Report

Regulation of Mammalian Tooth Cusp Patterning by Ectodin

See allHide authors and affiliations

Science  23 Sep 2005:
Vol. 309, Issue 5743, pp. 2067-2070
DOI: 10.1126/science.1116848

Abstract

Mammalian tooth crowns have precise functional requirements but cannot be substantially remodeled after eruption. In developing teeth, epithelial signaling centers, the enamel knots, form at future cusp positions and are the first signs of cusp patterns that distinguish species. We report that ectodin, a secreted bone morphogenetic protein (BMP) inhibitor, is expressed as a “negative” image of mouse enamel knots. Furthermore, we show that ectodin-deficient mice have enlarged enamel knots, highly altered cusp patterns, and extra teeth. Unlike in normal teeth, excess BMP accelerates patterning in ectodin-deficient teeth. We propose that ectodin is critical for robust spatial delineation of enamel knots and cusps.

Because cell differentiation and final pattern are directly linked in developing mammalian dentition, it is well suited for dissecting the molecular basis for induction and inhibition. The patterning of teeth involves iterative activation of the non-proliferative epithelial enamel knots at the places of future cusps (1, 2). The surrounding epithelium and the underlying mesenchyme continue to proliferate, which results in folding of the epithelium to produce cusps (1, 3). Spatial arrangements of cusps are typically species-specific and linked to evolution of diverse diets. Because erupted tooth crowns are fully mineralized, events during embryological patterning are critical for correct functional relations between occluding teeth. Consequently, classic studies have shown detailed coordination of development between occluding cusps (4), and that developing teeth are capable of self-regulation after external perturbations (3, 57). This combination of precision and robustness of tooth development presumably extends to the placement of the secondary enamel knots as they are the earliest developmental signs of species-specific cusp patterns (8). Mathematical modeling has shown that, like feather patterning, an activator-inhibitor loop of signaling molecules could explain how the enamel knots determine the morphological diversity seen in natural teeth (9). However, although not experimentally identified, this model suggested the existence of signaling molecules that inhibit the induction of the enamel knots.

Here we examined the functional role of ectodin (ectodermal inhibitor of BMP; GenBank accession number AB059271) in the regulation of mouse enamel knot formation following two lines of inquiry. First, we recently identified ectodin [the Xenopus ortholog is called Wise (10)] as a previously unknown protein (11) belonging to the Dan/Cerberus family of secreted BMP antagonists (1114). We were interested in the inhibition of BMP signaling during tooth cusp patterning because the earliest differentiation marker of enamel knot cells, a cyclin-dependent kinase inhibitor p21, is induced by BMPs (15). The second reason for our interest in ectodin was that it seemed to be expressed around the primary enamel knot, forming at the onset of tooth crown formation, rather than in it (11). This kind of expression pattern, if it was to be repeated around the secondary enamel knots, would be the first of its kind (16).

Because histological sections of developing lower molars show complex expression patterns of ectodin (Fig. 1A), we analyzed gene expression in three dimensions. The results show that although ectodin is expressed in most parts of the forming first molar, the expression is absent in two distinct locations (Fig. 1B). The anterior (the trigonid) and posterior (the talonid) ectodin-negative regions contain secondary enamel knots, visible as the epithelial areas expressing p21 (Fig. 1B). In mouse molar, the buccal and lingual cusps are joined by a transverse crest, and the areas forming these crests also express p21 but not ectodin (Fig. 1B). In contrast, the area forming a valley separating the anterior and posterior cusp pairs expresses ectodin (Fig. 1B), suggesting inverse roles for ectodin and p21 in cusp patterning. More distally, an ectodin expression domain separates the developing first and second molars, the latter being at the primary enamel knot stage of development (Fig. 1B). As in the first molar (11), the primary enamel knot in the second molar has no ectodin expression (Fig. 1, A and B).

Fig. 1.

Ectodin is expressed around the enamel knots and ectodin deficiency causes expansion of p21 expression domains. (A) Frontal sections of embryonic day 16 mouse lower molars show complex patterns of ectodin expression in the forming tooth crown (black arrowheads). (B) Three-dimensional reconstructions show that ectodin is expressed around the enamel knots that express p21 [occlusal views of the inner enamel epithelium marked with a white line in (A)]. (C) Ectodin-deficient mice show enlarged p21 expression domains with decreased separation between forming cusps and molars. White arrowhead in (B) and (C) mark the secondary enamel knot forming the protoconid cusp, and asterisk in (C) marks an extra molar anterior to the first molar. Anterior side is toward the bottom (B and C) and buccal side is toward the left; M1, first lower molar; M2, second lower molar. Scale bar, 100 μm.

The inverse expression patterns of ectodin and p21 (Fig. 1B) are intriguing because both are induced by BMPs (11, 15). BMPs are expressed in teeth throughout development (17), and mice lacking either type 1a BMP receptor in the epithelium (18) or, as a result of Msx1-null mutation, Bmp4 expression in mesenchyme (19), have their tooth development arrested before primary enamel knot induction. Thus, although these mutants are indicative of the requirement of BMPs for tooth development, they leave the issue of cusp patterning unresolved. Furthermore, mice lacking functional p21 have no reported tooth phenotypes (20), presumably as a result of redundancy with other cyclin-dependent kinase inhibitors. Ectodin, as an antagonist binding to BMPs (11), could be a feedback inhibitor of cusp development by interfering with the induction of p21. We first confirmed this by introducing ectodin-soaked beads together with BMP4-soaked beads on dental epithelium (21). Whereas BMP4 was sufficient to induce ectodin and p21 in isolated dental epithelium, ectodin inhibited BMP4-induced p21 expression (fig. S1).

We next investigated regulation of enamel knots by generating ectodin-deficient mice (21) (fig. S2). Although the overall development of teeth appears fairly normal in ectodin-deficient mice, tooth morphologies are highly altered. Changes in cusp patterns become visible as soon as p21 is up-regulated in the developing secondary knots. The results show enlarged expression domains of p21 with diminished intercuspal regions in the anterior portion of the first molar (Fig. 1C). Analyses of other enamel knot marker genes confirm that the induction of the enamel knot fate took place at the expense of intercuspal tissue (fig. S3). The distal p21 expression domain of the first molar extends without interruption to the area normally giving rise to the second molar with no clear morphological separation between these teeth (Fig. 1, B and C). Accordingly, in erupted tooth rows of ectodin-deficient mice, the first and the second molars were often fused (21/52 = 40%).

An extra molar forms anterior to the first molar in the ectodin-deficient mice (Figs. 1C and 2). This tooth appears to develop slightly faster than the more posterior teeth and, at day 16, the whole crown expresses p21 (Fig. 1C). To explore its developmental origin, we examined earlier stages using sonic hedgehog (Shh) expression as a marker for dental placodes. The results indicate that the extra tooth develops from a separate placode anterior to the first molar (Fig. 2A). Compared to the second molar, which develops as an early posterior extension of the first molar, the separate origin of the extra tooth conforms to the lack of fusion with the first molar. Similarities in placodes expressing Shh also suggest that the extra molar in ectodin-deficient mice is the same as that in mice that overexpress ectodysplasin (22). It remains to be shown whether ectodin and ectodysplasin function in the same or in parallel pathways regulating placode development.

Fig. 2.

Ectodin-deficient mice have extra teeth, fused molars, and highly altered cusp patterns. (A) The molar placodes indicated by Shh expression (open arrowhead) were larger in ectodin-null mutants than those in wild-type littermates, and at embryonic day 14.5, extra placodes (black arrowhead) appear anteriorly to the forming molars. (B) Three-dimensional laser confocal scans revealed that erupted tooth crowns of ectodin-deficient mice are broader than wild-type molars, have well-developed transverse crests, and have fused buccal cusps forming an ectoloph (open arrowhead). Extra cusps fused with the ectoloph, a lingual peg-shaped extra tooth (white arrowhead), and fused first and second molars are also observed. Anterior side is toward the left and buccal side is toward the top (B); M1, first lower molar; M 2, second lower molar; M3, third lower molar; M1 + M2, fused molars. Scale bars, 1 mm.

To study cusp patterns of the ectodin-deficient mice, we used high-resolution three-dimensional laser-confocal imaging (22). Although most aspects of tooth crowns are affected (Fig. 2B), the most pronounced change was in the buccal side of the crowns. Indicating decreased down-growth of valleys separating cusps, the buccal side had better developed connections or completely fused cusps anterior-posteriorly, forming a longitudinal crest with fused extra cusps (Fig. 2B). This phenotype is reminiscent of the ectoloph, a feature present in, for example, the upper molars of black rhinoceros (resemblance of “ectodin” to this classic morphological term is coincidental). This buccal bias is noteworthy because Bmp4 (17), ectodin, and p21 (Fig. 1) are all expressed strongly in the buccal side, and our results may suggest a role for regulation of ectodin in the evolution of lateral bias in teeth. Despite the extensively derived morphology, the ectodin-deficient teeth are functional and individual mice do not seem to have occlusal problems with their molar teeth.

Although the phenotype and enlarged enamel knots of the ectodin-deficient teeth are strongly suggestive about the role of ectodin in inhibiting enamel knot induction, these results do not directly implicate inhibition of BMP signaling. Next, to test the association of ectodin with BMP signaling in tooth crown development, we exposed ectodin-deficient teeth to BMP in culture.

We cultured isolated molars of both ectodin null-mutants and heterozygotes, which have wild-type dentition, with BMP4 in the culture medium for the first 2 days (21). The results show that in the absence of BMP4, the rates of tooth development of both null mutants and heterozygotes are similar (Fig. 3A). Whereas ectodin heterozygous teeth are not particularly sensitive to excess BMP4, the null-mutants show a markedly accelerated crown development, exceeding the rate of normal development in vivo by about 3 days (Fig. 3A). Already after 1 day in culture the null-mutant teeth show cusps and thickening of inner enamel epithelium (Fig. 3A). After 5 days of culture, ameloblasts and odontoblasts are fully differentiated in the null-mutants and the crowns are covered by dentine (Fig. 3B). Compared to both feather and limb patterning, which can be manipulated by BMPs (2326), our present results, together with our earlier observations (27, 28), suggest that normal tooth patterning is relatively robust against excess BMP.

Fig. 3.

Ectodin deficiency increases the sensitivity of developing teeth to excess BMP. Cap stage (day 14) teeth of ectodin heterozygous and null mutant embryos were cultured with BMP4 in the culture medium for the first 2 days. (A) (Left) BMP4 has little effect on tooth development of ectodin heterozygotes, and cusps appear normally during the third day of culture. (Right) In the ectodin-null mutants, which develop both the extra and the first molar, cusps are visible by the second day of culture with BMP4 (black arrowheads). The resulting crowns have sharp cusps, and all crowns were covered by dentine after 5 days of culture. (B) Histological sections of the regions in the black rectangles in (A) show that ectodin-null mutants cultured with BMP4 have fully polarized ameloblasts (white arrowhead) and odontoblasts (black arrowhead) that have already secreted a layer of dentine (asterisk). In heterozygote teeth cultured with BMP4, ameloblasts and odontoblasts are only starting to elongate. Some teeth cultured without BMP4 have begun to secrete dentine on cusp tips (not shown). All teeth (n = 28) were from the same homozygous-heterozygous crossing. Anterior side is toward the left (A). Scale bars, 100 μm.

Taken together, our results indicate that although the morphology of the ectodin-deficient teeth results from shifting the balance of induction toward larger enamel knots (Figs. 1 and 2), the systemic effect of lack of ectodin is the loss of self-regulation (Fig. 3). Thus, excess BMP administered to developing ectodin-deficient teeth caused relatively unchecked induction. In contrast, in normal teeth the cusp induction and inhibition cascades cancel each other out under excess BMP. Uniquely in mammals, once erupted, permanent dentition must last the whole lifetime of the animal. This alone suggests that there has been strong selective pressure to make tooth development as “fail-proof” as possible, and we propose that robustness of tooth development to external perturbations requires ectodin to integrate cusp induction and inhibition.

Although ectodin-deficient mice have the most extensively modified cusp patterns of mouse mutants studied to date, mice with superfluous production of ectodysplasin have some similarities with ectodin-deficient mice. Both have extra teeth and longitudinal crests connecting cusps. However, the ectodin-deficient mice have buccal crests (Fig. 2B), whereas mice overexpressing ectodysplasin have central crests (22). This single difference in morphology is enough to alter the mutant phenotypes to resemble rhinoceros and kangaroo teeth, respectively. These morphological alterations, although functional, are obviously larger than evolutionary transitions documented in the fossil record. Hence, it remains to be determined whether fine tuning of enamel knot formation could be sufficient, perhaps by allelic changes as shown for ectodysplasin in altering armor plates of sticklebacks (29), to account for the evolutionary diversity of teeth.

Supporting Online Material

www.sciencemag.org/cgi/content/full/309/5743/2067/DC1

Materials and Methods

Figs. S1 to S3

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article