INTRODUCTION

Vertebrate organ development, from the initiation to terminal differentiation, depends upon inductive interactions typically between epithelium and adjacent mesenchyme 1, 2. The inductive interactions consist of two main components: a tissue capable of producing the inducing stimulus and a tissue capable of receiving and responding to it. At the molecular level, these interactions involve complex signaling networks composed of various signaling molecules, their receptors, and the transcription control systems. It has been clear that embryonic inductions via cell and tissue interactions are mediated by diffusible protein signaling molecules known as growth factors. Prominent among these are Bone Morphogenetic Proteins (BMPs), Fibroblast Growth Factors (FGFs), Wnt, and Hedgehog (Hh) families. These growth factors function synergistically and/or antagonistically to organize and pattern tissues and organs during embryonic development. Growth factor functions through binding to its specific cell membrane-bound receptor that triggers intracellular signal pathway, leading to signal transduction into the nucleus where latent transcription factors are activated, and eventually causing change of gene expression. The expression of many transcription factors during development of many organs has been revealed. The overlapping expression patterns of transcription factors and growth factors in developing organs suggest a relationship between these two classes of gene products in inductive interaction. It was proposed that transcription factors in a tissue layer activate the expression of growth factors in response to the signaling of growth factors produced from the adjacent tissue, forming a signaling network that regulates organogenesis 3.

Mammalian tooth development shares many similarities with other embryonic organs. The mouse tooth has long served as a model organ for studying fundamental questions of developmental biology such as epithelial-mesenchymal inductive interactions, differentiation, and pattern formation 4. This is mainly due to the ease with which it can be accessed, manipulated, and cultured in vitro, and be transplanted to ectopic sites where it can develop and differentiate into a recognizable tooth organ.

Clinically, tooth allotransplantation, autotransplantation, and dental implants have been traditionally used as prosthetic procedures for tooth replacement for many years. Although missing or damaged teeth are not life-threatening, it apparently affects the quality of human's daily life. Stem cell-based tissue engineering approaches to generate organ and tissue for the clinical application appear very promising 5. Recent studies have greatly prompted the potential application of stem cells in tooth regeneration.

Embryology of mammalian tooth development

Sequential and reciprocal interactions between the cranial neural crest-derived mesenchymal cells and the stomadial epithelium regulate tooth morphogenesis and differentiation 6. In mice, the dental mesenchyme is attributed by neural crest cells derived from the midbrain and hindbrain region around embryonic day 8.5 (E8.5) 7, 8, 9, 10. The determination of tooth forming sites and tooth type occurs around embryonic day 10.5 (E10.5) 11, 12, 13. At E11.5, the first morphological sign of molar tooth development appears as a local thickening of the dental epithelium. During this process, the presumptive dental epithelial cells elongate along their apical-basal axis, change cell shape from cuboidal to columnar, assume an apical-basal polarity (polarization) and form a dental placode (Fig. 1A). At the E12.5 and E13.5 bud stage, the thickened dental epithelium proliferates and invaginates into the subjacent mesenchyme to form the epithelial tooth bud around which the mesenchymal cells condense (Fig. 1B, 1C). During these stages, the basal layer cells of the epithelial bud maintain columnar shape. At the E14.5 cap stage, the epithelial component undergoes specific folding, which is accompanied by the formation of the enamel knot, a transient cluster of non-dividing epithelial cells (Fig. 1D). A unique set of signaling molecules, including Shh, BMP2, BMP4, BMP7, FGF4, and FGF9, are expressed in the enamel knot. The enamel knot is therefore considered to be a signaling center to control the patterning of the tooth cusps 14. During the subsequent bell stage, the epithelially derived ameloblasts and mesenchymally derived odontoblasts differentiate. In addition, the mesenchymal cells differentiate into alveolar bone that forms the sockets for the teeth 15. The human tooth development shares great similarity to that of the mouse in the early process, including the lamina, the bud, and the cap stages (Fig. 1E, 1F, 1G, 1H). At the sixth week of human embryonic development, the cuboidal epithelial cells in the site of presumptive dentition elongate into columnar epithelial cells, and form a thickened horseshoe shaped epithelial band, namely the dental laminar, around upper and lower jaws (Fig. 1E). At the seventh week, successive but uneven proliferation in the thickened dental laminar leads to formation of 10 dental epithelial buds at intervals that will develop into the presumptive enamel organs of the primary teeth (decidous teeth) (Fig. 1F). From the eighth to the tenth week, the dental epithelial bud invaginates further into the subjacent dental mesenchyme, forming a tooth germ at the late bud stage, (Fig. 1G). Meanwhile, small diverticulums spring from the dental epithelial buds on the superficial side, and become the secondary tooth buds (permanent tooth buds) (Fig. 1G). These secondary tooth germs pause at the bell stage and resume until the primary teeth are replaced. At the tenth week, tooth germs assume cap-like structure (Fig. 1H).

Figure 1
figure 1

The early stages of molar tooth development in the mouse and human embryos. (A-D) the mouse molar tooth germ at the lamina stage (A), the early bud stage (B), the late bud stage (C), and the cap stage (D). (E-H) the human molar tooth germ at the lamina stage (E), the early bud stage (F), the late bud stage (G), and the cap stage (H). Note the presence of the secondary tooth germ (STG) in (G). Abbreviation: DE, dental epithelium; DM, dental mesenchyme; DP, dental papilla; EK, enamel knot; STG, secondary tooth germ.

The odontogenic potential represents an instructive induction capability of a tissue to induce gene expression in an adjacent tissue and to initiate tooth development. On the other hand, the odontogenic competence represents the capability of a tissue to respond to odontogenic signals and to support tooth formation. Tissue recombination experiments between isolated mouse dental epithelial and mesenchymal tissues have demonstrated that during early mouse molar tooth development, the odontogenic potential initially resides in the dental epithelium, and then shifts to the mesenchyme 16, 17. At the pre-bud stages of development (before and at E11.5) the presumptive dental epithelium possesses the potential, capable of inducing tooth formation when confronted with non-dental mesenchyme (Fig. 2). However, only the cranial neural crest derived mesenchyme is odontogenic competent. Teeth failed to form when dental epithelium was recombined with non-neural crest derived mesenchyme, such as limb mesenchyme 17. In contrast, at the early bud stage (E12.5), the odontogenic potential has switched to the mesenchyme (Fig. 2), which becomes able to instruct non-dental epithelium to form tooth-specific structures 16, 17, 18. The odontogenic potential then remains in the dental papilla that also controls tooth shape till birth 18, 19, 20.

Figure 2
figure 2

The odontogenic potential shifts from dental epithelium to mesenchyme during early mouse tooth development. The molar epithelium prior to E12.5 possesses the odontogenic potential and is capable of inducing tooth formation when recombined with non-dental mesenchyme. After E12.5, the odontogenic potential shifts to the molar mesenchyme which is able to instruct tooth formation when recombined with non-dental epithelium. Abbreviation: 2nd BA, secondary branchial arch; epi, epithelium; mes, mesenchyme.

GROWTH FACTORS MEDIATE INDUCTIVE INTERACTIONS DURING ODONTOGENESIS

Classic studies have demonstrated the importance of cross-talk between dental epithelium and mesenchyme during tooth development. The isolated component of embryonic mouse tooth germs, either epithelium or mesenchyme fails to undergo cytodifferentiation, and the epithelium dies when grown in isolation 21. However, physical separation by Millipore filter between epithelium and mesenchyme allows enamel organ and papilla differentiation with the deposition of matrix. These studies indicate that the cross-talk between these two tissue layers is required for their normal differentiation or survival, and the communications are mediated by diffusible signaling molecules. It is clear now that these diffusible molecules are growth factors.

Bmps

Recent studies have implicated BMPs in mammalian tooth development from the very beginning. Although the expression of several Bmp genes in tooth development has been reported in many studies, BMP4 is the one suggested to play a central role during tooth morphogenesis 14. Around E10, when the determination of the tooth forming site and tooth type occurs, the antagonistic effects of epithelially expressed Bmp4 and Fgf8 restrict Pax9 expression to the presumptive dental mesenchyme 11, 13 and the expression of Pitx2 to the presumptive dental epithelium 22. The tooth type is meanwhile determined. The expression domain of Barx1, a homeodomain transcription factor necessary for the molar tooth type specification, is restricted to the molar anlage by BMP4 that is expressed in the distal region of mandible 12, 23. On the other hand, this distally expressed BMP4 induces Islet1 expression in the future incisor mesenchyme 24. Subsequently, multiple genes encoding growth factors are expressed in the thickened dental epithelium around E11.5. They include Fgf8, Fgf9, Bmp2, Bmp4, Bmp7, Shh, Wnt10a, and Wnt10b 25, 26, 27, 28, 29, 30, 31. These epithelial signaling molecules are responsible for the induction of gene expression in the adjacent dental mesenchyme, including Msx1, Msx2, Lef1, Dlx1, Dlx2, Patched (Ptc), Gli1, and Syndecan-1 25, 28, 31, 32, 33, 34. At the following bud stage, in response to the epithelial inductive signals, the dental mesenchyme begins to express genes encoding signaling molecules, including BMP4, FGF3, activin-βA, and Wnt5a 25, 35, 36, 37. These mesenchymal growth factors act upon the dental epithelium as feedback signals, and also function within the dental mesenchyme, to regulate subsequent tooth development 3. It is at this stage that dental mesenchyme acquires odontogenic potential. Thus these mesenchymally expressed growth factors might represent components of the odontogenic potential of dental mesenchyme. At E14.5 tooth development reaches to the cap stage. A number of signaling molecules including BMP2, BMP4, and BMP7 are specifically expressed in the enamel knot which acts as signaling center that controls cell proliferation and apoptosis, and determine cusps number and position, and therefore the tooth patterning 38, 39, 40, 41.

Of particular interest was the observation that the shift of Bmp4 expression during early tooth development is associated with the shift of the odontogenic potential from the dental epithelium to the mesenchyme 25, suggesting that BMP4 is a component of the odontogenic potential. It was shown that the odontogenic potential of dental epithelium could be partially replaced by recombinant BMP4 protein in the organ culture of dental mesenchyme. Application of BMP4 to the molar dental mesenchyme in culture mimics the effects of the Bmp4-expressing dental epithelium, inducing morphogenetic change of the mesenchyme and activating the expression of transcription factors, such as Msx1, Msx2, Lef1, and Egr1 in the mesenchyme 25, 32. Similar results were obtained from investigations at molecular level on embryonic jaw of the tooth-less bird. It was demonstrated that the early odontogenic signaling pathway is conserved in Aves, and loss of odontogenic Bmp4 expression may be responsible for the early arrest of tooth development in living birds 42, 43. Furthermore, mesenchymal BMP4 was demonstrated to be responsible for the induction and maintenance of gene expression, such as Shh and p21, in dental epithelium, and for the formation of the enamel knot 39, 44. Expression of Bmp4, together with Bmp2 and Bmp7, in the enamel knot of the cap stage tooth germ may be responsible for apoptosis in the knot cells 4, 39, 45. Since mice deficient in Bmp4 die prior to the tooth formation 46, whether or not BMP4 functions as a component of the odontogenic potential in both dental epithelium and dental mesenchyme in vivo remains unknown.

Like other members of the TGF-β superfamily, BMP signaling is mediated via heterodimers of transmembrane serine/threonine kinase of type I and type II BMP receptors 47. In vertebrates, two type I BMP receptors, BMPR-IA (Alk3) and BMPR-IB (Alk6), have been identified, each of which can form heterodimers with type II BMP receptors and bind to BMPs 48. Once BMP ligands bind to the receptors, the type II receptor phosphorylates the type I receptor which in turn phosphorylates the DNA-binding proteins Smads. The latter translocates into nucleus to regulate target gene expression 49. BMP signals can also be transduced via ActRIIB and ActR-I (Alk2) 50. Both Bmpr-IA and Bmpr-IB are expressed in early developing tooth germ 51, 52, Z. Zhang and Chen, unpublished observations]. Mice deficient in Bmpr-IA die before tooth formation 53, while Bmpr-IB mutants do not exhibit a defect in teeth 54, 55. Conditional knockout of Bmpr-IA in dental epithelium results in tooth development arrested at the bud stage 56, 57. These observations further demonstrate that BMP signaling from the mesenchyme to the epithelium is required for the transition of tooth development from the bud to the cap stage 32. Although Bmpr-IA and Bmpr-IB do not show functional redundancy in embryonic hair follicle development, the case in tooth development is not clear yet 56.

There are eight vertebrate Smad proteins (Smad 1-8) that transducer TGF-β signaling. They can be divided into three distinct classes: the receptor-activated Smad (Smad1, Smad2, Smad3, Smad5, and Smad8), the common-mediated Smad (Smad4), and inhibitory Smad (Smad6, Smad7). Among them, Smad1, Smad5 and Smad8 are phosphorylated by BMP type I receptors 58. All eight Smads are expressed in the developing tooth germ 59, 60; X. Yu and Chen, unpublished observations]. However, the role of each of them in tooth development is largely elusive, since most knockout mice die before tooth development begins. Using an in vitro organ culture and antisense approach, Chai and his colleagues were able to show that attenuation of Smad2 and Smad7 expression result in abnormal tooth development, indicating that the effectiveness of TGF-β signaling in tooth development is highly sensitive to the level of Smad gene expression 59.

Fgfs

Several members of FGF family are expressed in the early developing tooth germ. They function at distinct steps of odontogenesis, from tooth initiation to the formation of the last tooth cusp 61. Intensive Fgf8 and much weaker Fgf9 expression are initially detected in the presumptive dental epithelium before tooth initiation and persist there until the early bud stage. FGF signaling participates in restricting tooth forming sites by inducing the expression of Pax9, Pitx1, and Pitx2 11, 13, 22. FGF8 is also responsible for the induction of Barx1 in the prospective molar mesenchyme 12. In addition, FGF8 is also primarily responsible for Lhx6 and Lhx7 expression in the odontogenic mesenchyme prior to the initiation and during tooth formation 62. At the late bud stage and the early cap stage, Fgf9 is up-regulated in the primary enamel knot, where Fgf4 is also activated by the Wnt signaling pathway 30, 38, 63. However, whether FGF8 is a component of the instructive odontogenic potential in the oral epithelium is still obscure, since in mice lacking FGF8 in the oral epithelium, most first branchial arch-derived structures were absent, including the molar, but incisors formed 64. Another FGF, perhaps FGF9, might rescue incisor formation in the absence of FGF8 64. FGF8 may also be responsible for the induction of Fgf3 expression in dental mesenchyme 37. FGF4 and FGF9 stimulate cell proliferation of both dental epithelium and mesenchyme 30, 38. On the other hand, Fgf3 is expressed in the dental mesenchyme from the E13.5 bud stage on, while Fgf10 expression can be detected in the dental mesenchyme around E14.0 37. FGF10 is able to stimulate cell proliferation only in the dental epithelium but not in the mesenchyme 37. FGF3, however, could stimulate cell proliferation in isolated dental mesenchyme 37. Mice deficient in either Fgf3 or Fgf10 did not exhibit major tooth defects 65, 66, 67. However, elimination of FGFR2b, the tyrosine kinase receptor for both FGF3 and FGF10, caused an arrest of tooth development at the bud stage 68, 69. Thus functional redundancy was proposed for FGF3 and FGF10 in tooth development 37, 63.

Hh

Among the three members of the hedgehog (Hh) family in vertebrates, Shh is the only Hh ligand to be expressed in teeth. Shh signaling has been implicated in the patterning and growth of a number of vertebrate organs by exerting its short- and long-range effect to activate downstream gene expression during organogenesis 70. In the developing tooth, Shh transcripts are first detected in the dental epithelium of the dental placode around E11.5 by in situ hybridization 27, 31, 71, 72, suggesting a role for Shh in the initiation of tooth development 27, 28, 29. However, Shh expression can be traced as early as E9.5 in the oral epithelium of transgenic mice bearing an Shh allele knocked-in with the LacZ reporter gene 73. The factors that induce Shh expression in dental epithelium are currently unknown, but BMP activity in both dental epithelium and dental mesenchyme has been shown to be necessary for the maintenance of Shh expression 44. The epithelially expressed Shh may regulate thickening and invagination of dental epithelium to form an epithelial tooth bud by stimulating epithelial cell proliferation 29, 74. This hypothesis is supported by the facts that the application of Shh-soaked beads to E10.5 mouse embryonic mandibles induced the initiation of ectopic epithelial thickening. Blocking of Shh activity in the E10.5 mandible with neutralizing antibody against Shh arrested tooth development at the lamina stage 29, 74. However, removal of the Shh gene, as well as the Smoothened (Smo) gene which encodes a membrane protein and is essential for the transduction of all Hh signals, in the dental epithelium at E11.75 caused altered cell proliferation, growth, and patterning of the teeth, but tooth initiation and subsequent development occurred anyway 75, 76. Although Shh is exclusively expressed in the epithelial compartment during early tooth development, accumulation of Shh protein is found in the mesenchyme of distema region 18 h after removal of oral epithelium. This observation implies that exclusion of Shh activity from distema mesenchyme may be required for correct tooth initiation at right position in the oral cavity 77. After the bud stage, Shh expression is confined to the enamel knot at the cap stage, and spreads to the surrounding inner enamel epithelium and the stratum intermedium cells during the following stages 40, 75, 78. Analyses of knockout mice that are mutants of other genes in the Hh signal pathway further revealed the function of Shh in tooth development. Removal of the Smo gene from the dental epithelium causes abnormally fusion of the first and second molars in both the maxilla and mandible, similar to those molars observed in the Shh conditional knockout mice, but obviously larger than them, suggest that Shh signaling may also operate intra-epithelially besides mediating epithelial-mesenchymal interactions during tooth development 76. On the other hand, removal of Smo activity from the neural crest derived cells of the first branchial arch (the dental mesenchyme) leads to different tooth phenotypes. In addition to the reduced molars and fused up incisor, the lower incisor is also missed 73. In the Hh signaling pathway, Dispatched (Disp) functions within the Hh secreting cells by regulating Hh trafficking within the synthesizing cells, controlling the levels of available Hh signal in the embryo 79, 80. Disp1 has been demonstrated essential for transduction of all Hh signals 81. By taking advantage of a hypomorphic mutant of Disp1, Shh signal in mice can be attenuated at several different levels. When Hh signal is decreased below certain degree, the maxillary incisors are missed 82. The same tooth phenotype is also characterized in human holoprosencephaly (HPE) that results from reduced Shh signaling 83. Gli genes encode the transcription effectors of Shh signaling. Single knockout of the three Gli genes did not show obvious tooth defects, but Gli2 and Gli3 double mutants exhibited arrest of tooth development prior to the bud stage 29. Shh signaling is thus required for the development of early tooth germ.

Wnts

The vertebrate Wnt gene family represents a large and diverse group of signaling molecules involved in the patterning, proliferation and differentiation of a variety of organs and cell types 84. They signal through the Frizzled family of receptors, 85, 86, 87. A second type of Wnt receptor is related to the low-density lipoprotein (LDL) receptor and known as LRP5/6 88, 89. Wnt binds to both LRP5/6 and Frizzled to form a functional ligand-receptor complex that activates the canonical Wnt/â-catenin pathway and an alternative planar cell polarity pathway 90. It has become clear that secreted Wnt antagonists play important roles in regulating Wnt signaling. Members of the secreted Frizzled-related protein (sFrp) family appear to function as soluble modulators of Wnt signaling, presumably by competing with membrane Frizzled receptors for the binding of secreted Wnt ligands 91, 92, 93, 94. SFrps sequester Wnts in the extracellular space, preventing activation of both the canonical and the planar polarity Wnt pathways 90. Members of another family of Wnt antagonist Dickkopf (Dkk) antagonize Wnt signaling by acting as inhibitory ligands of LRP5/6 co-receptors. Dkk binds to LRP5/6 and blocks interaction between Wnt-Frizzled and LRP5/6, preventing the canonical Wnt pathway, but the Wnt-Frizzled complex can still signal via the planar polarity pathway. Thus Dkk would block the β-catenin pathway (canonical pathway) but leave the cells still capable of responding via the planar polarity pathway.

A number of Wnt genes are expressed in the developing teeth, with most of them restricted solely to the dental epithelium. At E10.5 when tooth forming sites and tooth patterning are determined, Wnt7b is expressed in the oral epithelium except the presumptive dental epithelium 72. Ectopic expression of Wnt7b to the tooth-forming region represses Shh expression and subsequently inhibits tooth formation, but not vice versa 72. Wnt7b thus seems to interact with Shh signaling to set up the ectodermal boundaries between oral and dental ectoderm, positioning the sites of tooth formation 72. Wnt10a and Wnt10b are expressed in both molar and incisor epithelium from E11.5 when dental epithelium become thickened, and remain there throughout the bud stage 28. At the E14.5 cap stage, both genes are expressed in the enamel knot, a putative signaling center for tooth patterning. In addition, Wnt4, Wnt6, and one Wnt receptor MFz-6 were expressed in the dental epithelium 36. In contrast, Wnt5a, sFrp2, and sFrp3 are expressed restrictedly in the dental mesenchyme 36. The first evidence for a role of Wnt signaling in tooth development came from the analysis of Lef1 knockout mice. Mutation in Lef1, a critical component of Wnt signaling pathway, led to an arrest of tooth development at the bud stage 33, 63, 95. Ectopic expression of Dickkopf (Dkk) driven by the Keratin-14 (K14) promoter in the dental epithelium blocked tooth development before the bud stage, although thickening of the presumptive dental epithelium was observed 96. Application of exogenous sFRP-3 protein to the explanted presumptive molar region retarded tooth bud formation. Smaller teeth formed in the sFRP-3-treated tooth grafts after subrenal culture 97. Wnt3 is also expressed in the enamel knots. However, overexpression of Wnt3 by the K14 promoter in the epithelium did not result in significant change of tooth morphology 98. Since mice deficient in many Wnt genes, including Wnt1, Wnt2, Wnt3, Wnt3a, Wnt4, and Wnt5a, either die too early for tooth phenotype analysis or do not exhibit a tooth phenotype, the detailed function for each Wnt member in tooth development is not clear at present 84, 99, 100.

TNF

The most prominent structure of tooth morphogenesis is the temporal epithelium structure, the enamel knot, a signaling center thought to direct tooth morphogenesis and to determine the final tooth morphology 61. This hypothesis has been strongly supported by the evidence from the studies of Tumor Necrosis Factors (TNF) signal pathway in tooth development. As the first inherited ectodermal deficiency to be described in the mouse 101, Tabby mutant displays a characterized tooth phenotype. While the size and number of molar cusps are significantly reduced, the incisors and the third molar are also frequently missed in these mice 102. Later, identical tooth phenotype was found in two other spontaneous mouse mutants: Downless and Crinkled 103, 104, 105, which together with Tabby turn out to be the different genes in the same signaling pathway. Tabby is identified as Ectodysplasin A1 (EdaA1), a member of TNF family. Downless and Crinkled are the TNF receptor EDAR and EDAR death domain adaptor (EDARADD) respectively 102, 106, 107, 108, 109, 110, 111, 112. Both Edar and Edaradd are expressed in the enamel knot while Eda is expressed in the outer enamel epithelium and can be released by proteolysis 110, 111, 113. All three mutants show an abnormal enamel knot, and consequently, the molars formed in these mutants show reduced number of cusps with shallow depth 102, 110, 112. Moreover, the cusp phenotype in Tabby mice could be rescued by applying exogenous EDA protein to the pregnant Tabby mice 114, while overexpression of an activated form of Edar by the K14 promoter in the epithelium leads to extra cusps in the molars 115. Similarly transgenic expression of EdaA1 driven by the K14 promoter produces extra teeth and extra cusps in the molar 116, 117. Taken together, these data suggest that molar morphology is regulated by the activated level of EDA signaling 117. It has been demonstrated that the EDA signal pathway involves a kinase cascade leading to nuclear factor kappa B (NF-κB) activation 118, 119, 120, 121, 122. The activity of NF-κB is regulated by its inhibitor (IKB) that again is regulated by IkB kinase (IKK). IKK contains one regulatory subunit IKKg, and two catalyst subunits IKKα and IKKβ. During the cap stage of mouse tooth development, the expression of NF-κB, IkB and Ikkγ is found in the enamel knot while Ikkα and Ikkβ are expressed in the outer edges of dental epithelium. Molars in the Ikkα mutants exhibit abnormal cusp morphology that resembles those found in the Eda, Edar, and Edaradd mutant mice, respectively 123. Moreover, IKKα shows a NF-κB independent function in the incisor development. The activation of NF-κB by the EDA signal may be mediated by TNF receptor-associated factor 6 (TRAF6) 124, known to interact with the receptors of TNF family and then activate NF-κB 125. Several Trafs have been shown to be expressed in the dental epithelium 126. Analysis of the morphology of molar in Traf6 mutant mice reveals more severe cusp abnormalities than those found in the Eda/Edar/Edaradd mutants 124. Thus, it is proposed that Traf6 would function in two independent ways to activate NF-κB protein. First, Traf6 binds Troy directly 127, an orphan TNF receptor strongly expressed in the enamel knot 128, and then Troy recruits Traf6 for the NF-κB activation 129. On the other hand, Edaradd also shows the capability to recruit TRAF6 for the signal relaying 130. Since both pathways convert to the NF-κB activation, it explains why similar degree of cusp abnormality is observed in the Traf6 mutant mice and the NF-κB repressed mice. These studies clearly demonstrate the crucial role of the TNF signal pathway in molar cusp formation.

In addition to the function in regulating molar morphogenesis, EDA signaling may also be vital in the maintenance or expansion of the tooth placodes, although it is dispensable in the primary induction of tooth field. Overexpression of EdaA1 in the oral epithelium at the early stage rescues the third molar in Tabby mice 131. However, overexpression of a constitutively active TNF receptor in the oral epithelium occasionally causes missing of the third molar but produces an extra tooth before the first molar at the same time in the transgenic mice 115. It is thus suggested that the finely balanced signal from EDA/EDAR is required for the expansion of the initial tooth placode for the third molar development. Overexpression of EdaA1 in the oral epithelium also results in supernumerary teeth before the first molar in the transgenic mice 116. Coinciding with the supernumerary tooth formation is the appearance of an extra Shh positive field in the distema region. Hence, it is possible that overexpressed EdaA1 actually maintains this tooth field and consequently leads to tooth formation although it can expand the placodes in a dose-dependent manner in other ectodermal organs 132.

ActivinβA is able to induce Edar expression in the dental epithelium where Wnt6 also induces EdaA1 expression 112. ActivinβA knockout mice show incisor and mandible molar arresting at the bud stage while maxillary molars develop normally 35. Follistatin is an Activin-antagonizing protein, but can also bind and inhibit the functions of BMP proteins with low affinity 133, 134. Molars of Follistatin knockout mouse also display shallow and aberrant cusps 135. Overexpession of Follistatin by the K14 promoter in mice results in lost of the third molar and aberrant cusps of which the upper molars are affected more severely than the lower molars 135. Since ActivinβA can induce Edar expression, Follistatin may affect cusp pattern by regulating EDAR level through inhibiting the activity of ActivinbA, and may have the same inhibitory effect on the molecules that function redundantly with ActivinbA in the maxillary molar. Moreover, Follistatin also regulates enamel patterning in the incisors by asymmetrically inhibiting BMP signaling and ameloblast differentiation 136.

Cell-cell signaling

In addition to mediating tissue-tissue interactions, growth factor signaling also participates in mediating cell-cell interactions within epithelial and mesenchymal tissues during organogenesis. This short-range signaling between cells is usually accomplished by one cell possessing a transmembrane receptor and the neighboring cell possessing a membrane-bound ligand. Notch signaling is one such system implicated in tooth morphogenesis 14. Notch is a large transmembrane receptor known to control cell fate decision and the formation of tissue compartments during embryonic development 137. Jagged and Delta-like transmembrane proteins are ligands to the Notch receptors. Binding of the membrane-bound ligands to the Notch receptor, which requires intimate cell-cell contacts, triggers the release of the cytoplasmic domain of Notch that subsequently functions as a transcription factor in the nucleus. The expression of several Notch ligands, including Delta1 and Jagged-1, has been detected in developing teeth 138, 139, 140, 141. The expression of Notch-1, -2, and –3 is detected in the dental epithelium from the initiation stage to the later differentiation stage 138. However, the expression of these Notch genes is excluded from the basal layer epithelial cells that eventually form ameloblasts. It is thus suggested that ameloblasts may be determined very early at the tooth initiation stage, and the absence of Notch gene expression may be associated with this determination 138.

Crosstalk between different signaling pathways

As discussed above, many signaling molecules of different families are expressed and used repeatedly throughout the tooth development. They are often expressed either in the same tissue layer or in the adjacent one to mediate cell communication between tissues and within one tissue. These different signaling pathways must be integrated into a regulatory network that operates to orchestrate gene expression. They either function synergistically or antagonistically to regulate gene expression precisely. For example, BMPs and FGFs can up-regulate the same target genes, such as Msx1 and Dlx2, but BMPs also counteract the effects of FGFs in the induction of Pax9 and Pitx2 in the presumptive dental mesenchyme and epithelium, respectively 11, 22. Numerous lines of evidence have demonstrated the existence of crosstalk between different signaling pathways. Fgf4 has been shown to be a direct target of Lef1, and can rescue the Lef1 mutant teeth 63, indicating an integration of FGF and Wnt pathways. Wnt pathway is also associated with TNF pathway. This is evidenced by the facts that Wnt6 induces EdaA1 expression in the dental epithelium, and a downregulation of Eda is seen in the dental epithelium of the Lef1 mutant molar 112. Furthermore, it was shown that FGF4 or FGF10 can partially rescue the cusp abnormalities of Tabby molar 102. FGF signaling seems to mediate, at least partially, the Wnt and TNF pathways in tooth development. Together with the evidence that Activin induces Edar expression 112, these results provide evidence for the existence of a regulatory network that integrate FGF, Wnt, TNF, and TGF-β signaling pathways in tooth development. In addition, FGF signaling is also linked with the Notch pathway. FGF10 was shown to stimulate the expression of L-Fng, a secretory modulator of Notch signaling, in the stellate reticulum 142.

ROLE OF TRANSCRIPTION FACTORS IN TOOTH DEVELOPMENT

In situ hybridization experiments have revealed the expression of numerous transcription factors in the developing mouse tooth where the expression patterns of these transcription factors often overlap with those of growth factors, suggesting a potential relationship between the two classes of gene products in inductive interactions [3, 4, 13; also see: http://bite-it.helsinki.fi/]. Knockout studies have provided opportunities to analyze the functions of these transcription factors in tooth development. By far, the bud stage is the most frequent stage at which tooth development arrests in various gene knockout mice, such as Msx1-/-, Lef1-/-, Pax9-/-, and Pitx2-/- mice 33, 143, 144, 145, 146.

The homeobox-containing genes

In the mouse, prior to any morphological manifestation of tooth development, a number of homeobox genes, such as Pax9, Msx1, -2, Dlx-3, 5, 6, 7, Barx1, Pitx2, Lhx-6, -7, are expressed with a specific spatial pattern in the first branchial arch. Based on these observations, a homeobox code hypothesis for tooth pattern formation was proposed 147, 148. The overlapping expression domains of homeobox genes may subdivide the jaw into different regions, and specify each tooth's position.

Barx1 expression is found to be restricted to the molar mesenchyme before and during tooth formation 23. The antagonistic effects of BMP4 and FGF8 from the overlying oral epithelium are responsible for establishing Barx1 expression domains specific in the molar mesenchyme whereas it is implicated in the determination of molar tooth type 12. Barx1-deficient mice have not been reported, but a recent study using RNAi knock-down demonstrates a critical role for Barx1 in the development of molar tooth germ from the bud stage to the cap stage (Song et al., unpublished results).

The expression of Msx1 and Msx2 is detected in the developing tooth germ in patterns that correlate with morphogenetic steps in tooth development 149, 150. Msx1 is strongly expressed in the dental mesenchyme including dental papilla and dental follicle throughout the lamina, bud, cap and bell stages of odontogenesis. Msx2 is co-expressed with Msx1 in the dental mesenchyme, but a component of Msx2 expression also shifts to the dental epithelium and is restricted to the enamel knot. Mice deficient for Msx1 exhibited an arrest of tooth development at the bud stage, while Msx2 mutants manifested later defects in tooth development, including a marked reduction in the stellate reticulum and abnormal patterning of the cusps 143, 151. Furthermore, tooth development was arrested at the lamina stage in the Msx1-Msx2 double mutant mice, suggesting a functional redundancy for Msx1 and Msx2 in tooth formation 34. Analyses of gene expression in Msx1 mutant tooth germ revealed a down-regulation of a number of genes, including Bmp4, Fgf3, Lef1, Dlx-2, Ptc, syndecan-1, and tenascin 31, 32, 34. Application of exogenous BMP4 to the Msx1 mutant tooth germ in vitro, or ectopic expression of Bmp4 in the Msx1 mutant dental mesenchyme via a transgenic approach, partially rescued Msx1 mutant tooth phenotype and restored expression of several down-stream genes including Lef1 and Dlx2 32, 44, 152, 153. Evidence that mutations in the MSX1 gene in humans cause tooth agenesis and orofacial celfting further demonstrates a vital role of Msx1 in odontogenesis 154, 155, 156, 157, 158, 159, 160.

Pax9, a member of paired domain family genes, is initially expressed in the presumptive dental mesenchyme at the future sites of tooth formation 11, and continues to be expressed in the dental mesenchyme at the following stages 144. The confined expression of Pax9 in the presumptive dental mesenchyme, via the combined effects of BMP and FGF activities, represents an early molecular marker of the dental mesenchyme 11. Targeted disruption of Pax9 in mice results in an arrest of tooth development at the bud stage, which is accompanied by a down-regulation of Bmp4, Msx1, and Lef1 in the dental mesenchyme 144. Pax9 thus functions upstream of these genes, and controls Bmp4 expression coordinately with Msx1 in the dental mesenchyme. Similarly in humans, mutations in PAX9 cause oligodontia 161.

The expression of Pitx2, a member of the Pitx/RIEG family of the Bicoid-related homeobox genes, is regulated positively by FGF8 and negatively by BMP4, in the presumptive dental epithelium 22, 162, 163. The lack of the Pitx2 gene activity in mice leads to a down-regulation of Fgf8 in the dental epithelium 145, 146, indicating a positive feedback loop between Pitx2 and Fgf8. The mutant mice also exhibit an arrest of tooth development at the bud stage. Further investigation revealed that the tooth development is sensitive to the total quantity of Pitx2 regardless of its isoforms. While the dose of Pitx2 increases, intermediate tooth phenotypes that are arrested between the bud stage and normal development were found in different Pitx2 conditional mutant mice. Moreover, Pitx2 may also be involved in the regulation of tooth orientation 164.

Other transcription factors

Lef1 is a member of HMG box family and a nuclear mediator of Wnt signaling 166. In the developing tooth germ, Lef1 is initially expressed in the thickened dental epithelium at E11. At the subsequent bud, cap and bell stages, Lef1 transcripts are detected both in the dental mesenchyme and in the immediately adjacent dental epithelium 33. BMP4 was shown to be responsible for the expression of Lef1 in the dental mesenchyme 32, 33. Loss of Lef1 function resulted in mice lacking teeth. Histological analysis indicates an arrest of tooth development at the bud stage 95. It was further demonstrated that Lef1 function is needed only transiently in the dental epithelium to control Fgf4 expression in the enamel knot, thus relaying a Wnt signal reception to a cascade of FGF signaling activities 33, 63. Application of exogenous FGFs was able to rescue the phenotype of Lef1 mutant teeth 63.

The Cbfa1/Runx2 protein is a critical transcriptional regulator of the osteoblast differentiation 166. Null mutation of Cbfa1 in mice leads to a complete lack of osteoblast differentiation and bone formation 167, 168, 169. Cbfa1/Runx2 haploinsufficiency in humans causes cleidocranial dysplasia (CCD), a syndrome characterized by generalized bone defects and supernumerary teeth arising from the permanent dentition that fail to erupt 170. In mice, Cbfa1/Runx2 mRNA expression begins in the condensed tooth mesenchyme at the early bud stage (E12 mouse embryos) and continues in the dental papilla through the cap stage until the early bell stage (E16) 171. Mice deficient in Cbfa1/Runx2 exhibit an absence of teeth that are arrested at an aberrant cap stage, and the mandibular molars are more severely affected than the maxillary molars and incisors 171, 172. FGFs are able to stimulate Cbfa1/Runx2 expression in dental mesenchyme, and both Fgf3 and Fgf10 are down-regulated in the Cbfa1/Runx2 mutant dental mesenchyme 171, 173. However, exogenous FGFs failed to rescue the tooth phenotype of Cbfa1/Runx2 mutants in vitro 173. Additional studies further indicated that Cbfa1/Runx2 is a downstream gene of Msx1. This conclusion is supported by the facts that the expression of Cbfa1/Runx2 is down-regulated in the dental mesenchyme of Msx1 mutants while Msx1 maintains a normal expression pattern in the Cbfa1/Runx2 mutant dental mesenchyme 173, 174. Since Msx1 also controls Fgf3 expression in the dental mesenchyme 34, Msx1 most likely regulates Cbfa1/Runx2 expression via FGF3.

A common function of these transcription factors in the tooth development is to regulate the expression of growth factors. It was therefore suggested that transcription factors participate in epithelial-mesenchymal interactions through linking the signaling loops between tissue layers by responding to inductive signals and regulating the expression of other signaling molecules 3.

STEM CELLS AND TOOTH REGENERATION

Stem cells are generally defined as cells that have the capacity to self-renew as well as to give rise to differentiated progeny 175. Stem cells are present in small numbers in many vertebrate adult and fetal tissues, including the hematopoietic system, nervous system, gut, gonads, skin, and olfactory epithelium. They are responsible for normal tissue renewal and for regeneration following damage. It was traditionally thought that only embryonic stem (ES) cells are pluripotent, while adult stem cells are restricted in their differentiative and regenerative potential to the tissues in which they reside 176. However, this view of adult stem cell potential has been challenged by the recent discoveries that bone marrow stromal cells (BMSCs), when stimulated by differentiated microenvironment cues, have the capacity to differentiate into a range of cell types of different organs, including chondrocytes, osteoblasts, adipocytes, cardiac and skeletal muscle cells, neurons and astrocytes 177, 178, 179, 180, 181, 182, 183. BMSCs, also known as mesenchymal stem cells or colony-forming units fibroblastic, are a population of noncirculating bone marrow-derived cells with remarkable plasticity. They can be isolated based on their adhesive properties, and are capable of clonal expansion in culture 184, 185. BMSCs share characteristics with other multipotent stem cells, and give rise to differentiated progeny, including both mesenchymal and nonmesenchymal lineages. The properties of rapid expansion in vitro and multipotential of differentiation make BMSCs one of the most important adult stem cell sources for potential therapeutic use and tissue engineering.

Dentinal repair in adult animals takes place through the activity of odontoblasts that differentiate from a precursor cell population existing in the dental pulp tissue. Using methodology developed to isolate and characterize BMSCs, clonogenic and highly proliferative dental pulp stem cells (DPSCs) have been isolated from adult human teeth 186. These DPSCs maintained their high rate of proliferation even after extensive subculturing, and generated a dentin/pulp-like complex that is composed of mineralized matrix when grafted into the dorsal surface of immunocompromised beige mice 186, 187, 188. However, whether the DPSCs are able to support the formation of a functional tooth has yet to be tested. Both DPSCs and BMSCs are associated with mineralized tissue. Despite the different origins and developmental potentials in vivo, the human DPSCs and BMSCs share many features with respect to their expression of various proteins commonly present in the extracellular matrix of bone and dentin. Both of them are able to form calcified deposits in vitro 186. Furthermore, microarray analyses of gene expression profiles of the human DPSCs and BMSCs indicate that these two distinct precursor populations have similar level of gene expression for more than 4000 known genes, with only a few differentially expressed ones 189. In humans, the transition from the deciduous teeth to the permanent teeth is a postnatal development with very unique and dynamic processes. It was found that exfoliated human deciduous tooth contains multipotent stem cells [stem cells from human exfoliated deciduous teeth (SHED)] 190. SHED are distinct from DPSCs with respect to their capability to differentiate into odontoblasts and generate only dentin-like structure, instead of reconstituting a complete dentin-pulp-like complex as DPSCs do. In addition, as compared to DPSCs, SHED has higher proliferation rate and are able to differentiate into a variety of cell types including neural cells and adipocytes. SHED thus might represent a more immature population of multipotent stem cells 190.

Although the isolation of DPSCs in rodents has not been reported, a group of putative stem cells in dental epithelium has been identified in the continuously growing mouse incisor 142. These stem cells reside in the cervical loop epithelium and give rise to enamel forming ameloblasts. It was also demonstrated that proliferation of the epithelial stem cells is regulated by FGF signals derived from dental mesenchyme 142, and FGF10 in the dental mesenchyme is a survival factor that maintains the stem cell population in the developing incisor germs 191.

Tissue engineering is considered as one of the most powerful approaches to repair or replace an injured tissue or organ in the future. Although a 3-dimensional functional organ has not yet been generated from any stem cells, a functional pancreatic islet-like structure assembled from directed differentiating embryonic stem cells was recently reported 192. In vitro generation or assembly of an implantable human tooth from stem cells would have extremely significant implications for the dental practice. Animal models of tooth generation in vitro will provide the molecular and cellular basis for the future application in humans. In biodegradable scaffold seeded with single cell suspension dissociated from the third molar tooth germ of six-month old pig jaws, tooth structures formed containing dentin, odontoblast, a well-defined pulp chamber, putative Hertwig's rooth sheath epithelia, and enamel organ 193. However, the bioengineered teeth are very small and do not conform to the size and shape of the scaffold. It was also shown that a scaffold is not necessary for ex vivo development of tooth explants. Re-aggregates of dissociated mouse molar tooth germ are able to form well-differentiated teeth underneath the mouse kidney capsule without the need of a scaffold [194; Song et al, unpublished data]. Surprisingly, it was demonstrated recently that mouse stem cells, including ES cells, neural stem cells, and BMSCs, could be induced to reprogram into odontogenic fate to support tooth formation when proper odontogenic signals are provided 195. These cells, when aggregated and recombined with the E10.5 mouse molar epithelium which possesses the odontogenic potential, could all respond to the inductive signal from the dental epithelium, and initiate odontogenesis that result in tooth fromation 195. These studies support the idea that the odontogenic process can be initiated in stem cells with non-dental origin when proper odontogenic signals are provided.

CONCLUSION

The studies on molecular mechanisms underlying tooth morphogenesis in mice have proliferated in the past several years, providing a wealth of data on expression patterns of developmental regulatory molecules as well as their functions 61. Given the considerable homology between mouse and human odontogenesis as well as the profiles of gene expression, unveiling the molecular basis involved in mouse tooth morphogenesis will provide important insight for studying genetically related dental abnormalities and tooth regeneration in humans. The profiles of gene expression and function in normal tooth development are vital for studying tooth regeneration or reconstitution of a tooth organ because the process of morphogenesis, histogenesis, and cytodifferentiation as well as the molecules involved are likely to be similar. The knowledge gained from adult stem cell biology and molecular regulation of tooth morphogenesis makes it possible to manipulate stem cells to become odontogenic fate. The rapid expansion of our knowledge of experimental embryology, developmental and molecular biology, stem cell biology, and biomimetics makes tooth regeneration a realistic possibility in the near future, which would greatly improve the quality of human life 196.