Development, function and evolution of teeth mf teaford, mm smith, mwj ferguson
Development, Function and Evolution of TeethOver the past 20 years there has been an explosion ofinformation generated by scienti®c research. One of thebene®ciaries of this has been the study of morphology,where new techniques and analyses have led to insightsinto a wide range of topics. Advances in genetics,histology, microstructure, biomechanics andmorphometrics have allowed researchers to view teethfrom new perspectives. However, up to now there hasbeen little communication between researchers in thedifferent ®elds of dental research. This book bringstogether for the ®rst time overviews of a wide range ofdental topics, linking genes, molecules anddevelopmental mechanisms within an evolutionaryframework. Written by leading experts in the ®eld, thisbook will stimulate co-operative research in ®elds asdiverse as palaeontology, molecular biology,developmental biology and functional morphology.M A R K F. T E A F O R D is Professor of Anatomy at the JohnsHopkins University School of Medicine.M O YA M E R E D I T H S M I T H is Professor of EvolutionaryDentoskeletal Biology at the University of London, and isbased at the Dental Institute of Kings College London,Guys Campus.M A R K W . J . F E R G U S O N is Professor of Basic DentalSciences in the School of Biological Sciences at theUniversity of Manchester.
Development, Function andEvolution of Teeth Edited by Mark F. Teaford, Moya Meredith Smith and Mark W. J. Ferguson
ContentsList of contributors page viiAcknowledgements ixPart one Genes, molecules and tooth initiation 1 1 Homeobox genes in initiation and shape of teeth during development in mammalian embryos 3 P. T. Sharpe 2 Return of lost structure in the developmental control of tooth shape J. Jernvall and I. Thesleff 13 3 Molecules implicated in odontoblast terminal differentiation and dentinogenesis J. V. Ruch and H. Lesot 22 4 Enamel biomineralization: the assembly and disassembly of the protein extracellular organic matrix 37 A. G. Fincham, W. Luo, J. Moradian-Oldak, M. L. Paine, M. L. Snead and M. Zeichner-DavidPart two Tooth tissues: development and evolution 63 5 Evolutionary origins of dentine in the fossil record of early vertebrates: diversity, development and 65 function M. M. Smith and I. J. Sansom 6 Pulpo-dentinal interactions in development and repair of dentine A. J. Smith 82 7 Prismless enamel in amniotes: terminology, function, and evolution P. M. Sander 92 8 Two different strategies in enamel differentiation: Marsupialia versus Eutheria W. von Koenigswald 107 9 Incremental markings in enamel and dentine: what they can tell us about the way teeth grow M. C. Dean 119 v
vi ContentsPart three Evolution of tooth shape and dentition 13110 Evolutionary origins of teeth and jaws: developmental models and phylogenetic patterns 133 M. M. Smith and M. I. Coates11 Development and evolution of dentition patterns and their genetic basis 152 Z. Zhao, K. M. Weiss and D. W. Stock12 Evolution of tooth attachment in lower vertebrates to tetrapods P. Gaengler 17313 Tooth replacement patterns in non-mammalian vertebrates B. K. Berkovitz 18614 The evolution of tooth shape and tooth function in primates P. M. Butler 20115 `Schultzs Rule and the evolution of tooth emergence and replacement patterns in primates and 212 ungulates B. H. SmithPart four Macrostructure and function 22916 Developmental plasticity in the dentition of a heterodont polphyodont ®sh species A. Huysseune 23117 Enamel microporosity and its functional implications R. P. Shellis and G. H. Dibdin 24218 Pathways to functional differentiation in mammalian enamel J. M. Rensberger 25219 Trends in the evolution of molar crown types in ungulate mammals: evidence from the northern 269 hemisphere J. Jernvall, J. P. Hunter and M. Fortelius20 Function of postcanine tooth crown shape in mammals P. W. Lucas and C. R. Peters 28221 Primate dental functional morphology revisited M. F. Teaford 290Index 305
ContributorsBarry K. Berkowitz GKT School of Biomedical Science Ann Huysseune Instituut voor Dierkunde, University ofHenriette Raphael House, Guys Campus, London Bridge, Ghent, Ledeganckstraat 35, B-9000 Gent, BelgiumLondon SE1 1UL, UK Jukka Jernvall Institute of Biotechnology, PO Box 56,P. M. Butler 23 Mandeville Court, Strode Street, Egham, 00014 University of Helsinki, Finland and Department ofSurrey TW20 9BU, UK Anthropology, State University of New York at Stony Brook, Stony Brook, NY 11794, USAMike I. Coates Biological Sciences, University CollegeLondon, Gower Street, London WC1E 6BT, UK È Wighart von Koenigswald Institut fur Palaontologie, È Universitat Bonn, Nussallee 8, Bonn D-53115, Germany ÈM. C. Dean Evolutionary Anatomy Unit, Department ofAnatomy and Developmental Biology, University College Â H. Lesot Institute de Biologie Medical Universite LouisLondon, Gower Street, London WC1E 6BT, UK Â Pasteur, Strasbourg Faculte de Medicine, 11 Rue Humann, 67085 Strasbourg, FranceGeorge H. Dibdin MRC Dental Group, Dental School,University of Bristol, Lower Maudlin Street, Bristol Peter W. Lucas Department of Anatomy, University ofBS1 2LY, UK Hong Kong, Li Shu Fan Building, 5 Sassoon Road, Hong KongAlan G. Fincham Center for Cranofacial MolecularBiology, The University of Southern California, Wen Luo Center for Craniofacial Molecular Biology, TheSchool of Dentistry, 2250 Alcazar Street, Los Angeles, University of Southern California, School of Dentistry,CA, USA 2250 Alcazar Street, Los Angeles, CA, USAM. Fortelius Division of Geology and Palaeontology, Janet Moradian-Oldak Center for Craniofacial MolecularDepartment of Geology, University of Helsinki, PO Box Biology, The University of Southern California, School of11, FIN-00014 University of Helsinki, Finland. Dentistry, 2250 Alcazar Street, Los Angeles, CA, USAPeter Gaengler School of Dental Medicine, Conservative Michael L. Paine Center for Craniofacial MolecularDentistry and Periodontology, University of Witten- Biology, The University of Southern California, School ofHerdecke, D-58448 Witten, Germany Dentistry, 2250 Alcazar Street, Los Angeles, CA, USAJ. P. Hunter Department of Anatomical Sciences, State Charles R. Peters Department of Anthropology andUniversity of New York at Stony Brook, Stony Brook, NY Institute of Ecology, University of Georgia, Athens,11794±8081, USA Georgia 30602, USA John M. Rensberger Burke Museum of Natural and vii
viii List of contributorsCultural History and Department of Geological Sciences, Malcolm L. Snead Center for Craniofacial MolecularUniversity of Washington, Seattle, WA 98195, USA Biology, The University of Southern California, School of Dentistry, 2250 Alcazar Street, Los Angeles, CA, USA ÂJ. V. Ruch Institute de Biologie Medical Universite Louis ÂPasteur, Strasbourg Faculte de Medicine, 11 Rue David W. Stock Department of Anthropology,Humann, 67085 Strasbourg, France Pennsylvania State University, University Park, PA 16802, USA. Present Address: Department of Environmental, ÈP. Martin Sander Institut fur Palaontologie, Universitat È È Population and Organismal Biology, University ofBonn, Nussallee 8, Bonn D-53115, Germany Colorado, Boulder, CO80309-0334, USAIvan J. Sansom School of Earth Sciences, University of Mark F. Teaford Department of Cell Biology andBirmingham, Edgbaston, Birmingham B15 2TT, UK Anatomy, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USAPaul T. Sharpe Department of Craniofacial Development,UMDS, Guys Hospital, London Bridge, London SE1 9RT, Irma Thesleff Institute of Biotechnology, PO Box 56,UK 00014 University of Helsinki, FinlandR. Peter Shellis Division of Restorative Dentistry, Dental Kenneth M. Weiss Department of Anthropology,School, University of Bristol, Lower Maudlin Street, Pennsylvania State University, University Park, PA 16802,Bristol BS1 2LY, UK USA.B. Holly Smith Museum of Anthropology, 1109 Geddes, Margarita Zeichner-David Center for CraniofacialUniversity of Michigan, Ann Arbor, MI 48109, USA Molecular Biology, The University of Southern California, School of Dentistry, 2250 Alcazar Street, Los Angeles, CA,Moya Meredith Smith Department of Craniofacial USADevelopment, Dental Institute, Kings College London,Guys Campus, London Bridge, London SE1 9RT, UK Zhiyong Zhao Department of Anthropology, Pennsylvania State University, University Park,A. J. (Tony) Smith School of Dentistry, Oral Biology, PA 16802, USA Present address: Department ofUniversity of Birmingham, St Chads Queensway, Pediatrics, Yale University, New Haven, CT 06520,Birmingham B4 6NN, UK USA
AcknowledgementsThis volume is a product of a symposium entitled `Teeth:homeoboxes to function held at the 4th InternationalCongress of Vertebrate Morphology, at the University ofChicago in 1994. We wish to offer our sincere thanks toSue Herring and Jim Hanken for inviting us to organize asession for the Congress. Robin Smith, from CambridgeUniversity Press, provided enthusiastic guidance for thisvolume from the outset. Upon his leaving CambridgePress, Tracey Sanderson skilfully (and patiently!) guidedit through to completion. As is so often the case with avolume like this, many other people helped at variouspoints along the way, so many that it is impossible tothank them all here. However, special thanks must go toHoward Farrell and Jayne Aldhouse for their expert edi-torial help; to innumerable reviewers for their expertinsights and suggestions; and most of all, to each of thecontributors, whose efforts and patience helped bringthis volume successfully through its long gestation! The original symposium would not have taken placewithout support from the Johns Hopkins University De-partment of Cell Biology & Anatomy and NIH support forthe ICVM. Mark Teaford would also like to acknowledgethe continued support of the National Science Founda-tion for his work on teeth. Moya Meredith Smith wouldalso like to acknowledge numerous grants from theRoyal Society for travel to Europe, USA and Australia andto exchange information with her colleagues in palaeon-tology, in particular for exchange visits to Lithuania andcollaboration with Valia Karatjute-Talimaa. She wouldalso like to thank the Dental School at Guys Hospital andNERC for grants to pursue this work (GR/38543; GR3/10272), and to the Dean, Professor Ashley for his supportof her research. Finally, we could not have ®nished this projectwithout continued encouragement, understanding, andinsight from our families. ix
1 Homeobox genes in initiation and shape of teeth during development in mammalian embryos P. T. Sharpe1.1. Introduction discovered and the one which has produced the greatest interest is that of the homeobox genes. The discovery ofThe past decade has seen remarkable advances in our the homeobox as a small (180 bp) conserved region of DNAunderstanding of the genetic control of embryonic devel- found in homeotic genes of ¯ies (Drosophila) provided theopment. We now know that developmental processes are springboard for all subsequent advances in this ®eldinitiated and controlled by interacting pathways of extra- (reviewed in Duboule, 1994). The discovery of thecellular signalling molecules, receptors, intracellular sig- homeobox was an important milestone because it demon-nalling proteins and nuclear (transcription) factors. The strated that genes controlling ¯y development can also bedifferent types of proteins (genes) that carry out these present in vertebrates. This allowed progress in the under-functions most often occur as members of families of standing of vertebrate development to proceed at a farrelated proteins characterised by possessing conserved greater pace than was ever thought possible by cloningamino acid motifs but which do not necessarily have genes through nucleotide homology with ¯y genes.similar functions in development. Thus for example, the Homeotic mutations in ¯ies had long been suspectedtransforming growth factor-beta (TGFb) superfamily of of holding important clues to understanding morpho-secreted signalling proteins consists of a large family of genesis. Cells in an embryo differentiate into a limitedproteins that share some homology with TGFb, and in number of specialised types, around 200 different cellmany cases share cell surface receptors (Kingsley, 1994). types in mammals, and it is the arrangements of theseHowever, within this family different members have very cell types into de®ned structures with characteristicdifferent and speci®c functions in development. The forms that is the main achievement of embryogenesis.bone morphogenetic protein Bmp-4, for example, prob- The genetic control of morphogenesis of different struc-ably has multiple functions as a signalling molecule in tures was a mystery until molecular cloning becameembryogenesis, including a role in lung morphogenesis, possible, followed by analysis of the expression, functionbut targeted mutation of Bmp-4 (gene knock-out) shows a and control of homeotic genes.requirement for this protein for early mesoderm for- Homeotic mutations involve mutations in singlemation (Winnier et al., 1995; Bellusci et al.,1996). Bmp-7 genes that produce a phenotype where one body part ofon the other hand appears to have no direct role in the ¯y is replaced by another. Thus the Antennapediamesoderm formation but is required for skeletal develop- mutation is characterised by the development of legs inment (Luo et al., 1995). This illustrates a recurring theme place of antennae on the ¯ys head. Eight of the homeoticin development where similar molecules have multiple genes identi®ed in ¯ies were cloned and found to befunctions, some of which overlap with other family different except for one small 180 bp sequence towardsmembers whereas others are unique. This almost cer- the C terminus that was highly conserved. This sequence,tainly re¯ects the evolution of these gene families by or `box, was named the homeobox and was also found togene duplication resulting in some shared and some be present in other developmental genes in ¯ies, mostunique functions. notably the segmentation genes fushi tarazu and engrailed The ®rst of these families of developmental genes to be (reviewed in Duboule, 1994). 3
4 P. T. Sharpe The clues to the function of the homeobox came from teeth are formed from neural crest cells that migratestructural analysis that showed that the homeodomain from the rostral hindbrain (rhombomeres 1 and 2) and(60 amino acids) forms a helix-turn-helix structure which caudal midbrain. The anterior boundaries of expressionhas DNA binding characteristics (Laughon and Scott, of the most 3 Hox genes have been shown to correspond1984, Sherperd et al., 1984). Homeoproteins are thus DNA- to rhombomeric boundaries in the hindbrain and morebinding proteins that regulate gene transcription and as signi®cantly to be expressed in migrating neural crestsuch may exert a hierarchical control function over the cells at the appropriate axial level. Hoxb-2 expression, forexpression of genes required for morphogenesis of a example, has an anterior boundary at rhombomere 3/4particular structure. and is also expressed in neural crest cells that migrate The conservation of the homeobox sequence is not from rhombomere 4 to populate the second branchiallimited to ¯ies, and since 1984 many vertebrate homeo- arch (Prince and Lumsden, 1994). This feature has beenbox genes have been cloned by nucleotide homology. suggested as a mechanism of patterning the mesench-However, there is a clear and important distinction ymal cells of the branchial arches by positional speci®ca-between homeobox genes that are most closely related to tion of neural crest cells prior to their migration throughhomeotic genes and others that are not. Homeotic genes the combination of Hox genes they express, referred to ashave unique organisational and functional characteris- the `branchial Hox code.tics in ¯ies that distinguish them from other homeobox Since teeth in the ®rst branchial arch were ®rstgenes. Most signi®cantly eight of the homeotic genes are thought to develop from neural crest cells contributedclustered forming a complex (HOMC) and their linear from rhombomeres 1 and 2 that populate this arch, andorder in the clusters is re¯ected in their anterior±poster- the neuroepithelium of the second arch expresses Hoxa-2,ior expression domains in ¯y embryos, a feature known it was considered possible that tooth morphogenesis isas colinearity. Similarly the mammalian homeobox patterned by the same branchial Hox code. However, itgenes that most closely resemble homeotic genes, called appears that this is not so, since rhombomere 1 does notHox genes, also show the same feature of colinearity. In express a Hox gene and although Hoxa-2 is expressed inmammals there are 39 Hox genes arranged as four clus- rhombomere 2 it is unique in that this expression is notters on different chromosomes. Hox genes and HOMC transferred to neural crest cells that migrate from thisgenes probably had a single common evolutionary rhombomere. For both these reasons, expression of Hoxancestor in a segmented worm and the Hox genes were genes has not been observed in the ®rst branchial archduplicated during evolution of vertebrates (reviewed in mesenchyme and so patterning of the ®rst arch struc-Manak and Scott, 1994). tures cannot directly involve Hox genes (see also Chapter Hox gene expression is ®rst detected in ectoderm and 11). Also, we now know that, in the mouse it is the non-mesoderm cells in mammalian embryos during gastrula- segmental posterior midbrain crest which forms thetion (Gaunt et al., 1986). During organogenesis expression ectomesenchyme of the mandibular molars (Imai et al.,is found in the developing central nervous system and 1996). In the chick, and also the equivalent bones to thosehindbrain, in the developing prevertebrae (somites) and bearing teeth in the mouse, are formed from mid brainmore 5 genes are also expressed in limb buds. Targeted crest (Koentges and Lumsden, 1996). There are howevermutation analysis has con®rmed that Hox genes play a many non-Hox homeobox genes that are expressed in therole in development of the axial skeleton, mutation of ®rst branchial arch and also during tooth development.individual genes results in abnormal development of the This chapter describes these expression patterns andaxial skeleton at particular anterior±posterior levels. proposes how some of these genes might function toThus for example a null mutation of Hoxb-4 results in control tooth morphogenesis via an `odontogenicpartial homeotic transformation of the second cervical homeobox code (Sharpe, 1995).vertebra into a ®rst cervical vertebra, whereas mutationof Hoxc-8 produces a change in a more posterior vertebra(LeMouellic et al., 1992; Ramirez-Solis et al., 1993). A potentially important feature of Hox gene expres- 1.2. Homeobox genes and tooth budsion for tooth development is the expression of the most initiation3 Hox genes in cranial neural crest cells emanating fromthe developing hindbrain. Neural crest cells that form The ®rst morphological sign of tooth development is athe branchial arches of the embryo migrate from distinct narrow band of thickened epithelium (primary epithelialanterior±posterior positions in the hindbrain and caudal band) on the developing mandible and maxilla thatmidbrain. Thus the cells of the ®rst branchial arch which forms four zones, one in each quadrant. These bandsform the ectomesenchymal component of developing specify the area of epithelium from which teeth are
Homeobox genes in tooth development 5 Figure 1.1. In situ hybridisation of a section in a dorsoventral plane with anterior at the top, of a day 10.5 gestation mouse embryo head showing Msx-2 gene expression in the ectomesenchyme and oral epithelium in relation to the epithelial thickening. Left, dark ®eld, right, light ®eld. Section is sagittal (parallel to A/P axis, anterior to the top). Line, middle extent of epithelial band; arrowhead, lingual extent of epithelial band. Magni®cation6200. (Courtesy of Bethan Thomas.)capable of forming. The position of the bands thus deter- distinct early expression domains of Msx-1 and -2 inmines and restricts the location within the mandible and epithelium and mesenchyme suggest a possible role formaxilla of tooth development. Tooth buds form at dis- these genes in initiation of the primary epithelial bandcreet locations in these bands by secondary thickening of (Sharpe, 1995).the epithelium and invagination into the underlying Experimental evidence for a possible role of theseectomesenchyme. The ®rst important question in tooth genes in initiation of tooth development has come fromdevelopment, therefore, is which cells provide the infor- in vivo experiments using targeted mutagenesis and alsomation that speci®es the position of the oral epithelial in vitro experiments involving explant cultures of earlythickenings? Although results from recombination ex- tooth germs. Targeted mutation of the Msx-1 gene resultsperiments have largely supported the epithelium as the in development of all teeth being arrested at the earlysource of the initiation signals there is still some doubt bud stage (Satokata and Maas, 1994). Similarly mutationsas to whether it is the underlying neural crest-derived in MSX1 have been shown to be associated with toothectomesenchymal cells that are patterning the epithelial agenesis in humans (Vastardis et al., 1996). Since Msx-1 iscells (Mina and Kollar, 1987; Lumsden, 1988; and re- expressed at high levels in the condensing mesenchymeviewed by Ruch, 1995). at the bud stage, this suggests that Msx-1 is required for a One of the interesting observations from the localisa- signalling pathway from bud mesenchyme to dentaltion of homeobox genes expressed in tooth development epithelium in tooth histogenesis but not initiation of theis that for many genes, expression is not restricted to tooth bud. The recombination experiments have showneither the dental epithelium or mesenchymal cells. This that following an initial signal (possibly Bmp-4 or Shh),is particularly evident for the Msx-2 homeobox gene, from the thickened epithelium to the underlying me-which shows spatially restricted expression in both senchyme, the direction of communication is then re-epithelial and mesenchymal cells during tooth develop- versed and signals pass from the condensingment (Figure 1.1; MacKenzie et al.,1992). More signi®- mesenchyme to the epithelial bud (Figure 1.2A). Thecantly, the early expression of Msx-2 prior to tooth bud toothless phenotype of the Msx-1 mutants implies thatformation is also found in epithelium and mesenchyme. Msx-1 regulates the expression of these signalling mole-Msx-2 is expressed in the distal tips of the mandibular cules. A possible candidate signalling molecule is Bmp-4,and maxillary arch mesenchyme. Msx-1, a close relative of and in fact, the levels of Bmp-4 expression in tooth budMsx-2, is expressed in a similar domain of ectome- mesenchyme of Msx-1 mutants is reduced. Moreover,senchyme as Msx-2 but extends slightly more proximal tooth development in Msx-1 mutants can be rescued bythen Msx-2 (MacKenzie et al., 1992). Msx-1 is not expressed the addition of beads coated in Bmp-4; thus there is ain the oral epithelium (MacKenzie et al., 1991). These strong case for Msx-1 regulating Bmp-4 in condensing
6 P. T. Sharpe Figure 1.2. A. Proposed signalling pathways involved in interactions between primary odontogenic epithelium thickening and ectomesenchyme at the epithelial band stage. B. Gene expression in the enamel knot precursor cells induced at the early bud stage by signals from the mesenchyme which may include Bmp-4.mesenchyme (Chen et al., 1996). Although the Msx-1 the HMG-box family of DNA-binding proteins, of whichmutants give a clear tooth phenotype, the arrest does not the best known is the mammalian sex-determining geneoccur at the initiation stage, i.e. tooth buds are produced. SRY. Lef-1 is very closely related to another protein, Tcf-1,The earliest requirement for Msx-1 would thus appear to and these genes probably appeared by duplication of abe for mesenchymal to epithelial signalling at the bud single ancestor gene (Gastrop, et al., 1992). Lef-1 and Tcf-1stage. While it is possible that the targeted mutation is show a very similar pattern of expression during toothnot a complete null (only the third helix is deleted in the development, and both are expressed in T-lymphocytes.mutant allele) it is more likely that Msx-2 and/or Dlx-2 The function of both genes has been studied by targetedmay compensate for loss of Msx-1 in tooth initiation. Msx- gene disruption where, surprisingly, the phenotypes pro-2 targeted mutants have been generated and appear to duced are quite different. Lef-1 mutant mice have toothhave normal early tooth development. Signi®cantly, development arrested at the bud stage (similar to thehowever, tooth development in Msx-1/Msx-2 double Msx-1 -/- mice) and show no major defects in T-lympho-mutants is reported as being arrested earlier than the cytes (van Genderen et al., 1994). Tcf-1 mutants havetooth bud stage and initiation may not occur at all in the normal tooth development but severely impaired T-cellabsence of both genes (Maas et al., 1996). Thus, there is function (Verbeek et al., 1995). One possible hypothesis isstrong in vivo data that supports the role of these genes in that the archetypal gene is expressed and required forinitiation. The possible role of Dlx-2 in tooth development tooth development and that Tcf-1, duplicated from Lef-1is discussed below. and acquired a novel function in T-cells but which is not Lef-1 is a transcription factor which is a member of required for tooth development. Lef-1 is expressed in
Homeobox genes in tooth development 7secondary epithelial thickenings and in condensing me-senchyme of the tooth bud in a similar domain to Msx-1.Because tooth development in Lef-1 mutants is arrested atthe bud stage, it was originally believed that Lef-1 wouldbe required for initiating the signalling pathway of mole-cules from condensing mesenchyme to the epithelium ofthe tooth bud similar to Msx-1. However, detailed recom-bination experiments using Lef-1 mutant and wild-typeepithelium and mesenchyme have shown that the Lef-1 isrequired in the early thickened epithelium for toothdevelopment (Kratochwil et al., 1996). Thus a recombina-tion of Lef-1 mutant epithelium with wild-type me-senchyme does not give tooth development, whereas thereciprocal recombination of mutant mesenchyme andwild-type oral epithelium allows normal tooth develop-ment in vitro. Moreover, the requirement for Lef-1 in thethickened epithelium is transient, since mutant me-senchyme, cultured with wild-type epithelium and thendissociated and recultured with mutant epithelium,gives rise to normal tooth development, indicating thatthe signals regulated by Lef-1 have been initiated in the Figure 1.3. Whole mount in situ hybridisation of a day-13.5 mandiblemutant mesenchyme by the wild-type epithelium. Inter- showing expression of Shh in the epithelium of the developing incisor (arrow) and molar (arrowhead) tooth germs viewed from theestingly Tcf-1 expression is barely detectable in epithe- oral aspect.lium but mesenchymal expression overlaps that of Lef-1indicating why Tcf-1 may not be able to compensate forloss of Lef-1 function in tooth bud development. The important yet undetermined role for this pathway inpossibility that expression of Lef-1 in epithelium may be odontogenesis (Figure 1.3) (Bitgood and McMahon, 1995;suf®cient to induce tooth development has come from Koyama et al., 1996; Sharpe, unpublished).ectopic expression experiments in transgenic mice usinga keratin 14 promoter which produced ectopic teeth(Zhou et al., 1995). The genetic pathway in which Lef-1 belongs is starting 1.3. Patterning of tooth position and shapeto be elucidated, providing important insights into thecontrol of bud stage tooth development. Lef-1 interacts Mammalian teeth have characteristic shapes for eachwith b-catenin in the cell cytoplasm, resulting in trans- position in the jaws. The shape and position are impor-port of the Lef-1 protein into the nucleus. b-catenin is a tant for dietary requirements and have evolved and diver-component of the cell adhesion molecule E-cadherin, the si®ed for particular specialised feeding functions.expression of which is localised to tooth epithelial thick- Incisors are conical, or chisel-shaped and located at theenings (Kemler, 1993, Behrens et al., 1996, Huber et al., front of the jaws, where they are used not only for1996). b-catenin has been shown to be downstream of obtaining and cutting food, but also for grooming orWnt signalling in Xenopus and since some Wnt genes co- defence functions. Molars are triangular, rectangularlocalise with Lef-1 and E-cadherin in tooth epithelial and multicuspid in shape and are located towards thethickenings (A. McMahon, personal communication) a back of the jaws aand are for processing food, either bypathway involving Wnt regulation of Lef-1 expression cutting, grinding or crushing. Variations on these basicresulting in activation of E-cadherin with b-catenin as an shapes have evolved for specialised diets such as carnas-intermediate may be important in early tooth develop- sial teeth in carnivores which are a sectorial adaptationment (Yost et al., 1996; Molenaar et al., 1996; Luning et al., of tritubercular molars. The correct pattern of the denti-1994). tion is thus essential for animal survival. The dentition of The localisation of other signalling molecules and any animal species is as unique as its DNA and sincetranscription factors in tooth bud epithelial thickenings patterns are inherited the developmental mechanismsdemonstrates that the Lef-1 pathway is not the only one that direct pattern formation must be geneticallyof importance. The high levels of Shh gene expression in controlled.secondary tooth bud epithelial thickenings suggest an The importance of Hox genes in development of the
8 P. T. Sharpe Figure 1.4. Simpli®ed model of the `odontogenic homeobox gene code of dental patterning based on the expression of Msx-1, Msx-2, Dlx-1and Dlx-2 in the developing mandible. Overlapping domains of homeobox gene expression in the ectomesenchyme of the mandibular process can specify the shape of the tooth that develops at a particular position. Oral aspect, anterior to the top; I, incisor ®eld; M, molar ®eld.axial skeleton has been discussed. But it is possible that and Msx-2, which are predominantly expressed in distalhomeobox gene expression in the ®rst branchial arch may mesenchyme and not in proximal areas. In order to testbe involved in patterning the development of ®rst bran- this model, mice embryos with targeted disruption of thechial arch structures such as teeth. A molecular model of Dlx-1 and Dlx-2 genes using homologous recombination inpatterning the dentition, based on the expression of ES cells have been analysed. Mice with null mutations inseveral homeobox genes in neural crest-derived ectome- either gene have normal tooth development. It is onlysenchyme has been proposed and termed the `odonto- when double mutations are created, i.e. mice with nullgenic homeobox gene code (Sharpe, 1995). A simpli®ed mutations in both genes, that tooth development isform of this model is illustrated in Figure 1.4 where the affected. Dlx-1/ -2 -/- embryos have normal upper and lowerexpression of homeobox genes, such as Dlx-1, Dlx-2, Msx-1 incisors and lower molars but do not develop any upperand Msx-2 are proposed to specify the development of tooth molars (Qiu et al., 1997; Thomas et al., 1997).germs into either molars or incisors. Dlx-1 and Dlx-2 are Since the odontogenic mesenchyme expressing thesehomeobox genes belonging to a family of seven or eight homeobox genes is neural crest derived, this model pre-genes related to the Drosophila Distalless gene which is dicts that populations of cranial neural crest cells areinvolved in appendage development. Dlx-1 and Dlx-2 are speci®ed as odontogenic and further regionally speci®edlocated within 10 kb of each other on mouse chromosome as maxilla/mandible/molar/incisor. The failure of maxil-2 and their expression in mandibular and maxillary lary molar teeth to develop in the Dlx-1/-2 double mutantmesenchyme is almost indistinguishable (Bulfone et al., embryos thus supports the odontogenic homeobox code1993). Both genes are predominately expressed in the model for patterning of maxillary molar tooth develop-proximal ectomesenchyme of the mandible and maxilla ment and suggests that Dlx-1 and Dlx-2 are required forin the area where molars will develop prior to the start of the speci®cation of a subpopulation (maxillary molar) oftooth development (E10). Expression is absent in more odontogenic neural crest cells. The normal developmentdistal mesenchyme where incisors will develop. The ex- of mandibular molars implies that tooth patterning inpression of Dlx-1 and Dlx-2 is complemented by that of Msx-1 the lower and upper jaws is controlled independently, a
Homeobox genes in tooth development 9feature that was not originally predicted by the model produce cusps (morphogenesis). The physical processesbut one that has interesting implications for mechan- that direct cusp development have classically been sug-isms of evolution of dental patterns. gested to involve the differentiation of the stellate reti- Goosecoid (Gsc) was originally envisaged as part of the culum and differential cessation of mitosis in the dentalodontogenic homeobox code but has since been shown to epithelium. The importance of differentiated transientbe important for mandibular skeletal, but not tooth, epithelial structures, enamel knots, has recently gaineddevelopment. Targeted null mutations in the Gsc gene signi®cance. The enamel knot was originally identi®edproduce craniofacial defects that resemble ®rst arch syn- in the 1920s in molar cap stage tooth germs as a tran-drome in humans, but the development of teeth in these sient, small group of epithelial cells immediately abovemice is normal (Rivera-Perez et al., 1995; Yamada et al., the condensing mesenchyme forming the dental papilla.1995). A role for the enamel knot in cusp formation was pro- Gsc has been shown to be upregulated in Xenopus posed by Orban (1928) and Butler (1956) who suggestedmesoderm formation by the secreted signalling protein that enamel knot cells act as a local restraint causingactivin. Activin is a member of the TGFb superfamily of post-mitotic internal enamel epithelium to in¯ect at thegrowth factor-like signalling molecules that have wide- site of the future ®rst cusp and the external enamelranging functions in embryogenesis. Targeted null muta- epithelium to dimple as the swelling pressure of thetion of the activin-bA gene in transgenic mice was pre- developing stellate reticulum separated the external anddicted to result in defects in mesoderm formation. internal enamel epithelia everywhere else in the toothHowever, mesoderm formation in activin-bA mutants germ.was found to be normal but signi®cantly, the major Msx-2 was the ®rst gene whose expression was local-phenotype was abnormalities in craniofacial develop- ised to enamel knot cells and it was proposed that Msx-2ment, particular tooth development. Activin-bA -/- mice expression provided a molecular link between toothdevelop no incisor teeth and no mandibular molars but initiation and shape (MacKenzie et al., 1992). Subse-development of maxillary molars is always normal. Thus quently the expression of several more genes has beennull mutations in a molecule (activin-bA) produce a tooth shown to be restricted to enamel knot epithelial cells inpatterning phenotype. In common with the Dlx-1/2 muta- the tooth bud and the origin of enamel knot cells tracedtions, activin-bA is not required for development of all back to a few epithelial cells at the tip of molar toothteeth but only for initiation of incisors and mandibular buds. Expression of the genes for the secreted factorsmolars. Unlike the Dlx-1/2 mutations, activin-bA affects Shh, Bmp-2, Bmp-4, Bmp-7 and Fgf-4 are all localised inincisor tooth development to the same extent in the enamel knot cells (Chapter 2, Figure 1.2B) suggestingmandible and maxilla and this may either indicate that that this structure acts as a signalling centre similar toincisor development is controlled differently to molar both the AER and the zone of polarising activity (ZPA) indevelopment or that activin is involved in regulating limb development (Vaahtokari et al., 1996). Expression ofdifferent pathways in the mandible and maxilla (Matzuk these genes has also identi®ed the formation of sec-et al., 1995). ondary enamel knots in regions corresponding to the Other homeobox genes are expressed during tooth future cusp tips and it is proposed that the ®rst (primary)development and some of these may also contribute to enamel knot acts as a signalling centre to direct sec-the odontogenic homeobox code. Other members of the ondary knot formation which functions to control localDlx family such as Dlx-5 and Dlx-6 are expressed in prox- epithelial cell proliferation rates. Enamel knot cells haveimal regions of the developing mandible but not the a transient existence and rapidly undergo apoptosismaxilla and may form part of the code that is required which is probably controlled in part by early expressionfor mandibular molar development. Barx-1 is a homeobox of p21 in the enamel knot cells (I. Thesleff, personalgene that is expressed in the mesenchyme in areas where communication).molars form but not incisor mesenchyme (Tissier-Seta et The speci®cation of tooth patterning by the odonto-al., 1995). genic homeobox code suggests that one way the code acts is by the odontogenic mesenchyme cells that con- dense at the base of the epithelial tooth bud communi- cating to the epithelial tooth bud cells to initiate1.4. Regulation of tooth shape enamel knot cell differentiation at speci®c positions (see Chapter 2).Once the spatial information is provided to specify devel-opment of a tooth germ into an incisor or molar, genesmust be activated that control the shaping process to
10 P. T. Sharpe Figure 1.5. Development of ectopic teeth in a human ovarian teratoma. (Courtesy of the Anatomy Museum, Guys Hospital.)1.5. Ectopic tooth development SummaryThe correct development of teeth in the right place at the Homeobox genes are involved in the genetic control ofright time clearly involves many different interacting many different developmental processes, including orga-cellular and molecular processes. It comes as something nogenesis, in invertebrate and vertebrate embryogenesis.of a surprise, therefore, to see perfectly formed teeth Many different homeobox genes are expressed in thedevelop at ectopic sites. One of the most common sites of developing orofacial region and during tooth develop-ectopic teeth formation is in ovarian teratomas (Figure ment in mammalian embryos and the roles of several of1.5). Teratomas are de®ned as germ cell tumours and these genes is beginning to be elucidated. Homeoboxsince the normal development of ovaries and testes does genes were discovered in Drosophila where mutations innot involve any contribution from cells of neural crest certain homeobox genes, namely homeotic genes, pro-origin it is dif®cult to reconcile the development of teeth duced changes in embryonic patterning resulting inin these structures. Since other neural crest-derived transformations of body parts. The potential role oforgans such as pigmented hair also form in these tera- homeobox genes in positional information of cells intomas it seemed likely that cells of neural crest origin are organogenesis has led to investigation of the possiblepresent. Alternatively since germ cells are pluripotent it role of homeobox genes expressed in the developingis possible that in certain circumstances these may con- orofacial region of mammalian embryos in patterningtribute to tooth mesenchyme, and that the neural crest for tooth position and shape.cell phenotype, although only present normally indevelopmental stages, is strongly expressed in this ab-normal development of these germ cells, neural crestbeing a vertebrate synapomorphy. Whatever the cellular Acknowledgementsorigin of these teeth the fact remains that their shape,mineralisation, etc. is perfectly normal and clearly the Work in the authors laboratory is supported by themechanisms operating in normal tooth development in Medical Research Council and the Human Frontierthe oral cavity are working in this ectopic site. If it were Science Programme.only possible to study the formation of these ectopicteeth during the early developmental stages they couldteach us a great deal about the molecular control ofnormal tooth development.
Homeobox genes in tooth development 11 References tooth development in embryonic and postnatal mouse. Developmental Dynamics, 206, 59±72. ÈBehrens, J., von Kries, J. P., Kuhl, M., Bruhn, L., Wedlich, D., Ä Kratochwil, K., Dull, M., Farinas, I., Galceran, J. and Grosschedl, R. Grosschedl, R. and Birchmeier, W. (1996). Functional inter- (1996). Lef±1 expression is activated by Bmp-4 and regulates action of b-catenin with the transcription factor LEF-1. Nature, inductive tissue interactions in tooth and hair development. 382, 638±642. Genes and Development, 10, 1382±1394.Bellusci, S., Henderson, R., Winnier, G., Oikawa, T. and Hogan, Laughon, A. and Scott, M. P. (1984). Sequence of a Drosophila B. L. M. (1996). Evidence form normal expression and tar- segmentation gene: protein structure homology with DNA- geted misexpression that bone morphogenetic protein-4 binding proteins. Nature, 310, 25±31. (Bmp-4) plays a role in mouse embryonic lung morphogen- Ã LeMouellic, H., Lallemand, Y., and Brulet, P. (1992). Homeosis in esis. Development, 122, 1693±1702. the mouse induced by a null mutation in the Hox-3.1 gene.Bitgood, M. J. and McMahon, A. P. (1995). Hedgehog and Bmp genes Cell, 69, 251±264. are coexpressed at many diverse sites of cell±cell interaction Lumsden, A. G. S. (1988). Spatial organisation of the epithelium in the mouse embryo. Developmental Biology, 172, 126±138. and the role of neural crest cells in the initiation of theBulfone, A., Kim, H. J., Puelles, L., Porteus, M. H., Grippo, J. F. and mammalian tooth germ. Development (Supp.), 103, 155±169. È È Luning, C., Rass, A., Rozell, B., Wroblewski, J. and Obrink, B. Rubenstein, J. L. R. (1993). The mouse Dlx-2 (Tes-1) gene is expressed in spatially restricted domains of the forebrain, (1994). Expression of E-cadherin during craniofacial develop- face and limbs in midgestation mouse embryos. Mechanisms of ment. Journal of Craniofacial Genetics and Developmental Biology, Development, 40, 129±140. 14, 207±216.Butler, P. M. (1956). The ontogney of molar pattern. Biological Luo, G., Hofmann, C., Bronckers, A. L. J. J., Sohocki, M., Bradley, A. Reviews, 31, 30±70. and Karsenty, G. (1995). Bmp-7 is an inducer of nephrogen-Chen, Y., Bei, M., Woo, I., Satokata, I. and Maas, R. (1996). Msx1 esis, and is also required for eye development and skeletal controls inductive signalling in mammalian tooth morpho- patterning. Genes and Development, 9, 2808±2820 genesis. Development, 122, 3035±3044. Maas, R. L., Chen, Y., Bei, M., Woo, I. and Satokata, I. (1996). TheDuboule, D. (1994). Guidebook to the Homeobox Genes. Oxford: role of Msx genes in mammalian development. Annals of the Oxford University Press. New York Academy of Sciences, 785, 171±181.Gastrop, J., Hoevenagel, R., Young, J. R. and Clevers, H. C. (1992). A MacKenzie, A., Leeming, G., Jowett, A., Ferguson, M. W. J. and common ancestor of the mammalian transcription factors Sharpe, P. T. (1991). The homeobox gene Hox-7.1 has speci®c RCF-1 and TCF-1 alpha/LEF-1 expressed in chicken T cells. regional and temporal expression patterns during early European Journal of Immunology, 22, 1327±1330. murine craniofacial embryogenesis, especially tooth develop-Gaunt, S. J., Miller, J. R., Powell, D. J. and Duboule, D. (1986). ment in vivo and in vitro. Development, 111, 269±285. Homeobox gene expression in mouse embryos varies with MacKenzie, A. L., Ferguson M. W. J. and Sharpe P. T. (1992). position by the primitive streak stage. Nature, 324, 662±664. Expression patterns of the homeobox gene Hox-8 in theGehring, W. J. (1987). Homeo boxes and the study of develop- mouse embryo suggest a role in specifying tooth initiation ment. Science, 236, 1245±1252. and shape. Development, 115, 403±420. Ävan Genderen, C., Okamura, R., Farinas, I., Quo, R. G., Parslow, R. Manak, J. R. and Scott, M. P. (1994). A class act: conservation of G., Bruhn, L. and Grosschedl, R. (1994). Development of homeodomain protein functions. In The Evolution of Develop- several organs that require inductive epithelial±mesench- mental Mechanisms, eds. M. Akam, P. Holland, P. Ingham and ymal interactions is impaired in LEF-1 de®cient mice. Genes G. Wray pp. 61±77. Cambridge: The Company of Biologists. and Development, 8, 2691±2703. Matzuk, M. M., Kumar, T. R., Vassaill, A., Bickenbach, J. R., Roop,Huber, O., Korn, R., McLaughlin, J., Ohsugi, M., Herrmann, B. G. D. R., Jaenisch, R. and Bradley, A. (1995). Functional analysis and Kemler, R. (1996). Nuclear localization of b-catenin by of activins during mammalian development. Nature, 374, interaction with transcription factor LEF-1. Mechanisms of 354±356. Development, 59, 3±10. Mina M. and Kollar E. J. (1987). The induction of odontogenesis inImai, H., Osumi-Yamashita, N.,Ninomiya, N. and Eto, K. (1996). non-dental mesenchyme combined with early murine man- Contribution of early-emigrating midbrain crest cells to the dibular arch epithelium. Archives of Oral Biology, 32, 123±127. dental mesenchyme of mandibular molar teeth in rat Molenaar, M., van de Wetering, M., Oosterwegel, M., Peterson- embryos. Developmental Biology, 176, 151±165. Maduro, J., Godsave, S., Korinek, V., Roose, J., Destree, O. andKemler, R. (1993). From cadherins to catenins: cytoplasmic Clevers, H. (1996). XTcf-3 transcription factor mediates beta- protein interactions and regulation of cell adhesion. Trends in catenin-induced axis formation in Xenopus embryos. Cell, 86, Genetics, 9, 317±321. 391±399.Kingsley, D. M. (1994). The TGF-B superfamily: new members, new Orban, B. (1928). Dental Histology and Embryology. Chicago: Rogers receptors, and new genetic tests of function in different Printing. organisms. Genes and Development, 8, 133±146. Prince, V. and Lumsden, A. (1994). Hoxa-2 expression in normalKoentges, G. and Lumsden, A. (1996). Rhombencephalic neural and transposed rhombomeres: independent regulation in crest segmentation is preserved throughout craniofacial on- the neural tube and neural crest. Development, 120, 911±923. togeny. Development, 122, 3229±3242. Qiu, M., Bulfone, A., Ghattas, I., Menses, J. J., Sharpe, P. T., Presley,Koyama, E., Yamaai, T., Iseki, S., Ohuchi, H., Nohno, T., Yoshioka, R., Pedersen, R. A. and Rubenstein, J. L. R. (1997). Role of the H., Hayashi, Y., Leatherman, J. L., Golden, E. B., Noji, S. and Dlx homeobox genes in proximodistal patterning of the bran- Paci®ci, M. (1996). Polarizing activity, Sonic Hedgehog, and chial arches: mutations of Dlx-1, Dlx-2 and Dlx-1 and -2 alter
12 P. T. Sharpe morphogenesis of proximal skeltal and soft tissue structures scription factor expressed in cranio-facial ectomesenchyme derived from the ®rst and second arches. Developmental and the stomach. Mechanisms of Development, 51, 3±15. Biology, 185, 165±184. Ê Vaahtokari, A., Aberg, T., Jernvall, J., Keranen, S. and Thesleff, I. ÈRamirez-Solis, R., Zheng, H., Whiting, J., Krumlauf, R. and (1996). The enamel knot as signalling centre in the devel- Bradley, A. (1993). Hoxb-4 (Hox 2.6) mutant mice show oping mouse tooth. Mechanisms of Development, 54, 39±43. homeotic transformation of a cervical vertebra and defects in Vastardis, H., Karimbux, N., Guthua, S. W., Seidman, J. G. and the closure of the sternal rudiments. Cell, 73, 279±294. Seidman, C. E. (1996). A human MSX1 homeodomain mis-Rivera-Perez, J. A., Mallo, M., Gendron-Maguire, M., Grindley, T. sense mutation causes selective tooth agenesis. Nature Ge- and Behringer, R. R. (1995). goosecoid is not an essential netics, 13, 417±421. component of the mouse gastrula organiser but is required Verbeek, S., Izon, D., Hofhuis, F., Robanus-Maandag, E., te Riele, for craniofacial and rib development. Development, 121, H., van de Wetering, M., Oosterwegel, M., Wilson, A., Mac- 3005±3012. Donald, H. R. and Clevers, H. (1995). An HMG-box-containingRuch, J. V. (1995). Tooth crown morphogenesis and cytodifferen- T-cell factor required for thymocyte differentiation. Nature, tiations: candid questions and critical comments. Connective 374, 70±74. Tissue Research, 32, 1±8. Winnier, G., Blessing, M., Labosky, P. A. and Hogan, B. L. M. (1995).Satokata, I. and Maas, R. (1994). Msx1-de®cient mice exhibit cleft Bone morphogenetic protein-4 is required for mesoderm palate and abnormalities of craniofacial and tooth develop- formation and patterning in the mouse. Genes and Develop- ment. Nature Genetics, 6, 348±356. ment, 9, 2105±2116.Sharpe, P. T. (1995). Homeobox genes and orofacial development. Yamada, G., Mansouri, A., Torres, M., Stuart, E. T., Blum, M., Connective Tissue Research, 32, 17±25. Schultz, M., De Robertis, E. M. and Gruss, P. (1995). TargetedSherperd, J. C. W., McGinnis, E., Carrasco, A. E., DeRobertis, E. M. mutation of the murine goosecoid gene results in craniofacial and Gehring, W. J. (1984). Fly and frog homeo domains show defects and neonatal death. Development, 121, 2917±2922. homology with yeast mating type regulatory proteins. Nature, Yost, C., Torres, M., Miller, J. R., Huang, E., Kimelman, D., and 310, 70±71. Moon, R. T. (1996). The axis-inducing activity, stability, andThomas, B. L., Tucker, A. S., Qiu, M., Ferguson, C. A., Hardcastle, subcellular distribution of b-catenin is regulated in Xenopus Z., Rubenstein, J. L. R. and Sharpe, P. T. (1997). Role of Dlx-1 embryos by glycogen synthase kinase 3. Genes and Development, and Dlx-2 genes in patterning of the murine dentition. Devel- 10, 1443±1454. opment, 124, 4811±4818. Zhou, P., Byrne, C., Jacobs, J. and Fuchs, E. (1995). LymphoidTissier-Seta, J.-P., Mucchielli, M.-R., Mark, M., Mattei, M.-G. and enhance factor 1 directs hair follicle patterning and epithe- Brunet, J.-F. (1995). Barx1, a new mouse homeodomain tran- lial cell fate. Genes and Development, 9, 570±583.
2 Return of lost structure in the developmental control of tooth shape J. Jernvall and I. Thesleff2.1. Introduction Rather, mouse, the standard laboratory model of mam- malian development, has to represent all the rest ofPerhaps one of the main attractions, both scienti®cally mammalian evolutionary diversity. This has of courseand aesthetically, of mammalian teeth is their diversity. prevented studies on, for example, the developmentalAmong other things, differently shaped teeth are a good control of premolars, which mice lack.reminder of both evolutionary ¯exibility and precision of Recently, however, the problem of biological diversitydevelopmental control mechanisms. In this respect, has resurfaced among developmental biologists. This is,mammalian teeth offer a good opportunity to use the rather paradoxically, because new molecular discoveriesfossil record to test our models of development. However, have demonstrated an increasing number of develop-although the diversity of tooth shapes has been described mental regulatory processes that are shared amongin great detail and their evolutionary history has been different organs and organisms (which has also revivedrelatively well reconstructed, our current knowledge of interest in homology, see Gilbert et al., 1996). Indeed, nowgeneral mechanisms controlling tooth shape develop- the question is how to de®ne a tooth using develop-ment is limited. As a ®rst approximation, a great deal of mental parameters of regulatory processes. And this ismammalian molar tooth diversity derives from different where the development of cusps becomes important. Ifcombinations of cusps (Chapter 19). Cusp number, size we simplify matters and de®ne teeth as a set of cusps thatand shape can vary from one tooth to the next and often have suf®cient developmental plasticity to facilitatecusps are connected by crests (lophs). Even this `gross morphological evolution, we have restated the researchtooth diversity constitutes a considerable challenge to a programme to specify the developmental basis ofdevelopmental biologist wishing to study the develop- morphological diversity (or evolutionary developmentalment of tooth shapes. Indeed, the list of students of biology, see Gilbert et al., 1996). As a great deal of mam-comparative tooth development is extensive and extends malian molar tooth diversity comprises differentback in time to Richard Owen (e.g. Owen, 1845; Leche, arrangement, size and shape of cusps, a great many1915; Bolk, 1920±22; Butler, 1956; Gaunt, 1961). speci®c developmental mechanisms are presumably fa- While the enthusiasm to study comparative ontogeny cilitating this diversity. So what is, developmentallyof different tooth shapes appears to have slowly declined, speaking, a tooth cusp anyway? As a partial answer, wemolecular studies of mouse tooth development have shall discuss in this chapter the role of the `epithelial¯ourished (see Weiss, 1993; Thesleff et al., 1995; Thesleff enamel knot in tooth shape development. The enameland Sahlberg, 1996 for reviews). As a subject of study, the knot is an embryological structure whose origins go backdiversity of biological shapes has been transposed by the to classical descriptive studies, but which has laindiversity of genes and their products. In practical terms dormant in the literature for several decades. Recentthis implies that, of total mammalian tooth diversity, we discoveries concerning gene expression patterns,are left with the dentition of a mouse. This does not however, have suggested that, in a developing tooth, theimply that mouse teeth are somehow uninteresting. enamel knot is very far from being dormant. 13
14 J. Jernvall and I. ThesleffFigure 2.1A ± C. Transverse sections of tooth germs. A drawing (A) of late bud-stage bottlenose whale tooth germ (drawing from Ohlin, 1896)showing the characteristic appearance of the enamel knot cell cluster (x). The bud- (B) and cap-staged (C) ®rst mouse lower molar histologicalsections showing the formation of the primary enamel knot. The ®rst sign of the enamel knot (arrow in B) is marked by the cessation of the cellproliferation (unlabelled cells, proliferating cells have dark nuclei). The basement membrane separating epithelium and mesenchyme ismarked with a white line. ek, enamel knot; dm, dental mesenchyme. Scale bar represents 100 mm in B and C.2.2. Found and lost: the enamel knot enamel knots in crocodilians. The authors of this chapter also admit to ignoring the enamel knot for a long time.Enamel knots can be seen in several old descriptions of For us, the rediscovery started with the cells of thetooth development, starting more than a century ago enamel knot itself.(e.g. Ohlin, 1896). In these drawings of histological sec-tions, the enamel knot is a clump of cells, usually in thecentre of the tooth germ (Figure 2.1A). The packing and¯attening of the cells in a histological section gives the 2.3. Cells different from all others: theenamel knot its characteristic look, hence names like enamel knot found again`spherical bodies (Tims, 1901), `Kugelkorper (Heinick, È1908) or `cellanhopningar (Ohlin, 1896) have been used The enamel knot is formed during early tooth morpho-to describe the appearance of the tooth germs with the genesis in the centre of the tooth germ epitheliumenamel knot (`a halved cabbage head being quite accur- (Figure 2.1). It forms at the late bud stage and this alsoate; Jernvall, 1995). Although usually recognized, their marks the beginning of tooth shape development. More-possible signi®cance for tooth development was not con- over, the cells of the enamel knot cease to proliferatesidered. The enamel knot enjoyed a more prominent (Jernvall et al., 1994). This was our initial discovery (orplace in the literature during the ®rst three decades of rediscovery, there was disagreement in the literature onthe twentieth century. Functional interpretations for the this matter, see Butler, 1956 for review), which drove ourrole of the enamel knot in tooth development include attention to this area of the tooth germ.functioning as a reservoir of cells for the rapidly growing Before the actual cell division into two daughterenamel organ, in particular for the stellate reticulum cells, the genome has to be duplicated (S-phase of a cellabove the inner enamel epithelium (Butler, 1956). It has cycle). The proliferating cells in the tooth germ can bealso been suggested to cause the epithelial-mesenchymal localized by mapping cells that are replicating their DNAinterface to fold by pushing or acting as a restraint and incorporating a labelled DNA-base analogue.(Ahrens, 1913). However, although the problem of its However, the standard histological sections used havesigni®cance or even its existence never seems to quite the problem of leaving the three-dimensional associ-subside, the enamel knot was gradually reduced to a ations of cell proliferation and tooth shape to an investi-developmental curiosity, possibly originating as a histo- gators imagination. Therefore, because one mouselogical artifact (reviewed in Butler, 1956; also Mackenzie tooth germ can consist of up to a hundred sections, weet al., 1992). The notion that the enamel knot is actually a decided early on to use three-dimensional reconstruc-widespread structure in developing teeth was proposed tions of serial sections to relate the cellular level pro-by Westergaard and Ferguson (1987) where they reported cesses to morphology. After the reconstructions, the
Developmental control of tooth shape 15 Figure 2.2. A three- dimensional reconstruction of a mouse cap-stage ®rst lower molar. The basement membrane side of the inner enamel epithelium is viewed from above at an oblique angle. Note how cell proliferation patterns (as incorporated BrdU) and expression pattern (detected messenger RNA) of ®broblast growth factor 4 gene (Fgf-4) are inverted: only the non- proliferating enamel knot cells express Fgf-4. Below corresponding transverse histological sections. The basement membrane is marked with a white line. (Adapted from Jernvall, 1995.)enamel knot was impossible to miss (Figure 2.2). Morpho- example of allometric growth which happens aroundgenesis of the tooth crown begins as lateral growth of the enamel knot.the tooth bud epithelium forms a characteristic `cap. All The return of the enamel knot from oblivion wasteeth, incisors as well as molars, pass through this stage. ®nally clinched by the discovery that their cells have aThe enamel knot sits along the midline of this cap stage distinct gene expression pattern that differs from thegrowth and one easily gets the impression that the tip of rest of the tooth germ. First, Mackenzie et al., (1992)the tooth bud is ®xed and grows laterally instead. More- found that the transcription factor Msx-2 had an expres-over, the enamel knot is present only during the for- sion pattern largely restricted to the cells of the enamelmation of the tooth germ cap stage. The knot disappears knot in a cap staged tooth. Second, ®broblast growthand the cells of the inner enamel epithelium resume factor 4 (Fgf-4) was reported to be expressed in a speci®cproliferation, and the whole life span of the enamel knot manner in a tooth germ (Niswander and Martin, 1992).is less that 2 days in a mouse molar (Jernvall, 1995). The Particularly the expression of Fgf-4 turned out to betransition of a tooth bud to a cap offers us a very clear remarkably restricted to the cells of the enamel knot
16 J. Jernvall and I. Thesleff Figure 2.3 Schematic comparison of expression patterns of some signalling molecules in a limb bud and in a cap-staged tooth. Limb and tooth mesenchyme are marked in grey. While different sets of signalling molecules are expressed in the limb bud in one mesenchymal and one epithelial domain, in the teeth these signalling molecules are expressed in the epithelial enamel knot.(Jernvall et al., 1994). So much so, that now we had three many signalling molecules are present in two separatemarkers for the enamel knot cells; the histology, cessa- signalling tissues: the apical ectodermal ridge (AER) andtion of cell proliferation and Fgf-4 expression (Figure 2.2). the zone of polarizing activity (ZPA), which have beenWhile this should have been enough to convince most shown to control proximodistal growth and anteropos-embryologists of the existence of the enamel knot, it also terior patterning, respectively (Tabin, 1991). Fgf-4, whichresurrected the question about its functions during is expressed in the tooth germ in the enamel knot, istooth morphogenesis. However, because of the new tools restricted to the posterior AER in the limbs (Niswanderin molecular biology, we now had a little bit more to and Martin, 1992; Suzuki et al., 1992). Other knownwork with. signalling molecules include sonic hedgehog (Shh) and Fgf-4 expression by the non-proliferating cells of the bone morphogenetic proteins (Bmp-2, -4, -7). In the limbenamel knot is intriguing because secreted FGF-4 protein buds, Shh is expressed only in the ZPA (Echelard et al.,is a known mitogen in vertebrate limbs stimulating cells 1993) while the Bmps are expressed both in the AER andof the limb mesenchyme to divide (Niswander and in the ZPA (Francis et al., 1994; Lyons et al., 1995). PositiveMartin, 1993; Vogel and Tickle, 1993). Our experiments feedback loops operate between the AER and the ZPA,con®rmed this to be the case for both the dental epithe- and for example, once induced, Shh expression in the ZPAlium and mesenchyme in vitro (Jernvall et al., 1994). is maintained by FGF-4 signal from AER (Laufer et al.,However, these results implied a slightly paradoxical 1994; Niswander et al., 1994).situation for the cells of the enamel knot itself; the same Contrasting with limbs, there seems to be only onecells that do not divide, are manufacturing a protein that signalling centre in the cap stage tooth germ, the enamelstimulates cells to divide. On the other hand, this could knot. And, at least on the molecular level, the enamelful®ll the minimum requirement for the control of tooth knot performs the duties of both limb signalling centresbud folding and formation of the cap stage: the enamel (Figure 2.3). Fgf-4, Shh and Bmp-2, -4, and -7 are all expressedknot may cause the unequal growth of the epithelium by in the enamel knot (Vaahtokari et al., 1996b). This veryconcurrently remaining non-proliferative and by stimu- centralized expression pattern of several signalling mole-lating growth around it (Jernvall et al., 1994; Jernvall, cules also explains why the role of the enamel knot has1995; Vaahtokari et al., 1996a). The combination of the been dif®cult to demonstrate experimentally. There hasnon-dividing enamel knot cells and surrounding dividing been no way to `dissect the roles of different centrescells could create packing of cells around the knot re- because there is only one present. Hence, one gets ansulting in folding and downgrowth of the epithelium individual tooth or no tooth.around it. While this kind of `physical pushing has been The similarities between different signalling centresargued to play an important role in the initiation of cusp are not only limited to signalling molecules. Also thedevelopment (e.g., Butler, 1956; Osborn, 1993), molecular same transcription factors are expressed in the enamelevidence supporting it had been lacking. knot and limb signalling centres. The already mentioned But genes never function alone. Vertebrate limb devel- homeobox-containing Msx-2 gene (Mackenzie et al., 1992)opment is an excellent example of the multigene cas- is expressed both in the limb AER and in the enamelcades involved in shape development. In the limb bud, knot. Transcription factors function like cells internal
Developmental control of tooth shape 17switches that can cause the cell to progress to the next signalling centres may have evolved with the endoske-developmental phase (differentiation, mitosis) and to leton resulting from a partly pirated developmentalmanufacture new sets of signalling molecules. These pathway from an epithelial±mesenchymal centre.signalling molecules, on the other hand, may trigger theneighbouring cells to proceed to the next developmentalphase, which may or may not be the same as that of thesignalling cells. Also the signalling cells themselves may 2.4. Back to cusps: how many enamel knotsrespond to a signal called autocrine induction (in con- are there?trast to paracrine ± between populations of cells). Theexact biochemical role of Msx-2 is not known, but it A reasonable hypothesis is that the enamel knot may beappears to be associated with programmed cell death important for tooth shape development. Indirect evi-(apoptosis) (Graham et al., 1994). Recent ®ndings indicate dence of this is the discovery that multicusped teeth havethat the signalling of apoptosis is another universal more than one enamel knot (Jernvall et al., 1994). Themolecular cascade in the embryo. Both BMP-4 and Msx-2 classical enamel knot that forms at the end of the budhave been shown to be involved in apoptosis of interdi- stage correlates with the beginning of tooth crown shapegital tissue in chick limbs and rhombomeres (Graham development. The tooth germ grows around it and theet al., 1994; Zou and Niswander, 1996). While Msx-2 tran- knot sits at the tip of the folding. At least for rodentscripts are present in the cells that are going to go into incisors and crocodilian teeth, this is the one and onlyapoptosis, the actual (autocrine) signalling molecule that enamel knot present. But in teeth with several cusps (e.g.mediates apoptosis is BMP-4 (Graham et al., 1994; Zou and mouse and opossum molars; Figure 2.4), secondaryNiswander, 1996). And, perhaps no surprise, the disap- enamel knots form later at the tips of each cusp (Jernvallpearance of the enamel knot happens via apoptosis et al., 1994; Jernvall, 1995). The secondary knots can be(Vaahtokari et al., 1996a) and Bmp-4 expression marks the detected at the onset of, or even slightly prior to, thedeath of its cells as well (Jernvall et al., 1998). visible development of the cusps (Figure 2.4). Therefore, a While the enamel knot cells appear distinctly differ- secondary enamel knot `marks the cusp and once cuspent from all the other cells of the tooth germ, no `enamel development is initiated, the knot disappears (Jernvall etknot genes have been discovered that would put the al., 1994; Jernvall, 1995). This, again, happens via apop-enamel knot apart from other embryonic signalling tosis (Vaahtokari et al., 1996a).centres and their molecular cascades. Indeed, the enamel The existence of secondary enamel knots has beenknot nicely demonstrates how different organs can be even more elusive than the ®rst (primary) enamel knot.made using quite similar developmental processes or Actually, early workers sometimes recognized them, butmodules (Wray, 1994; Gilbert et al., 1996, Raff, 1996). no distinction was made between the primary and sec-These developmental modules are homologous as pro- ondary knots. And the secondary knots can look quitecesses even though the structures themselves (such as a similar to the primary knot: the cells are packed, post-limb and a tooth) are not homologous (Bolker and Raff, mitotic, and they express Fgf-4 gene (Jernvall et al., 1994).1996; Gilbert et al., 1996). Roth (1988) proposed that new The secondary enamel knots are not, however, as differ-morphological structures can result from `genetic pir- ent from the rest of the tooth germ as the primaryatism. The signalling functions and the apoptotic path- enamel knot. While the expression of many signallingways may have risen only once (or a few times, molecules is tightly restricted to the primary enamelconvergence and parallelism are rampant in the fossil knot, during the formation of secondary enamel knots,record, e.g., Hunter and Jernvall, 1995), but these mole- the expression domains of these signals are broadercular cascades have been co-opted (pirated) repeatedly to (Vainio et al., 1993; Bitgood and McMahon, 1995). Indeed,generate new structures. Whereas the enamel knot may only Fgf-4 remains restricted to enamel knots throughoutnot have any speci®c genes of its own, the combination of the odontogenesis. This may re¯ect a more restrictedactive signalling molecules and transcription factors molecular function for secondary enamel knots as com-may well be unique (Figure 2.3). Overtly simplifying, we pared with the primary enamel knot.can formulate that limb AER + ZPA = EK (the enamel The short lifespan of individual enamel knots mayknot). That is, the molecular identity of the enamel knot mean that the control for cusp development is neededis the sum of molecular activities of two limb signalling only intermittently. In order for a cusp to develop, it hascentres. Evolutionarily speaking though, the enamel to be initiated, and the initiation of a cusp appears to beknot may represent a signalling centre for general epithe- the most important event in its development. Thelial-mesenchymal organs (e.g. similar cascades are likely sequence of cusp formation closely follows the evolution-to be found in feathers, Nohno et al., 1995), and limb ary appearance of the cusps, and usually the ®rst cusp
18 J. Jernvall and I. Thesleff Figure 2.4. Three-dimensional reconstructions of the opossum (Monodelphis domestica) ®rst lower molar development (obliquely lingual view, mesial is toward the viewer) showing the formation of metaconid (arrows). In the histological section of the 11-day-old tooth (11d), the secondary enamel knot is visible as a maximum of 90 mm thickening (average 68 mm) of the inner enamel epithelium (white dashed circle). No secondary enamel knot is detected in the 9- (9d) or 13-day-old (13d) tooth germs (maximum thickness of the metaconid area inner enamel epithelium = 53 mm and average thickness = 40 mm). Note the mineralization of protoconid tip (marked as white) by the time of the initiation of metaconid. The basement membrane, from which the 3-D reconstructions were made, is marked with a white line in the histological sections. Scale bar represents 200 mm.(the primary cusp) to develop is the protoconid in the 2.5. Developmental control of tooth shape?lower, and the paracone in the upper teeth (e.g. Butler,1956; Luckett, 1993). However, because cusps grow from We certainly do not yet know what controls tooth shapethe tip down, the order of initiation of cusp development development. Even in itself, tooth shape is a ratheris the main determinant of relative cusp size and phylo- elusive concept. Tooth shape is not the same as toothgenetically `recent but large cusps can start to form family (incisor, canine, premolar or molar). Tooth familyprior to phylogenetically older cusps (Berkovitz, 1967; is de®ned by relative position, and individual teeth canButler, 1956, 1967). Therefore, for a cusp to become the evolve shapes that are found in adjacent tooth familiesright size, its development has to be initiated at the right (e.g. molarization of premolars in many ungulates, `pre-moment. The control of cusp initiation by the enamel molarization of molars in seals). The existence of differ-knots is perhaps all that is required for the cusps to form. ent tooth families manifests developmental decouplingSo, a very simplistic view of tooth shape development has of these sets of teeth, and may be as old as heterodontybasically two variables; where and when. The placement itself. The determination of tooth families may be underof secondary enamel knots determines the cusp pattern, homeobox-containing genes which may create a patch-and their timing determines the relative height of the work of overlapping gene expression domains (for Dlxcusps. genes see Chapters 1, 11). The actual determination of One complication is the function of the primary tooth shape during development is probably a down-enamel knot in the initiation of cusp development. Does stream process from the speci®cation of tooth family.it initiate the paracone on the upper molars and proto- The primary enamel knot can be considered as one candi-conid on the lower molars, the phylogenetically oldest date for the control of actual tooth shape. Its timelycusps? Currently we do not really know this. The primary appearance, and cellular functions suggest that it isenamel knot establishes the whole crown area (including needed for tooth morphogenesis. Indeed, mutant micetalonid, Figure 2.4), and the temporal initiation of the with experimentally inactivated transcription factors®rst cusp is initiated from the `remnants of the primary Msx-1 and Lef-1 (Satokata and Maas, 1994; van Genderen etenamel knot. Thus, it is possible that the enamel knot of al., 1994) have their tooth development arrested to thethe primary cusp is actually a portion of the primary bud stage and no enamel knot forms. Both these genesknot. This is an issue that can be helped with careful are normally expressed in the tooth bud (Msx-1 is neededmapping of different molecular markers. in the mesenchyme, Lef-1 in the epithelium; Satokata and