Identification of putative dental epithelial stem cells in a lizard with life-long tooth replacement

Most vertebrate species, from fish to humans, can replace lost teeth. Snakes and lizards, for example, generally form multiple sets of teeth throughout life (polyphyodonty), whereas mammals form only two. Despite being a topic of study for over a century (Bolk, 1922; Edmund, 1960; Leche, 1895; Woerdeman, 1919), the cellular basis of tooth replacement is poorly understood. Two reasons account for this. First, tooth replacement occurs postnatally and so is challenging to study. Second, the mouse, the predominant animal model for studying tooth development, does not actually replace its teeth. Tooth renewal in mice is limited to the continuous growth of incisor teeth. Extra teeth can be induced in mice by genetic manipulation, particularly of Wnt pathway genes (Järvinen et al., 2006; Wang et al., 2009), but these experiments reveal little about tooth replacement as it naturally occurs.

Compared with tooth replacement, a great deal more is known about the renewal of the mammalian hair follicle. Hair is continually renewed by a population of stem cells located in the bulge, an area near the base of the follicle. Bulge stem cells were first located by a pulse-chase strategy, whereby retention of tritiated thymidine revealed the presence of cells with exceedingly slow cell cycling times (Cotsarelis et al., 1990). Bulge stem cells have since been shown to express many of the same genes as other adult stem cells (Blanpain et al., 2004; Morris et al., 2004; Tumbar et al., 2004). Lgr5, for instance, marks adult stem cells of the intestine (Barker et al., 2007), the mouse incisor (Suomalainen and Thesleff, 2010) and the hair follicle (Jaks et al., 2008). Likewise, mammalian adult stem cells share a requirement for Wnt pathway signaling. When the pathway is active, stem cells divide to reconstitute a tissue, and when inactive, stem cells remain undifferentiated (Haegebarth and Clevers, 2009).

Some authors have suggested that epithelial stem cells regulate tooth replacement in vertebrates as they do hair renewal in mammals (Huysseune and Thesleff, 2004; Smith et al., 2009). Here, we test this hypothesis using a novel animal model, the leopard gecko Eublepharis macularius. The gecko replaces its teeth continuously throughout life and is amenable to postnatal experimentation. We present pulse-chase and gene expression data that imply that the gecko dental lamina houses cells that resemble mammalian adult stem cells in terms of cell cycling, gene expression and molecular regulation.

Juvenile geckos (see Fig. 1A) were reared at 30°C (Thorogood and Whimster, 1979). Once a gecko reached a mass of 3 g, we administered 5-bromo-2′-deoxyuridine (BrdU; 1 mg/ml) twice daily for 1 week by squirting the solution into the mouth. BrdU was also added to drinking water (1 mg/ml). Two geckos were collected immediately after the pulse to gauge the initial degree of BrdU saturation. Three other BrdU-treated animals were chased for 4, 9 or 20 weeks prior to euthanasia. A segment of the right maxillary tooth row of the 20-week-chased gecko was digitally reconstructed from frontal sections using WinSurf software (Buchtová et al., 2008).

Paraffin sections were steamed in 10 mM sodium citrate for 10 minutes, incubated with antibodies against either BrdU (GE Healthcare) or mouse PCNA (1:50; Vector Laboratories), and visualized using Alexa Fluor 488 (1:200). Radioactive in situ hybridization was performed as described (Buchtová et al., 2008) using probes against E. macularius Lgr5, Igfbp5, Dkk3, Tcf3 and Tcf4 (see Table S1 and Figs S1-S4 in the supplementary material). cDNAs were amplified by degenerate RT-PCR (see Table S2 in the supplementary material) and cloned into pGEM-T Easy (Promega).

Dental tissues from stage 39 gecko embryos (Wise et al., 2009) were grown in Trowell-type culture (see Handrigan and Richman, 2010) for 5 days. The GSK3β inhibitor 6-bromoindirubin-3′-oxime (BIO; EMD Biosciences) was added to experimental cultures (20 μM) to activate Wnt signaling (Sato et al., 2004); controls received DMSO. Three hours prior to collection, cultures were treated with BrdU (10 μM) for proliferation assays.

Juvenile geckos bear ∼100 erupted marginal teeth and a greater number of replacement teeth organized into tooth families in series along the jaw margins (Fig. 1B). At hatching, each family comprises up to three generations of teeth connected by the dental lamina (Fig. 1B-D; see Movie 1 in the supplementary material). Replacement teeth form lingual to the erupted tooth and increase in size until they ultimately erupt and displace the predecessor tooth, a process that takes ∼3-4 months (Edmund, 1960). All replacement teeth initiate from the successional lamina, an outgrowth of the outer enamel epithelium of the predecessor tooth (Fig. 1C-E). The lamina consists of lingual and labial layers and, in between them, loose interstitial cells (Fig. 1F).

The successional lamina makes an obvious candidate for the site of dental stem cells as it gives rise to all replacement teeth in amniotes and in amphibians. However, we showed previously that the successional lamina of the gecko and other squamates comprises a population of proliferative cells (Handrigan and Richman, 2010). This defies a key criterion of adult stem cell populations: that they turnover slowly (Fuchs, 2009). It is unclear, however, where else dental stem cells might be; we found no structure in the gecko dental epithelium that resembles the mouse hair follicle bulge or the distended cervical loop of the mouse incisor.

To determine whether the gecko dental epithelium contains slow-cycling cells, we performed the first BrdU pulse-chase experiment on a non-avian reptile. In geckos fixed immediately following the pulse (0 week), BrdU had been incorporated into ∼57% of dental epithelial cells (Fig. 2A; see Table S3 in the supplementary material). Similarly, Cotsarelis et al. achieved less than 100% saturation of slow-cycling cells in the hair follicle when they pulsed mice for 7 days (Cotsarelis et al., 1990). They reasoned that this was because not all hair follicles cycled during the pulse. Likewise, only some of the 150 or more replacement teeth developing in a juvenile gecko are likely to draw on stem cells during 1 week. BrdU signal in the pulsed geckos was lower in areas of the dental lamina between tooth buds (see Fig. S5 in the supplementary material). Presumably, some of these cells have cell cycling times that exceed 1 week.

Next, we analyzed BrdU retention in geckos chased for 4, 9 and 20 weeks (see Fig. S6 and Table S3 in the supplementary material). After 4 weeks, overall BrdU saturation of the dental epithelium had decreased nearly threefold to 21%, but the distribution of BrdU-positive cells remained unchanged compared to the 0-week geckos (Fig. 2B). After 9 weeks, overall BrdU saturation did not change appreciably compared with 4 weeks (Fig. 2C). However, the fluorescent signal appeared fragmented, indicating that BrdU had been metabolized (compare insets in Fig. 2B,C). We detected signal in the labial, lingual and interstitial cells of the dental lamina.

We extended the chase period to 20 weeks to whittle the BrdU signal down to only the most quiescent cells. After this lengthy chase, the percentage of BrdU-labeled cells in the dental epithelium (regions A-E, Fig. 2D) decreased to ∼3%. The few remaining BrdU-positive cells were disproportionately found in regions B and C along the oral-aboral axis (Fig. 2D,E; see Fig. S7 in the supplementary material). There were virtually no labeled cells in the highly proliferative successional lamina (region D, Fig. 2F).

To determine the spatial distribution of label-retaining cells along the anterior-posterior axis of the jaw, we analyzed BrdU labeling in sagittal sections (see Fig. S5 in the supplementary material) and in a three-dimensional reconstruction of the maxillary tooth row (Fig. 2G,G′). In sagittal sections, BrdU-retaining cells occurred in comparable intertooth areas of the dental lamina that showed relatively low BrdU uptake in the 0-week-chased geckos (see Fig. S5 in the supplementary material). In the reconstruction, BrdU-retaining cells were clustered in spaces between second-generation teeth (Fig. 2G; see Movie 2 in the supplementary material) and were typically found close to, but not in the tip of, the dental lamina (Fig. 2G′).

To test the `stemness' of label-retaining cells in the lingual dental lamina, we scrutinized them for expression of genes that mark hair bulge stem cells: Dkk3, Igfbp5 and Lgr5 (Morris et al., 2004; Tumbar et al., 2004). All three genes were expressed in a restricted domain in the dental lamina, overlapping a subset of BrdU-retaining cells (Fig. 3A-D). These BrdU-retaining, Lgr5/Dkk3/Igfbp5-positive cells are likely to be stem cells. We tested this hypothesis by checking whether the cells share molecular regulation with mammalian adult stem cells.

The canonical Wnt pathway is sufficient to induce quiescent stem cells to form transit-amplifying cells in the mouse intestine and hair follicle (Blanpain et al., 2004; Haegebarth and Clevers, 2009; Lowry et al., 2005). We characterized the dental expression of Tcf3 and Tcf4, two targets of the Wnt pathway, to determine whether the dental lamina cells of the gecko are Wnt active. The dental expression pattern of these genes has not previously been examined in any species. In the mouse hair follicle, Tcf3 and Tcf4 are co-expressed by bulge stem cells and transit-amplifying cells (Nguyen et al., 2009). We predicted that the expression domains of the two genes would overlap with each other and with BrdU-retaining cells (Fig. 4B) and stem cell marker gene expression (Fig. 4C) in the dental lamina. Unexpectedly, they did not overlap (Fig. 4D,E): Tcf4 was expressed in the enamel organ and successional lamina overlapping proliferating cells (regions D, E, Fig. 4E,F), whereas Tcf3 was expressed throughout the dental lamina, including where the putative dental stem cells are located (Fig. 4B,D). These expression data imply that Wnt signals act broadly throughout the dental epithelium and that Tcf3 and Tcf4 mediate different functions. Tcf3 might maintain dental lamina stem cells, whereas Tcf4 might induce them to proliferate. In the mouse, Tcf3 maintains stem cells in the hair follicle (Nguyen et al., 2006), whereas Tcf4 mediates proliferation in the intestinal crypts (Korinek et al., 1998). Although neither gene is known to function in mouse tooth development, a growing body of data implicates canonical Wnt signaling in promoting tooth renewal in mammals (Järvinen et al., 2006; Wang et al., 2009).

To determine how canonical Wnt signaling affects the lingual dental lamina cells, we stimulated the pathway in gecko dental explants using the GSK3β inhibitor BIO. BIO caused the dental lamina to thicken at intertooth and tooth levels (Fig. 4G,H). This phenotype is due to a preferential increase in cell proliferation on the lingual side of the dental lamina (Fig. 4I-N; see Fig. S8 and Table S4 in the supplementary material). Interestingly, the effect of Wnt gain of function on mouse dental cell proliferation has not been explored (Järvinen et al., 2006; Wang et al., 2009) despite the presumed requirement for increased cell proliferation in supernumerary tooth formation.

Since the lingual dental lamina is the primary location of slow-cycling cells, we suggest that Wnt gain of function induced dental stem cells on that side to proliferate. Tcf4 might have mediated the mitogenic effect of BIO as its expression was upregulated throughout the culture, including in the lingual dental lamina (Fig. 4K,L, insets). It is unclear whether BIO acted directly on the lingual cells or indirectly via neighboring mesenchymal cells. Further work is needed to clarify this uncertainty and to dissect how other pathways (e.g. FGF, BMP, Hh) interact with Wnt to determine stem cell fate. The sources of the signals that regulate stem cell fate should also be explored. We suggest that adjacent tooth buds and neural crest-derived mesenchyme are two potential sources (see Fig. S9 in the supplementary material).

On the basis of our pulse-chase and gene expression data (Fig. 4O), we hypothesize that the lingual layer of the gecko dental lamina bears a population of putative dental stem cells. The lingual dental lamina constitutes a fitting home for stem cells as it is a quiescent environment and is not involved in tooth morphogenesis. Thus, lingual dental lamina cells are sequestered from signals that might otherwise instruct them to proliferate or differentiate. At the same time, in the lingual layer of the lamina, stem cells are conveniently positioned to give rise to the successional lamina. We suggest that the successional lamina is built from the proliferative descendents of the stem cells, whereas the stem cells themselves remain fixed in position and sheltered from molecular signals in the lingual dental lamina (see Fig. S9 in the supplementary material). The distribution pattern of BrdU-retaining cells in our reconstruction of the gecko tooth row implies that stem cells are arranged as discrete plaques or `hubs' along the anterior-posterior axis of the dental lamina. Each hub may provide progenitors for a single tooth family or a group of families (see Fig. S9 in the supplementary material). These hypotheses await experimental confirmation by cell lineage analyses.

Finally, we predict that the possession of dental epithelial stem cells is a prerequisite for tooth renewal and, furthermore, that it is a conserved feature of all animals that replace their teeth. Once this has been confirmed, the molecular network that governs the fate of dental epithelial stem cells should be worked out in detail. In particular, more needs to be determined about the signaling that enables stem cells to self-renew indefinitely in non-mammalian vertebrates and why this capacity is reduced in mammals.

We thank Katherine Fu, Marcia Graves and Arthur Sampaio for help with in situ hybridization, provision of reagents and micro CT, respectively. Operating funds from the Natural Sciences and Engineering Research Council (NSERC) to J.M.R. and postdoctoral fellowships to G.R.H. (Michael Smith Foundation for Health Research; NSERC) supported this work.