ABSTRACT
The expression of genes involved in the Sonic Hedgehog signalling pathway, including Shh, Ptc, Smo, Gli1, Gli2 and Gli3, were found to be expressed in temporal and spatial patterns during early murine tooth development, suggestive of a role in early tooth germ initiation and subsequent epithelial-mesenchymal interactions. Of these Ptc, Smo, Gli1, Gli2 and Gli3 were expressed in epithelium and mesenchyme whereas Shh was only detected in epithelium. This suggests that Shh is involved in both lateral (epithelial-mesenchymal) and planar (epithelial-epithelial) signalling in early tooth development. Ectopic application of Shh protein to mandibular mesenchyme induced the expression of Ptc and Gli1. Addition of exogenous Shh protein directly into early tooth germs and adjacent to tooth germs, resulted in abnormal epithelial invagination, indicative of a role for Shh in epithelial cell proliferation.
In order to assess the possible role of this pathway, tooth development in Gli2 and Gli3 mutant embryos was investigated. Gli2 mutants were found to have abnormal development of maxillary incisors, probably resulting from a mild holoprosencephaly, whereas Gli3 mutants had no major tooth abnormalities. Gli2/Gli3 double homozygous mutants did not develop any normal teeth and did not survive beyond embryonic day 14.5; however, Gli2−/−; Gli3+/− did survive until birth and had small molars and mandibular incisors whereas maxillary incisor development was arrested as a rudimentary epithelial thickening.
These results show an essential role for Shh signalling in tooth development that involves functional redundancy of downstream Gli genes.
INTRODUCTION
Sonic Hedgehog (Shh), a member of the vertebrate family of hedgehog signalling proteins, has been shown to play a critical role in development (Echelard et al., 1993). Shh signalling has been implicated in establishing the polarity of the floorplate, neural tube, somites and limbs (reviewed by Hammerschmidt et al., 1997; and references therein) and Shh null mutant mice die before birth with extensive defects in these areas and are cyclopic (Chiang et al., 1996). Cyclopia is a feature of holoprosencephaly, which in humans has recently been attributed to point mutations with SHH (Belloni et al., 1996; Roessler et al., 1996).
Patched (Ptc) has been shown through biochemical assays (Marigo et al., 1996a; Stone et al., 1996) to be a receptor for the Shh ligand. Ptc is a transmembrane protein that is thought to act with Smoothened (Smo), also a transmembrane protein (Alcedo et al., 1996; Van Den Heuvel and Ingham, 1996), to form a receptor complex for Shh. Much of the detail of component parts of the Shh signalling pathway has been elucidated in Drosophila from studies of the role of hedgehog in segmentation. The current model (Nusse, 1996; Stone et al., 1996; Kalderon, 1997) is that hh binds to ptc and, in response to this binding, ptc which normally represses smo, releases this inhibition, allowing smo to activate the transcription of downstream target genes via the cubitus interruptus (ci) transcription factor. Ci is a member of the Gli family of zinc finger transcription factors (Domingeuz et al., 1997; Hepker et al., 1997), of which the mouse homologues are Gli1, Gli2 and Gli3 (Hui et al., 1994). Mutant studies have shown that Gli2 and Gli3 are essential for development (Johnson, 1967; Vortkamp et al., 1991, 1992; Hui and Joyner, 1993; Mo et al., 1997). Both Gli2 and Gli3 mutant mice die at birth and have severe skeletal abnormalities. Loss of Gli2 is associated with defects of the teeth, palate, limbs, sternum, vertebral column and the skull (Mo et al., 1997). Although Gli3 mutant mice also exhibit skeletal defects, these defects are generally different to those of the Gli2 mutants (Johnson, 1967; Schimmang et al., 1993; Hui et al., 1994; Mo et al., 1997). Gli3 homozygous mutant mice have defects in skull vault formation, cleft palate, severe polysyndactyly and shortening of the tibia, Gli3 mutants also show abnormal development of the neural arches.
Although the precise role of the Gli genes in Shh signalling remains to be determined, in vitro studies have indicated that a Gli-binding site can function as a Shh-responsive element in cultured cells (Sasaki et al., 1997). Several observations suggested that different Gli genes may possess distinct functions in Shh signalling. In chick limb development, Shh has been shown to regulate Gli1 and Gli3 expression (Marigo et al., 1996b) and studies in the mouse limb have suggested that Gli3 is a negative regulator of Shh (Büscher et al., 1997; Masuya et al., 1997). Gli1 has been suggested as a potential mediator of the Shh signal (Platt et al., 1997) and has been shown to be activated transcriptionally by Shh in developing mouse (Hynes et al., 1997) and Xenopus embryos (Lee et al., 1997). The lack of floor plate differentiation and reduced Ptc and Gli1 expression in the Gli2 mutant neural tube also indicates that Gli2 is a downstream mediator of Shh signalling (C-c. Hui, unpublished).
Shh expression has previously been shown to be highly restricted to epithelial cells during tooth development (Bitgood and McMahon, 1995; Koyama et al., 1996). The early expression of Shh at the epithelial thickening stage suggests a possible role for this signalling pathway in tooth germ initiation and epithelial-mesenchymal interactions. Shh expression in the enamel knot (Vaahtokari et al., 1995), a proposed signalling centre within developing teeth, may be involved in the cuspal patterning of the tooth germ by regulating proliferation. In order to investigate the role of Shh signalling in tooth development, we have mapped the expression of components of the pathway including Ptc, Smo, Gli1, Gli2 and Gli3. Tooth development and expression of the Shh signalling pathway genes has been studied in Gli2 and Gli3 mutant embryos, together with expression of a number of other genes such as Activin βA, Bmp-4, Lef-1 and Msx-1. Application of Shh protein on beads shows that downstream genes in the pathway can be activated in mesenchyme and that tooth bud morphology is altered suggesting a role for Shh signalling in epithelial cell proliferation involved in tooth bud formation. The results further show that, although an intact Shh signalling pathway is required for normal tooth development, there is functional redundancy between Gli genes such that tooth development is only primarily affected when both Gli2 and Gli3 genes are absent.
MATERIALS AND METHODS
Generation of mice
Gli2, Gli3 and Gli2;Gli3 mutant mice have been described previously (Mo et al., 1997). Some of the Gli2;Gli3 mutant mice were maintained in a background containing a gene trap line gtC101 that carries a lacZ transgene insertion at the cordon bleu gene (Gasca et al., 1995). Wild-type embryos were generated by mating CD-1 mice. On the following day, if a vaginal plug was present, midday was taken as E0.5.
Histology
Mutant embryos, newborns and their litter-mates were fixed as described by Mo et al. (1997). Newborn heads were decalcified in 0.5 M EDTA pH 7.6. Wild-type embryos were fixed at 4°C overnight in 4% paraformaldehyde in PBS and then washed in PBS. All embryos and newborns were then dehydrated through a graded series of ethanols, embedded in paraffin and sectioned frontally at 7-10 μm. Whenever possible, adjacent or nearby sections of mutant embryos were used for in situ hybridisation. Sections not used for in situ hybridisation were stained with haematoxylin and eosin to visualise tooth morphology. For sections from embryos containing gtC101, β-galactosidase staining was performed as described by Gasca et al. (1995).
In situ hybridisation
Whole-mount in situ hybridisation was carried out according to Wilkinson (1992). In situ hybridisation using 35S-radiolabelled riboprobes was carried out as described by Angerer and Angerer (1992). The radioactive riboprobes were synthesised with 35S-labelled UTP (ICN).
Explant cultures
E10.5 mandibles were dissected from freshly harvested embryos. The mandibles were treated with 2 units/ml of Dispase (GibcoBRL) in order to separate the epithelial component from the mesenchyme. The mesenchyme was then placed on a 0.1 μm Millipore filter, on a metal grid in an organ culture dish containing media consisting of DMEM (GibcoBRL), 10% Foetal Calf Serum and 20 units/ml penicillin/streptomycin (GibcoBRL). Affi-Gel agarose beads (heparin acrylic beads were used for Fgf-8 protein) were prepared by thoroughly washing the beads in 1× PBS, allowing them to dry and then incubating them in protein for 1 hour at 37°C. The beads were then applied to the mesenchyme and cultured for 24 hours at 37°C, 5% CO2 and 40% O2. The beads used in this study were soaked in either Shh protein (1.25 μg/μl rat Shh, 19.6 kDa amino terminal fragment, a kind gift from Ontogeny), Fgf-8 protein (1 mg/ml recombinant mouse Fgf-8b, R&D Systems), Bmp-4 protein (1 mg/ml Bmp-4, a gift from the Genetics Institute) or Bovine Serum Albumin (BSA). Intact mandibles (epithelial and mesenchymal components still in contact) were also cultured with Shh protein-soaked beads from E10.5 for 3 days. The culture medium was as described above and this media was changed on the second day of culturing.
RESULTS
Expression of Shh, Ptc, Smo, Gli1, Gli2 and Gli3 in developing murine teeth
The expression patterns of the Shh pathway genes were examined in developing teeth from E11.5 to E14.5 using in situ hybridisation.
At E11.5 Shh expression was localised to the epithelial thickenings of all the tooth germs (Fig. 1B). Ptc expression in the molars E11.5 was restricted to the mesenchyme underlying the tooth thickening (Fig. 1C) and, in the mandibular and maxillary incisors at this stage, the expression domain of Ptc appeared to have extended slightly more laterally but was still localised in the odontogenic region of the mandible. Smo was found ubiquitously throughout the mandibular and maxillary processes (data not shown). The expression of Gli2 and Gli3 was widespread at E11.5 (Fig. 1E and F, respectively) and Gli1 was similar to Ptc expression (Fig. 1D).
Radioactive in situ hybridisation of Shh, Ptc, Gli1, Gli2 and Gli3 in wild-type mandibular and maxillary processes. (A-F) E11.5 molar thickenings. (A) H&E (B)Shh. (C) Ptc. (D) Gli1. (E) Gli2. (F) Gli3. (G-L) E12.5 molars. (G) H&E (H) Shh. (I) Ptc. (J) Gli1. (K) Gli2. (L) Gli3. (M-R) E13.5 molars. (M) H&E (N) Shh. (O) Ptc. (P) Gli1. (Q) Gli2. (R) Gli3. (S-X) E14.5 molars. (S) H&E (T) Shh. (U) Ptc. (V) Gli1. (W) Gli2. (X) Gli3. (Y-D′) E14.5 enamel knot. (Y) H&E (Z) Shh. (A′) Ptc. (B′) Gli1. (C′) Gli2. (D′) Gli3. E.K., enamel knot. H&E, haemotoxylin and eosin.
Radioactive in situ hybridisation of Shh, Ptc, Gli1, Gli2 and Gli3 in wild-type mandibular and maxillary processes. (A-F) E11.5 molar thickenings. (A) H&E (B)Shh. (C) Ptc. (D) Gli1. (E) Gli2. (F) Gli3. (G-L) E12.5 molars. (G) H&E (H) Shh. (I) Ptc. (J) Gli1. (K) Gli2. (L) Gli3. (M-R) E13.5 molars. (M) H&E (N) Shh. (O) Ptc. (P) Gli1. (Q) Gli2. (R) Gli3. (S-X) E14.5 molars. (S) H&E (T) Shh. (U) Ptc. (V) Gli1. (W) Gli2. (X) Gli3. (Y-D′) E14.5 enamel knot. (Y) H&E (Z) Shh. (A′) Ptc. (B′) Gli1. (C′) Gli2. (D′) Gli3. E.K., enamel knot. H&E, haemotoxylin and eosin.
The localised expression of Shh was intensified at the early bud stage in both the maxillary and mandibular incisors. The molar Shh expression was, however, decreased in intensity (Fig. 1H). At this stage, Ptc was expressed uniformly in the odontogenic epithelium and mesenchyme of the incisors, whereas in the molars the medial pattern of expression was maintained (Fig. 1I). At E12.5, Smo was again widely expressed (data not shown) and Gli1 was expressed in the same regions as Ptc only at slightly lower levels (Fig. 1J). Gli3 at this stage was significantly more localised in the mesenchyme around the tooth germs than at previous stages of development (Fig. 1L). Gli2 was widely expressed at the early bud stage of tooth development (Fig. 1K).
By E13.5, Shh transcripts were found to be restricted to cells that will form the enamel knot (Fig. 1N). At E13.5, the Ptc molar expression resembled that, at E12.5 (Fig. 1O), in that Ptc was expressed in the dental papilla and also in regions of the dental epithelium. In the dental follicle, Ptc was expressed more strongly on the lingual side of the tooth germs. In the incisors at E13.5, Ptc was also expressed most strongly on the lingual side of the tooth germs. Smo was again found to be ubiquitously expressed. At E13.5, Gli3 was expressed more strongly than at previous stages (Fig. 1R). In the molars and mandibular incisors, Gli3 was expressed uniformly in the mesenchyme surrounding the dental organ. In the maxillary incisors, the mesenchymal expression was greater on the buccal side of the tooth germ. Gli2 transcripts in the molars and maxillary incisors were now more localised around the tooth germs and were still detected in both the epithelial and mesenchymal components of the developing teeth (Fig. 1Q). However, at this stage, Gli2 transcripts in the maxillary incisors were observed to be more strongly expressed medial to the tooth germs. Gli2 continued to be expressed uniformly around the mandibular incisors. Gli1 expression continued to mimic that of Ptc at the bud stage of development; in the maxillary and mandibular incisors and in the molars, it was apparent in the dental papilla, part of the dental organ and in the dental follicle (Fig. 1P).
By the cap stage, E14.5, the enamel knot has fully formed and Shh expression clearly marks this structure (Fig. 1T), as previously described by Vaahtokari et al. (1995). At E14.5, the pattern of expression of Ptc in the molars (Fig. 1U) and incisors resembled that at E13.5. The expression of Smo was slightly more specific at E14.5 (data not shown), being more strongly expressed in the epithelial component, but absent from the enamel knot. Smo was also found to be expressed much more specifically in the tooth germs at E15.5 (data not shown). At the cap stage of tooth development, Gli1 mRNA could be detected in the incisors in similar regions as at E13.5. In the molars Gli1 was expressed in a region of the tooth germ that appeared to surround the area of the enamel knot that was marked by Shh expression (Fig. 1V). Gli3 was found in the mesenchyme surrounding the dental organ in all the teeth (Fig. 1X) and was also expressed weakly in the dental organ, but like the rest of the Shh signalling pathway genes, was absent from the enamel knot. Gli2 was expressed weakly in the incisor mesenchyme and in the maxilla at the site of fusion between the nasal septum and the primary palate. In the molars Gli2 could be seen in the dental follicle surrounding the dental organ and dental papilla and only weakly expressed in the epithelial component of the tooth germ (Fig. 1W).
A closer look at the expression patterns of this pathway of genes in the enamel knot at E14.5, the cap stage of tooth development introduces some interesting data. As previously known Shh is strongly expressed in the enamel knot at this stage (Fig. 1Z), whereas Ptc (Fig. 1A′), Smo (data not shown),Gli1 (Fig. 1B′), Gli2 (Fig. 1C′) and Gli3 (Fig. 1D′) are all expressed in the developing tooth germ but are, strikingly, absent from the enamel knot structure.
The expression patterns of Shh, Ptc, Smo, Gli1, Gli2 and Gli3 from E11.5 to E14.5 are summarised in Fig. 2.
Schematic diagram summarising the overlapping expression patterns of Shh, Ptc, Smo, Gli1, Gli2 and Gli3. (A-C) Summary of the expression of Shh, Ptc and Smo. Shh expression is denoted in red, Ptc expression in blue and Smo in green. (A,D) Epithelial thickening stage of tooth development. (B,E) Early bud stage. (C,F) Cap stage. (D-F) Summary of the expression of Gli1, Gli2 and Gli3. Gli1 expression is shown in orange, Gli2 in pale blue and Gli3 in purple.
Schematic diagram summarising the overlapping expression patterns of Shh, Ptc, Smo, Gli1, Gli2 and Gli3. (A-C) Summary of the expression of Shh, Ptc and Smo. Shh expression is denoted in red, Ptc expression in blue and Smo in green. (A,D) Epithelial thickening stage of tooth development. (B,E) Early bud stage. (C,F) Cap stage. (D-F) Summary of the expression of Gli1, Gli2 and Gli3. Gli1 expression is shown in orange, Gli2 in pale blue and Gli3 in purple.
Induction of mesenchymal gene expression by Shh
In order to investigate the response of genes in the Shh signalling pathway to exogenous Shh ligand, beads soaked in Shh protein were applied to murine mandibular mesenchyme from which the epithelium had been removed (and hence the endogenous source of Shh). After 24 hours of culturing the mesenchyme with Shh beads, in situ hybridisation was performed and it was established that Shh beads induced the expression of Ptc and Gli1 in E10.5 mandibular mesenchyme cultures (Fig. 3B and D respectively). In the control mandibular mesenchyme cultures in which beads soaked in BSA protein were applied to mesenchyme with its epithelium removed, after 24 hours of culturing in identical conditions to that of the Shh bead cultures, there was no evidence of these beads switching on the expression of either Ptc or Gli1 (Fig. 3A and C, respectively). This experiment was repeated for Gli2, Gli3 and Pax9 and it was found that ectopic Shh did not induce the expression of any of these genes. In control experiments at E10.5, the epithelium was removed from the mesenchyme and both Ptc (Fig. 3E) and Gli1 (Fig. 3F) expression was lost from the mesenchyme. E11.5 mandibular mesenchyme was also cultured with Shh protein-soaked beads for 24 hours. In situ hybridisation showed that Shh did not induce the expression of either Msx-1 or Bmp-4 (data not shown). E10.5 mesenchymal explant cultures were also performed with beads soaked in Fgf-8 or Bmp-4, which are co-expressed with Shh in the epithelium, and in situ hybridisation showed that neither of these signalling molecules could induce the expression of Ptc, Gli1, Gli2 or Gli3 (data not shown).
E10.5 mandibular mesenchyme cultured for 24 hours with protein soaked beads, (black arrow). (A) Mesenchyme cultured with a BSA bead followed by in situ hybridisation shows that Ptc expression has not been induced. (B) Mesenchyme cultured with Shh beads. In situ hybridisation shows that Ptc expression has been induced by the Shh protein. (C) Mesenchymal cultures with a BSA bead. In situ hybridisation with Gli1 shows that this gene has not been induced by a BSA bead. (D) Mesenchymal cultures for 24 hours with beads soaked in Shh protein. In situ hybridisation shows that Shh protein does induce expression of Gli1. (E and F) E10.5 mesenchymal cultures for 24 hours followed by in situ analysis for Ptc and (F) Gli1. At E10.5 the expression of these genes is lost in the mesenchyme when the epithelium is removed.
E10.5 mandibular mesenchyme cultured for 24 hours with protein soaked beads, (black arrow). (A) Mesenchyme cultured with a BSA bead followed by in situ hybridisation shows that Ptc expression has not been induced. (B) Mesenchyme cultured with Shh beads. In situ hybridisation shows that Ptc expression has been induced by the Shh protein. (C) Mesenchymal cultures with a BSA bead. In situ hybridisation with Gli1 shows that this gene has not been induced by a BSA bead. (D) Mesenchymal cultures for 24 hours with beads soaked in Shh protein. In situ hybridisation shows that Shh protein does induce expression of Gli1. (E and F) E10.5 mesenchymal cultures for 24 hours followed by in situ analysis for Ptc and (F) Gli1. At E10.5 the expression of these genes is lost in the mesenchyme when the epithelium is removed.
Effect of ectopic Shh on developing tooth germs
In order to determine the role of Shh in developing tooth germs, Shh protein-soaked beads and control BSA beads were placed in presumptive molar tooth forming regions at E10.5 and explants were cultured for 3 days. In cases where the beads were found in the tooth germ epithelium, Shh beads produced a reproducible change in tooth bud morphology. Buds with Shh implanted beads formed multiple invaginations into the mesenchyme (Fig. 4A) which were never seen in BSA controls (Fig. 4C). Radioactive in situ hybridisation with Msx-1 was used to show that the epithelial invaginations had induced expression in condensing (odontogenic) mesenchyme around the buds, as seen in wild-type tooth buds. Msx-1 was expressed in the mesenchyme around the entire abnormal Shh-treated tooth buds, but interestingly was absent from the mesenchyme closest to the tip of the tooth germ (Fig. 4D). When a Shh-soaked bead was placed next to the molar tooth germ and cultured for 3 days, ectopic epithelial invaginations were observed (Fig. 4E).
Ectopic application of Shh to tooth germs. (A,B) BSA bead applied to an E10.5 presumptive molar tooth germ region and then cultured for 3 days (sections stained with Haematoxylin and eosin (H & E)). (B) Adjacent section to A, showing Msx-1 expression. (C,D) Shh bead placed in an E10.5 presumptive molar tooth germ region and then cultured for 3 days. (D) Adjacent section to C, showing Msx-1 expression (arrow = area of mesenchyme devoid of expression). (E) Shh bead placed in oral epithelium next to a molar tooth germ and cultured for 3 days; an ectopic epithelial invagination can be observed around the bud.
Ectopic application of Shh to tooth germs. (A,B) BSA bead applied to an E10.5 presumptive molar tooth germ region and then cultured for 3 days (sections stained with Haematoxylin and eosin (H & E)). (B) Adjacent section to A, showing Msx-1 expression. (C,D) Shh bead placed in an E10.5 presumptive molar tooth germ region and then cultured for 3 days. (D) Adjacent section to C, showing Msx-1 expression (arrow = area of mesenchyme devoid of expression). (E) Shh bead placed in oral epithelium next to a molar tooth germ and cultured for 3 days; an ectopic epithelial invagination can be observed around the bud.
Tooth phenotypes in Gli2 and Gli3 mutants
Gli2 null mutants were found to have tooth defects predominantly associated with maxillary incisors (Fig. 5). In all the embryos examined (n=19), the morphology of the molars was unaffected by the loss of the Gli2 gene. In only one of the embryos examined were the mandibular incisors affected where an additional ectopic mandibular incisor was observed medial to one of the normal incisor germs. In situ hybridisation with Msx-1 and Ptc revealed that this epithelial bud was indeed odontogenic (data not shown). The effect of the Gli2 null mutation on the maxillary incisors varied between embryos. The most common phenotype (n=15/19) was the partial fusion of the two maxillary incisors (Fig. 5F-J). In three Gli2 mutant embryos, both maxillary incisors remained but were in close proximity to each other (Fig. 5K). In another embryo maxillary incisors were absent. Close examination of the histology of the single central maxillary incisors at E13.5 (Fig. 5G) showed that the incisors had arisen through the fusion of two maxillary incisors and were not mesiodens. Although the incisors were fused, at all the stages studied the basic histology remained mostly unaffected (Fig. 5). At E13.5, the mesenchymal condensations appeared normal (Fig. 5G) and the enamel knots were present at E14.5, shown by the presence of Fgf-4 and Shh (data not shown). Newborn maxillary incisors had odontoblasts and ameloblasts but these teeth were small and severely misshapen. The location of ameloblasts in particular was abnormal (Fig. 5M). Mutant maxillary incisors that had not fused but were close together had correct positioning of ameloblasts. In Gli3 null mutants (n=5), tooth development appeared to be normal (Fig. 6B).
Tooth phenotype of Gli2 null mutants. The epithelial component of the maxillary incisors is outlined in red. (A-C) Gli2+/+ embryos stained with malachite green. (D,E) Gli2+/+ stained with Haematoxylin and eosin. (F-H) Gli2−/− embryos stained with malachite green. (I,J) Gli2−/− embryos stained with Haematoxylin and eosin. (A,F) E11.5; one central epithelial thickening forms in the Gli2 mutants (F) in contrast to the two thickenings observed in the wild-type embryo (A).(B,G) E13.5; the mutant maxillary incisors have fused (G). (C,H) E15.5; one central maxillary incisor can be seen in the mutant (H). (D,I) E16.5; At this stage and also at newborn, a central maxillary incisor in the mutant can be seen. (E,J) Newborn. (K) E15.5 Gli2−/− maxillary incisors. The maxillary incisor tooth germs have not fused but are in close proximity, outlined in red. (L) Newborn wild-type maxillary incisors showing odontoblasts (od) and ameloblasts (am). (M) Newborn Gli2−/− maxillary incisors showing the unusual positioning of the ameloblasts. (od) odontoblasts, (am) ameloblasts.
Tooth phenotype of Gli2 null mutants. The epithelial component of the maxillary incisors is outlined in red. (A-C) Gli2+/+ embryos stained with malachite green. (D,E) Gli2+/+ stained with Haematoxylin and eosin. (F-H) Gli2−/− embryos stained with malachite green. (I,J) Gli2−/− embryos stained with Haematoxylin and eosin. (A,F) E11.5; one central epithelial thickening forms in the Gli2 mutants (F) in contrast to the two thickenings observed in the wild-type embryo (A).(B,G) E13.5; the mutant maxillary incisors have fused (G). (C,H) E15.5; one central maxillary incisor can be seen in the mutant (H). (D,I) E16.5; At this stage and also at newborn, a central maxillary incisor in the mutant can be seen. (E,J) Newborn. (K) E15.5 Gli2−/− maxillary incisors. The maxillary incisor tooth germs have not fused but are in close proximity, outlined in red. (L) Newborn wild-type maxillary incisors showing odontoblasts (od) and ameloblasts (am). (M) Newborn Gli2−/− maxillary incisors showing the unusual positioning of the ameloblasts. (od) odontoblasts, (am) ameloblasts.
Tooth phenotypes in Gli3 null mutant embryos and in the compound heterozygotes Gli2−/−Gli3+/−. A-D are stained with Haematoxylin and eosin. The only tooth abnormality in the Gli3 null mutant affects the maxillary incisors which are slightly elongated (B) in comparison to wild-type embryos of the same stage, newborn, (A). Maxillary incisors were absent at E16.5 in Gli2−/−Gli3+/− embryos (D), compared to E16.5 wild-type maxillary incisors (C) (E,F) E12.5 wild-type and Gli2−/−;Gli3+/− embryos respectively. Maxillary incisors can be seen in both embryos, outlined in red. The maxillary incisors of the Gli2−/−;Gli3+/− arrested at the thickening stage of tooth development (F). (G) shows E14.5 molar tooth germs exhibiting a lacZ transgene, cordon-bleu. The transgene is expressed in part of the epithelial component of the tooth germ and in some of the underlying condensing mesenchyme cells. (H,J) show the developing maxillary (H) and mandibular (I) incisors in E13.5 Gli2−/−;Gli3−/− mutant embryos. The β-gal-stained epithelial component of the tooth germ is indicated by arrows and the β-gal-stained condensing mesenchyme cells with an arrowhead.
Tooth phenotypes in Gli3 null mutant embryos and in the compound heterozygotes Gli2−/−Gli3+/−. A-D are stained with Haematoxylin and eosin. The only tooth abnormality in the Gli3 null mutant affects the maxillary incisors which are slightly elongated (B) in comparison to wild-type embryos of the same stage, newborn, (A). Maxillary incisors were absent at E16.5 in Gli2−/−Gli3+/− embryos (D), compared to E16.5 wild-type maxillary incisors (C) (E,F) E12.5 wild-type and Gli2−/−;Gli3+/− embryos respectively. Maxillary incisors can be seen in both embryos, outlined in red. The maxillary incisors of the Gli2−/−;Gli3+/− arrested at the thickening stage of tooth development (F). (G) shows E14.5 molar tooth germs exhibiting a lacZ transgene, cordon-bleu. The transgene is expressed in part of the epithelial component of the tooth germ and in some of the underlying condensing mesenchyme cells. (H,J) show the developing maxillary (H) and mandibular (I) incisors in E13.5 Gli2−/−;Gli3−/− mutant embryos. The β-gal-stained epithelial component of the tooth germ is indicated by arrows and the β-gal-stained condensing mesenchyme cells with an arrowhead.
To determine whether there is a functional redundancy of Gli2 and Gli3 in tooth development, we examined the phenotypes in Gli2;Gli3 double mutants. Tooth development was not affected in Gli2+/−;Gli3+/− mice, while Gli2−/−;Gli3+/− mutants (n=7) had smaller than normal mandibular incisors and molars and maxillary incisors were absent (Fig. 6D). At E12.5 two central epithelial thickenings were visible, but these were fused and did not progress beyond this stage (Fig. 6F). Most Gli2−/−;Gli3−/− mutants died around E10.5 and only a few could survive up to day E14.5 (C.-c. H., unpublished). Examination of one E13.5 and one E14.5 Gli2−/−;Gli3−/− embryo revealed no signs of tooth development beyond a rudimentary bud stage, equivalent to about E13.0. In E13.5 double mutant embryos, incisor buds were severely malformed, with single, very small epithelial invaginations present (Fig. 6H,I). In gtC101 background (Gasca et al., 1995), β-gal staining marked both the epithelial and condensing mesenchymal cells in the developing tooth (Fig. 6G). β-gal staining in these mutant tooth buds (Fig. 6H and I) was similar to that in wild-type buds suggesting that some interactions between the epithelium and mesenchyme had taken place. No sign of any molar tooth development could be detected in Gli2−/−;Gli3−/− embryos suggesting that their development was more severely affected than incisors.
Epithelial-mesenchymal interactions in tooth development of Gli2 and Gli3 mutants
In order to determine whether the early interactions between epithelium and mesenchymal cells that are known to be required for initiation and bud formation could occur in the mutant embryos, expression and genes involved in these interactions was examined. Lef-1 expression in epithelial thickenings has been shown to be essential for tooth development (Kratochwil et al., 1996), expression of Msx-1 and Bmp-4 in early tooth bud mesenchyme has been shown to be involved in signalling interactions at this stage (Chen et al., 1996) and expression of activin βA in mesenchyme prior to epithelial invagination has been shown to be essential for development of incisors and mandibular molars (Ferguson et al., 1998). Expression of each of these genes in Gli2−/−; Gli3+/− embryos was examined and found to be normal (Fig. 7).
Radioactive in situs on E12.5 Gli2−/−;Gli3+/− mutant maxillary incisors. (A) Light-field. Expression of the following genes is unchanged in the Gli2−/−;Gli3+/− embryos. (B) Lef1; (C) Msx1; (D) Activin; (E) Bmp4.
Expression of Shh pathway genes in Gli mutants
In Gli2−/− embryos, the expression of Gli1 and Ptc were found to be altered. Between E11.5 and E15.5, Gli-1 expression was found to be downregulated specifically in the epithelial component of all the tooth germs, but remained normal in the mesenchymal component (Fig. 8D). The expression of Ptc in Gli2 mutants was not as straightforward as that of Gli1. Ptc expression was normal at the stages examined except at E13.5-E14 where Ptc was downregulated in the epithelium only (Fig. 8C). Adjacent sections of Gli2 mutants hybridised with Gli1 and Ptc showed that Gli1 and Ptc expression was absent from the epithelium in identical areas. In Gli3−/− embryos at E13.5 the expression of Ptc and Gli1 was significantly upregulated (Fig. 8E,F). However, in the Gli2−/−;Gli3+/− embryos at E13.5, Ptc and Gli1 expression was only slightly weaker in the epithelial component (Fig. 8G,H).
In situs showing changes in the expression of Ptc and Gli1 in molar tooth germs of Gli mutant embryos. (A,B) Wild-type expression patterns of Ptc and Gli1 at E13.5. (C,D) Gli2 mutants at approximately E14. Ptc (C) and Gli1 (D) expression is absent in the epithelial component of the tooth germ. (E,F) Gli3 mutant embryos at E13.5. Ptc (E) and Gli1 (F) are upregulated in these mutants. (G,H) Gli2−/−; Gli3+/− mutants at E13.5. Both Ptc (G) and Gli1 (H) expression is slightly weaker in the epithelial component of the tooth germs.
In situs showing changes in the expression of Ptc and Gli1 in molar tooth germs of Gli mutant embryos. (A,B) Wild-type expression patterns of Ptc and Gli1 at E13.5. (C,D) Gli2 mutants at approximately E14. Ptc (C) and Gli1 (D) expression is absent in the epithelial component of the tooth germ. (E,F) Gli3 mutant embryos at E13.5. Ptc (E) and Gli1 (F) are upregulated in these mutants. (G,H) Gli2−/−; Gli3+/− mutants at E13.5. Both Ptc (G) and Gli1 (H) expression is slightly weaker in the epithelial component of the tooth germs.
DISCUSSION
The expression of genes in the Shh signalling pathway during murine tooth development suggests that this pathway is active and has some direct role during the early stages of development of this organ. Expression of Ptc is considered to be an indication of active Shh signalling because when Shh binds to Ptc it derepresses Ptc transcription (Goodrich et al., 1997). Since Ptc expression at E10.0-E12.5 is localised to cells close to Shh-expressing cells in areas where teeth will develop implies that this pathway has a role in the early formation of tooth germs. Moreover, the restricted expression of all three downstream Gli genes supports this role for Shh signalling. The expression of Shh in the enamel knot and Ptc and Gli genes in surrounding tooth germ cells confirms the postulated role of the enamel knot as a signalling centre although the role of the centre remains unresolved.
The ability of an exogenous source of Shh, applied to mesenchyme, to induce the expression of Ptc and Gli1 indicates that the early odontogenic mesenchymal cells are a target for Shh signalling. Indeed this induction of Ptc and Gli1 in mesenchymal cells appears to be specific for Shh, since neither Fgf-8 nor Bmp-4, which are co-expressed with Shh in dental epithelium, could induce the expression of these genes. Addition of an exogenous source of Shh to the epithelium of early intact tooth germs produced abnormal epithelial invaginations resulting in ‘star-shaped’ tooth buds and ectopic epithelial invaginations were formed when Shh was added close to but not directly in tooth germs. This implies that Shh activity affects epithelial cell proliferation and that the role of Shh in early tooth development may be to direct epithelial cell proliferation to produce a tooth bud. This role for the Shh signalling pathway was confirmed by the result of ectopic epithelial invaginations produced when recombinant Shh was placed in oral (not dental) epithelium. Such a role is consistent with that proposed for Shh in lung development where overexpression of Shh expressed in endoderm induces Ptc and Gli1 expression in mesenchyme and results in increased proliferation of epithelium and mesenchyme (Grindley et al., 1997). Indeed the expression of Shh signalling pathway genes at E14.5 also supports a proposed role for this pathway in cell proliferation. Vaahtokari et al. (1995) showed that in the tooth germ the enamel knot cells do not proliferate. Although Shh is expressed in the enamel knot the signalling pathway genes are all absent from this structure, implying that the action of Shh is outside the enamel knot.
Because of the severity of the facial defects in Shh mutant mice, it was not possible to investigate the role of the ligand in this pathway for tooth development. Mice with targeted mutations in Gli2 and Gli3 genes were therefore used to evaluate the importance of the Shh pathway in tooth development. Gli3 mutants had no obvious tooth defects indicating that this gene function is not essential for tooth development (Mo et al., 1997). Gli2 mutants were found to have a consistent tooth defect affecting development of maxillary incisors. Maxillary incisors were either absent or more often present as a single central incisor. Examination of maxillary incisor tooth germ development showed that the tooth germs formed very close together, indicating that their fusion is the probable cause of the single central incisor. Overall the Gli2 phenotype resembles that of a mild holoprosencephaly and suggests that the incisor tooth phenotype observed is a result of abnormal midfacial development. Abnormalities of maxillary incisors are a common feature of holoprosencephly where even in its mildest form a single central maxillary incisor is present. In more severe forms of holoprosencephly, no maxillary incisors develop but development of other teeth is not affected (Cohen and Sulik, 1992). The differentiation of ameloblasts and odontoblasts in fused incisors in Gli2 mutants indicates that cytodifferentiation has occurred. The small size, distorted position and shape of the fused incisors, together with the abnormal positioning of ameloblasts suggests that fusion of the two tooth germs has resulted in a ‘mixing’ of the inner enamel epithelium and dental papilla cells, which give rise to ameloblasts and odontoblasts, respectively.
Because the other tooth types in Gli2 mutants form normally this suggested that Gli2 is not essential for tooth development. However, since Gli2 and Gli3 are co-expressed to a large extent during early tooth development, it seemed possible that the lack of a tooth phenotype in the mutants was due to functional redundancy. Functional redundancy of transcription factor genes in tooth development has been described for both the Msx and Dlx classes of homeobox genes (Thomas et al., 1997, Richard Maas, personal communication). Functional redundancy between Gli2 and Gli3 was confirmed by analysis of double mutant embryos. Gli2−/−;Gli3−/− mice showed no signs of any molar tooth development and only single, very small incisor tooth germs could be detected. The presence of these rudimentary tooth germs indicates that initiation of tooth development can occur in the absence of both Gli2 and Gli3 genes suggesting that either Shh signalling is not involved in tooth initiation or that there is functional redundancy between all three Gli genes.
Gli2−/−;Gli3+/− embryos had no maxillary incisors, smaller mandibular incisors and smaller molars. Epithelial thickenings were present in the maxillary incisor region but these failed to develop further. Although Gli2−/−;Gli3+/− mice had a more severe facial phenotype than Gli2 mutants, which may have been the indirect cause of failure of maxillary incisor development, this seems unlikely. Epithelial-mesenchymal signalling interactions involving other pathways that are known to be important for tooth initiation, namely Lef-1, and for signalling at the early bud stage, namely Bmp-4, Msx-1 and activin βA, were found to be normal in Gli2−/−;Gli3+/− maxillary incisor tooth germs. The arrest in maxillary incisor development is more likely directly related to Shh signalling and failure of the epithelium to invaginate and form buds is consistent with a role for Shh signalling in epithelial invagination. Although molars and mandibular incisors formed in the Gli2−/−;Gli3+/− mutants, these were consistently smaller than in controls, supporting some role for this pathway in early cell proliferation but also indicating a differential effect on tooth development according to position. Thus maxillary incisor tooth development is more affected than development of the other teeth by loss of both Gli2 alleles and one Gli3 allele. Similar position-dependent effects have been described for loss of Dlx-1 and Dlx-2 genes, which only affect development of maxillary molars, and activin βA, which only affects development of incisors and mandibular molars, despite being expressed in all developing teeth (Thomas et al., 1997; Ferguson et al., 1998).
Interestingly, although loss of both alleles of Gli3 had no effect on tooth development, loss of one Gli3 allele on a background of no Gli2 alleles produced a more severe phenotype than Gli2−/− alone. Thus, although Gli2 is able to compensate completely for loss of Gli3 in facial development, Gli3 can only partially compensate for loss of Gli2. Moreover, the dosage of both Gli2 and Gli3 alleles seems to be important because the tooth phenotype is more penetrant in Gli2−/−; Gli3+/− than in Gli2+/−;Gli3−/− embryos (data not shown). The failure of tooth development to progress beyond a rudimentary bud stage in Gli2−/−;Gli3−/− embryos supports the notion that these two genes are functionally redundant for tooth development.
Changes in expression of Shh pathway genes were detected in molar tooth development of Gli2 and Gli3 mutants even though these teeth developed normally. In Gli2−/− there was downregulation of Gli1 and Ptc specifically in epithelial cells and in Gli3−/− there was upregulation of Gli1 and Ptc in epithelial and mesenchymal cells. It is difficult to explain these changes in the context of normal tooth development. Clearly increased local activity of the Shh pathway has little discernible effect on tooth development in contrast to direct increase in Shh ligand, which affects epithelial cell invagination. This suggests extensive feedback pathways operating to control the cell proliferative action of Shh. Thus, in Gli2 mutant epithelium, activity of Gli3 may compensate for the loss of Gli1 and Ptc expression in some way. Alternatively, a second recently identified Ptc gene, a potential Shh receptor, Ptch2, may also have some role here (Motoyama et al., 1998). Since Ptch2 is only expressed in the epithelial component of the tooth germ (Motoyama et al., 1998), it could be compensating for the loss of epithelial Ptc expression in the Gli2 mutants; this could explain why the loss of Ptc does not have any drastic effects on Gli2 mutant tooth development. In Gli3 mutants, Gli2 may exert a control role in the pathway limiting the effects of increased expression of Gli1 and Ptc.
Acknowledgement
This work was supported by HFSPO, MRC and the British Heart Foundation (PTS) and by a grant to C.-c.H. from the National Cancer Institute of Canada, with funds from the Terry Fox Run. We thank Andy McMahon for the Shh cDNA, Matthew Scott for Ptc cDNA, Alex Joyner for Gli1 cDNA, Genentech for Smo cDNA, Rudi Balling for Pax-9 cDNA, Brigid Hogan for Bmp-4 cDNA and Ontogeny for Shh protein. We would also like to thank Joy Preece and Stalin Kariyawasam for technical support and Bethan Thomas for her comments on the histology.