Wnt signaling pathway, an important regulator of embryonic morphogenesis, has essential functions in the maintenance and differentiation of stem cells in adult tissues, including the intestine and hair follicle (Chu et al., 2004; Clevers, 2006; Gordon and Nusse, 2006; Mikkola and Millar, 2006; Blanpain et al., 2007; Haegebarth and Clevers, 2009). Wnt signaling is also one of the several conserved signaling families regulating tooth morphogenesis (Thesleff and Tummers, 2009). Many Wnts and Wnt pathway mediators are expressed during the development of embryonic teeth, and the indispensable role of Wnt signaling in tooth morphogenesis has been demonstrated in mouse and human studies (Dassule and McMahon, 1998; Sarkar and Sharpe, 1999; Kratochwil et al., 2002; Obara et al., 2006; Adaimy et al., 2007). Tooth development in mouse embryos is arrested at an early stage when Wnt signaling is inhibited by expressing the Wnt inhibitor Dkk1 and by inhibiting epithelial β-catenin, and when Lef1 function is deleted (van Genderen et al., 1994; Andl et al., 2002; Liu et al., 2008). On the other hand, continuous activation of the Wnt/β-catenin pathway in the oral epithelium induces ectopic teeth and continuous tooth formation in mice (Järvinen et al., 2006; Kuraguchi et al., 2006; Liu et al., 2008). In human, heterozygotic loss of Wnt10a or Axin2 function causes tooth agenesis in association with other ectodermal defects and colorectal cancer, respectively (Lammi et al., 2004; Adaimy et al., 2007).
Although most mammals have lost the potential of tooth renewal, some mammalian teeth have maintained the ability to grow continuously. One example is the rodent incisor, in which the sharpness of the cutting edge is maintained by continuous eruption and asymmetric deposition of the hard enamel only on the labial (anterior) surface of the tooth. The continuous growth and differentiation of the dental hard tissue producing cells in the rodent incisors are sustained by stem cells, present in the proximal end of the incisor in the cervical loop (Fig. 1A). The epithelial stem cells are thought to reside in the stellate reticulum compartment and to invade the surrounding loop of basal epithelial cells, the enamel epithelium. The cells then proliferate as transient amplifying cells, migrate toward the apex of the tooth and differentiate into enamel forming ameloblasts (Harada et al., 1999). The asymmetric deposition of enamel only on the labial surface is associated with a large cervical loop and stem cell compartment on the labial side compared with a very small cervical loop on the enamel-free lingual side.
The maintenance and activation of the epithelial stem cells in the incisor is regulated by a network of signaling molecules including the mesenchymally expressed fibroblast growth factor 3 (Fgf3), Fgf10 and Activin stimulating epithelial proliferation, and bone morphogenetic protein 4 (Bmp4) inhibiting progenitor cell proliferation but inducing their differentiation into ameloblasts (Harada et al., 1999, 2002; Wang et al., 2004, 2007; Plikus et al., 2005; Klein et al., 2008). However, the possible functions of Wnt signaling in epithelial stem cell maintenance and differentiation have not yet been explored.
The characteristics of Wnt pathway have given a challenge into analyzing the signaling pathways activated by Wnts. According to the current view, the signaling output depends more on the receptors and the identity of the signal-receiving cell rather than on the identity of the Wnt ligand (Clevers, 2006; van Amerongen et al., 2008). This indicates that the same ligand can activate different pathways (Gordon and Nusse, 2006; Mikels and Nusse, 2006a). The most studied and best characterized pathway is the canonical β-catenin pathway. Less characterized are the pathways earlier called as noncanonical Wnt pathways, including planar cell polarity (PCP), Ca2+ related pathways or signaling mediated through receptors Ror2 and Ryk (van Amerongen et al., 2008). When Wnt/β-catenin signaling is inactive, β-catenin is phosphorylated and targeted to degradation by a protein complex consisting of several molecules including Axin, APC, CK1, and GSK3. Wnt ligand binding to a Frizzled receptor and a lipoprotein receptor related protein (Lrp) 5/6 co-receptor activates the interaction between receptor complex and intracellular Dishevelled. This leads to the inactivation of the degradation protein complex and accumulation of β-catenin into the nucleus where it interacts with members of Lef/Tcf family of transcription factors and activates gene expression (Clevers, 2006).
To get insight into the role of Wnt signaling in the regulation of the continuous growth and hard tissue formation in the mouse incisor and in particular in the maintenance and differentiation of the epithelial stem cells and their progeny, we examined the expression patterns of several Wnt ligands, inhibitors, signal mediators, and targets, including Lgr5, a Wnt target gene in the intestine, in the embryonic mouse incisors (Barker et al., 2007; van der Flier et al., 2007). To locate the cells involved in active Wnt signaling, we used three Wnt/β-catenin reporter mouse lines, BATgal, TOPgal, and Axin2/ConductinLacZ/LacZ and localized also Axin2 mRNA expression (DasGupta and Fuchs, 1999; Lustig et al., 2002; Maretto et al., 2003). We observed that the expression patterns of Wnt ligands were mainly associated with the epithelial and mesenchymal cell differentiation and that the ligands were absent from the stellate reticulum cells in the labial cervical loop, i.e., the putative stem cell compartment. Active Wnt/β-catenin signaling was detected in the mesenchymal tissues, and in the dental epithelium in the lingual side as well as weakly in the preameloblasts and ameloblasts in the labial side. Of interest, we did not detect Wnt/β-catenin signaling in the epithelial stem cells or stellate reticulum in the labial cervical loop. However, we found restricted expression of Lgr5 in the putative epithelial stem cells, suggesting that Lgr5 is a stem cell marker for the incisor epithelial stem cells as has been earlier shown for the intestinal and hair follicle stem cells and that it is not regulated by Wnt signaling in the incisor.
The Expression Patterns of Wnt Ligands
The expression patterns of 12 Wnt ligands were analyzed at embryonic day (E) 16 and E18 mouse incisors: Wnt2b, Wnt3, Wnt3a, Wnt4, Wnt5a, Wnt5b, Wnt6, Wnt7a, Wnt7b, Wnt10a, Wnt10b, and Wnt11. Of these ligands, Wnt2b and Wnt3 were not detected in the incisors (data not shown), while all other Wnt ligands were expressed in varying temporal and spatial patterns in the incisors. The expression patterns are shown in Figure 1 and summarized in Table 1.
Table 1. Expression of Wnt Ligands in Different Cells and Tissues in the Incisor at E16 and E18
Most Wnt ligands were expressed in the dental epithelium at E16 and/or E18, yet none of them were expressed in the labial cervical loop epithelium. Wnt4 was the only ligand which was expressed intensely in the epithelium in the lingual cervical loop at E16 and E18 (Fig. 1D,E). Wnt6 expression became intense in the lingual cervical loop at E18 (Fig. 1K). Wnt4 was also expressed in the labial outer enamel epithelium and ameloblasts, and Wnt3a, Wnt6, and Wnt10a were expressed in the preameloblasts and ameloblasts (Fig. 1B–E,J,K,P,Q).
Most Wnt ligands were expressed also in the dental mesenchyme. Only Wnt5a showed strong expression in the cervical loop area (Fig. 1F,G). It was intensely expressed around the lingual cervical loop at E16 and E18, and weakly around the labial cervical loop at E16. Wnt5a as well as Wnt4, Wnt5b, Wnt6, and Wnt10a were expressed intensely in the papilla mesenchyme at E16 (Fig. 1D,F,H,J,P). At E18, the expression of Wnt5a continued in the papilla mesenchyme extending to the cervical loop areas especially in the lingual side, whereas Wnt6 and Wnt10a had become restricted to the odontoblasts (Fig. 1G,K,Q). Wnt11 was expressed mainly in the dental follicle (Fig. 1T,U).
Expression Patterns of Wnt Pathway Mediators and Inhibitors
Lef1 and Tcf1 are transcription factors mediating Wnt/β-catenin pathway (Behrens et al., 1998). Lrp4 is thought to regulate Wnt signaling, but many aspects of its functions remain still unknown (Ohazama et al., 2008). Dickkopf (Dkk) family members are mostly known as antagonists of Wnt/β-catenin pathway; however, the results concerning the interaction between Dkk3 and Wnt/β-catenin signaling are controversial (Caricasole et al., 2003; Hoang et al., 2004; Niehrs, 2006; Lee et al., 2009). The expression patterns are shown in Figure 2 and summarized in Table 2.
Table 2. Expression of Wnt Pathway Mediators and Inhibitors in Different Cells and Tissues in the Incisor at E16 and E18
All genes were predominately expressed in the dental mesenchyme, with the exception of Dkk4, which was not detected in the incisor (data not shown). Lef1 and Tcf1 were expressed intensely in the mesenchyme next to labial and lingual cervical loops (Fig. 2A–D). Tcf1, Lrp4, and Dkk1 were expressed in the papilla mesenchyme and especially in the preodontoblasts and odontoblasts (Fig. 2C–H).
None of the analyzed Wnt pathway mediators and inhibitors was detected in the stellate reticulum of the labial cervical loop, i.e., in the putative stem cells. Lrp4 was the only gene expressed in the epithelium of the labial cervical loop (Fig. 2E,F). However, it was expressed only at E16 and restricted to the basal epithelium and absent from the central core, the stellate reticulum, of the loop. Dkk3 was expressed in the lingual cervical loop and in the differentiated ameloblasts (Fig. 2K,L). Expressions of the other analyzed genes were not detected in the dental epithelium.
Wnt/β-Catenin Reporter Activity Is Mainly Confined to Dental Mesenchyme and Is Completely Absent From the Epithelial Stem Cell Compartment
We examined the locations of active Wnt/β-catenin signaling in the incisors by localizing Axin2 expression by in situ hybridization and by using three different Wnt/β-catenin reporter mouse lines, BATgal, TOPgal, and Axin2LacZ/LacZ (DasGupta and Fuchs, 1999; Lustig et al., 2002; Maretto et al., 2003). Axin2 is a negative regulator and also a downstream target of the Wnt/β-catenin pathway and, hence, reports pathway activity (Lustig et al., 2002). The reporter mouse lines indicate active Wnt signaling by expressing the enzyme β-galactosidase from the LacZ gene in cells where the transgene is activated. BATgal and TOPgal lines have several multimerized TCF/Lef consensus binding sites and siamois or c-fos minimal promoter driving LacZ expression (DasGupta and Fuchs, 1999; Maretto et al., 2003). In Axin2LacZ/LacZ mice the Axin2 gene has been deleted by inserting LacZ to exon 2 of the gene and because Axin2 is a direct Wnt target gene the expression of LacZ indicates active Wnt/β-catenin signaling (Lustig et al., 2002). Wnt signaling has many target genes depending on the responding cell, but currently one of the best candidates to indicate active Wnt/β-catenin signaling is Axin2 (Clevers, 2006). Deficiency in Axin2 expression causes only minor cranial defects and no dental abnormalities, despite the important role in the GSK3/APC/Axin complex (Yu et al., 2005; and our unpublished observations).
Axin2 was expressed intensely in the mesenchyme next to labial and lingual cervical loops and especially in the preodontoblasts and odontoblasts at E16 and E18 (Fig. 3A,B). In the epithelium, Axin2 expression was detected in the lingual epithelium and also very faint expression was detected in the labial side in the preameloblasts and ameloblasts (Fig. 3A,B arrow).
BATgal, TOPgal, and Axin2LacZ/LacZ reporter lines did not show any Wnt/β-catenin reporter activity in the incisor epithelium either in the labial or lingual side. The reporter activity was restricted to the mesenchymal tissue both at E16 and E18 in all three reporter lines (Fig. 3C–F, and data not shown). However, there were differences in the intensities and patterns of expression between the lines. The BATgal reporter was very active at E16 and E18 in scattered mesenchymal cells, including the preodontoblasts, odontoblasts, and cells lining the cervical loops, and at E16, positive cells were present throughout the dental papilla (Fig. 3C,D). TOPgal activity was only detected in odontoblasts and in the cells underlying the odontoblasts (Fig. 3E,F). Axin2LacZ/LacZ reporter activity was weaker but it was apparent in the cervical loop areas in the mesenchyme surrounding both the labial and lingual cervical loops as well as in the preodontoblasts and odontoblasts, thus resembling the signaling activity detected in BATgal and TOPgal mice (Fig. 3C–F, and data not shown). When the reporter activity in the incisor is compared with the distribution of Axin2 transcripts detected by the in situ hybridization analysis, the Axin2 transcripts show a markedly intense and wider distribution covering all positive areas in the three reporter lines (Fig. 3A–F). In addition, only Axin2 mRNA was detected in the dental epithelium.
Taken together, the reporter activities in the different analyses varied in intensity and pattern. All reporters showed activity in the odontoblasts, and the BATgal and Axin2LacZ/LacZ reporters as well as Axin2 transcripts indicated activity in the mesenchyme surrounding the cervical loops. In the epithelium Wnt/β-catenin activity was detected only as the expression of Axin2 mRNA in the lingual dental epithelium and faintly in the preameloblasts and ameloblasts in the labial epithelium. Thus, neither the reporter lines nor Axin2 expression indicated Wnt activity in the stellate reticulum of the labial cervical loop, indicating that Wnt/β-catenin signaling may not directly regulate the maintenance of epithelial stem cells in the incisors. However, Wnt signaling might have a role in the differentiation of preameloblasts and ameloblasts.
Lgr5, a Stem Cell Marker and a Wnt Target Gene in the Intestine, Is Expressed in the Putative Epithelial Stem Cells
Lgr5 is an orphan G protein-coupled receptor consisting of a seven transmembrane protein with a large N-terminal extracellular domain and leucine-rich repeats (McDonald et al., 1998; Hsu et al., 1998; Hermey et al., 1999). We analyzed Lgr5 expression in the incisors because of its recently identified role as a Wnt target gene and a marker of the epithelial stem cells in the intestine and hair follicle (Barker et al., 2007; Jaks et al., 2008). In situ hybridization analysis revealed a remarkably localized pattern of Lgr5 expression in the epithelium of E16 and E18 incisors. The Lgr5-expressing cells were restricted to a subset of cells in the stellate reticulum compartment in the labial cervical loop at the site where the epithelial stem cells are thought to reside (Fig. 4A,B). No Lgr5 expression was detected elsewhere in the dental epithelium or mesenchyme. Because Lgr5 was expressed in the cells where we did not detect Wnt reporter activity, Lgr5 appears not to be a Wnt target gene in the incisor stem cells unlike in the intestine (Barker et al., 2007).
The continuously growing mouse incisor is an advantageous model for stem cell research because the progressive differentiation of the epithelial and mesenchymal cell lineages occurs as a gradient advancing from the proximally located stem cell niche toward the incisor apex and the process can be easily observed in longitudinal tissue sections. In the present study, we addressed the role of Wnt signaling by localizing Wnt pathway gene expression and Wnt/β-catenin reporter activity in incisor tissue sections. We observed remarkably intense and developmentally regulated expression of 10 different Wnt ligands in the incisors at the late embryonic stages. The patterns varied between different Wnts, but all were expressed in epithelial and/or mesenchymal cell lineages. Surprisingly, active Wnt/β-catenin signaling was observed almost exclusively in mesenchymal tissue in the reporter mice, and the Wnt signal mediators Lef1 and Tcf1 were also restricted to mesenchymal cells. Also the Wnt inhibitor Dkk1 and possible inhibitor Dkk3 were found exclusively in dental mesenchyme, notably in the cells lining the Wnt-expressing dental epithelium at the dental papilla side (Dkk1) and at the follicle side (Dkk3).
Because some of the studied Wnt ligands are known to signal by means of other pathways than β-catenin, it is conceivable that there is β-catenin–independent Wnt signaling in the incisors. Wnt5a, which is often associated with the noncanonical pathways, but depending on context may either inhibit or activate the Wnt/β-catenin pathway (Mikels and Nusse, 2006b), was intensely expressed in the dental mesenchyme. Wnt5a was the only Wnt ligand expressed in the mesenchyme around the cervical loops. Because the expression was remarkably intense around the lingual cervical loop, it might negatively regulate the epithelial stem cells in their niche by means of a noncanonical pathway not detected in our reporter studies.
While the three Wnt reporter mouse lines we used are established and known to faithfully report β-catenin–dependent Wnt signaling (DasGupta and Fuchs, 1999; Lustig et al., 2002; Maretto et al., 2003), we found distinct differences in the activities between the lines. Similar observations have been reported in other tissues earlier, and it is speculated that the variation is due to differences in the transgenes and failure to detect low levels of Wnt/β-catenin activity (Barolo, 2006; Boras-Granic and Wysolmerski, 2008). However, BATgal and TOPgal reporter lines were found to have similar reporter activity in embryonic teeth (Liu et al., 2008).
The most intense Wnt/β-catenin reporter activity was localized in the mesenchymal odontoblasts in all the reporter lines. Based on the expression patterns of Wnt ligands, Wnt6 and Wnt10a are the most likely candidates to induce reporter activity in odontoblasts because they were expressed in the dental epithelium that is known to regulate the differentiation of odontoblasts (Thesleff et al., 2001). Recently Wnt10a was proposed to have a function in odontoblast differentiation in molars (Yamashiro et al., 2007). Because the odontoblasts also expressed remarkably intensely both Wnt6 and Wnt10a and no reporter activity was seen in the surrounding cells, it seems apparent that these Wnts have both autocrine and paracrine effects in odontoblasts.
Wnt has been the most actively studied pathway in several epithelial stem cells, in particular in the skin and intestine where the Wnt/β-catenin pathway has established functions in stem cell regulation (Lowry et al., 2005; Barker et al., 2007; Jaks et al., 2008). The most intriguing observation in our study was the apparent lack of any signs of active Wnt/β-catenin signaling in the stellate reticulum in the labial cervical loop, which is the region where the putative stem cells reside. Wnt/β-catenin signaling is also very low in the hair follicle stem cells, but it is high in the transit amplifying cells, and β-catenin stabilization promotes the transition between stem cells and transit amplifying cells in the hair follicle (Lowry et al., 2005). In the incisors, however, no BATgal, TOPgal, or Axin2LacZ/LacZ reporter activity was detected in the progeny of stem cells, and only faint Axin2 expression was detected in the proliferating preameloblasts and differentiated ameloblasts on the labial side epithelium. Instead, Axin2 was expressed more strongly in the lingual incisor epithelium, which fails to differentiate into ameloblasts and to form enamel, and participates in the development of the enamel-free root-like lingual surface of the incisor. Thus, Wnt signaling in the epithelium might rather have an inhibitory role on stem cell proliferation and differentiation.
Of interest, the overexpression of Wnt3 in the epithelium of postnatal mouse incisor results a progressive loss of ameloblasts and inhibition of enamel formation (Millar et al., 2003). This could be explained either by the depletion of stem cells or inhibition of their proliferation/differentiation to ameloblasts, or by both. Because we associated high Wnt signaling with the lack of stem cell maintenance as well as with absence of ameloblast differentiation, it is not possible to conclude which stage of epithelial stem cell development was affected in the Wnt3 overexpressing mice. Also because high Wnt signaling was present both in the mesenchyme and epithelium in the lingual enamel-free side, the epithelially expressed Wnt3 may have affected the epithelium as well as the mesenchyme.
The cervical loop stem cell niche bears morphological similarity with the stem cell niches in the hair follicle and intestine (Thesleff and Tummers, 2009). The current concept according to which the epithelial stem cells reside in the core of the labial cervical loop in the incisor is based on the morphology, in vitro tissue ablation experiments, the presence of long-term label-retaining cells, and analyses of cell division kinetics and gene expression patterns (Smith and Warshawsky, 1975; Harada et al., 1999, 2002; Wang et al., 2007). As direct evidence is lacking, the location and identity of the incisor stem cells have remained elusive. Lgr5 is a specific marker of epithelial stem cells in the intestine and hair follicle, and was identified in a screen of Wnt/β-catenin target genes in the intestinal stem cell niche (Barker et al., 2007; van der Flier et al., 2007; Jaks et al., 2008). The expression of Lgr5 in the region of the putative stem cells in the labial cervical loop in the incisor is an intriguing observation and can be taken as additional support to the proposed stemness of these cells. Because we did not observe any signs of Wnt activity in the Lgr5-positive cells in the incisor stem cell niche, there are apparently differences between the dental and the intestinal stem cells in the regulation of Lgr5 expression. However, it is reasonable to suggest that the function of Lgr5 in the different epithelial stem cells, i.e., hair follicle, intestine, and tooth, may be conserved. The Lgr5−/− mice die at birth, and the role of Lgr5 in adult stem cells still remains uncharacterized (Morita et al., 2004).
Although our data indicate that Wnt/β-catenin signaling does not directly regulate the epithelial stem cells of the incisor, the active Wnt signaling in the mesenchyme surrounding the cervical loop epithelium may well affect indirectly the epithelial stem cells. BATgal and Axin2LacZ/LacZ reporter activity and particularly Axin2 mRNA expression were intense in the mesenchyme around the cervical loop, and also Lef1 and Tcf1 were highly expressed there. All these indicators of Wnt activity were more prominent around the lingual cervical loop, which is devoid of stem cells as compared to the labial cervical loop, and at the more advanced E18 stage Wnt/β-catenin activity around the labial cervical loop was diminishing. These observations suggest that mesenchymal Wnt/β-catenin signaling may have an inhibitory role on the stem cell maintenance. Several of the Wnts analyzed in our experiments were expressed in the lingual aspect of the incisor including cervical loop. These were Wnt3a, Wnt4, Wnt6, Wnt7a, and Wnt11 in the lingual epithelium and Wnt5a in the mesenchyme. The expression of Wnt4 was remarkably intense and was confined to the lingual dental epithelium, including the cervical loop.
We propose that the observed Wnt/β-catenin activity in the mesenchyme lining the lingual cervical loop regulates the mesenchymal signals controlling the maintenance and proliferation of epithelial stem cells. Such signals include FGF3, FGF10, and Activin, which have been identified as positive regulators and BMP4, which is a negative regulator of epithelial stem cells. These signals act in an integrated network, fine-tuned by specific inhibitors including Follistatin and Sprouty (Harada et al., 1999; Plikus et al., 2005; Wang et al., 2007; Klein et al., 2008). The asymmetry between the labial and lingual cervical loops, i.e., the presence of stem cells in the labial and their absence in the lingual cervical loop, result from modulation of the balance of positive and negative signals (Wang et al., 2007; Klein et al., 2008). Based on our observations, we suggest that Wnt/β-catenin signaling in the cervical loop mesenchyme is a negative effector in this signal network and contributes to the deficiency of epithelial stem cells in the lingual cervical loop.
Animals and Preparation of Embryonic Tissues
NMRI mice were used at various embryonic stages. Transgenic mouse lines used in this study have been described earlier: BATgal mice were kindly provided by Stefan Piccolo, TOPgal mice were from Jackson Laboratories and Axin2/ConductinLacZ/LacZ (Axin2LacZ/LacZ) mice were kindly provided by Walter Birchmeier (DasGupta and Fuchs, 1999; Lustig et al., 2002; Maretto et al., 2003).
Embryos were staged according to morphological criteria, plug day was embryonic day (E) 0. Whole heads or lower jaws were dissected, fixed in 4% paraformaldehyde, dehydrated, and embedded in paraffin. Sagittal paraffin sections of 7 μm were processed for in situ hybridization.
LacZ activity was detected with X-gal staining as described earlier (Järvinen et al., 2006). The E16 and E18 mandibles were fixed in 2% paraformaldehyde in 0.2% mM glutaraldehyde in phosphate buffered saline (PBS) for 60 to 90 min, respectively, and stained with X-gal staining solution for 20 hr for LacZ activity. Background staining for sectioned samples was done with Fast Red.
In Situ Hybridization
Radioactive in situ hybridization for paraffin sections was carried out using 35S-UTP labeling (Amersham) as described previously (Wilkinson and Green, 1990). The following probes were used: Wnt2b, Wnt3a, and Wnt7a (M. James), Wnt3, Wnt4, Wnt5a, Wnt5b, Wnt6, Wnt7b, and Wnt11 (Sarkar and Sharpe, 1999), Wnt10a and Wnt10b (Wang and Shackleford, 1996), Tcf1 (James et al., 2006), Lef1 (Kratochwil et al., 1996), Lrp4 and Dkk4 (Fliniaux at al., 2008), Axin2 (Lammi et al., 2004), Dkk1 (Andl et al., 2002), Dkk2 (Diep et al., 2004), Dkk3 (Fjeld et al., 2005), Lgr5 (Barker et al., 2007).
We thank Marja Mikkola for critical reading of the manuscript, Mark Tummers for valuable comments, and Merja Mäkinen, Riikka Santalahti, and Raija Savolainen for their excellent technical assistance.