Primary cilia regulate Shh activity in the control of molar tooth number

Tooth number is highly regulated in mammals. Species such as mice have a highly reduced dentition consisting of only molars and incisors, whereas the development of other mammalian tooth types has been lost during evolution. Evidence from three-dimensional reconstructions of early tooth development in mouse embryos has identified tooth primordia in the diastema, a toothless region between the incisors and molars, which are initiated but are later suppressed by apoptosis (Tureckova et al.,1996; Peterkova et al.,2002). These primordia are believed to be the vestiges of teeth lost during mouse evolution and suggest that a mechanism for suppression of tooth formation in the diastema involves selective apoptosis of tooth primordia (Peterkova et al.,2003). In the early oral cavity, Shh expression is restricted to the thickenings of oral epithelium (dental placodes), from which the tooth germs arise (Bitgood and McMahon,1995; Dassule and McMahon,1998; Hardcastle et al.,1998). A loss of Shh activity in these regions at embryonic day(E)10 results in a lack of epithelial cell proliferation and a failure of tooth bud formation (Cobourne et al.,2001). Once the early tooth bud has formed, continued Shh activity is required within discrete regions of the tooth germ epithelium and mesenchyme for normal growth and morphogenesis of the developing tooth(Dassule et al., 2000; Gritli-Linde et al., 2002; Jeong et al., 2004). Significantly, Shh transcriptional activity is lacking in the diastema and, thus, during normal development of the murine dentition,suppression of Shh pathway activity in the diastema may play an important role in controlling primary tooth number(Cobourne et al., 2004).

Primary cilia have recently been shown to play a crucial role in Shh signalling (Rohatgi et al.,2007; Caspary et al.,2007; Singla and Reiter,2006; Huangfu and Anderson,2005). The Shh receptor Ptch1, its activator Smo and the transcriptional effector Gli1 are all found to localize in cilia, suggesting that the Shh signalling pathway is active in cilia and that cilia are required for Shh signal transduction (Rohatgi et al., 2007; Haycraft et al.,2005). Intraflagellar transport (IFT) proteins are highly conserved in all ciliated eukaryotic cells, and mutation of the proteins that comprise the IFT process result in defects in cilia formation in all organisms studied to date (Rosenbaum and Witman,2002; Pan et al.,2005; Scholey and Anderson,2006). Mice with null mutations in Ift88, which encodes the IFT protein polaris, lack cilia on all cells and die mid-gestation with severe defects in neural tube patterning and closure, polydactyly, and left-right axis determination (Murcia et al., 2000; Zhang et al.,2003). Defects in the neural tube have been associated with loss of Shh signalling regulation through disruption of Gli activation, whereas the limb patterning defects are associated with a gain of Shh activity as a result of a loss of Gli3 repression (Huangfu and Anderson, 2005; Haycraft et al., 2005; Michaud and Yoder,2006). Similar defects have also been reported for mice with severe mutations in other IFT proteins(Huangfu and Anderson, 2005; Liu et al., 2005; May et al., 2005). Tg737^orpk is a hypomorphic allele of polaris; homozygous Tg737^orpk mice exhibit a complex pathology, including kidney and pancreatic cysts, preaxial polydactyly, and supernumerary teeth(Zhang et al., 2003).

We examined the development of vestigial diastema tooth primordia in Tg737^orpk mice and found ectopic tooth formation that correlates with ectopic Shh signalling activity in the diastema mesenchyme. Using Wnt1-Cre and keratin(K)5-Cre to conditionally inactivate polaris in tooth mesenchyme and epithelium, respectively, we show that this cilia-mediated Shh signalling is required only in the mesenchyme. We further show that mutants in the Shh regulatory protein Gas1 also develop diastema teeth as a result of ectopic Shh activity in diastema mesenchyme. Therefore,in the context of control of tooth number, cilia are involved in the repression of Shh activity in diastema mesenchyme cells.

Tg737^orpk and Tg737^Δ2-3β-galwere produced as described previously(Moyer et al., 1994). Gas1 mutant mice, polaris^LoxP mice, K5-Cre and Wnt1-Cre mice were produced as described previously(Chai et al., 2000; Lee et al., 2001; Ramirez et al., 2004; Haycraft et al., 2007).

Radioactive section in situ hybridisation using ³⁵S-UTP radiolabelled riboprobes was carried out as described previously(Ohazama et al., 2008). Whole-mount in situ hybridisation was carried out as described previously(Pownall et al., 1996).

Heads were scanned with Explore Locus SP (GE Pre-clinical imaging) high resolution Micro-CT with a voxel dimension of 8 μm. Three-dimension reconstruction was performed by the three structure analysis software Microview (GE Pre-clinical imaging).

Embryo heads were processed as previously described(Taulman et al., 2001).

Epithelial three-dimension reconstruction of mandible dental and adjacent oral epithelium were performed as previously described(Lesot et al., 1996).

Heads were fixed in 2.5% glutaraldehyde (phosphate buffer) overnight at 4°C and postfixed in 2% osmium tetroxide (Millonigs buffer) after washing by buffer. Specimens were dehydrated through a graded series of ethanols and embedded in Epon 812-equivalent (TAAB Lab). Semithin sections (1 μm) were stained with Toluidine Blue for light microscopy analysis. Ultrathin sections(40-90 nm) were cut, stained with uranyl acetate and lead citrate and examined with a Hitachi H7600 transmission electron microscope.

Sections were incubated with antibody to acetylated α-tubulin orγ-tubulin (Sigma). Alexa488 or Alexa594 was used (Molecular probe) for detecting primary antibody. Immunohistochemistry for Polaris or Shh were performed as previously described(Rosenbaum and Witman 2002; Martinelli and Fan, 2007; Gritli-Linde et al.,2001).

Primary cilia have been shown to be important mediators of Shh activity and, as Shh controls tooth initiation, we examined tooth development in Tg737^orpk adult mice, which contain a hypomorphic mutation in the IFT protein polaris. We identified teeth mesial to the first molar in all four jaw quadrants with 100% penetrance that were not present in wild-type animals (Fig. 1A-D)(Zhang et al., 2003). Because the development of ectopic digits in Tg737^orpk mutants is due to alternations in Shh signalling in the limb bud, we examined the expression of Ptch1 and Gli1, which are upregulated in response to Shh signal transduction, in the region of ectopic tooth development in the Tg737^orpk homozygous mutants at E12.5 and E13.5. To determine the expression pattern of Ptch1, we generated Tg737^orpk homozygous mutants that were also heterozygous for a Ptch1-lacZ allele and examined the activity of β-gal. At E13.5, Ptch1-lacZ staining was increased in the incisor and molar areas of Tg737^orpk homozygous mutants and expanded into the diastema, which is normally devoid of Shh signalling activity(Fig. 1E,F). Gli1expression was also expanded into the diastema region(Fig. 1G-H′) and expression of Gas1, a Shh antagonist in tooth development that is downregulated by Shh activity, was reduced(Fig. 1I-J′)(Cobourne et al., 2004; Martinelli and Fan, 2007). The expansion of Ptch1 and Gli1 expression into the diastema correlates with the placement of premolar-like tooth buds that develop in the Tg737^orpk mutants. These observations suggest that an absence of the IFT protein polaris results in a Shh gain-of-function phenotype. In limb buds, hypomorphic mutation of polaris leads to a series of changes in Shh signalling that include disruption of Gli3 processing,resulting in severe polydactyly similar to that found in Gli3 mutants(Haycraft et al., 2005; Litingtung et al., 2002). The expression of Shh is downregulated in the midline of polaris null mutants, which show severe midline defects(Murcia et al., 2000). It has been suggested that the loss of Gli activators, which play a major role in neural tube patterning, is consistent with the loss of ventral neural tube cells in the IFT protein mutant mice, whereas loss of Gli3 repressor function,which is more important in digit patterning, is consistent with the formation of extra digits in the IFT protein mutants(Huangfu and Anderson, 2005; Michaud and Yoder, 2006). The absence of an IFT protein can therefore affect both positive and negative aspects of Shh signalling, depending upon the particular Gli activity. In the context of the control of tooth number, increased Shh activity is thus consistent with loss of Gli3 repressor which is prominently expressed in the wild-type diastema (see Fig. S1 in the supplementary material)(Huangfu and Anderson, 2005; Haycraft et al., 2005; Michaud and Yoder, 2006).

To investigate the cells that require fully functional cilia during tooth formation, we first determined the location of cilia in tooth germs using a cilia marker, acetylated α-tubulin and a marker of basal bodies of cilia, γ-tubulin. Cilia were found in tooth epithelium and mesenchyme cells at early stages of development (Fig. 2A,B,D). Cilia were observed on both epithelial and mesenchymal cells of supernumerary and endogenous tooth germs in Tg737^orpk mice (Fig. 2C,E). Transmission electron microscopy (TEM) was performed to investigate the structure of the cilia on tooth cells. Cilia were found on dental epithelial and mesenchymal cells and no obvious morphological differences were visible between Tg737^orpk and wild-type cells (Fig. 2F-I). We next examined Tg737 expression and polaris protein localisation in tooth development to confirm that this protein is a component of cilia on dental cells. Tg737 was expressed throughout the dental epithelium and in the underlying mesenchyme, as determined by expression of β-gal in embryos heterozygous for the Tg737^Δ2-3β-galallele (Fig. 2J,K). In agreement with the β-gal expression pattern, immunolocalisation of polaris showed cilia were present on both tooth epithelial and underlying dental mesenchyme cells (Fig. 2L,K).

As cilia are present on both epithelial and mesenchymal cells of early tooth germs, the effect of reduction in polaris in Tg737^orpk mice could be mediated in either or both cell types. Conditional loss-of-function of polaris in epithelial cells was investigated by crossing polaris floxed mice with K5-Cre mice. The K5-Cre/polaris^flox/flox mice survived but there was no evidence of supernumerary teeth, although they did show polydactyly(Fig. 3A,B; data not shown). To test whether polaris is required in dental mesenchyme, we crossed polaris floxed mice with Wnt1-Cre mice. Wnt1-Cre/polaris^flox/flox mice died at birth and showed severe craniofacial abnormalities (data not shown). Supernumerary tooth germs were observed mesial to first molars in the Wnt1-Cre/polaris^flox/flox mice(Fig. 3C,D). The expansion of Gli1 expression into the diastema was seen in the Wnt1-Cre/polaris^flox/flox mice as also observed in Tg737^orpk mice (Fig. 3E-F′). This confirmed that the formation of supernumerary teeth results from loss of polaris in dental mesenchyme. Although Shhis only expressed in the epithelium during tooth development, Ptch1and other Shh pathway molecules are expressed in both epithelial and mesenchymal cells in early tooth primordia. Moreover, a role for Shh in dental epithelial cells is confirmed by abnormal cell differentiation in K14-Cre/Shh and K14-Cre/Smo conditional mutant mice; however, the teeth that develop in K5-Cre/polaris^flox/flox mice undergo apparently normal morphogenesis and cytodifferentiation(Dassule et al., 2000;Gritli-Linde, 2002). This suggests that cilia-mediated Shh activity mediated by polaris is dispensable in dental epithelial cells, whereas Shh signalling in early dental mesenchyme requires polaris-mediated cilia function. Despite the expectation that cilia would be absent in tooth mesenchymal cells in Wnt1-Cre/polaris^flox/flox mice, α-tubulin immunohistochemistry did show cilia present but in reduced numbers (see Fig. S2 in the supplementary material). This may represent incomplete expression of Cre or a contribution of non-neural crest cells(Chai et al., 2000).

In order to confirm that ectopic Shh signalling in the diastema was the direct cause of supernumerary tooth formation, we examined the dentition of mice with a targeted disruption in the Gas1 gene, which we has previously identified as an inhibitor of Shh in diastema mesenchyme. Gas1 is expressed in diastema mesenchyme and non-dental mesenchyme,but not in tooth germs (Fig. 4G,H) (Cobourne et al.,2004). Supernumerary teeth in the diastema region of both the maxilla and mandible were observed with 100% penetrance in Gas1mutants (Fig. 4A-F). These teeth, positioned mesial to the first molars, were identical to those observed in Tg737^orpk mice. The presence of these teeth was associated with ectopic Shh signalling activity within the diastema region,demonstrated by analysis of Ptch1, Gli1 and Shh expression,and Shh protein localization, as observed in the Tg737^orpkmutants (Fig. 4I-U). Significantly, we also found that Gas1 expression was downregulated in diastema mesenchyme of Tg737^orpk mutant embryos(Fig. 1I-J′). Ectopic Shh activity in the diastema mesenchyme was thus observed in two different mutants, both resulting in ectopic tooth formation mesial to the first molars. In Gas1 mutants, the effect on Shh signalling can be attributed to a direct consequence of loss of an antagonist, whereas Tg737^orpk may be indirect following reduction in polaris that results in loss of Gas1.

In the developing tooth, Shh signalling is required for initiation and subsequent downgrowth of the dental epithelium into the underlying mesenchyme to control development of the teeth(Sarkar et al., 2000; Cobourne et al., 2001); in the diastema, Shh signalling is normally specifically inhibited by Gas1(Cobourne et al., 2004). When Gas1 is lost either directly, or indirectly in the mesenchyme of Tg737^orpk mice, Shh signalling is increased. As a result,the vestigial tooth rudiments mesial to the first molar do not undergo apoptosis, but survive and develop into a functional tooth. This suggests that modulation of Shh signalling has played an important part in the evolution of dental patterns by regulating tooth number. Interestingly, in the talpid² chick mutant, ectopic activation of Shh signalling in the developing oral cavity correlates with the formation of reptile-like teeth [reminiscent of avian ancestors(Harris et al., 2006)]. The requirement for Shh signalling in the mesenchyme to regulate events in the epithelium also identifies a mesenchyme-to-epithelium interaction in this process.

Ectopic tooth formation mesial to the first molars has been observed in mice with mutation of the BMP/Wnt antagonist ectodin (Wise) and following misexpression of Eda and Edar (Mustonen et al., 2003; Kassai et al.,2005; Tucker et al.,2004). In both these cases, development of the molars is impaired. The formation of diastema teeth in the context of normal molar development is observed in Sprouty2 and Sprouty4 mutant mice, and has been interpreted as a role for suppression of FGF signalling in suppression of diastema tooth formation. In Sprouty4 mutants, ectopic teeth are restricted to one quadrant with low penetrance; in Sprouty2 mutants,92% have bilateral, 5% have unilateral and 3% have no ectopic teeth in the mandible, whereas ectopic teeth are observed in less than 5% of the maxilla(Klein et al., 2006) (J.-M. Courtney and P.T.S, unpublished). We show here that ectopic Shh signalling in the diastema produces ectopic teeth with 100% frequency in all four jaw quadrants without affecting molar development and thus we propose that Shh is the direct determinant of tooth initiation and that the roles of FGF, BMP and other signalling pathways in tooth suppression are mediated through Shh.

Mouse first molar tooth buds form from a fusion of four epithelial swellings, the most mesial of which is excluded from the first molar tooth bud and undergoes apoptosis (Peterkova et al.,2002). We used histology and 3D reconstruction of E15.5 embryos to show that the supernumerary tooth primordia were observed in the position of the former mesial swellings, indicating that rather than being suppressed,these swellings continued to develop into independent teeth(Fig. 1K-N). The cyclin-dependent kinase inhibitor p21 provides a marker for diastema bud apoptosis and in Tg737^orpk homozygous embryos we found that p21 expression was downregulated in diastema epithelial buds(see Fig. S3 in the supplementary material). The premolar-like teeth in Tg737^orpk thus form from the survival of vestigial swellings.

To verify whether the supernumerary teeth developed with a molar-like or incisor-like programme, we performed in situ hybridisation at E15.5 for Barx1, which is specifically expressed in developing molars, in addition to dHAND (Hand2 – Mouse Genome Informatics)and Islet1 (Isl1–Mouse Genome Informatics), which are expressed in incisors (Thomas et al.,1998; Tucker et al.,1998; Mitsiadis et al.,2003). The supernumerary teeth expressed Barx1 in a similar manner to the developing first and second molars, but did not express Hand2 or Isl1 (see Fig. S4 in the supplementary material). These data support the identification of these ectopic teeth as molar-like,and confirm they develop from oral epithelium in the molar region.

In several other mouse mutants that have been reported to develop extra teeth, the form and position of the molars are highly abnormal, which makes analysis of the morphology of diastema teeth difficult(Mustonen et al., 2003; Kassai et al., 2005). Interestingly, the molar and incisor teeth of homozygous Tg737^orpk mutants showed no major abnormalities in shape,size or position. The supernumerary teeth were smaller than the first molars,had fewer cusps and a cusp pattern that was consistent with a premolar-like identity. The lingual cusps of maxillary supernumerary teeth were more prominent in comparison with the lingual cusps of mandibular supernumerary teeth (Fig. 1O-T). Furthermore,the supernumerary teeth in the maxilla showed a concave-shape root, whereas those in the mandible had a round root shape(Fig. 1U,V). These differences in root and cusp shapes are a consistent feature of human premolars(Ash and Nelson, 2003; Berkovitz et al., 2002).

The fact that the extra teeth that develop as a result of ectopic Shh form a shape that is appropriate for their position indicates that despite having lost the ability to form a premolar tooth shape over 50-100 million years ago,the mouse embryo has retained all the genetic information necessary to make this tooth type (Meng et al.,1994; Ji et al.,2002). This supports the concept that loss of tooth types during evolution results from suppression of tooth initiation. This is also consistent with the specification of tooth shape being provided by expression of specific homeobox genes such as Dlx and Barx1 in the mesenchyme prior to initiation (Tucker and Sharpe, 2004). These gene expression domains in the mesenchyme of the facial primordia provide cells with positional information. This information is used to direct cells to follow particular pathways of hard tissue morphogenesis. In animals where tooth development is suppressed, such as birds or the mouse diastema, this information is still required for bone and cartilage morphogenesis. Thus, what has been `lost' in the evolution of mice is the ability to initiate and maintain epithelial tooth buds, mesial to the first molars. When ectopic tooth development is stimulated as a result of ectopic Shh signalling, the tooth primordia develop from mesenchyme cells with the information appropriate for that position.