Signaling Networks Regulating Tooth Organogenesis and Regeneration, and the Specification of Dental Mesenchymal and Epithelial Cell Lineages

SUMMARY

Teeth develop as ectodermal appendages from epithelial and mesenchymal tissues. Tooth organogenesis is regulated by an intricate network of cell–cell signaling during all steps of development. The dental hard tissues, dentin, enamel, and cementum, are formed by unique cell types whose differentiation is intimately linked with morphogenesis. During evolution the capacity for tooth replacement has been reduced in mammals, whereas teeth have acquired more complex shapes. Mammalian teeth contain stem cells but they may not provide a source for bioengineering of human teeth. Therefore it is likely that nondental cells will have to be reprogrammed for the purpose of clinical tooth regeneration. Obviously this will require understanding of the mechanisms of normal development. The signaling networks mediating the epithelial-mesenchymal interactions during morphogenesis are well characterized but the molecular signatures of the odontogenic tissues remain to be uncovered.

Unlike many human epithelial appendages, teeth have no regenerative capacity. To reprogram nondental cells for clinical tooth regeneration, a detailed understanding of tooth organogenesis is required.

Teeth are one of the most diverse organs in vertebrates both morphologically and functionally. Mammalian teeth belong to four tooth families: incisors, canine, premolars, and molars, and they are replaced either once or not at all. Humans have teeth from all four tooth families and excluding molars, the teeth are replaced once (Fig. 1A). The laboratory mouse (Mus musculus), which is the most common model animal in tooth development studies, has a much derived dentition. It lacks the canine and premolars and has only one incisor and three molars separated by a toothless diastema in each half of the jaw (Fig. 1B). Furthermore, the mouse incisors grow continuously but the teeth are not replaced. In contrast, reptiles, fish, and amphibians can replace their teeth multiple times during the life of the animal. Teeth of these nonmammalian species are usually simpler in shape. Thus, during evolution the complexity of tooth shape has increased, whereas the replacement capacity has been reduced.

Figure 1.

The dental formula of human and mouse, and a schematic representation of tooth development. The permanent dentition of human consists of two incisors, a canine, two premolars, and three molars in each half of the jaw (A). Mice have one incisor and three molars separated by a toothless diastema in each half of the jaw (B). Tooth development starts from the dental lamina, a thickening of the epithelium. Individual placodes form within the dental lamina. The growing epithelium forms a bud and the dental mesenchyme condenses around the epithelium. During morphogenesis, the epithelial tissue folds to cap and bell shapes. Primary and secondary enamel knots in the enamel organ regulate the growth and shape of the tooth. During cell differentiation, enamel-secreting ameloblasts and dentin-secreting odontoblasts mature from the epithelial and mesenchymal cell compartments. The permanent tooth develops lingually to the deciduous tooth from an extension of the dental lamina (C).

The same conserved signaling pathways that regulate most aspects of embryonic development are required for tooth development, and the core regulatory network seems to have been in place already when teeth appeared in evolution (Fraser et al. 2009; Tummers and Thesleff 2009). It is noteworthy that teeth develop as epithelial appendages and share the same regulatory molecules during the first steps of initiation and morphogenesis with other ectodermal organs. However, unlike many other human epithelial appendages, human teeth have no regenerative capacity. The adult human teeth contain stem cells that are capable of differentiating to cells producing the extracellular matrix of tooth-specific mineralized tissues, but so far they have not been shown to have morphogenetic potential. The replacement of adult teeth in humans by tissue engineering appears still a distant goal and it is obvious that more research on stem cell regulation and the molecular control of early tooth development is required. In this article we review the current knowledge about the mechanisms involved in tooth morphogenesis and replacement, and how the epithelial and mesenchymal cell lineages acquire odontogenic competence and differentiate into tooth-specific cells depositing the dental hard tissues, and discuss the future challenges and scenarios of tooth bioengineering.

1. MORPHOGENESIS AND CELL DIFFERENTIATION DURING TOOTH DEVELOPMENT

Teeth are initiated from two tissue components: the surface epithelium and the underlying mesenchyme. The dental mesenchyme derives from cranial neural crest cells that migrate into the frontonasal process and first branchial arch. In mammals the epithelium originates from ectoderm, whereas in fish and some amphibians, pharyngeal teeth derive from the endoderm.

The first sign of tooth development is the formation of the dental lamina, a horseshoe-shaped epithelial stripe along the mandible and maxilla (Fig. 3B). The teeth form within the dental lamina, where their development starts from placodes, local thickenings of the epithelium (Figs. 1C and 3B,C). Probably all teeth in one tooth family are initiated sequentially from a single placode. For instance, the mouse molars develop successionally, starting from the first molar and followed by initiation of the second and third molars from a posterior extension of the dental epithelium.

Figure 3.

Shift of the odontogenic potential from epithelium to mesenchyme between dental lamina and placode stages as shown by reciprocal tissue recombinations (Mina and Kollar 1987). Epithelium is capable of inducing tooth development when recombined with nondental mesenchyme until E11 stage of mouse development. At E12 the odontogenic potential has shifted to mesenchyme, and it can induce tooth development when recombined with nondental epithelium (A). Pitx2 is expressed in the dental lamina of the mouse lower jaw at E11 (t = tongue) (B). At E12.5 the Pitx2 expression is restricted to the placode epithelium of the incisors (arrows) and molars (asterisks) (C).

The individual teeth develop from an epithelial bud that grows down to the underlying mesenchyme. The neural crest-derived mesenchyme becomes specified as the dental mesenchyme condenses around the bud and gives rise to all the dental tissues except enamel. The epithelial bud invaginates at its tip and its cervical loops grow to encompass the dental papilla mesenchyme, which gives rise to the dental pulp and odontoblasts forming dentin (Fig. 1C). The mesenchyme surrounding the epithelium and dental papilla becomes the dental follicle and gives rise to the periodontal tissues and cementoblasts forming the cementum.

During morphogenesis the epithelium acquires cap and bell shapes and is called enamel organ. It consists of several cell types: the inner enamel epithelium facing the dental papilla and differentiating to enamel producing ameloblasts, the outer enamel epithelium facing the dental follicle, and the stellate reticulum and stratum intermedium cells in between. The growth and folding of the inner enamel epithelium during the bell stage determine the size and shape of the tooth crown. The shape becomes fixed when the organic matrices of dentin and enamel mineralize because no remodeling of either dentin or enamel takes place later.

Root formation is initiated after crown development when ameloblast differentiation reaches the future crown-root boundary, and the cells of the inner enamel epithelium no longer differentiate into ameloblasts. Instead they form the Hertwig’s epithelial root sheath (HERS) with the outer enamel epithelium. HERS proliferates and migrates downward guiding root formation, and it also induces the differentiation of odontoblasts forming root dentin. HERS has a limited growth potential, which determines the length of the root. The disintegration of HERS results in the formation of an epithelial network called epithelial rests of Malassez (ERM) and this allows the cells of dental follicle to come in contact with root dentin and their differentiation into cementoblasts depositing cementum on the root surface. The periodontal ligament that connects the tooth to the bone is formed by fibroblasts differentiating from the dental follicle cells. In addition, the dental follicle gives rise to osteoblasts that form the alveolar bone where the fibers of the periodontal ligament are embedded (Nanci 2008). The dental follicle has an important function later in tooth eruption as it regulates bone remodeling around the tooth (Marks and Cahill 1987).

2. SIGNAL NETWORKS AND SIGNALING CENTERS

All aspects of tooth morphogenesis are regulated by epithelial-mesenchymal interactions, which are mediated by the conserved signaling pathways including Hedgehog (Hh), Wnt, Fibroblast growth factor (FGF), Transforming growth factor β (Tgfβ), Bone morphogenic protein (Bmp), and Ectodysplasin (Eda) (Fig. 2). Their interactions, targets, and expression patterns have been elucidated in considerable detail in teeth (http://bite-it.helsinki.fi; Bei 2009b; Tummers and Thesleff 2009). Epithelial signaling centers play a pivotal role regulating the different steps of tooth development. There are three sets of such centers: the placodes, the primary enamel knots, and the secondary enamel knots. Their formation is regulated by epithelial-mesenchymal interactions and they all express largely the same array of multiple growth factors.

Figure 2.

Cross talk between epithelium and mesenchyme through the conserved signaling pathways regulates all aspects of tooth development. When tooth development is initiated, signals from the epithelium activate a set of transcription factors in the mesenchyme, leading to condensation of the mesenchyme and formation of the epithelial placode (A). The enamel knot is a signaling center expressing multiple signaling molecules that induce reciprocal signals from the mesenchyme. Enamel knots determine the position of the cusps and initiate differentiation of odontoblasts (B). Tgfβ, Bmp, and Shh signaling regulate epithelial-mesenchymal interactions in the cervical loop of the mouse incisor. They support the maintenance and proliferation of the stem cells as well as ameloblast differentiation and enamel production (C).

All ectodermal organs begin to develop from a placode, and the molecular mechanisms of tooth placode formation and signaling are shared to a great extent with placodes of other organs such as hairs (Mikkola 2009b). One of the important genes regulating placode formation is the transcription factor p63 that is expressed throughout the surface ectoderm. When p63 function is deleted in mice, the placodes of teeth and other ectodermal appendages do not develop, but the dental lamina forms (Laurikkala et al. 2006). Key signaling pathways including Bmp, Fgf, Notch, and Eda are impaired in the absence of p63 (Laurikkala et al. 2006). The importance of the signaling that takes place at the placode stage is further highlighted by the phenotype of several mouse mutants where tooth development stops before epithelial budding (Bei 2009b).

Ectodysplasin (Eda) is a signal of the tumor necrosis factor family and signals through its receptor Edar that is locally expressed in the placodes of all ectodermal appendages as well as in primary and secondary enamel knots (Mikkola 2009b). Mutations in the Eda pathway genes cause the human syndrome hypohidrotic ectodermal dysplasia (HED) manifesting multiple missing teeth as well as defects in other ectodermal organs, e.g., sparse hair and reduced sweating (Mikkola 2009b). Mice lacking functional Eda often lack third molars or incisors and the cusp patterning of molars is abnormal, indicating a requirement of Eda in the function of placodes and enamel knots (Pispa et al. 1999). Mice that overexpress Eda in epithelium (under keratin14-promotor) develop an extra tooth in front of the molars as well as supernumerary hairs and mammary glands (Fig. 5A,B) (Mustonen et al. 2003). The targets of Eda signaling include molecules from all the other important signaling pathways (e.g., Shh, Fgf20, Dkk4, ctgf, Follistatin) making Eda a key regulator of ectodermal organ development (Mikkola 2009b).

Figure 5.

Supernumerary teeth, tooth replacement, and continuous tooth renewal. Overexpression of ectodysplasin in the surface epithelium results in development of a supernumerary tooth (arrow) in front of the molars (m1-3) in the K14-Eda mice (A). The rudiment of the supernumerary tooth (blue arrow) in front of the first molar (red arrow) can be visualized by Shh expression in E14.5 lower jaw of K14-Eda embryo (B). The permanent canine (C) of the ferret develops as an extension of the dental lamina (dl) on the lingual side of the deciduous canine (dC) at E33 (C). Both deciduous and permanent canine express Pitx2 in the epithelium (D). Sostdc1 is expressed in the intersection between the dental lamina and the deciduous third premolar (dP3) at the time when permanent P3 is initiated in the E35 ferret embryo (arrow) (E). Stimulation of Wnt signaling by stabilized β-catenin in mouse oral epithelium leads to the development of multiple small teeth from a single E14 mutant tooth germ cultured under the kidney capsule (F). Fgf20 is expressed in the enamel knots of upper and lower molars of E14.5 wild-type mouse embryos (G). Multiple enamel knots expressing Fgf20 have been induced in the dental epithelium of β-cat^Δex3K14/+ embryos (H).

The primary enamel knot appears in the dental epithelium at the transition from bud to cap stage. In addition to directing crown formation, in molars it determines the positions of the secondary enamel knots which in turn mark the positions of the cusp tips in the molar crown (Fig. 2B) (Jernvall et al. 2000). Wnts are important upstream regulators of enamel knots as shown by the requirement of Lef1 for Fgf4 expression in the enamel knot (Kratochwil et al. 2002) and the induction of new enamel knots and placodes by forced activation of Wnt/β-catenin signaling in oral epithelium (Järvinen et al. 2006; Wang et al. 2009). More than a dozen different signal molecules belonging to all four conserved signal families are locally expressed in the primary and secondary enamel knots. The enamel knots initiate and regulate the folding of the epithelium by stimulating the surrounding epithelium to proliferate through Fgfs (Fgfs 3, 4, 9, and 20) and remaining nonproliferative themselves. They express the cyclin-dependent kinase inhibitor p21 and lack Fgf receptors making them insensitive to the proliferative signals (Jernvall et al. 1998; Kettunen et al. 1998). The Fgfs also signal to dental mesenchyme and induce e.g., Runx2, and Fgf3, which signals back to epithelium illustrating the bidirectional Fgf signaling between epithelium and mesenchyme regulating tooth morphogenesis (Klein et al. 2006). Shh from the enamel knot stimulates epithelial morphogenesis indirectly via the mesenchyme (Gritli-Linde et al. 2002).

Important aspects of enamel knot signaling are the modulation and fine-tuning, which affect the patterning of the secondary enamel knots and thereby the pattern of molar cusps via lateral inhibition and reaction diffusion mechanisms. Different shapes of molars can be generated by mathematical modeling using parameters of activating and inhibiting enamel knot signals, and it has been suggested that changes in signaling during evolution are responsible for the species-specific cusp patterns (Salazar-Ciudad and Jernvall 2002). This hypothesis has gained experimental support from phenotypes of transgenic mice where signal modulation has resulted in phenotypes resembling teeth of other species. Examples include molars of K14-Eda resembling kangaroo teeth, and of Sostdc1 knockout (inhibitor of Wnt and Bmp signaling) resembling rhino teeth (Kangas et al. 2004; Kassai et al. 2005). Furthermore, epithelial deletion of Dicer, which is required for processing of microRNA (miRNA), results in the modulation of molar cusp pattern (Michon et al. 2010).

Fine-tuning of the activity of the conserved signaling pathways controls many other aspects of tooth formation as well. For example, a supernumerary tooth forms in front of the first molar in several mutant mouse lines when signaling activity is modulated. These teeth do not represent de novo tooth induction. Instead they form from activation of the development of a vestigial tooth rudiment found in wild-type mice in the diastema and represent premolars lost during the evolution of rodents. Examples are the K14-Eda mouse (Fig. 5A,B) (Mustonen et al. 2003) and the Sprouty mutants (Klein et al. 2006). In the Osr2 knockout an extra tooth develops lingually to the first molar (Zhang et al. 2009). This is accompanied by spreading of Bmp4 expression to the lingual mesenchyme and results probably from a subsequent broadening of the dental field (Mikkola 2009a). The relative sizes of the mouse molars are influenced by activation and inhibition between successionally developing teeth (Kavanagh et al. 2007), the size and number of mouse incisors is affected by fine-tuning Bmp signaling in the placodes (Munne et al. 2010), and the continuous growth and enamel deposition in incisors can be modulated by the levels of Fgf, Activin, and Bmp signaling in the epithelial stem cell niche (Fig. 2C) (Wang et al. 2007).

3. REGULATION OF THE IDENTITY AND DIFFERENTIATION OF ODONTOGENIC MESENCHYMAL AND EPITHELIAL CELL LINEAGES

Classical recombination experiments have shown that the odontogenic potential shifts from the epithelium to mesenchyme in mouse teeth between embryonic days 11 and 12, i.e., around the time of placode formation (Fig. 3). When epithelium of the first branchial arch from an E9-11 mouse embryo was recombined with second arch mesenchyme, a tooth formed (Fig. 3A) (Mina and Kollar 1987). Similarly, first arch epithelium from an E9-10 embryo induced tooth formation when recombined with cranial neural crest cells that normally form the dental mesenchyme and, interestingly, also when combined with premigratory trunk neural crest cells (Lumsden 1988). At E12 the epithelium no longer has inductive potential and now the first arch mesenchyme can induce tooth formation when recombined with second arch epithelium (Fig. 3A). The mesenchyme from E13 or older tooth germs has the information on the tooth identity as the shape of the tooth in the reciprocal recombinations between incisor and molar epithelium and mesenchyme will form according to origin of the mesenchyme (Kollar and Baird 1969). In addition, the dental papilla can induce tooth formation when recombined with limb epithelium (Kollar and Baird 1970). It has been proposed, based on in vitro experiments, that the incisor versus molar identity of teeth is determined by the level of Bmp signaling (Tucker et al. 1998). However, this conclusion was challenged recently by the observation that inhibition of Bmp signaling caused partial splitting of the incisor placode resulting in the formation of two fused incisors rather than incisor to molar transformation (Munne et al. 2010).

The molecular basis of odontogenic competence in early jaw epithelium and later in the condensed dental mesenchyme remains elusive. As all the genes that are known to regulate tooth development are also expressed in other developing organs, it seems that there is no single tooth-specific gene that defines the odontogenic tissues. Currently only few genes such as Sonic hedgehog (Shh) and the transcription factor Pitx2 are known to be restricted to the dental lamina (Fig. 3B,C) (Keränen et al. 1999). It is not known how the dental lamina becomes established, and to date there is no mouse mutant reported where the dental lamina would be missing. All teeth develop within the dental lamina, even in mice where extra teeth are induced. Activation of Wnt signaling in the epithelium induces supernumerary placodes throughout the surface epithelium and they give rise to various epithelial appendages. Yet extra teeth form only in the region of dental arches and mostly in connection with other teeth (Järvinen et al. 2006; Wang et al. 2009). These observations indicate that the odontogenic competence is present only in the oral region.

It is likely that spatiotemporal patterns of the epithelial signals are involved in the shift of competence to mesenchyme (Fig. 2A). In addition to Shh, which is restricted to the dental lamina, many Wnt ligands are expressed in the oral epithelium (Sarkar and Sharpe 1999). Wnt/β-catenin signaling regulates Fgf8 expression in the early epithelium (Wang et al. 2009), and placodes do not form in mice overexpressing the Wnt inhibitor Dkk1 (Andl et al. 2002). Bmp4 and Fgf8 are expressed in the jaw epithelium in overlapping patterns before any morphological sign of tooth development. Bmp4 is expressed more distally at the site where molars will develop and Fgf8 more proximally in the incisor region (Neubüser et al. 1997). Epithelial Bmps induce the expression of Bmp4 in the mesenchyme before bud stage correlating with the shift in the odontogenic potential (Vainio et al. 1993). Also, Wnt/β-catenin signaling in the incisor mesenchyme stimulates the expression of Bmp4 that in turn regulates Shh in the epithelium (Fujimori et al. 2010). Furthermore, Wnt signaling was shown to be required in the molar mesenchyme for the bud to cap stage transition and primary enamel knot formation (Chen et al. 2009). The signals induced in the mesenchyme by epithelial Fgfs and acting reciprocally to the epithelium include Activin, Fgf3, and Fgf10 (Ferguson et al. 1998; Kettunen et al. 2000). These signals regulate the subsequent epithelial morphogenesis and the enamel knot formation (Fig. 2A).

The shift of the odontogenic competence from epithelium to mesenchyme is accompanied by the induction of important transcription factors in the dental mesenchyme (Fig. 2A). The deletion of the function of several of these either alone or together with another transcription factor in the same family results in tooth arrest at placode or bud stage. All four signal pathways have been shown to be involved in the regulation of these transcription factors (Bei 2009b). For example, Bmp4 induces the expression of Msx1 and Fgf8 induces the expression of Pax9 (Vainio et al. 1993, Neubüser et al. 1997). Other targets of Bmp and Fgf signaling in mesenchyme at this stage include Lhx6,7, Barx1, Dlx1,2, and Runx2 (Bei 2009b, Tummers and Thesleff 2009). The Shh mediators Gli2 and Gli3 are expressed in the mesenchyme and are required for tooth formation (Hardcastle et al. 1998). In addition, the expression of Lef1, a Wnt effector, shifts from the epithelium to the mesenchyme together with the shift in the odontogenic potential and is regulated by Bmp4 in mesenchyme (Kratochwil et al. 1996). Perhaps a combination of these transcription factors constitutes the code for the odontogenic identity of the mesenchyme.

The differentiation of the tooth-specific cell types is intimately linked with epithelial morphogenesis. Odontoblasts and cementoblasts differentiate from the lineage of dental mesenchyme, the odontoblasts from the dental papilla, and cementoblasts from the dental follicle, whereas ameloblasts differentiate from the epithelial lineage. They are responsible for the formation and the deposition of the extracellular matrices of the tooth-specific mineralized tissues, dentin, cementum, and enamel, respectively. It is not known exactly at which stage of tooth formation the cells become committed, but the final steps of odontoblast and ameloblast differentiation have been analyzed in detail during the bell stage of tooth formation.

The mesenchyme is first induced to differentiate into odontoblasts by the inner enamel epithelium. The differentiation starts from the cusp tips and proceeds downward to cervical and intercuspal directions. Signals in Tgfβ/Bmp families have been implicated in odontoblast induction (Ruch et al. 1995), and it was shown recently that the conditional loss of Smad4, a mediator of Tgfβ/Bmp signaling, from the dental papilla prevents the terminal differentiation of odontoblasts and dentin deposition (Li et al. 2011a). As the formation of enamel knots is temporally associated with the initiation of odontoblast differentiation at the cusp tips, the enamel knot signals have been suggested to play a role (Fig. 2B) (Thesleff et al. 2001). One of these signals, Wnt10b, was suggested to regulate the expression of dentin sialophosphoprotein (Dspp) and odontoblast differentiation (Yamashiro et al. 2007). The localization of Wnt reporter activity in odontoblasts is also in line with the role of Wnts in the process (Suomalainen and Thesleff 2010). In addition, the basement membrane is important for the polarization and differentiation of the odontoblasts and serves presumably as a reservoir of signal molecules (Thesleff and Hurmerinta 1981; Ruch et al. 1995). Dentin is composed mainly of type I collagen, dentin phosphoprotein, and Dspp, and mutations in these genes cause dentinogenesis imperfecta in humans (Shields et al. 1973).

After the odontoblasts have been induced to differentiate, they signal back to the epithelium. The signals from the mesenchyme involved in the ameloblast induction include Bmp2, Bmp4, and Tgfβ1 (Fig. 4) (Coin et al. 1999; Wang et al. 2004). In addition, Shh from the epithelial stratum intermedium cells is required to support ameloblast differentiation and maturation (Dassule et al. 2000; Gritli-Linde et al. 2002). Other epithelial growth factors regulating ameloblasts are TFGβ1, Wnt3, Eda, and Follistatin (Bei 2009a). Ameloblasts express transcription factors such as Sp6 and Msx2 that have been shown to play a role in amelogenesis in mice (Bei 2009a). Mutations in ameloblast-specific genes including ameloblastin, amelogenin, enamelin, and Mmp20 cause human amelogenesis imperfecta (Bei 2009a). Very little is known about the molecular regulation of cementoblast development. Bmp signaling was reported to induce cementoblast differentiation from dental follicle cells, whereas Wnt signaling promotes their proliferation (Zhao et al. 2002; Nemoto et al. 2009).

Figure 4.

Bmp4 is one of the signals regulating ameloblast induction. A schematic view of the postnatal mouse incisor shows the asymmetrical deposition of enamel only on the labial side of the tooth and the cervical loop stem cell niche (A). Amelogenin protein is present in the ameloblasts (a) and in the first enamel matrix on the labial side of newborn (NB) incisor (arrow) but not on the lingual side [asterisk; o, odontoblasts]) (B). Bmp4 is expressed in the mesenchyme and is intense in the odontoblasts (arrows) of the developing incisor at E16. The white line surrounds the epithelium (C). A bead soaked in Bmp4 protein induces ameloblastin expression in E16 incisors (D). (B and D reprinted, with permission, from Wang et al. 2004.)

4. REGULATION OF TOOTH REPLACEMENT, CONTINUOUS GROWTH, AND STEM CELLS IN TEETH

As the mouse teeth are not replaced, relatively little is known about the mechanisms of tooth replacement in ±mammals. Histological observations in nonmodel animals indicate that replacement teeth develop from the dental lamina associated with their predecessors (Luckett 1993; Järvinen et al. 2009). The ferret (Mustela putorius furo) replaces its incisors, canines, and premolars, and it was shown that the deciduous teeth are connected to each other by a continuous dental lamina, and the permanent teeth start to grow from the lingual side of each deciduous tooth as an extension of the dental lamina (Fig. 5C–E) (Järvinen et al. 2009). Similarly, in the reptiles the replacement tooth arises from an outgrowth of the dental lamina each time the previous tooth has grown to a certain size (Richman and Handrigan 2011). In contrast, in the fish species studied, there seems to be no successional dental lamina, and the new teeth are initiated directly from the epithelium of the previous tooth or from the oral epithelium (Smith et al. 2009).

Wnt signaling has been associated with tooth replacement both in mammals and reptiles and may be a key factor regulating tooth renewal across vertebrates (Järvinen et al. 2009; Richman and Handrigan 2011). In the ferret, the expression of Sostdc1, an inhibitor of Wnt and Bmp signaling, marks the border between the deciduous tooth and the dental lamina that gives rise to the permanent tooth (Fig. 5E) (Järvinen et al. 2009). The expression of Axin2, a feedback inhibitor of Wnt signaling, was also detected in the mesenchyme between the tooth and the growing dental lamina (Järvinen et al. 2009). During snake tooth replacement, there is Wnt activity in the tip of the dental lamina and it is promoted by Shh and Bmp signaling from the mesenchyme (Richman and Handrigan 2011).

The phenotypes of some human syndromes and their mouse models support the role of Wnt signaling in tooth replacement. Mutations in the human AXIN2 gene cause oligodontia, which specifically affects permanent teeth (Lammi et al. 2004). On the other hand, supernumerary teeth are common in familial adenomatous polyposis (FAP), which is caused by mutations in APC, an inhibitory component of the Wnt pathway, and the patients also develop odontomas, benign tumors composed of numerous small teeth (Wang and Fan 2011). A similar phenotype is seen in mice when Wnt signaling is activated in the epithelium either by deletion of APC or stabilization of β-catenin (Fig. 5F–H) (Järvinen et al. 2006; Liu et al. 2008; Wang et al. 2009). Sp6^−/− (Epiprofin) mutants have a similar phenotype but this gene has not been associated with human conditions (Nakamura et al. 2008; Wang and Fan 2011). The teeth in the Wnt gain-of-function mouse models were shown to develop successionally from previous teeth resembling the continuous generation of the simple-shaped replacement teeth in fish and reptiles. This led to a suggestion, that Wnt signaling may have been involved in the reduction in the replacement capacity and in the gain in tooth complexity during evolution (Järvinen et al. 2006).

Interestingly, the phenotypes of two syndromes suggest that the capacity for continued tooth replacement can be unlocked in humans. The supernumerary teeth were suggested to represent a third dentition in cleidocranial dysplasia (CCD) and in a novel craniosynostosis syndrome, caused by mutations in the transcription factor RUNX2 and interleukin receptor IL11RA, respectively (Jensen and Kreiborg 1990; Nieminen et al. 2011). Unfortunately, the mouse models of these syndromes do not show supernumerary teeth, likely because the mouse teeth are not normally replaced, and they are therefore not suitable for studies on tooth replacement (D’Souza et al. 1999; Nieminen et al. 2011). Runx2 has been associated with Fgf as well as Wnt signaling in tooth development as the induction of the Wnt inhibitor Dkk1 in dental mesenchyme by epithelial Fgf4 requires Runx2 (James et al. 2006).

Putative epithelial stem cells have been identified during tooth replacement in gecko, a reptile (Handrigan et al. 2010). These cells reside in the lingual side of dental lamina and express some known stem cell marker genes. It is possible that the reduction of tooth replacement capacity in mammals to maximally one replacement has involved depletion of such stem cells and that they are maintained in the cleidocranial dysplasia and craniosynostosis syndromes.

Although the human teeth do not regenerate and their development is completed already during adolescence, there are stem cells in the adult teeth. Human dental mesenchymal stem cells were first isolated from the dental pulp and when transplanted they formed dentin (Gronthos et al. 2000). Stem cells in periodontal ligament were shown to produce cementum and periodontal ligament-like structures (Seo et al. 2004). Similar stem cells were also characterized in exfoliated deciduous teeth and third molars (Rodriguez-Lozano et al. 2011). Epithelial stem cells may reside within the epithelial cell rests of Malassez as these cells can be induced to ameloblastlike cells (Shinmura et al. 2008).

Some mammals have teeth that grow continuously throughout life and thus have stem cells. The most studied such tooth is the mouse incisor, which harbors epithelial stem cells in a niche situated in the cervical loop at its proximal end (Fig. 2C) (Harada et al. 1999). The mouse incisor has an asymmetric structure with enamel deposited only on the labial side, whereas the lingual side is covered by dentin (Fig. 4A). The asymmetry arises from differences between the lingual and labial cervical loops as only the larger labial cervical loop contains label-retaining stem cells and transient amplifying (TA) cells (Harada et al. 1999; Seidel et al. 2010). The stem cells reside within the stellate reticulum cells in the core of cervical loop, which is surrounded by dental mesenchyme. The progeny of stem cells invades the basal epithelium and proliferates as TA cells before differentiating into ameloblasts (Figs. 2C and 4A,B).

Mesenchymal signals play key roles in the regulation of the epithelial stem cells and their progeny (Fig. 2C). Fgf10 is the key mesenchymal signal required for epithelial stem cell maintenance and proliferation, and Fgf3 has a partly redundant function as stimulator of TA cell proliferation (Wang et al. 2007). Mesenchymal Fgfs act in a regulatory loop with epithelial Fgfs, notably Fgf9, and stimulation of Fgf signaling by deleting the function of Sprouty genes results in extensive growth of the incisors and ectopic deposition of enamel on the lingual surface (Klein et al. 2008). Lingual enamel also forms in Follistatin knockout mice, whereas enhanced Follistatin expression results in complete absence of enamel from labial side as well as in growth inhibition (Wang et al. 2004). Follistatin is expressed in the lingual epithelium and it antagonizes Activin function in the cervical loop epithelium while, interestingly, inhibiting Bmp4 function in the zone of differentiation, which prevents enamel formation. It was shown that Bmp signaling is required for ameloblast differentiation (Fig. 4) (Wang et al. 2004). Accordingly, when the Bmp inhibitor Noggin is overexpressed in epithelium, the mouse incisors grow extensively and lack enamel (Plikus et al. 2005). Bmps and Activin function in a regulatory network with Fgf3, which is inhibited by Bmp4, which is in turn repressed by Activin that is strongly expressed in the labial dental mesenchyme but not on the lingual side. This contributes to the asymmetric production of TA cells only in the labial cervical loop (Wang et al. 2007).

Fgf and Shh signaling play a role in the postnatal homeostasis of the TA cell production in the mouse incisors but not in stem cell survival (Parsa et al. 2010; Seidel et al. 2010). On the other hand, Wnt signaling activity is not detected in the stem cells in the cervical loops (Suomalainen and Thesleff 2010). Some stem cell marker genes, such as Lgr5, Bmi1, Oct3/4, and Yap, have been localized in the incisor stem cells (Suomalainen and Thesleff 2010; Li et al. 2011b). Despite the intense investigations on the signal pathways regulating the incisor stem cell niche, the characterization of these stem cells is still on its way.

5. FUTURE CHALLENGES: STEM CELL-BASED BIOENGINEERING OF TEETH

Different scenarios have been proposed for bioengineering of human teeth. One possibility could be the direct induction of tooth development in the jaws with activators such as Wnt and Eda. However, although supernumerary teeth are induced in mouse models and in human syndromes by modulation of signal pathways, it is not likely that this approach would function in the adult jaws. The main reason is that the supernumerary teeth in mice—as well as in the rare human syndromes—form from the tissue associated with developing teeth and such teeth would not be present in adult jaws anymore. The stem cells discovered in adult teeth described above, including the mesenchymal stem cells in dental papilla, pulp, and periodontal ligament have the capacity to generate cells forming dentin, cementum, and periodontal ligament, but it is unlikely that they have morphogenetic potential, and even less likely that the cells could be targeted in vivo to undergo tooth morphogenesis.

A more realistic approach would be to engineer a tooth in vitro and implant it to patient’s mouth. It has been proposed that such teeth could be generated by growing cells in tooth-shaped scaffolds. However, taking into account the complex structure and organization of the hard and soft tissues of teeth and the fact that tooth size and shape emerge during the multistep process of morphogenesis together with the periodontal tissue attaching the tooth to the bone, it is difficult to imagine how a functional tooth could be developed within a scaffold. Therefore, the preferable way would be to trigger the initiation of tooth development program in progenitor cells and let the tooth develop itself.

The classical tooth bud transplantation and tissue recombination experiments have shown that the program for tooth morphogenesis is present very early in the jaws and that a tooth bud can form a complete tooth when transplanted to various ectopic sites. The proof of principle experiments in mouse have already shown that even dissociated embryonic dental epithelial and mesenchymal cells can regenerate a tooth germ in vitro and that this forms a functional tooth when implanted to the jaw of an adult mouse (Nakao et al. 2007; Oshima et al. 2011).

To use such an approach in human therapy, one would need to replace the embryonic dental cells with adult cells preferably with tooth forming capacity. Obviously, both epithelial and mesenchymal cell lineages are needed, but based on the classical recombination experiments only one cell type needs to have the odontogenic potential. Although there is some evidence that adult mouse bone marrow stem cells can form a tooth together with embryonic branchial arch epithelium (Ohazama et al. 2004), it is probable that in the recombinations the nonodontogenic mesenchyme should have properties of cranial neural crest and the epithelium should be ectodermal. The inductive potential of the current human dental stem cell lines has not been explored. However, although dental stem cells from adult teeth could perhaps be used for tooth bioengineering because they are likely to share characteristics with embryonic dental tissue, collecting dental stem cells from adults is challenging as it would imply sacrificing a tooth from the patient needing a new tooth.

Odontogenic cells might be produced from adult somatic cells by iPS cell technology or direct reprogramming (Hanna et al. 2010). Thus they could be first reprogrammed to embryonic stem cells by iPS technology and programmed further to dental epithelial or mesenchymal cell fates. Oral mucosal epithelium and stromal cells could be feasible sources for reprogramming because their developmental history is likely more similar to dental tissues. Alternatively, the oral mucosal cells or other adult somatic cells could perhaps be directly converted to tooth epithelium and mesenchyme as recently shown in other tissues (Zhou et al. 2008; Hanna et al. 2010).

The knowledge lacking at the moment is the molecular signatures of the epithelial and mesenchymal lineages that could be used in reprogramming. The key genes likely include transcription factors expressed by the early embryonic tissues such as Pitx2 in the branchial arch epithelium and Msx1,2, Dlx1,2,5, Runx2, Pax9, Lhx6,7,8, and Prx1,2 in the odontogenic mesenchyme (Fig. 2A) (Thesleff and Tummers 2008). So far these are only educated guesses based on expression patterns and mutant phenotypes (http://bite-it.helsinki.fi; Bei 2009b).

Finally, it should be noted that it is unrealistic to aim at generating a perfect tooth crown, because it is rather obvious that the right shape and size of the crown as well as the color and proper structure of enamel cannot be generated by bioengineering. Therefore, the crown needs to be completed prosthetically. The most important aspect of the bioengineered tooth would be a functional root providing physiological anchorage of the tooth to jaw bone. However, initiation of root formation, without a preceding crown appears impossible, at least by mimicking normal developmental mechanisms. The physiological anchorage is lacking in the titanium implants that otherwise are successfully used for tooth replacement. To this end interesting experiments have been performed in minipigs, where stem cells from the root apical papilla and periodontal ligament stem cells were used. When the cells were seeded on a root-shaped cylindrical hydroxyapatite/tricalcium phosphate scaffold, and implanted in the jaw, dentin, cementum, and periodontal ligament were generated and a structure resembling root developed (Sonoyama et al. 2006). It is not yet known whether the bioengineered root has adequate physiological properties to be used in clinical tooth replacement.

6. CONCLUDING REMARKS

Studies on the laboratory mouse have given a great deal of knowledge on the molecular regulation of tooth initiation, morphogenesis, and stem cell maintenance. However, before the building of teeth by tissue engineering becomes a reality, more detailed understanding of the process of tooth development and regeneration is required. In particular, the gene regulatory networks during cell lineage specification in dental epithelium and mesenchyme need to be understood more thoroughly and the origins of the two types of progenitor cells to be used for tooth bioengineering should be determined.