Teeth are seen in all jawed vertebrate classes from fish to mammals. There is such tremendous diversity among vertebrates that it is possible to find an animal representing every possible type of dentition, ranging from a complete loss of teeth (edentulism) as in birds and turtles to partial loss as in rodents to multiple sets of teeth occurring in the oral and pharyngeal cavities in fish. In the face of this enormous variation, most of our understanding of tooth development is derived from studies on just a few model organisms, including the mouse and, more recently, zebrafish. There is a great need to develop additional non-model organisms in order to gain further insight into questions relating to the evolution of dental pattern, tooth anatomy, and tooth replacement capacity. For this reason we began to study the tooth ontogeny in squamates (snakes and lizards). Our goal is to reintroduce some of the older literature in order to highlight the excellent descriptive work carried out on reptilian dentitions. In addition, we will discuss recent work from our own lab and others that has provided molecular insights into the development of snake and lizard teeth.
Global Patterning Decisions—Positioning and Number of Tooth Rows
The dentition is formed as a result of an epithelial-mesenchymal interaction between oral ectoderm and neural crest-derived mesenchyme. There is no overt thickening present in the oral epithelium at the earliest stage, but instead there is a band of gene expression marking the position of the future tooth row/s. This has been termed the odontogenic band (Smith et al.,2009b). The band is characterized by expression of the secreted signal Shh (Sonic hedgehog; (Buchtová et al.,2008; Cobourne et al.,2004; Fraser et al.,2004,2006,2009; Vonk et al.,2008) as well as the transcription factor Pitx2 (paired-like homeodomain transcription factor 2; Fraser et al.,2009; Jernvall et al.,2000). Odontogenic-specific gene expression in the epithelium and the presence of a neural crest-derived mesenchyme are required to form teeth. It is immaterial whether the interacting mesenchyme is Hox-negative from the face or Hox-positive from the second and more caudal pharyngeal arches. Both types of neural crest-derived mesenchymes are capable of forming teeth (Fraser et al.,2009; Mina and Kollar,1987; Van der Heyden and Huysseune,2000). The participating epithelium can either be ectodermal as in the case of oral teeth (Buchtová et al.,2008; Cobourne et al.,2004) or endodermal as in the case of pharyngeal teeth in fish or amphibian (Fraser et al.,2009; Soukup et al.,2008; Van der Heyden and Huysseune,2000).
In amniotes (reptiles and mammals), as opposed to other vertebrates, teeth are restricted to the oral surface. In the lower jaw, the dental pattern invariably consists of a single, marginal row of teeth. In the upper jaw, however, there are a wide range of patterns ranging from a single row of marginal teeth as seen in crocodilians to parallel rows of marginal and palatal teeth mainly found in snakes (Mahler and Kearney,2005). The exceptions within these taxa come mainly from the burrowing or fossorial snakes, which generally lack palatal teeth (Mahler and Kearney,2005). Ultimately, the marginal teeth attach to the maxillary bone, and the palatine teeth are affixed to the palatine bone, pterygoid bone or both. In most snakes, the palatine and pterygoid teeth exist as one continuous straight row. In many lizards, however, the palatine teeth are disorganized and spread out over much of the bone surface (Mahler and Kearney,2005).
The presence of palatine teeth is documented in the most primitive extinct amniotes and thus it is considered a basal condition for both the mammalian and reptilian lineages (Carroll,1988). Modern mammals do not have palatal teeth, but they may have retained a latent ability to form these teeth. In a mouse mutant lacking functional Osr2, embryos develop lingual supernumerary teeth in both the upper and lower jaws (Zhang et al.,2009). Thus, Osr2 may be one of the genes whose function is to repress the formation of palatine teeth in mammals. We hypothesize that in reptiles there is relatively lower Osr2 expression in the medial side of the maxillary prominence, where the palatal tooth row originates. Osr2 may be higher in the mandible than in the maxillary prominence of amniotes since no species bears a second, more medial tooth row in the lower jaw that is analogous to the palatal row of the upper jaw.
What is the correlation between adult dental pattern and expression of Shh in the palatal odontogenic band? Our work in snakes has shown that in the upper jaw there are two mirror-image bands of Shh expression on each side of the head and these expression domains mark the positions of the future marginal and palatal tooth rows (Buchtová et al.,2007,2008). Interestingly, we noted that a single odontogenic band spans both the pterygoid and palatine bones, implying that the teeth associated with these bones have a common embryonic origin. The extensive wholemount in situ analyses of Shh expression in snake embryos carried by Richardson et al. shows that nearly all species have a continuous domain of Shh expression in the palatal odontogenic band that spans the future palatine and pterygoid teeth (Vonk et al.,2008). This is also what we found in analysis of serial sections of pythons (Buchtová et al.,2008). The only exception to this rule was Calloselasma rhodostoma (a viper). This species bears a palatal odontogenic band that overlies the future site of the pterygoid, but does not extend anteriorly to span the palatine bone (Vonk et al.,2008). Interestingly, in this species, the presence of palatine bone teeth is variable (Mahler and Kearney,2005), and thus once again gene expression closely matches the eventual dental pattern in the jaws. Thus far, the data support the existence of a single developmental field for palatal teeth, but that this is sometimes truncated at the anterior end.
In lizards there also is a correlation between Shh expression and the position of the tooth rows. Neither of the two species that we have studied, Eublepharis macularius and Pogona vitticeps, has palatal teeth, and as expected, they both lack palatal Shh expression (Richman and Handrigan, unpublished data). It will be interesting to sample other lizards that do have palatal teeth in order to see whether Shh is expressed in a distinct band as in snakes or in a more disorganized, punctate pattern. With all these variations on a theme, it is hard to determine the most basal condition for palatal teeth. One possibility, based on the more frequent appearance of pterygoid teeth (Mahler and Kearney,2005), is that the pterygoid field is more evolutionarily conserved than the anterior palatine field.
Although there is strong comparative data showing a link between Shh and specification of the tooth row, other epithelial signals could be involved. In the mouse, there is a complementary and antagonistic relationship between the Wnt and Shh signaling pathways during odontogenic band specification (Sarkar et al.,2000). Our preliminary data for the Wingless-related growth factors Wnt10b and Wnt6 show that they are expressed in complementary patterns to Shh in the oral epithelium. We have demonstrated antagonistic interactions at slightly later stages within the dental lamina of the ball python (Handrigan and Richman,2010b); therefore it is likely that these pathways regulate each other during odontogenic band specification in squamates as they do in the mouse.
Participation of the Bmp pathway is also possible since Bmp2 and 4 are also expressed in the odontogenic band in mice, partially overlapping the Shh domain (Bitgood and McMahon,1995). However several different gene targeting experiments designed to reduce Bmp signaling in the epithelium have not been able to disrupt the formation of the odontogenic band (Liu et al.,2005; Plikus et al.,2005). The most profound phenotype was produced with a K14-Cre driving expression of Noggin, a BMP antagonist, in the epithelium (Plikus et al.,2005). In these mutant mice, mandibular molars failed to form, but it was still possible to see placode-like thickenings in the epithelium. This implies that an odontogenic band formed, but additional gene expression analysis with a band marker (e.g., Shh, Pitx2) is needed to confirm this possibility. Until an epithelial Cre-driver with earlier onset of expression than K14 is discovered, we cannot determine the role of Bmp—or for that matter Shh or Wnt signaling—in specifying the position of the odontogenic band in the mouse. In the future, studies in reptiles or other model organisms such as the fish may be able to clarify the role of Bmp and other signals in tooth row positioning.
The Lack of a Relationship Between Placodes and Individual Teeth in Reptiles
In the mouse, the odontogenic band is transient and breaks down into placodes that will mark the positions of individual teeth. Shh expression is found in all the placodes and is not specific to a tooth type. This pattern of Shh expression in dental placodes has also been observed in diphyodont shrews (Miyado et al.,2007; Yamanaka et al.,2007) and is predicted to exist in all dentate mammals. In contrast, in snakes and lizards, the odontogenic band persists and goes on to form a continuous dental lamina in the jaw. First there is a shallow thickening of the epithelium and then later this extends deep into the jaw mesenchyme to form a multilayered dental lamina (see Fig. 1). A similar dental lamina is present in crocodilians although when the first-generation teeth are formed they evaginate rather than grow into the jaw (Harris et al.,2006; Westergaard and Ferguson,1986a,b,1987). In the upper jaws of snakes, there are three dental laminae: marginal, palatal, and premaxillary. When the dental lamina extends into the jaw, the Shh expression domain in the oral epithelium is broken down into discrete foci (Vonk et al.,2008), which persist until the latest stages examined in pythons (Stage 6; Buchtová et al.,2008). Similar oral epithelial expression is present in P. vitticeps and E. macularius (Handrigan and Richman,2010a). These punctuated Shh expression domains in the snake embryo are not equivalent to the mammalian tooth placodes. Instead, transverse sections clearly show that Shh is restricted to the non-dental, oral epithelium that is immediately adjacent to the dental lamina (lingual in marginal teeth, labial in palatal teeth). The function of this oral expression domain is unknown; however, sagittal sections show a correlation between Shh expression and the spacing of tooth families along the anterior–posterior axis of the jaw (Vonk et al.,2008). It is unlikely that signals from this remote oral ectoderm influence the replacement of teeth deep within the jaw. In crocodilians, Shh also appears to be expressed in periodically spaced foci along the tooth row (Harris et al.,2006). No sections were provided in this study, but since first-generation teeth are very superficial (Westergaard and Ferguson,1986b), we suspect the expression is actually within the inner enamel epithelium of each forming tooth bud rather than in placodes. Thus the lack of Shh expressing placodes in reptiles may be a feature of the continuous dental lamina, a character that is distinct from mammals.
Setting up Asymmetries in the Early Squamate Dental Lamina
The dental lamina of squamates (snakes and lizards) is a simple epithelial ribbon of cells that grows into the neural crest-derived mesenchyme. Prior to tooth bud formation, the lamina appears to be homogeneous along the length of each tooth row. However, we have found that even from the earliest stages of invagination from the odontogenic band, there are already asymmetries at the molecular level. For simplicity, from now on, we will restrict ourselves to describing expression in the marginal teeth, but the same patterns occur in mirror image in the palatal tooth row. The dental lamina begins to take on a distinct lingual angulation that is obvious in three-dimensional reconstructions of lizard and snake dental laminae (Buchtová et al.,2008; Handrigan and Richman,2010a). In this review we will use the words labial, lingual, mesial, and distal when discussing the axes of teeth/dental lamina and the labels anterior, posterior, medial, and lateral when discussing the relationship of the teeth to the jaw bones. The oral–aboral axis will refer to the third axis passing through the vertical axis of the dental lamina or teeth.
The side of the dental lamina at an obtuse angle to the oral epithelium is always the side that participates in tooth formation (see Fig. 1). The marginal teeth form on the labial side of the marginal dental lamina, whereas palatal teeth form on the lingual side of the palatal dental lamina. Almost as soon as the dental lamina begins to grow into the underlying mesenchyme, Shh and its receptor Ptc1 become preferentially expressed on the lingual side (Fig. 2b,d). Meanwhile, the gene ectodysplasin receptor (Edar; Fig. 2l) and the Wnt pathway genes Lef1, Axin2, Wnt6, and Wnt10b (Fig. 2e–h) are preferentially expressed on the labial side of the dental lamina in a complementary pattern to the Hh pathway genes. There are also asymmetries in mesenchymal gene expression. Bmp2 (bone morphogenetic protein) is more highly expressed on the lingual side under the oral ectoderm, whereas Bmp4 is more evenly expressed (Fig. 2j,k). Proliferation is also asymmetric with higher levels on the labial rather than the lingual surface (Handrigan and Richman,2010b). There are identical patterns of gene expression and proliferation in palatal teeth although they are in mirror image in relation to the marginal teeth (i.e., proliferation is highest on the labial side of marginal dental laminae, but on the lingual side of palatal laminae). It is noteworthy that until such molecular data could be collected on non-venomous snakes (Buchtová et al.,2007,2008; Handrigan and Richman,2010a,2010b) and venomous snakes (Vonk et al.,2008), there was no direct proof of homology between these two tooth rows. The well-developed palatal teeth in the snake have allowed us to conclude that the same developmental mechanisms are used regardless of where teeth are positioned within the oral cavity.
Two possible functions for the labial-lingual asymmetries in gene expression are (1) to direct growth at an angle into the jaws, thus creating more space for the forming teeth or (2) to impart differential responsiveness to growth factors in lingual and labial cells.
We have carried out one study that supports a role for Shh in directing the growth of the dental lamina. In organ cultures of python jaw explants, we showed that blocking Hh signaling with cyclopamine leads to a straightening and shortening of the dental lamina (Buchtová et al.,2008). Treatment of gecko teeth also shortens the dental lamina (Handrigan and Richman,2010a). In other studies, we perturbed the Wnt signaling pathway (Handrigan and Richman,2010b) and, in contrast to cyclopamine treatment, the angulation of the dental lamina was not affected. Thus the data point to a specific role for Hh signaling in orienting the dental lamina within the jaw mesenchyme.
There is also asymmetry in the oral–aboral axis of the python dental lamina. Genes that are excluded from the free- end or growing tip of the dental lamina include Shh (Fig. 2d), Ptc1 (Fig. 2b), Wnt6 (Fig. 2g) and Wnt10b (Fig. 2h). In contrast, the Wnt pathway readout genes Lef1 and Axin2 (Fig. 2e,f) are more strongly expressed at the aboral tip of the dental lamina, implying that the pathway is more active there than near the oral surface. Bmp signaling is also more active at the tip of the lamina than at the base as demonstrated by the expression of phospho-Smads (Handrigan and Richman,2010b). Finally, there is always higher proliferation at the tip of the dental lamina (Fig. 2i; Handrigan and Richman,2010b). Thus, high proliferation is correlated with increased Wnt and Bmp signaling in the python dental epithelium. The significance of all these patterns of gene expression and proliferation is that tip cells are engaged in molecular signaling that may specify them towards a different fate than cells elsewhere in the dental epithelium.
The periodic spacing of teeth in reptiles predicts that there are molecular signals that set up the anterior-posterior spacing of teeth along the length of the tooth row. Osborn (1973) proposed a clone theory in which teeth of a similar morphology are established by clones of cells coming from the previously formed tooth, moving from anterior to posterior in the jaw. Thus a group of similarly shaped teeth are derivatives of one clone. Once a tooth forms, a zone of inhibition is created around it thus controlling the distance between teeth (Osborn,1973,1978,1998). This is an interesting concept to reconcile with a dental lamina that appears morphologically homogeneous during early dental development. At this time we can summarize by saying that genes enriched on the labial or ‘tooth-forming’ side of the dental lamina (e.g., Wnt6, Wnt10b, Bmp4, Bmp2) do not appear to be expressed any differently at one level along the length of the lamina compared to another. A priority of our future work will be to describe gene expression along the entire length of the dental lamina. We are particularly interested in studying pathway inhibitors, such as secreted frizzled protein 2 (Wnt pathway), Dickkopf (Wnt pathway), Noggin (Bmp pathway), and Sprouty (Fgf pathway). Localized expression of these antagonists around the labial dental lamina in areas that are going to form teeth would support the clone or zone of inhibition theory (Osborn,1978). We are just entering an exciting period of research where the molecular tools are available to test these decades-old hypotheses.
Crown Morphogenesis—The Presence of Enamel Knots in Reptile Teeth
The fully formed or mature dental lamina is defined as one that is studded with teeth at various stages of development. In this section, we will focus on morphogenesis of individual teeth and compare this process to that occurring in mammalian teeth. Reptilian teeth pass through bud, cap, and bell stages resembling those described in mammals (Luckett,1993). The epithelium thickens on the labial side of the dental lamina and surrounding it the odontogenic mesenchymal condensation forms. Next, a multilayered enamel organ is formed. It consists of the inner enamel epithelium, which gives rise to ameloblasts, the stellate reticulum, an intermediate supportive layer, and the outer enamel epithelium, which gives rise to the successional dental lamina on the lingual side of the tooth. There is no stratum intermedium in the reptilian teeth we have examined, possibly due to the smaller size of the enamel organs. The neural crest-derived mesenchyme enclosed by the enamel organ gives rise to the dental papilla, the source of dentin-forming odontoblasts. During bell stage the shape of the crown is established, and differentiation of odontoblasts and then ameloblasts occurs. This is followed rapidly by the secretion of dentin and enamel going in an incisal-to-apical direction. In other words, the first matrix to be laid down is always at the cusp tip and this is conserved across vertebrate taxa.
In mammals, the primary enamel knot is found at the tip of bud-stage teeth and is incorporated into the inner enamel epithelium of cap-stage teeth. The enamel knot epithelium is thickened and can be distinguished morphologically. The other features that define the mammalian enamel knot are the absence of proliferation and higher levels of apoptosis (Jernvall et al.,1994,2000; Matalova et al.,2004). The prevailing hypothesis is that the primary enamel knot is a signaling center (Thesleff et al.,2001) and serves three important functions. First, the enamel knot secretes mitotic factors that diffuse outwards to stimulate proliferation of the adjacent inner enamel epithelial cells and cervical loops. Second, the dental papilla is induced by the primary enamel knot of bud-stage teeth. Third, the primary enamel knot induces secondary enamel knots, which then form additional cusps and induce odontoblasts in the adjacent dental papilla. In this way, matrix deposition begins at the cusp tips. Deletion of several of these signaling molecules or components of their pathways results in arrest at the bud or early cap stage confirming their importance in tooth formation.
The extent of evolutionary conservation of primary and secondary enamel knots outside of the mammals is not known. It is difficult to pinpoint the location of enamel knots in sharks (Smith et al.,2009b), fish or amphibians by morphology alone. In reptiles, one study reported seeing a slightly thickened enamel knot in the American alligator (Westergaard and Ferguson,1987), but in our opinion this is not as obvious as in the mammal. The situation in squamates is similar in that no localized thickening of the inner enamel epithelium can be detected. However we have now gathered molecular evidence from snakes and lizards that allow us to reevaluate whether or not squamates bear an evolutionary homologue of the mammalian enamel knot.
The mammalian enamel knot is the only region of the inner enamel epithelium that is non-proliferative. The inner enamel epithelial cells of squamates are also non-proliferative as seen with both PCNA and BrdU labeling (Buchtová et al.,2007,2008; Handrigan and Richman,2010a,b). However, unlike mammals, it is the entire inner enamel epithelium that is non-proliferative, extending down to the cervical loops. Apoptosis is also different in reptile teeth compared to mammals. There is no evidence of apoptosis in the inner enamel epithelia of bearded dragon (Handrigan and Richman,2010a) or python teeth (Buchtová et al.,2008). Instead apoptosis was focused in the stellate reticulum. Thus in the reptile, it appears to be unnecessary to decrease the number of cells in the inner enamel epithelium. Interestingly, studies on mice deficient in enzymes that mediate apoptosis can form normally shaped molars in the apparent absence of apoptosis in the enamel knot (Matalova et al.,2006; Setkova et al.,2007). These data, together with our reptile data, imply that there is no functional requirement for apoptosis to take place in the inner enamel epithelium, as it relates to crown formation.
The other defining feature of the mouse enamel knot is that it expresses many signaling molecules, including members of the Bmp, Fgf (Fibroblast Growth Factor), Hh, and Wnt families. Are similar sets of signaling molecules expressed in the inner enamel epithelium of reptilian teeth? We have examined Shh expression as well as its receptor Ptc1 and find that, like in mammals, Shhis expressed in the inner enamel epithelium in cap stage teeth and signal is lost at the cusp tip once ameloblasts begin to polarize (Buchtová et al.,2008; Handrigan and Richman,2010a). There is also relatively lower expression of Shh in the cervical loops. The cells that sense the Shh signal, identified on the basis of Ptc1 expression, are located in the cervical loops and everywhere in the enamel organ except for the most central area of the inner enamel epithelium (Handrigan and Richman,2010a). Expression of Gli2 is somewhat complementary to Shh in that it is very low in the central inner enamel epithelium (Fig. 3a). There is particularly high expression of Gli2 in the interstitial cells of the dental lamina and in the stratum intermedium suggesting this transcription factor mediates Hh signaling in these regions. Other genes are also expressed in the inner enamel epithelium of reptiles, including Wnt7a (Handrigan and Richman,2010a), Wnt6 (Fig. 3b), Bmp4 (Fig. 3e), and Edar (Fig. 3f). Bmp4 transcripts are transiently expressed; they are only found in early cap stage teeth (Fig. 3e), but are lost as soon as the inner enamel epithelium begins to differentiate (Handrigan and Richman,2010b). We also reported that there is a small group of cells with phospho-Smad activity within the inner enamel epithelium (Handrigan and Richman,2010b). This suggests that these cells are receiving and responding to the Bmp signal either from the inner enamel epithelium (autonomous) or from the adjacent dental papilla (non-cell autonomous). Finally, our finding that Axin2 is expressed in the inner enamel epithelium (Fig. 3d) implies that the canonical Wnt pathway is also active there. In summary, even though the enamel organ of a squamate tooth lacks a localized swelling of cells akin to the mammalian enamel knot, the inner enamel epithelium appears to have similar signaling properties to those of the mammalian enamel knot.
Successional Lamina and Diphyodont Versus Polyphyodont Dentitions
The cycle of tooth replacement in reptiles has been a subject of interest for over a century. The careful work of Edmund, Osborn, and others on lizards (Cooper et al.,1970; Edmund,1960; Osborn,1971; Westergaard,1988; Woerdeman,1919) and Westergaard on alligator teeth (Westergaard and Ferguson,1986a,b,1987) have described in detail the formation of generations of teeth and how they replace the functional set. The common theme to these studies is that tooth replacement in reptiles is a similar process to diphyodont mammals since teeth are shed on a schedule, independent of wear or injury. There is no indication that unexpected tooth loss is the trigger for tooth replacement to begin.
Reptilian replacement teeth arise from the successional lamina, an epithelial outgrowth derived from the outer enamel epithelium of the previous generation tooth. Almost as soon as each new tooth is specified a new outgrowth can be seen for the successional lamina (see Fig. 1). The successional lamina is in effect the continuation of the dental lamina. The histological structure of the reptilian successional lamina is similar to the early dental lamina in that there are two rows of cuboidal epithelial cells each contacting a basement membrane with loosely arranged interstitial cells (Handrigan et al.,2010).
We have observed several commonalities between the successional lamina and the free end of the early dental lamina. Both demonstrate high levels of cell proliferation. Both are negative for Hh activity while being positive for Wnt and Bmp activity. The odontogenic condensation adjacent to each structure is a source of Bmp ligands. We have dissected the interactions between these three signaling pathways during early dental lamina development. Mesenchymal Bmp signals promote Wnt activity in the dental epithelium and mesenchyme, while Hh signaling restricts Wnt activity to the tip of the lamina. A similar negative role of Hh signaling on Wnt activity was recently demonstrated in the vestigial teeth in the mouse diastema (space between the incisors and molars where no teeth are formed; Ahn et al.,2010). On the basis of these conserved features, we have suggested that the behavior of the successional lamina will be identical to the tip of the early dental lamina and, moreover, that the dental lamina tip cells are specified to form the successional lamina even before the first tooth is formed (Handrigan and Richman,2010b).
Just how similar is the successional lamina among different vertebrates that replace their teeth? What is known about proliferation patterns and gene expression and, importantly, is there any functional data that indicate if the mechanisms are conserved with reptiles? We should begin by saying that many of the fish models including zebrafish and trout do not form their teeth from a successional lamina. Instead, replacement teeth form directly from a thickening of the outer enamel epithelium (Smith et al.,2009b). In cichlids, replacement teeth originate from an epithelial downgrowth which originates directly from the oral epithelium and not the predecessor tooth bud (Vandervennet and Huysseune,2005). Unfortunately, in animals with true successional laminae, such as sharks (Smith et al.,2009a) or diphyodont mammals (Järvinen et al.,2008,2009), no descriptive molecular data have been published. Thus for the moment, ours are the only data to explore the properties of this important dental structure in any vertebrate.
The Evolutionary Loss of Polyphyodonty in Mammals
What can be the reasons for the loss of tooth replacement capacity seen in mammals? Indeed, early reptilian-like mammals (e.g., pelycosaurs) could replace their teeth, indicating that this ability has been subsequently lost (Carroll,1988). Why are no other teeth formed after the second generation? There are at least two possible explanations. The first is that a successional lamina is never formed from the enamel organs of the permanent dentition. The second is that a successional lamina forms but remains vestigial.
At the outset we should clarify that the mouse dentition is not a good model in which to distinguish between these two possibilities. The mouse dental formula is typical of the apomorphic (derived or advanced) rodent dentition, which is characterized by a loss of canines and bicuspids. The rodent dental pattern consists of continuously erupting incisors and molars, neither of which arises from a successional lamina. The mouse molar is homologous to human molars, which are not successional teeth (Cobourne and Sharpe,2010). The permanent molars are derived from a continuation of the dental lamina that extends distally just under the oral ectoderm (Ooë,1981). No deciduous teeth are shed to make way for permanent molars to erupt. Similarly, the murine molars are not derived from a successional lamina, but instead are formed directly from the molar placode. In mutant mice where extra teeth have been induced, these are not a result of de novo induction of a successional lamina from the existing molar teeth. Sometimes the supernumerary teeth are a result of increased cell survival of vestigial buds in the diastema (Ahn et al.,2010; Klein et al.,2006; Mustonen et al.,2003; Ohazama et al.,2009; Peterkova et al.,2009) and, in other cases, due to participation of the oral/vestibular ectoderm in tooth formation (Järvinen et al.,2006; Wang et al.,2009). Nevertheless, even though the origins of supernumerary teeth in mice are different than in humans, they often arise from defects in the same signaling pathways that are involved in successional tooth formation (Bmp, Wnt).
Is there evidence for a “post-permanent” dentition in diphyodont mammals? In the recent literature on mammalian tooth development, there are two diphyodont mammals as well as a marsupial that have been studied in detail, the shrew (Järvinen et al.,2008; Yamanaka et al.,2007), ferret (Järvinen et al.,2009) and opossum (Moustakas et al.,2010). From the published figures, it is not possible to discern a successional lamina coming from the permanent tooth buds. Although Bolk dismissed the idea of a post-permanent dentition (Bolk,1922), there is intriguing morphological evidence of a rudimentary successional lamina in bell-stage permanent incisors (Ooë,1981). Such histological material is relatively rare (derived from post-natal jaws), so this is partly why not much is known about the presence of a successional lamina extending from the permanent enamel organs. The other evidence that might support the existence of a vestigial successional lamina in humans is the very frequent occurrence of supernumerary teeth in the incisor and premolar regions of the jaw (Gardiner,1961; Solares and Romero,2004). Gardiner proposed that supernumerary teeth could arise between the primary and permanent tooth or terminal to the permanent tooth. There are two clinical features of human supernumerary teeth that support the idea they derive from a successional lamina derived from the permanent tooth—they are usually lingual to the normal tooth and they are developmentally delayed with respect to the permanent tooth. We speculate that certain permanent teeth, such as lateral incisors and bicuspids, have a greater propensity to form this vestigial successional lamina, and this is why supernumerary teeth most commonly occur in these areas of the mouth.
In reptiles we have much better evidence for the existence of a vestigial successional lamina. Not all dentate reptiles are endowed with indefinite tooth replacement. It turns out that in agamids, such as the bearded dragon, there is no tooth replacement of the functional teeth (Cooper et al.,1970). Thus, in sharp contrast to other reptiles we have studied, we see only a single fully-formed set of teeth in pre-hatching specimens of the bearded dragon. These teeth are associated with a finger-like successional lamina on their lingual side (Fig. 4a, inset). It is possible that the successional lamina is a transient structure, disappearing some time after hatching. Developmental arrest of the successional lamina could be the result of increased apoptosis (Fig. 4b,c), which would counteract the proliferation occurring within the tissue (Fig. 4d,e). Finally, the cessation of development could be due to a decrease in Wnt pathway activity which is high in the successional lamina of the polyphyodont species we have studied (Handrigan and Richman,2010b; Handrigan et al.,2010). Indeed, we have observed that Wnt pathway readout genes Axin2 and Tcf7 are expressed at negligible levels in the bearded dragon's successional lamina (Fig. 4f–i). This may be due to decreased ability of the successional lamina cells to respond to Wnts or due to decreased availability of the ligand. Thus our evidence from a reptile with a vestigial successional lamina supports the idea that such structures could be present in mammals, but for reasons such as the ones we encountered, development would not proceed.
Dental Epithelial Stem Cells and the Replenishment of the Successional Lamina
As the molecular basis of tooth replacement becomes clear, a key question about the process remains: how can the successional lamina continuously give rise to new teeth? Regeneration of adult tissues is typically the work of stem cells, either mesenchymal or epithelial. Hair cycling in mammals, for example, is driven by a population of adult epithelial stem cells residing in the bulge area of the hair follicle. These cells lie quiescent until an appropriate signal (e.g., Wnt) stimulates them to divide and give rise to progenitor cells that will lead to the formation of a new hair (Lo Celso et al.,2004). Inspired in part by these hair studies, we recently investigated whether there are epithelial stem cells in the dental lamina of squamates. The animal model we selected for this study was the adult leopard gecko E. macularius, a polyphyodont lizard. Using a pulse-chase strategy, we identified label-retaining cells (LRCs) in the dental lamina of this species (Handrigan et al.,2010). These LRCs cycle much slower than neighboring dental epithelial cells. Furthermore, the LRCs express genes in common with mammalian adult stem cells (Lgr5, Dkk1, Igfbp5) and appear to be activated by the canonical Wnt pathway in the same way. On the basis of these similarities, we contend that the slow-cycling cells of the gecko dental lamina represent a population of stem cells.
The gecko's putative dental stem cells are located exclusively on the lingual face of the dental lamina (Handrigan et al.,2010). This layer of the lamina makes a good home for stem cells because it is a quiescent environment. Tooth buds never form on the lingual side of the dental lamina (only on the labial side; Handrigan and Richman,2010a). LRCs appear to be clustered into periodically spaced “hot spots” along the length of the dental lamina. The 3D reconstruction data indicates that the clusters mainly line up with an inter-dental region (Summarized in Fig. 5a). Returning to the clone model proposed by Osborn, it may be that tooth spacing and control of the “waves” of tooth replacement can be controlled by groups of stem cells that feed into tooth families (one cluster might feed two adjacent families, Fig. 5a). Others discussing the work of Osborn (Mitsiadis and Smith,2006) had thought that mesenchymal cells comprised the clones rather than epithelial cells. Here we suggest that epithelial cells provide the clonal cells for the next generation of teeth. The role of the mesenchyme may instead be to activate stem cells, directing them to form new teeth (Fig. 5a,b).
What signals may induce dental lamina stem cells to activate? We propose that the canonical Wnt pathway induces quiescent stem cells to divide and become transit-amplifying cells. Two lines of evidence support this hypothesis. First, gain-of-function of the canonical Wnt pathway in the gecko (Handrigan et al.,2010) and in the python (Handrigan and Richman,2010b) caused a preferential increase in cell proliferation on the lingual side of the dental lamina, implying that the putative stem cells contained there had been activated. Secondly, we noted expression of Wnt target genes (Tcf3 and Tcf4) on the lingual side of the dental lamina in the gecko (Handrigan et al.,2010). These data imply that the molecular network we have described for python tooth development may also apply to the process in the gecko—and be key to regulating the fate of stem cells in the dental lamina.
If Wnt is indeed the “on/off switch” for dental lamina stem cells, what other signals control whether Wnt activity is turned on or off? The answer to this question may lie with our work on the ball python (Buchtová et al.,2008; Handrigan and Richman,2010a,2010b). In this species, Wnt activity in the dental lamina is regulated by the Bmp and Hh pathways. When Hh signaling is blocked in python jaw explant cultures with cyclopamine, we found that Wnt activity spreads throughout the dental lamina, including its lingual side. This was associated with a qualitative increase in the thickness of the dental lamina. We saw a similar phenotype when we exposed python jaw cultures to Wnt agonist LiCl. The convergence of the phenotypes of the two cultures could suggest that the increased Wnt activity was responsible for both. Further work is needed to confirm whether cyclopamine treatment also leads to ectopic lingual proliferation.
If the LRCs in the gecko dental lamina are indeed stem cells, then we would expect to find labeled cells in the successional lamina or enamel organs of tooth germs in the pulse-chase experiment as well. These cells would represent the direct descendents (i.e., transit-amplifying cells) of the stem cells. However, the number of LRCs left in the gecko dental epithelium after 5 months, the longest chase we performed, is so few that it is difficult to find any in the cells in teeth or successional laminae at all. Interestingly, for shorter chase periods (e.g., 9 weeks), we noted that LRCs were spread throughout the dental lamina, including regions near the oral epithelium as well as in the successional lamina. We suggest that at least some of these cells may be derived from the true stem cells. This is merely indirect evidence for the existence of dental epithelial stem cells in the gecko. Our future efforts will be directed at determining whether the gecko's dental lamina LRCs are de facto stem cells (i.e., they self-renew and are pluripotent). This will require developing methods to stimulate the LRCs in the adult gecko in vivo, so that the fate of these cells can be followed into the next generation of teeth.