cranium and paired fins in the zebrafish Danio rerio (Ostariophysi, Cyprinidae). J Morphol 229:121–160 Domon T, Sugaya K, Yawaka Y, Osanai M, Hanaizumi Y, Takahashi S, Wakita M (1994) Electron microscopic and histochemical studies of the mononuclear odontoclast of the human. Anat Rec 240:42–51 Driever W, Stemple D, Schier A, Solnica-Krezel L (1994) Zebrafish: genetic tools for studying vertebrate development. Trends Genet 10:152–159 Driever W, Solnica-Krezel L, Schier AF, Neuhauss SCF, Malicki J, Stemple DL, Stainier DYR, Zwartkruis F, Abdelilah S, Rangini Z, Belak J, Boggs CA (1996) Genetic screen for mutations affecting embryogenesis in zebrafish. Development 123:37–46 Ekker M, Akimenko MA (1991) Le poisson zèbre (Danio rerio), un modèle en biologie du développement. Méd Sci 7:553– 560 Fink SV, Fink WL (1981) Interrelationships of the ostariophysan fishes (Teleostei). Zool J Linn Soc 72:297–353 Fink WL (1981) Ontogeny and phylogeny of tooth attachment modes in actinopterygian fishes. J Morphol 167:167–184 Hanken J, Wassersug R (1981) The visible skeleton. A new double-stain technique reveals the nature of the “hard” tissues. Funct Photogr 16:22–26 Huysseune A (1983) Observations on tooth development and implantation in the upper pharyngeal jaws in Astatotilapia elegans (Teleostei, Cichlidae). J Morphol 175:217–234 Huysseune A (1989a) Development of pharyngeal replacement teeth in a cichlid. Progr Zool 35:456–458

Huysseune A (1989b) Morphogenetic aspects of the pharyngeal jaws and neurocranial apophysis in postembryonic Astatotilapia elegans (Teleostei: Cichlidae). Acad Analecta Brussels 51:11–35 Huysseune A (1990) Development of the anterior part of the mandible and the mandibular dentition in two species of Cichlidae (Teleostei). Cybium 14:327–344 Huysseune A, Sire JY (1992) Bone and cartilage resorption in relation to tooth development in the anterior part of the mandible in cichlid fish: a light and TEM study. Anat Rec 234:1–14 Huysseune A, Sire JY (1997a) Structure and development of teeth in three armoured catfish, Corydoras aeneus, C. arcuatus and Hoplosternum littorale (Siluriformes, Callichthyidae). Acta Zool 78:69–84 Huysseune A, Sire JY (1997b) Structure and development of firstgeneration teeth in the cichlid Hemichromis bimaculatus (Teleostei: Cichlidae). Tissue Cell 29:679–697 Huysseune A, Sire JY (1998) Evolution of patterns and processes in teeth and tooth-related tissues in non-mammalian vertebrates. Eur J Oral Sci 106 [Suppl 1]:437–481 Huysseune A, Aerts P, Verraes W (1989) Pharyngeal tooth replacement in Astatotilapia elegans. Progr Zool 35:480–481 Kallenbach E, Piesco NP (1978) The changing morphology of the epithelium-mesenchyme interface in the differentiation zone of growing teeth of selected vertebrates and its relationship to possible mechanisms of differentiation. J Biol Buccale 6: 229–240 Kemp A (1995) Marginal tooth-bearing bones in the lower jaw of the recent Australian lungfish, Neoceratodus forsteri (Osteichthyes, Dipnoi). J Morphol 225:345–355 Lele Z, Krone PH (1996) The zebrafish as a model system in developmental, toxicological and transgenic research. Biotechnol Adv 14:57–72 Linde A, Goldberg M (1993) Dentinogenesis. Crit Rev Oral Biol Med 4:679–728 Nakajima T (1979) The development and replacement pattern of the pharyngeal dentition in the Japanese Cyprinid fish, Gnathopogon coerulescens. Copeia 1979:22–28 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 of cypriniform dentitions. Copeia 1987:208–213 Nakajima T (1990) Morphogenesis of the pharyngeal teeth in the Japanese Dace, Tribolodon hakonensis (Pisces: Cyprinidae). J Morphol 205:155–163 Nakajima T, Yue P (1989) Development of the pharyngeal teeth in the big head, Aristichthys nobilis (Cyprinidae). Jpn J Ichthyol 36:42–47 Ørvig T (1951) Histologic studies of Placoderms and fossil Elasmobranchs. I. The endoskeleton, with remarks on the hard tissues of lower vertebrates in general. Ark Zool 2:321–454 Ørvig T (1967) Phylogeny of tooth tissues: evolution of some calcified tissues in early vertebrates. In: Miles AEW (ed) Structural and chemical organization of teeth, vol 1. Academic Press, London, pp 45–110 Osborn JW (1971) The ontogeny of tooth succession in Lacerta vivipara Jacquin (1787). Proc R Soc Lond B Biol Sci 179: 261–289 Osborn JW (1978) Morphogenetic gradients: fields versus clones. In: Butler PM, Joysey KA (eds) Development, function and evolution of teeth. Academic Press, London, pp 171–201 Peyer B (1968) Comparative odontology. University Press, Chicago, 347 pp Reif WE (1982) Evolution of dermal skeleton and dentition in vertebrates. The odontode regulation theory. Evol Biol 15: 287–368 Sahara N, Okafuji N, Toyoki A, Suzuki I, Deguchi T, Suzuki K (1992) Odontoclastic resorption at the pulpal surface of coronal dentin prior to the shedding of human deciduous teeth. Arch Histol Cytol 55:273–285

305 Sahara N, Okafuji N, Toyoki A, Ashizawa Y, Deguchi T, Suzuki K (1994) Odontoclastic resorption of the superficial nonmineralized layer of predentine in the shedding of human deciduous teeth. Cell Tissue Res 277:19–26 Sahara N, Toyoki A, Ashizawa Y, Deguchi T, Suzuki K (1996) Cytodifferentiation of the odontoclast prior to the shedding of human deciduous teeth: an ultrastructural and cytochemical study. Anat Rec 244:33–49 Sasaki T, Shimizu T, Suzuki H, Watanabe C (1989) Cytodifferentiation and degeneration of odontoclasts in physiologic root resorption of kitten deciduous teeth. Acta Anat 135:330– 340 Schilling TF, Piotrowski T, Grandel H, Brand M, Heisenberg CP, Jiang YJ, Beuchle D, Hammerschmidt M, Kane DA, Mullins MC, Eeden FJM van, Kelsh RN, Furutani-Seiki M, Granato M, Haffter P, Ödenthal J, Warga RM, Trowe T, Nüsslein-Volhard C (1996) Jaw and branchial arch mutants in zebrafish I: branchial arches. Development 123:329–344

Sire JY, Allizard F, Babiar O, Bourguignon J, Quilhac A (1997) Scale development in zebrafish (Danio rerio). J Anat 190: 545–561 Smith AJ, Cassidy N, Perry H, Bègue-Kirn C, Ruch JV, Lesot H (1995) Reactionary dentinogenesis. Int J Dev Biol 39:273–280 Smith MM, Coates M (1998a) Evolutionary origins of teeth and jaws: developmental models and phylogenetic patterns. In: Teaford M, Ferguson M, Smith MM (eds) Teeth: development, function and evolution. Cambridge University Press, Cambridge (in press) Smith MM, Coates MI (1998b) Evolutionary origins of the vertebrate dentition: phylogenetic patterns and developmental evolution. Eur J Oral Sci 106 [Suppl 1]:482–500 Thesleff I, Hurmerinta K (1981) Tissue interactions in tooth development. Differentiation 18:75–88 Thesleff I, Vaahtokari A, Partanen AM (1995) Regulation of organogenesis. Common molecular mechanisms regulating the development of teeth and other organs. Int J Dev Biol 39:35–50