The continuously growing mouse incisor is an excellent model to analyze the mechanisms for stem cell lineage. We designed an organ culture method for the apical end of the incisor and analyzed the epithelial cell lineage by 5-bromo-2′-deoxyuridine and DiI labeling. Our results indicate that stem cells reside in the cervical loop epithelium consisting of a central core of stellate reticulum cells surrounded by a layer of basal epithelial cells, and that they give rise to transit-amplifying progeny differentiating into enamel forming ameloblasts. We identified slowly dividing cells among the Notch1-expressing stellate reticulum cells in specific locations near the basal epithelial cells expressing lunatic fringe, a secretory molecule modulating Notch signaling. It is known from tissue recombination studies that in the mouse incisor the mesenchyme regulates the continuous growth of epithelium. Expression of Fgf-3 and Fgf-10 were restricted to the mesenchyme underlying the basal epithelial cells and the transit-amplifying cells expressing their receptors Fgfr1b and Fgfr2b. When FGF-10 protein was applied with beads on the cultured cervical loop epithelium it stimulated cell proliferation as well as expression of lunatic fringe. We present a model in which FGF signaling from the mesenchyme regulates the Notch pathway in dental epithelial stem cells via stimulation of lunatic fringe expression and, thereby, has a central role in coupling the mitogenesis and fate decision of stem cells.
Stem cells are present in many vertebrate regenerative tissues including the hematopoietic system, nervous system, gut, gonads, skin, and olfactory epithelium. The stem cells are generally defined as cells that have the capacity to self-renew as well as to give rise to differentiated progeny. In invertebrates, stem cells have been shown to undergo asymmetric cell division resulting in one daughter cell remaining in the stem cell compartment and the other undergoing further cell divisions and giving rise to differentiated tissues (Morrison et al. 1997). The direct identification of stem cells in most mammalian tissues has been problematic because of the lack of specific stem cell markers.
Neural and hematopoietic stem cells have been the subject of specific interest recently, and multipotent neural cells that could generate new neurons and glia were identified in the ependymal cells of the adult brain ventricles (Johansson et al. 1999). These cells were shown to divide rarely and asymmetrically, and the other daughter cell was shown to stay as an undifferentiated stem cell in the ependymal layer, whereas the other cell moved to the subventricular layer and gave rise to a rapidly dividing progenitor cell pool, the so-called transit-amplifying cells providing the source for neuronal and glial precursors. Hematopoietic stem cells have been isolated using antibodies to cell surface antigens (Spangrude et al., 1998), and their stem cell properties have been demonstrated by transplantation into irradiated host animals under conditions where the progeny of a single stem cell can be identified (Morrison et al., 1994). Recently, evidence was presented from similar transplantation experiments that neuronal stem cells may have the competence to give rise to hematopoietic cells (Bjornson et al. 1999).
The molecular mechanisms regulating the maintenance of the stem cells, their cell division, and differentiation remain largely unknown, but it has been suggested that there are common mechanisms shared by stem cells in different tissues that underlie their specific properties. The maintenance of the undifferentiated state of the stem cells has been thought to be regulated by cell–cell interactions involving the Notch signaling pathway (Fortini and Artavanis-Tsakonas 1993; Bray 1998; Carlesso et al. 1999). There is no experimental evidence for this assumption in vertebrate tissues, but it is supported by in situ expression data and the established role of Notch in the regulation of cell differentiation in invertebrates. Notch receptors have been shown to be expressed in undifferentiated neural cells as well as in keratinocytes, but expression is not confined to stem cells only (Zagouras et al. 1995; Johansson et al. 1999). The hematopoietic stem cells express Notch receptors, and the Notch ligand Jagged1 (Serrate1) has been shown to be expressed by stromal cells in culture, suggesting roles in mediation of interactions regulating stem cell maintenance and/or differentiation (Jones et al. 1998; Varnum-Finney et al. 1998).
It is generally thought that stem cells respond to environmental cues that stimulate cell division both in stem cells and their multipotent progeny, the transit-amplifying cells. Some conserved diffusible signal molecules have been proposed to function in the regulation of self-renewal of the stem cells. Neural stem cells can be stimulated to divide in vitro by EGF and FGF (Reynolds and Weiss 1996).
We have used the mouse incisor tooth as a model to analyze certain aspects of stem cell regulation and functions. The mouse incisor differs from mouse molars as well as from all human teeth in that it erupts continuously throughout the life of the animal. Cells in the apical end of the tooth proliferate and differentiate into the various tooth forming cells including the mesenchyme-derived odontoblasts producing dentin and the epithelium-derived ameloblasts producing enamel. The epithelial tissue at the labial aspect of the incisor apex forms the cervical loop, which consists of a core of stellate reticulum cells surrounded by basal epithelial cells contacting the dental mesenchyme (see Fig. 1, a–d). The cells in the apex of the tooth divide rapidly as compared with the more incisal region, and there is a gradient of cell differentiation from the apex towards the incisal direction (Smith and Warshawsky 1975, Smith and Warshawsky 1977). Hence, it is generally thought that the apex is the reservoir for the cells producing the enamel and dentin of the erupting tooth. There is no evidence of stem cells among the differentiated ameloblasts (Nataatmadja et al. 1990). Therefore, the epithelial stem cells presumably reside within the cervical loop, but this has not been demonstrated directly and their identity has remained completely unknown.
A specific advantage of the tooth as a model is that the basic mechanisms regulating morphogenesis and differentiation have been analyzed in great detail both in classical experimental embryological studies as well as recently by molecular methods. Hence, although the previous studies have not focused on stem cells, there is a wealth of data on expression patterns of developmental regulatory molecules as well as their functions (Gene expression in tooth: www database, 1999). Several conserved signal molecules have been implicated in the mediation of the epithelial–mesenchymal interactions regulating morphogenesis and cell differentiation in teeth. These include FGFs, BMPs, Wnts, and Shh (Thesleff and Sharpe 1997; Peters and Balling 1999). In addition, several Notch receptors and ligands have been localized in the dental tissues and they have been implicated in the regulation of cell determination and differentiation both in dental epithelium and mesenchyme (Mitsiadis et al. 1995, Mitsiadis et al. 1997, Mitsiadis et al. 1998). Furthermore, it is known from tissue recombination experiments that the mesenchymal tissue controls advancing morphogenesis after the initiation of tooth development (Kollar and Baird 1969), and that mesenchyme also determines whether the tooth grows continuously or not (Jernvall, J., personal communication). Hence mesenchymal signals apparently regulate the continuous production of progenitor cells in the epithelium.
In this study, we have designed an organ culture system for the apical end of the mouse incisor and used it for the analysis of cell division and cell fate in the epithelium. Although there is no universally accepted definition of a stem cell, we have identified a cell population in the dental epithelium that fulfills several criteria of the stem cells. We have shown by 5-bromo-2′-deoxyuridine (BrdU)1 and [1,1′-dioctadecyl-6,6-di(4-sulfophenyl)-3,3,3′,3′-tetramethylindocarbocyanine] (DiI) labeling that stem cells are present in the cervical loop epithelium. They give rise to transit-amplifying cells and their progeny differentiate into ameloblasts forming the enamel matrix. We have performed a detailed in situ hybridization analysis of the expression of certain genes in the Notch and FGF signaling pathway and studied experimentally the effects of FGF-10 on the epithelium. Slowly dividing putative stem cells expressing Notch1 were located in close proximity with basal epithelial cells expressing lunatic fringe. Lunatic fringe is a secretory molecule involved in the modulation of Notch signaling in invertebrates (Cohen et al. 1997; Fleming et al. 1997; Panin et al. 1997; Klein and Arias 1998) and in chicks (Sakamoto et al. 1998). Our studies also indicate that FGF-10 is a mesenchymal signal that in addition to stimulating division in stem cells and/or transit-amplifying cells, may regulate the fate of the stem cells by affecting the Notch pathway via stimulation of lunatic fringe expression. Our data are in line with FGF-10 being a signal that may couple the division of stem cells and their determination to the ameloblast cell fate.
Materials and Methods
Dissection and Culture of Tissues
The incisors were dissected carefully from the lower jaws (mandibles) of 2-d-old mice (CBA × NMRI) (see Fig. 1 b). This stage was selected for experiments because it was easy to dissect incisors from calcified mandibular bone. The apical ends of the incisors were dissected and cultured in Trowell-type organ cultures on 0.1-μm pore-size. Nuclepore filters (Costar Corp.) supported by metal grids in a humidified atmosphere of 5% CO2 in air at 37°C. The culture medium consisted of DME (GIBCO BRL) supplemented with 10% FCS (GIBCO BRL), penicillin/streptomycin, glutamate I (GIBCO BRL), and 100 μg/ml ascorbic acid (Sigma Chemical Co.). Calcification of the enamel and dentin matrices was visualized by alizarin red staining.
For tissue recombination experiments, the apical ends of the incisors were incubated for 10 min in 2% collagenase (GIBCO BRL) in DME at 37°C, and the epithelia were mechanically separated from mesenchyme in PBS. The epithelia were placed in contact with mesenchyme and cultured for 24 h on the filters as described above. Heparin acrylic beads (Sigma Chemical Co.) releasing FGF-10 (25 ng/μl; Amgen) were placed in cultures of whole apical ends of the incisors and on isolated epithelial tissue. Control beads were incubated in BSA (Sigma Chemical Co.). The preparation of the beads has been described in detail (Kettunen et al. 1998).
Processing of Tissues for In Situ Hybridization
Freshly dissected as well as cultured tissues were fixed in 4% paraformaldehyde in PBS, pH 7.2, overnight at 4°C, dehydrated, embedded in paraffin, and serially sectioned at 7 μm. Sections were placed on triethoxysilane- and acetone-treated slides, dried overnight at 37°C, and stored at 4°C until used. For histological analysis, the sections were stained with hematoxylin/eosin.
Cryostat sections were incubated with rabbit anti-CVADR, (1:50; Tomko et al. 1997) as a primary antibody for 1 h at room temperature, rinsed in PBS, and incubated with peroxidase-labeled anti–rabbit antibodies for 45 min at room temperature. For color reaction DAB (Vector Laboratories, Inc.) as chromogen was used and the sections were counterstained with hematoxylin.
DiI Labeling Analysis
The fluorescent DiI label was microinjected to the epithelium of dissected apical ends of the incisors that were subsequently cultured as described above. DiI [1,1′-dioctadecyl-6,6-di(4-sulfophenyl)-3,3,3′,3′-tetramethylindocarbocyanine; Molecular Probes, Inc.] was diluted in DMSO (0.2% wt/vol). The explants were fixed after 5 d and observed using a fluorescent microscope.
BrdU Incorporation Analysis
Cell division was analyzed by culturing the tissues in the presence of BrdU (Amersham Life Science, Inc.). BrdU (1:1,000) was present in the medium for 3, 24, or 72 h, and the tissues were fixed in 4% paraformaldehyde in PBS, pH 7.4, either directly after labeling or after a 7-d chase period. The tissues were embedded in OCT compound (Tissue Tek), and 14-μm frozen sections were cut. The incorporated BrdU was detected by indirect immunoperoxidase method with mouse mAb against BrdU (Amersham Life Science, Inc.) and peroxidase-labeled anti–mouse secondary antibody (ENVISION; DAKO). For color reaction, DAB (Vector Laboratories, Inc.) as chromogen was used and the sections were counterstained with hematoxylin.
Construction of Probes and In Situ Hybridization
For in situ hybridization analysis, single stranded [35S]UTP- and digoxigenin-labeled antisense riboprobes for mouse Notch1, Notch2, Notch3 (Mitsiadis et al. 1995), Fgf-3, Fgfr1b, and Fgfr2b (Kettunen et al. 1998) were synthesized as described. Lunatic fringe antisense probe of 570 bp was obtained by linearizing the plasmid (pBS SKII(−) with 860 bp 5′ coding region of mouse lunatic fringe) with BamHI and transcribing with T7 RNA polymerase. For the sense probe, XhoI and T3 RNA polymerase were used, respectively. Serrate 1 probe was a gift from Domingos Henrique (University of Lisbon), Fgf-3 probe from David Wilkinson (NIMR, London), and Fgf-10 probe from Nobuyuki Itoh (Kyoto University). The procedures for in situ hybridization of paraffin sections and whole mounts of cultured explants have been described previously (Kettunen and Thesleff 1998; Kettunen et al. 1998).
The Progenitors of Ameloblasts Reside in the Cervical Loop Epithelium
The appearance of the dissected incisor tooth of a 2-d-old mouse in a stereo microscope is shown in Fig. 1 b. The cervical loop epithelium appears translucent as compared with the mesenchymal tissue. The histology of the apical end of the incisor and its schematic representation show the different cell types (Fig. 1c and Fig. d). The cervical loop is composed of basal epithelium and a central core of stellate reticulum. A gradient of an increasing level of cell differentiation is evident in the basal epithelium contacting the pulp mesenchyme from apical to incisal direction. The inner enamel epithelial cells differentiate into tall columnar ameloblasts secreting the enamel matrix.
We devised an organ culture method for the dissected apical ends of the incisors to analyze cell kinetics and fate. The appearance of an explant as photographed through a stereomicroscope during a 7-d culture period is shown in Fig. 1e–i. The cervical loop produced new dental epithelium, which was seen as an upward movement of the apical end of the epithelium to encompass the apical mesenchyme. This differs from the in vivo situation where the growing epithelium moves towards the incisal direction, probably because of mechanical factors as in vitro the tissues are attached to the filter. After 3 d of culture, production of new enamel and dentin by differentiated ameloblasts and odontoblasts, respectively, was evident (Fig. 1 g). The thickness of the enamel and dentin matrices increased with time when the explants were cultured for 5 and 7 d, respectively (Fig. 1, g–i). Histological sections revealed normal morphology of the ameloblasts that had differentiated in vitro into tall columnar cells with nuclei polarized to their distal ends facing the normal looking stratum intermedium cells. Tomes processes were evident in the secretory ends of the ameloblasts and they appeared as a picket fence in the enamel matrix (Fig. 1 m). Whole mount staining of the explants after 7 d of culture with alizarin red showed that the extracellular enamel and dentin matrices, formed in vitro, had calcified (Fig. 1k and Fig. l).
To visualize the fate of the cervical loop cells, they were labeled with the fluorescent DiI stain at the onset of culture, and the labeled cells were examined after 5 d of culture. Care was taken to microinject DiI to the central cells in the cervical loop, and the restriction of the dye was checked at the onset of culture (not shown). After 5 d of culture, fluorescent cells were seen to extend from the cervical loop to the differentiated ameloblasts (Fig. 2, a and b). The original site of DiI injection could still be seen (Fig. 2, a and b, arrows), but, interestingly, the highest intensity of fluorescence was observed at some distance incisally from the original site. These presumably represent daughter cells, which proliferated rapidly. Extending incisally from this site were labeled cells that had entered the stage of ameloblast differentiation. In conclusion, the DiI injection experiments showed that the cells in the cervical loop give rise to the differentiated dental epithelial cells, in particular the ameloblasts.
Only Cervical Loop Epithelium Has Regenerative Capacity
The DiI labeling experiments suggested that the stem cells for differentiated dental epithelium reside in the cervical loop, but they did not rule out the possibility that there would be stem cells also among the more differentiated cells. To answer this question, we experimentally removed either the cervical loop epithelium or the more differentiated epithelial cells before culture of the apical ends of the incisors. Excision of the epithelium was done at the junction of the cervical loop and the inner enamel epithelium. When the cervical loop was removed (five explants) and the explants were cultured for 9 d, it was evident that the wounded end of the epithelium did not grow like in the controls (Fig. 3, a–e, compare with Fig. 1, e–i). Instead, the epithelial cells underwent differentiation into ameloblasts. The cells produced enamel matrix, and after 9 d of culture, a sharp end of matrix production was seen and its calcification was visualized by alizarin red staining (Fig. 3 f). These observations indicate that the inner enamel epithelial cells did not exhibit regenerative capacity, and that they were already committed to differentiation into enamel-producing ameloblasts.
In the second set of experiments, the inner enamel epithelium and ameloblast zones were removed from the apical ends of the incisors and only the cervical loop epithelium was left intact (five explants). During the 7-d culture, the cervical loop was shown to grow remarkably and to give rise to inner enamel epithelium (Fig. 3, g–k). After 7 d of culture, matrix production by new epithelium was seen and its calcification was visualized by alizarin red staining (Fig. 3 l, small arrowheads). Therefore, these results gave evidence that cells with significant growth potential reside exclusively in the cervical loop epithelium.
Slow Cycling Putative Stem Cells Reside in the Cervical Loop Epithelium
The kinetics of epithelial cell division was examined by labeling the apical ends of the incisors with BrdU for various times and examining the BrdU incorporating cells in tissue sections after culture. When the explants were labeled for 3 h and immediately fixed, labeled cells were seen throughout the cervical loop epithelium extending to the inner enamel epithelium zone (Fig. 4, a–c). Presumably both stem cells and transit-amplifying cells were labeled with BrdU during this period, as recently shown for skin keratinocytes in vitro (Jensen et al. 1999). When the explants were cultured for 24 h in the presence of BrdU, labeled cells were observed in the same locations as in the 3-h cultures and in the ameloblast zone, indicating that the inner enamel epithelium, which had undergone their last divisions during the culture, had differentiated into ameloblasts (Fig. 4d and Fig. f). When the labeling period was extended to 72 h, BrdU incorporation was seen to extend further to the zone of differentiated, postmitotic ameloblasts (Fig. 4 i). These results are in line with the observations on DiI-labeled explants, indicating that proliferating cells in the apical region of the incisor give rise to ameloblasts.
To analyze kinetics of cell division, we cultured the explants for 7 d after they had been incubated for 3 h with BrdU. The rapidly proliferating cells were, thereby, given time to dilute the label. After the 7-d chase period, very few labeled cells were seen in the inner enamel epithelium area, indicating that the vast majority of cells had diluted the label by repeated divisions or differentiated into ameloblasts after their last divisions. BrdU-labeled ameloblasts were present near the inner enamel epithelium zone, and they were grouped into clusters. Hence, they appeared as clones of cells that presumably were descendants of individual inner enamel epithelial cells, which had undergone their last cell divisions and incorporated BrdU during the 3-h labeling period (Fig. 5, a and c). Similarly, clusters of labeled cells were seen in the epithelial cells overlying the labeled ameloblasts, suggesting that they had differentiated from precursor cells that had been labeled during the first 3 h of culture.
Interestingly, intensely labeled epithelial cells were observed within the cervical loop. They were located among the peripheral stellate reticulum cells in close vicinity to the basal epithelial cells that contact the basement membrane and dental mesenchyme at their basal end (Fig. 5, a and b). We suggest that the labeled cells represent a population of slowly dividing stem cells in the cervical loop. They did not have neighboring labeled cells, suggesting that the stem cells had underwent asymmetric cell division. It is conceivable that one daughter cell had remained as a stem cell and the other daughter cell had generated a transit-amplifying population of cells, which are the progenitors of ameloblasts.
As a potential stem cell marker, we used the antibody against the adenovirus receptor CVADR (previously known as CAR), which has been associated with stem cells in the brain (Tomko et al. 1997; Johansson et al. 1999). Strong expression was seen in the cervical loop epithelium and, at higher magnification, the expression appeared more intense in the stellate reticulum cells than in the basal epithelium. The transit-amplifying cells of the inner enamel epithelium and the differentiated ameloblasts did not express CVADR (Fig. 5d and Fig. e). Hence the expression of CVADR correlated well with the suggested stem cell distribution within the cervical loop epithelium.
Expression of Genes in the Notch Signaling Pathway Is Associated with the Localization of Stem Cells
Because the Notch pathway has an important role in cell fate determination in invertebrates and it has been specifically implicated in the determination of stem cell fate in vertebrates, we performed a careful in situ hybridization analysis of the expression patterns of several genes in this pathway. Notch receptors and ligands previously have been shown to be associated with early tooth morphogenesis and cell differentiation (Mitsiadis et al. 1995, Mitsiadis et al. 1997, Mitsiadis et al. 1998), and we have recently shown that lunatic fringe, a homologue of a gene that is linked with Notch signaling in Drosophila, has an expression pattern during early tooth development suggesting developmental roles (Mustonen et al., in preparation).
Expression of the receptors Notch1, 2, and 3 mRNA, the Notch ligand Serrate1 mRNA as well as lunatic fringe mRNA was localized in longitudinal sections through the 2-d mouse incisors. In the cervical loop, Notch1 was restricted to stellate reticulum cells and was most intense in the cells facing inner enamel epithelium. On the other hand, Notch2 was expressed in the outer enamel epithelium and the underlying stellate reticulum cells, and Notch3 was not detected in the cervical loop (Fig. 6, a–c and a′–c′′). Expression of Serrate1 was not detected in the cervical loop (Fig. 6d and Fig. d′). Interestingly, lunatic fringe was expressed in the inner enamel epithelium starting from the cervical loop (Fig. 6e and Fig. e′). This correlates closely with the distribution of BrdU incorporating cells in the enamel epithelium (Fig. 4 a). Furthermore, the localization of the slowly dividing putative stem cells in our BrdU incorporation study in the peripheral stellate reticulum cells correlates with the boundary between lunatic fringe and Notch1, suggesting that the maintenance and fate of the stem cells may be influenced by Notch signaling.
The expression patterns of the Notch pathway genes were also correlated with terminal cell differentiation. In the zone of ameloblast differentiation, Notch1 was expressed in the stratum intermedium cells, whereas Notch2 was intensely expressed in the stellate reticulum cells (Fig. 6, a′′ and b′′). Notch3 was expressed in stellate reticulum cells and, in addition, intense expression was seen in blood vessels and in the subodontoblastic mesenchyme (Fig. 6c and Fig. c′′). In contrast to Notch genes that were absent from ameloblasts, the ligand Serrate1 was intensely expressed in terminally differentiated ameloblasts (Fig. 6d and Fig. d′′). These findings are in line with the recent data by Mitsiadis et al. 1998 and, thus, support a role for Notch signaling in the maintenance of the differentiated state of ameloblasts. Lunatic fringe was not expressed in the zone of differentiated ameloblasts (Fig. 6 e′′), and its functions, therefore, appear to be confined to the regulation of their determination and differentiation.
Fgf-3 and Fgf-10 Are Expressed in Dental Mesenchyme and Their Receptors in Overlying Cervical Loop Epithelium
Earlier tissue recombination experiments have shown that dental mesenchyme controls the morphogenesis of dental epithelium (Kollar and Baird 1969) including the continuous growth of the incisor epithelium (Jernvall, J., personal communication), and our recent studies on early tooth morphogenesis have implicated FGFs-3 and -10 as mesenchymal signals regulating the early morphogenesis of dental epithelium (Kettunen, P., N. Itoh, and I. Thesleff, manuscript submitted for publication). Hence, we analyzed the patterns of the expression of these signals as well as their receptors by in situ hybridization in the 2-d-old mouse incisors. Both Fgfs were intensely expressed in a restricted area of the dental mesenchyme in the apical end of the tooth. Fgf-10 expressing cells surrounded the whole cervical loop epithelium and extended to the zone underlying inner enamel epithelium (Fig. 7, a and c). Fgf-3 showed a more restricted pattern of expression. It overlapped with Fgf-10 in the mesenchyme under the inner enamel epithelium, but was not expressed in the mesenchyme surrounding the cervical loop (Fig. 7 e). Fgfr1b was intensely expressed in the cervical loop epithelium including the basal epithelial cells and stratum intermedium, but was less intense in the stellate reticulum cells (Fig. 7b and Fig. d), and Fgfr2b (Kgfr) showed a similar pattern in the cervical loop (Fig. 7 f). Hence, the patterns of the expression of Fgf-3 and Fgf-10 and their receptors were in line with the suggestion that these FGFs may be mesenchymal signals regulating the development of cervical loop epithelium.
FGF-10 Stimulates the Growth of Cervical Loop Epithelium
The effects of FGF-10 on the growth of cervical loop epithelium were analyzed in explant cultures of apical ends of the 2-d-old mouse incisors. FGF-10 recombinant protein was applied with beads on the explants, and the control explants received beads soaked in BSA. Two beads were placed on each explant in contact with the cervical loop epithelium, one at its apical end and one at the inner enamel epithelium. When the development of the explants with FGF-10 and BSA beads was compared during 48 h of culture, a clear acceleration of the growth of the epithelium was observed by FGF-10 (Fig. 8, a–f).
The effects of the FGF-10 and BSA beads on cell proliferation were analyzed by culturing them on isolated cervical loop epithelium. The beads were placed on different areas of the dental epithelium, and the explants were cultured in the presence of BrdU for 6 h, fixed, and processed for immunohistological analysis of BrdU incorporation. A clear zone of BrdU incorporating cells was evident around the FGF-10 beads that were placed on the inner enamel epithelium and cervical loop epithelium (Fig. 8g and Fig. h), but not when placed on the zone of differentiated ameloblasts (not shown). No increase in BrdU incorporation was seen around the control BSA beads (Fig. 8g and Fig. h).
Lunatic Fringe Expression in Cervical Loop Epithelium Is Stimulated by the Dental Mesenchyme and FGF-10
The expression of lunatic fringe in the basal epithelial cells of the cervical loop suggested that it may be regulated by the underlying mesenchyme, and the close correlation of Fgf-10 and lunatic fringe expression in mesenchyme and epithelium was suggestive of a role for FGF-10 in the regulation of lunatic fringe. We examined these possibilities in vitro in isolated epithelial and mesenchymal tissues from the apical ends of 2-d-old mouse incisors. When the isolated epithelium and mesenchyme were placed in contact on the filter and cultured for 24 h, intense expression of lunatic fringe was seen in the basal epithelial cells of the cervical loop and in the inner enamel epithelium contacting the mesenchyme (Fig. 9 a). No expression was seen in the epithelium of the differentiated zone contacting mesenchyme. When the epithelium was cultured alone, lunatic fringe expression was not seen, indicating that expression had been downregulated in the absence of mesenchyme. When FGF-10 releasing beads were placed on the cervical loop epithelium and inner enamel epithelium, they caused an upregulation of lunatic fringe expression in the contacting epithelium (Fig. 9 b). When FGF-10 beads were placed on the ameloblast zone, the beads had no effect (Fig. 9 c). Control BSA beads had no effect on expression. Hence, these experiments demonstrated that the maintenance of lunatic fringe in the cervical loop epithelium depends on mesenchymal signals, and that this effect is mimicked by FGF-10.
The incisors of rodents erupt throughout the lifetime of the animals as the wear at the incisal edge is compensated for by renewal in the apex of the tooth located deep in the jawbone. The generation of the highly specialized dental tissues including the epithelially derived enamel and mesenchymal dentin involves proliferation of progenitor cells and their differentiation, matrix deposition, and subsequent mineralization. This process is seen as an increasing gradient of cell differentiation starting from the germinative apical area and extending towards incisal direction. Although it is generally assumed that there are stem cells in the apical end of the tooth, this problem has not been actively studied and the identity of the stem cells has remained enigmatic. We present evidence that epithelial stem cells possibly reside in a specific location in the cervical loop at the apex of the tooth, and that they produce a transit-amplifying cell population that gives rise to the differentiated dental epithelial cells, in particular, the enamel-producing ameloblasts. We also show data suggesting that Notch and FGF signaling are associated with specification of these stem cells and present a putative molecular mechanism, which may be applicable to the proliferation and determination of cell fate in stem cells also in other vertebrate tissues.
Putative Dental Epithelial Stem Cells Reside in the Cervical Loop, Divide Slowly, and Give Rise to Transit-amplifying Cells Differentiating into Ameloblasts
Our in vitro model system for the culture of the apical end of the mouse incisor supported the growth and differentiation of the germinative cells of the tooth. Like in vivo, the cervical loop epithelium proliferated, differentiated into ameloblasts, and produced enamel matrix, which underwent calcification in organ culture. A special advantage of the system was that the stereomicroscopic examination of the advancing development was easy during culture. As the cervical loop and enamel-producing cells are located at the labial side of the incisor, and the explants were oriented so that the labial surface was on one side of the explant, the growth of the cervical loop epithelium and the gradient of differentiation and matrix deposition could be visualized almost two dimensionally.
Using the in vitro culture of the incisor apex, we produced two lines of evidence indicating that the epithelial stem cells reside within the cervical loop epithelium. First, when cells in the center of the cervical loop were labeled with the fluorescent dye DiI and cultured in vitro, the dye was seen in the differentiating cells after 1 d and in the ameloblasts after 2 d of culture, indicating that the ameloblast lineage starts in the cervical loop. This was supported by the analysis of BrdU incorporation during extended time periods showing that the labeled cells extended progressively more incisally and occupied the zone of postmitotic mature ameloblasts. Second, when the cervical loop was removed mechanically from the apex of the tooth, the remaining epithelium could not regenerate the cervical loop in vitro. In these explants, all epithelial cells differentiated and produced mineralized matrix forming a stunt end with no indication of new cervical loop formation. In the opposite experiment where the differentiated epithelium was removed, the remaining cervical loop generated new epithelium differentiating into secretory ameloblasts, thus, indicating that it contains a pool of immature cells competent to regenerate the dental epithelium.
We identified the slowly dividing putative stem cells in organ culture experiments by labeling the cells of the incisor apex with BrdU for 3 h, followed by a 7-d chase period. Labeled cells were detected in the cervical loop among the stellate reticulum cells in close vicinity to the basal epithelium. This location is consistent with the observed strong expression of the adenovirus receptor CVADR, a putative stem cell marker in brain cells (Johansson et al. 1999). We suggest that the BrdU-labeled cells represent the daughter cells of stem cells that had divided during the 3-h labeling period and had remained in the stem cell pool and not divided thereafter. We suggest that the other daughter cell had entered the pathway of cell differentiation, and that the cell division therefore was asymmetric. This is supported by the finding that the BrdU-labeled cells did not have labeled neighboring cells (unlike the labeled ameloblasts, which appeared in clusters as they presumably had been generated by inner enamel epithelium after the last division). Asymmetric cell division of stem cells has been demonstrated in invertebrates (Hawkins and Garriga 1998; Jan and Jan 1998), and in vertebrates there is evidence that neural stem cells undergo asymmetric cell division (Chenn and McConnell 1995; Johansson et al. 1999).
Association of Notch Signaling Pathway with Dental Epithelial Stem Cells
The Notch signaling pathway is an evolutionarily conserved cell interaction mechanism that controls fundamental aspects of cell determination during development (Artavanis-Tsakonas and Simpson 1991; Artavanis-Tsakonas et al. 1995; Lewis 1998), and it has been suggested that Notch signaling may provide a microenvironment where the stem cells are maintained. Although Notch has been shown to function in Caenorhabditis elegans and in Drosophila in the maintenance of the potential of stem cells and their undifferentiated state, respectively (Fortini and Artavanis-Tsakonas; 1993; Bray 1998), definitive roles for Notch have so far not been demonstrated in vertebrate stem cells (Carlesso et al. 1999). The expression of Notch1 in skin keratinocytes and in immature neural cells including neural stem cells has been suggested to be associated with stem cell functions (Powell et al. 1998; Johansson et al. 1999). Furthermore, asymmetric distribution of Notch1 has been observed in immature neural cells before their cell division, suggesting a role in regulation of different cell fates in the daughter cells, i.e., whether they are maintained as stem cells or give rise to differentiated progeny (Chenn and McDonnell, 1995).
During tooth development, Notch signaling has been associated with the differentiation of dental epithelial and mesenchymal cells. Notch1, -2, and -3 are downregulated in presumptive dental epithelium already during tooth initiation and, thereafter, these genes are not expressed in the ameloblast cell lineage, but they are intensely expressed in the other dental epithelial cells, suggesting roles in the determination and maintenance of cell differentiation of the ameloblast cell lineage (Mitsiadis et al. 1997, Mitsiadis et al. 1998). Therefore, these previous studies suggested to us that Notch signaling may be involved in the regulation of the stem cells in the continuously growing incisor. Interestingly, the slowly dividing potential stem cells in the cervical loop epithelium were located within the Notch1 and Notch2 expressing cells facing basal epithelial cells not expressing Notch, which is in line with the possibility that these Notch receptors regulate fate of the stem cells at this border.
The possible role of Notch signaling in the regulation of the dental stem cells is further strengthened by the expression pattern of lunatic fringe, a gene encoding a secretory molecule resembling a glycosyltransferase that modulates Notch signaling in Drosophila (Yuan et al., 1997). Three vertebrate fringe genes are known and of these lunatic fringe has a developmentally regulated expression pattern during early tooth morphogenesis (Mustonen, T., unpublished results). We localized lunatic fringe expression to the cervical loop epithelial cells neighboring Notch expressing cells. Such expression borders are characteristic to molecules in the Notch signaling pathway in Drosophila tissues, and have been shown to be sites of specific signaling activities (Irvine and Wieschaus 1994; Johnston et al. 1997).
An important piece of evidence supporting a role for the Notch pathway in the regulation of stem cells is the finding that the lunatic fringe expressing basal epithelial cells that contact Notch1 expressing stem cells on one side are facing the dental mesenchyme at the opposite side, and that the mesenchyme regulated lunatic fringe expression. This is significant because the dental mesenchyme controls the epithelial morphogenesis in teeth, including the type of tooth formed, i.e., whether it develops to a molar with roots or to an incisor that grows continuously (Kollar and Baird 1969; Jernvall, J., personal communication). Hence, our demonstration that the mesenchyme controls epithelial lunatic fringe expression is in line with the possibility that the mesenchymal control of the continuous epithelial growth involves modulation of the Notch pathway affecting fate determination of the stem cells.
FGFs May Be Mesenchymal Factors, which Couple the Proliferation and Determination of Epithelial Stem Cells by Stimulating Cell Division and Lunatic Fringe Expression
It is not clear whether the maintenance and differentiation of stem cells and their cell division are regulated independently by distinct signals or whether there are signals that couple the different processes. It is also not known how similar the molecular mechanisms are between stem cells in different tissues, although there is evidence suggesting that there may be significant similarities (Morrison et al. 1997). Of the various regenerating tissues in vertebrates, the tooth resembles developmentally other derivatives of the epithelium such as skin, gut and, in particular, hair follicles. In all these tissues, the growth and differentiation of the regenerating epithelium is controlled by interactions with the underlying mesenchymal cells (Thesleff et al. 1995). The continuously growing incisor bears significant similarities histologically to the hair follicles as in both organs the regenerating epithelium encompasses a mesenchymal papilla, which is necessary for the growth of the epithelium. Stem cells have been located in skin keratinocytes (Jones et al. 1995) and hair follicle epithelium (Cotsarelis et al. 1990), but possible mesenchymal factors involved in their regulation have not been examined. Interestingly, the stem cells in skin and hair follicle epithelium as well as in the gut are located among the basal epithelial cells and are associated with the basement membrane, whereas the putative dental stem cells were located in the stellate reticulum in close vicinity to the suprabasal aspect of the basal epithelial cells and, hence, did not contact the basement membrane.
Our localization of the expression of Fgf-10 and Fgf-3, two signal molecules that we have recently identified as proliferative signals from mesenchyme to epithelium during early tooth morphogenesis (Kettunen, P., N. Itoh, and I. Thesleff, manuscript submitted for publication), suggested that they might function in the regulation of the continuous growth of the incisor epithelium. The expression of both Fgfs was restricted to the mesenchyme underlying the rapidly proliferating inner enamel epithelium. In addition, Fgf-10 expression extended more apically and surrounded the entire cervical loop. Furthermore, the FGF receptors that are known to bind these particular FGFs, namely the IIIb splice forms of FGFR1 and FGFR2 were expressed in the cervical loop epithelium, Fgfr1b mRNA being specifically intense in the basal epithelial cells facing the mesenchyme. These findings supported roles for FGF-3 and FGF-10 as signals mediating the effects of mesenchyme on epithelium, and our bead implantation experiments showed that FGF-10, in fact, stimulated proliferation in isolated cervical loop epithelium in vitro. It is apparent that they stimulated cell division in the transit-amplifying cell population in the cervical loop, and we also suggest that they stimulated proliferation of the stem cells (Fig. 10 a).
Interestingly, FGF-10 also stimulated lunatic fringe expression in the isolated cervical loop epithelium and, thereby, mimicked the effect of the dental mesenchyme in recombination cultures. We have recently shown that FGF-10 has a similar stimulatory effect on dental epithelial lunatic fringe expression during early budding morphogenesis of mouse teeth (Mustonen, T., unpublished results). As discussed above, the regulation of lunatic fringe expression likely affects Notch signaling (Cohen et al. 1997; Fleming et al. 1997; Panin et al. 1997; Klein and Arias 1998) and, thereby, perhaps the cell fate of the stem cells in the cervical loop (Fig. 10 b). We propose that FGFs may be regulatory signals for stem cells in the tooth and perhaps also in other tissues. FGF stimulates cell division in neural stem cells, whereas removal of FGF permits their differentiation to neurons (Gritti et al. 1996). The expression of Fgf-3 and Fgf-10 in the dental mesenchyme was downregulated exactly in the region where lunatic fringe was downregulated in the inner enamel epithelium differentiating into ameloblasts, and it is possible that this arrest in FGF signaling contributes to terminal differentiation of the cells.
In conclusion, we suggest that the Notch 1 expressing stellate reticulum cells in the cervical loop of the continuously erupting mouse incisor are stem cells that have the capacity to self-renew and to generate progeny that are fated to differentiate into the various cell types of the dental epithelium, most importantly the enamel-producing ameloblasts. Based on the localization of slowly dividing cells among the peripheral stellate reticulum cells, the expression of genes in the Notch and FGF pathways, and the findings that FGF-10 stimulates both cell proliferation and expression of lunatic fringe in the cervical loop epithelium, we propose a model for the molecular mechanism of the putative dental stem cell lineage (Fig. 10 b). We propose that Notch receptors influence the maintenance of the stem cell state, and that the dental mesenchyme controls their proliferation and differentiation by FGF signals directly stimulating stem cell division and regulating the Notch pathway via stimulation of lunatic fringe expression.
We wish to thank Ms. Riikka Santalahti (Institute of Biotechnology, University of Helsinki) for skillful technical assistance. We also thank Dr. Lennart Philipson (Karolinska Institutet, Stockholm) for anti-CVADR antiserum, Dr. Domingos Henrique for Serrate1 probe, Dr. Nobuyuki Itoh for Fgf-10 probe, David Wilkinson for Fgf-3 probe, and Amgen for FGF-10 protein.
This work was supported by the Academy of Finland and the Ministry of Education, Science and Culture of Japan.
1.used in this paper: BrdU, 5-bromo-2′-deoxyuridine; DiI, 1,1′-dioctadecyl-6,6-di(4-sulfophenyl)-3,3,3′,3′-tetramethylindocarbocyanine; FGF, fibroblast growth factor
The present address of Hidemitsu Harada is Second Department of Oral Anatomy and Cell Biology, Kyushu Dental College, 2-6-1 Manazuru, Kokurakita-ku, Kitakyushu-shi, 803-8580, Japan.