Observations on continuously growing roots of the sloth and the K14-Eda transgenic mice indicate that epithelial stem cells can give rise to both the ameloblast and root epithelium cell lineage creating distinct tooth patterns

INTRODUCTION

The teeth can be roughly subdivided into three groups. The first group consists of brachydont or low-crowned teeth where the root is relatively long compared with the crown. This is the tooth type usually described in textbooks when describing root formation (Nanci 2003). The second group consists of hypsodont, or high–crowned, teeth, where the crown is high compared with the root. The third group consists of the hypselodont teeth, ever-growing or open-rooted teeth that grow continuously during the lifetime of the animal. Open-rooted refers solely to the large apical opening present in all continuously growing teeth and does not imply that the tooth actually needs to have a root in a classical sense as described in the textbooks.

It is thought that brachydonty is the ancestral state of mammalian teeth. During evolution the shift from brachydont to hypsodont teeth is a common phenomenon (Macfadden 2000). This trend is often initiated by environmental pressures. Teeth with higher crowns last longer with abrasive diets. For instance, a significant increase in the prevalence of hypsodonty in commonly found mammals of many taxonomic groups occurred during the Neogene due to an increased aridity in the environment of Europe (Jernvall and Fortelius 2002). Hypselodonty can be seen as an extreme form of hypsodonty. The crown never stops growing and root formation is postponed indefinitely, but often with small tracts of dentin covered with cementum acting as regions attaching the tooth to the jaw bone with a periodontal ligament.

Within closely related species there can be a variation between brachydont, hypsodont, and hypselodont teeth, indicating that the regulation of crown height is rather flexible. For instance in closely related rodent species, the Mouse (Mus musculus) molar is brachydont, the molars of the Bank vole and the Southern Red-backed vole (Clethrionomys glareolus, Clethrionomys gapperi) are hypsodont, and the molars of the Sibling vole and Meadow vole (Microtus rossiaemeridionalis and Microtus clethrionomys) are hypselodont (Phillips and Oxberry 1972; Tummers and Thesleff 2003). It has been proposed that the switch from hypsodont to hypselodont between the microtine genera of Clethrionomys and Microtus is caused by the maintenance of a regenerative unit, possibly due to a simple mutation (Phillips and Oxberry 1972), or that increased crown height results from delayed termination/cytodifferentiation and that hypselodonty is an extreme outcome of such a delay (von Koenigswald 1982). We have more recently proposed that the increase in crown height is a result of prolonging the period during which the epithelial stem cell niche is maintained (Tummers and Thesleff 2003).

During classic root formation the dental epithelium of the cervical loop undergoes some major structural changes (Fig. 1). The cervical loop is created during crown morphogenesis and with root initiation loses the central core of stellate reticulum and stratum intermedium cells, including the putative epithelial stem cells (Ten Cate 1961; Starkey 1963; Harada et al. 1999; Harada et al. 2002). A double layer of basal epithelium is left that is known as Hertwig's epithelial root sheath (HERS) (Thomas 1995). As HERS proliferates, the growing epithelial sheet becomes discontinuous and forms a fenestrated network lining the root surface known as the epithelial cell rests of Malassez (ERM) (Ten Cate 1996). Through this network the follicular mesenchyme cells can migrate to the dentin surface and differentiate into cementoblasts depositing the cementum. The ERM functions in the induction of cementoblast differentiation and regulation of their function (Thomas 1995; Bosshardt and Schroeder 1996; Kagayama et al. 1998). Fibers of the periodontal ligament are embedded in the cementum and connect the root to the jaw bone. HERS forms in brachydont and hypsodont teeth when root formation is initiated and crown formation ends, and its transition to ERM is generally regarded as a typical feature of root formation. Interestingly, the continuously growing rodent incisor is subdivided into two halves. The labial side is called crown analog because it produces ameloblasts and enamel, whereas the lingual half is called root analog, because its epithelium fragments and forms ERM and cementum is produced. Both root and crown analogs are generated continuously by the apical end of the incisor. It has been suggested that the labial cervical loop of the crown analog in the incisor is a specialized stem cell niche providing the epithelial progeny for the entire incisor and that lingually HERS is formed (Ohshima et al. 2005).

This last notion is questioned by the existence of a special type of continuously growing or hypselodont teeth as is represented by the sloth molar. The dentition of the contemporary sloth species is heavily modified, lacking both incisors and canines. Sloth teeth are open-rooted, grow continuously, and at the same time lack enamel (Naples, 1982). In juvenile specimens the tooth erupts as a simple cone. In adult specimens the cap of the dentin is worn off, leaving a hard shell of dentin with a soft pulp in the center. The edges of the dentin get sharper with age due to wear (Naples 1982). Similarly, the dentition of the mouse, as a representative of the rodents, is also heavily modified during evolution, with only two incisors in each jaw, followed by a diastema region lacking teeth, and three molars.

The transgenic K14-Eda mouse has ectodysplasin (Eda) expressed under the keratin 14 promoter leading to an excessive production of Eda throughout the ectoderm from E10 onwards, including the oral and dental epithelium (Mustonen et al. 2003). The constitutive expression of Eda in the dental epithelium leads to the formation of supernumerary molars and the loss of enamel on crown analog of the incisors. This transgenic mouse line therefore has possibly transformed its incisor into a continuously growing root, and serves in this paper as a model system for continuously growing roots. If HERS is required for the production of root epithelium, these teeth would not have a cervical loop and a stem cell niche.

Here we investigate the structure of the cervical loop area of continuously growing roots of a sloth molar that has erupted into the oral cavity and shows the typical cone-shaped morphology of a juvenile stage, and in the incisors of the wild-type and the K14-Eda transgenic mouse to investigate if root formation is truly linked to HERS formation characterized by the loss of the stellate reticulum containing the putative stem cells. Furthermore, we analyzed molecular markers of the incisor stem cell niche and differentiation in the K14-Eda incisor in order to check the state of the stem cell niche and the fate of the epithelial progeny. We show that stem cells exist in continuously growing roots in the sloth molar, in the K14-Eda incisor, and in the root analog side of the wild-type incisor. This indicates that the crown vs. root formation, i.e., the production of enamel vs. cementum, and the differentiation of the epithelial cells into ameloblasts vs. ERM, can be regulated independently from the maintenance of the stem cells, and that the maintenance of stem cells does not indicate an implicit ameloblast fate for the progeny.

MATERIAL AND METHODS

The sloth histological sections are of a Bradypus tridactylus specimen. The teeth have started to erupt and resemble the juvenile stage (Naples 1982). The sections were collected and processed by the Dutch researcher Van den Broek in 1913 and are part of the historical collection of the Hubrecht Laboratory in Utrecht. The K14-Eda mouse is a transgenic mouse that overexpresses the signal molecule Ectodysplasin-A1 under the Keratin14 promoter and has been previously described (Mustonen et al. 2003). Wild-type tissue was used from 1, 4, and 12 days, and 4-week post-natal NMRI mice. K14-Eda tissue was from 4-week-old specimens.

Radioactive in situ hybridization with 35 S labeled RNA probes was performed on serial paraffin sections as described previously (Tummers and Thesleff 2003). Immunohistochemistry was performed on 7-μm-thick paraffin sections. After deparaffination the sections were microwaved for 10 min in 10 mm natrium citrate buffer, pH 6.0, and then treated for 20 min in Proteinase K 7 μg/ml in phosphate-buffered saline (PBS). After washes in PBS the sections were incubated for 1 h in 3% BSA in PBS and then with polyclonal rabbit anti-human keratin (Dako, Glostrup, Denmark, A0575) 1:250 overnight at 4°C. The Vectastain ABC kit was used for detection and the sections were stained with DAB (Vector Laboratories, Burlingame, CA).

For histology the tissues were sectioned at 4 and 7 μm thickness, deparaffinized and stained with hematoxylin–eosin. The histological structures were identified based on definitions and examples in Ten Cate's Oral Histology (Nanci 2003).

RESULTS

The sloth molar

Figure 2A shows the general histology of a frontal section of the sloth molar (Bradypus tridactylus) from an unspecified stage, showing a conical-shaped molar of which the tip has erupted into the oral cavity, similar to the juvenile stage (Naples 1982). This molar is characterized by a prominent thick cap of dentin at the tip. This cap was not covered by enamel typical of the crown of brachydont and hypsodont teeth. Also the side surface of the tooth seemed to lack enamel and we confirmed this with a close-up of a representative area (Fig. 2B). The sloth molar lacked enamel-producing ameloblasts and the surface of this molar was entirely covered with dentin and cementum, with occasional cementoblasts visible within the cementum, all typical features of a root surface (Fig. 2C). The sloth molar therefore lacked any enamel from the tip to the base of the tooth and instead had acquired a root surface.

The general overview (Fig. 2A) showed a thin epithelial structure at the base of the tooth, where normally the HERS is found in brachydont roots. However, a close-up of this area showed that the typical structure of the HERS, consisting only of inner and outer enamel epithelium, was not found in the sloth. Instead we found that the cervical loop was maintained and it contained a core of cells surrounded by inner and outer enamel epithelium (Fig. 2C).

Histological structure of the cervical loop of the wild-type incisor

The mouse incisor is subdivided into two domains, the labial crown analog and the lingual root analog. The crown analog is characterized by an enamel surface whereas the lingual side has a cementum surface. An overview of the apical end of the incisor was dominated by the presence of the prominent labial cervical loop and a reduced epithelial structure on the lingual side. It has recently been suggested that the lingual aspect of the incisor consists of HERS instead of a cervical loop (Ohshima et al. 2005) and therefore we closely examined the structural organization of the epithelium on the lingual side (Fig. 3B). We observed that there are indeed two epithelial cell layers, apparently representing the inner and outer enamel epithelium, but also that a small core of stellate reticulum is retained between these layers. We checked this at older stages as well, and this phenotype did not change from 1 day post-natal to 4 weeks post-natal. In HERS this core is lost; hence the lingual side of the mouse incisor has maintained the cervical loop structure although diminished in size. The labial cervical loop is much enlarged as described previously due to a large amount of stellate reticulum in the core of the cervical loop and here the inner enamel epithelium proliferates actively and subsequently differentiates into ameloblasts (Fig. 3C).

Molecular markers of the stem cell niche

In the wild-type incisor, notch1 has been shown to be specifically expressed in the stellate reticulum and stratum intermedium cells of the labial cervical loop (Harada et al. 1999) (Fig. 4A, and B). We showed that Notch1 was also expressed in the central cells of the lingual cervical loop of the wild-type incisor (Fig. 4C) as well as in the lateral cervical loop area (data not shown). Also in the K14-Eda incisor notch1 was expressed in the central epithelial cells of the cervical loop (Fig. 4D, and E) resembling the lingual wild-type pattern (Fig. 4C). In the K14-Eda incisor Fgf10 was expressed in the mesenchyme directly surrounding the cervical loop similar to the wild type (Fig. 4F). Fgf3 expression is restricted in the wild-type incisor to the labial mesenchyme and this pattern was similar in the K14-Eda incisor (Fig. 4G). Lunatic fringe is a marker for the transit-amplifying epithelial cells of the inner enamel epithelium (Harada et al. 1999) and it was expressed in the K14-Eda incisor similar to the wild type in the inner enamel epithelium of the labial cervical loop (Fig. 4H).

Differentiation in wild-type and K14-Eda incisor

We compared cell differentiation in the K14-Eda and wild-type incisor by means of histology and markers for different cell types. The labial aspect of the K14-Eda incisor showed an identical picture to the lingual side of the wild-type incisor with a layer of odontoblasts, dentin, cementum, and periodontal ligament typical of a root. The transformation of the crown analog into a root analog in the K14-Eda incisor was confirmed by frontal sections labeled with a pan-keratin antibody. The distinctive cap of tall ameloblasts was obvious in the wild-type incisor on the labial aspect, and fragmented ERM covered the lingual side (Fig. 5D), whereas ERM surrounded the entire circumference of the K14-Eda incisor (Fig. 5E). The absence of ameloblasts was confirmed by the differentiation marker jagged1, which is normally expressed in differentiating ameloblasts (Harada et al. 1999), and it was absent in the epithelium of the K14-Eda incisor (Fig. 5F). Bsp1 is a marker for cementoblasts and odontoblast differentiation (Yamashiro et al. 2003), and in the K14-Eda incisor cementoblasts were present on both the lingual and labial aspects of the incisor (Fig. 5G) while in the wild type they were restricted to the lingual side.

DISCUSSION AND CONCLUSIONS

The rodent incisor is functionally and morphologically subdivided into the labial crown analog and the lingual root analogue. Each side shows a typical differentiation pattern where the crown analog is covered by enamel deposited by ameloblasts and the root analog is covered by cementum deposited by cementoblasts. It has been suggested that the large cervical loop at the labial aspect of the incisor represents the sole epithelial stem cell niche supplying epithelial stem cells for the growth of all aspects, and that HERS typical of roots in molars, forms lingually and is responsible for root formation there (Ohshima et al. 2005). However, we found no presence of HERS at the lingual side; instead there were stellate reticulum cells present in the core of the lingual cervical loop, as was confirmed by the notch1 expression in these cells. Moreover, the cervical loop was shown to be a continuous structure around the base of the incisor.

Similarly, no HERS typical of brachydont teeth (Ten Cate 1996) was found in the continuously growing sloth molar or the K14-Eda incisor. Both these teeth were covered by a typical root surface consisting of dentin covered by cementum, and they had cervical loops in their apical ends that had maintained stellate reticulum cells in the center. Moreover, the cervical loop was present in all sections of the apical aspect, indicating that it is present in the entire circumference of the base of the tooth. A similar situation is present in the continuously growing molar of the sibling vole where the cervical loop is not restricted to a specific local area (Tummers and Thesleff 2003). The continuously growing molar of the guinea pig shows a morphology comparable to that of the vole (Hunt 1958).

The 3D reconstruction of the apical end of the K14-Eda incisor allowed the observation of the cervical loop structures and this indicated that the labial cervical loop had a similar sized core of stellate reticulum as the cervical loop at other aspects of the tooth (Fig 3F). In addition, the lateral cervical loops were seen to protrude slightly from the apical end and the labial cervical loop was folded inwards. At the most lingual aspect the lateral loops did not meet and close. The total lack of epithelium here may indicate that the tooth is subdivided into individual sections representing a lineage from stem cell to differentiated cell similar to that of the crypt of the gut (Crosnier et al. 2006). Lack of the cervical loop in a specific area of the incisor therefore means local depletion of stem cells and eventually all epithelial structures deriving from those stem cells, because no replenishment takes place from neighboring areas. The husbandry of the K14-Eda transgenic mice shows that the reduced cervical loop of the K14-Eda incisor is indeed a functional stem cell niche because the incisors need to be clipped regularly to prevent misalignment due to constant regeneration of this tooth. In conclusion, our data clearly showed that the HERS is not an obligatory structure for root formation, that no specialized stem cell niche existed in the sloth molar or mouse incisor that is restricted to a local area, and that a cervical loop with a reduced core of stellate reticulum cells can still act as a stem cell niche.

It is known that Fgf10 is important for the maintenance of the stem cell niche in the incisor (Harada et al. 1999; Harada et al. 2002), and we have suggested that Fgf10 signaling is maintained in all continuously growing teeth to maintain the epithelial stem cell niche based to the similarities in the continuously growing molar of the sibling vole and the rodent incisor (Tummers and Thesleff 2003). It has also been proposed that the disappearance of Fgf10 signaling leads to the transition from crown to root formation due to a loss of the dental epithelial stem cell compartment (Yokohama-Tamaki et al. 2006). Based on our observations we would however like to suggest that although lack of Fgf10 can lead to a reduction of the stem cell niche and switch to root fate as can be seen in the mouse molar (Tummers and Thesleff 2003), differentiation into root can also take place in the presence of a functional epithelial stem cell niche. In the K14-Eda incisor, Fgf10 and Fgf3 expression was continued and although the amount of stellate reticulum, containing the putative stem cells, was reduced, it was not lost. At the same time the stem cell niches in the K14-Eda incisor and sloth molar give rise to root epithelium, suggesting that maintenance of the stem cells has no default effect on the differentiation of the progeny, which apparently can differentiate into either ameloblasts or ERM. It does not seem that the size of the niche determines the fate of the progeny, because the enlarged cervical loops in the K14-noggin transgenic mouse form no enamel (Plikus et al. 2005). We propose that in the wild-type incisor the labial cervical loop is enlarged due to the functional requirement to produce a large amount of ameloblast progeny, whereas the lingual cervical loop merely provided progeny for fragmented epithelium of the ERM.

We do not suggest that modification of Eda signaling is the mechanism used in the sloth tooth to acquire the continuously growing root phenotype and lack of enamel. The acquisition of this phenotype may occur at different regulatory levels. Recent studies on the regulation of the asymmetric development of the mouse incisor have revealed a central role for follistatin, an inhibitor of TGFβ signaling. The expression of follistatin in the lingual epithelium prevents enamel formation by inhibiting the inductive effect of BMPs on ameloblast differentiation (Wang et al. 2004). Interestingly, follistatin also inhibits the proliferation of the epithelial cells in the cervical loop, but this effect is due to inhibition of the positive effect of activin on stem cell proliferation (Wang et al. 2007). Recombinant Eda protein induces the expression of follistatin as well as another BMP inhibitor CCN2 and prevents BMP-induced ameloblastin expression in vitro, showing that the lack of enamel in the K14-Eda mice may result from inhibited BMP signaling (Pummila et al. 2007). These studies indicate that the maintenance of stem cells and their differentiation are regulated by different molecular mechanisms supporting the findings we have presented here. Taken together the different models show that there are many possible ways to create a sloth tooth phenotype.

In conclusion, there appears to be regulatory flexibility in the decision between crown and root fate that is independent of the depletion of the stem cells in the niche. The differentiation compartment and stem cell compartment of the niche can be regulated independently, giving rise to multiple patterns (Fig. 6): the brachydont pattern with low crown and high roots, the hypsodont pattern with high crown and low roots, the crown hypselodont pattern with a continuously growing crown domain and root domain, and the exclusively hypselodont root pattern. In the brachydont tooth, the disappearance of the stem cells coincides with the switch to root fate of the epithelial progeny during late development. Hypsodonty can be seen as a simple extension of the brachydont pattern where the stem cells are maintained longer during crown development, and root formation is postponed leading to a higher crown. In sharp contrast, the fate of root and crown domains in continuously growing teeth is probably already determined during early development, and is independent of the maintenance of the stem cells. We propose that the diversity of tooth patterns is possible because the differentiation of the progeny of the epithelial stem cells in the cervical loop is not restricted to one specific fate, the ameloblast cell lineage, but also root epithelium can form.