Transformation of Tooth Type Induced by Inhibition of BMP Signaling

See allHide authors and affiliations

Science  06 Nov 1998:
Vol. 282, Issue 5391, pp. 1136-1138
DOI: 10.1126/science.282.5391.1136


Mammalian dentitions are highly patterned, with different types of teeth positioned in different regions of the jaws. BMP4 is an early oral epithelial protein signal that directs odontogenic gene expression in mesenchyme cells of the developing mandibular arch. BMP4 was shown to inhibit expression of the homeobox gene Barx-1 and to restrict expression to the proximal, presumptive molar mesenchyme of mouse embryos at embryonic day 10. The inhibition of BMP signaling early in mandible development by the action of exogenous Noggin protein resulted in ectopic Barx-1 expression in the distal, presumptive incisor mesenchyme and a transformation of tooth identity from incisor to molar.

BMP4 that is produced from the early oral epithelium is able to activate expression of the homeobox gene (Msx-1) in the underlying ectomesenchyme cells of the mandibular arch, which is initially expressed in the distal ectomesenchyme that marks presumptive incisor cells (1). The proximal extent of Msx-1 expression coincides with the distal extent of another homeobox gene (Barx-1),which is restricted to the proximal mesenchyme that marks presumptive molar cells (Fig. 1A) (2). The boundary between cells that express Msx-1 andBarx-1 thus demarcates the distinction between presumptive incisor (Msx-1 positive:Barx-1 negative) and molar (Msx-1 negative:Barx-1 positive) tooth-forming regions in mouse embryos.

Figure 1

Regulation of Barx-1expression. (A) Schematic representation of the expression domains of Barx-1, Msx-1, Bmp-4,and Fgf-8. (B through G) Barx-1 in situ hybridization. E10 (B and C) and E9.5 (D and E) mandibular arch explants that were cultured for 24 hours are shown. (B) Induction of Barx-1 (epithelium removed; Fgf8 bead present). (C) Repression of Barx-1 (epithelium intact; BMP4 bead present). (D) Induction occurring only around the proximally placed bead (epithelium intact; Fgf8 beads present). (E) Induction in the distal region (epithelium removed; Fgf8 beads present). The mandibular arch in (F) Msx-1+/– and (G) Msx-1–/–embryos is shown by the frontal section of an E10.5 head. The expression ofBarx-1 is unchanged. Scale bars, 500 μm.

The boundary between Msx-1 and Barx-1 expression was found to correspond to the junction between Fgf-8 andBmp-4 expression in the overlying oral epithelium (schematically represented in Fig. 1A) (3, 4). The ability of these signaling proteins to regulate the spatial expression of Barx-1 was investigated by implanting protein-soaked beads into mandibular arch explants (5). At embryonic day 10 (E10), the removal of the epithelium resulted in a dramatic down-regulation of Barx-1 expression (6). Fgf8 beads that were placed in mesenchymal explants from which the epithelium had been removed resulted in a reexpression of Barx-1 (Fig. 1B). In contrast, when BMP4 beads were implanted into intact mandible explants, BMP4 was found to have a marked inhibitory effect on the endogenousBarx-1 expression (Fig. 1C). The ability of endogenous BMP4 to inhibit Barx-1 expression was investigated by implanting Fgf8 beads into the distal (Barx-1 negative) mesenchyme of intact explants at E9.5. In the presence of endogenous BMP4, exogenous Fgf8 was unable to induce ectopic Barx-1 expression (Fig. 1D). However, if the endogenous source of BMP4 (the epithelium) was removed, Fgf8 beads were able to induce Barx-1expression in the medial region at E9.5 (Fig. 1E). At E10.5, Fgf8 beads were no longer able to induce Barx-1expression in this presumptive incisor region, indicating a change in competence of the distal neural crest-derived ectomesenchyme cells (6). In order to determine whether the BMP4 inhibition of Barx-1 was direct or through the activation ofMsx-1, which might repress Barx-1 expression, the expression of Barx-1 was investigated inMsx-1 –/– embryos at E10 (7). No change in the medial-lateral boundary ofBarx-1 was detectable in theMsx-1 –/– mandibular mesenchyme, which indicates that the inhibitory action of BMP4 on Barx-1 does not occur through the activation of Msx-1 (Fig. 1, F and G).

To investigate the outcome of inhibiting BMP signaling from the oral epithelium to the mandibular arch mesenchyme, we implanted Noggin beads into the distal, presumptive incisor mesenchyme of mandibular arch explants at E10 (5). Noggin protein binds BMP4 with high affinity and can abolish BMP4 activity by blocking binding to cell surface receptors (8). After a culture period of 24 hours,Barx-1 expression was considerably extended distally into the presumptive incisor domain and, as shown previously,Msx-1 expression was lost (Fig. 2, A through C) (3). The consequence of ectopic expression of Barx-1, a molar marker, in presumptive incisor regions was investigated by examining the effect that Barx-1 had on incisor marker Tlx-1.Tlx-1 (Hox-11) is expressed specifically in the mandibular arch incisor tooth epithelium at E10 and is not expressed in the molar tooth epithelium (Fig. 2D) (9). Noggin beads that were implanted into the distal mesenchyme of E10 mandibular arch explants produced a marked repression of Tlx-1 expression (Fig. 2, G and H). BMP4 beads were not able to induce Tlx-1expression in the epithelium, which indicates that the Noggin-induced loss of expression in the presumptive incisor epithelium is not due to the loss of a direct inducer (BMP4) of expression (Fig. 2, I through K).

Figure 2

Inhibition of Bmp-4 signaling by Noggin beads. E10 mandibular arch explants that were cultured for 48 hours are shown in (A) through (C) and in (E) through (H). (A) DIG in situ hybridization showing the extension of the domain of Barx-1 into the distal regions of the arch after the addition of Noggin beads (blue) and BSA beads (white). (B and C) Radioactive in situ hybridization showing serial sections through an arch that was cultured with a Noggin bead. (B) The expression of Barx-1 extends toward the distal region, in comparison with the control side with no bead. (C) The expression of Msx-1 is inhibited in the distal region, in comparison with the control side with no bead. (D throughK) β-galactosidase–stainedTlx-1 +/− embryos. Expression in the presumptive incisor epithelium and the mesoderm core of the first and second branchial arches is shown by the E10.5 head (D). The mandibular arch explant and BSA control beads are shown (E). In (F), early tooth buds expressing Tlx-1 are shown in a section through the incisor region of the explant at the level of F′ (dashed white line) in (E). A loss of Tlx-1 expression is shown by the mandibular arch explant and Noggin beads (G). In (H), a section through the incisor region of the explant at H′ in (G) is shown. The E10 mandibular arch explant that was cultured for 24 hours with BMP4 beads is shown in (I) through (K). BMP4 beads cannot induce expression of Tlx-1 in the epithelium (I). Sections through the explant at the level of J′ and K′ (white dashed lines) show endogenous expression in the epithelium (J) but show no new expression in the epithelium near to the bead (K). Arrows indicate the position of beads. Scale bars, 500 μm.

To determine whether these Noggin-induced changes in gene expression in the distal part of the mandibular arch had any affect on incisor tooth development, we dissected mandibular explants [which had been cultured with Noggin or bovine serum albumin (BSA) beads] into molar and incisor regions and transplanted these explants into renal capsules (10). This technique allows for the full morphological development of teeth after E10 (11). Normal molar and incisor teeth developed from control explants (Fig. 3, A, D, E, and H), but in Noggin-treated explants at E10, multicuspid molar teeth developed from incisor explants at the expense of incisor teeth (Table 1 and Fig. 3, B, C, F, and G). In these treated explants, the expected number of molar teeth developed from the molar explants, indicating that molars had indeed formed from incisor explants. The same experiment (when repeated with E11 explants) showed no effect on tooth identity (Table 1), indicating that the ability of Noggin beads to affect tooth patterning is limited to a small temporal window of development. This temporal window was further confirmed by showing that, at E11, Noggin had no effect on Barx-1 expression but still inhibitedMsx-1 expression (Fig. 3, I and J). All the molars obtained from the incisor regions were morphologically and histologically normal, except for one that had a molar-incisor hybrid appearance (Fig. 3, C and G). Macrodontia and microdontia are relatively common human craniofacial abnormalities, and macrodontia of premolars has frequently been described as molarized premolars. True transformations of incisors to molars have rarely occurred in humans (12).

Figure 3

Transformation of tooth identity from incisor to molar shape. (A throughH) Teeth obtained from E10 tooth germs that were cultured for 2 days in vitro and implanted under host kidney capsules. A section of the multicuspid tooth shown in (E) that was formed from a molar tooth germ cultured with a BSA bead is shown in (A). Multicuspid teeth that were formed from incisor tooth germs cultured with Noggin beads are shown in (B), (C), (F), and (G). (F) A tooth that has a distinctive broad molar shape and three cusps; a section of (F) is shown in (B). (G) A tooth that has a narrow, incisor shape but has started to develop multiple cusps that are more appropriate to molar development; a section of (G) is shown in (C). This tooth is therefore classified as a molar-incisor hybrid. (H) A conical-shaped tooth that was formed from an incisor tooth germ cultured with a BSA bead; a section of (H) is shown in (D). (I and J) E11 mandibular arch explants that were cultured for 2 days with Noggin beads and were subjected to DIG whole-mount in situ hybridization. The proximal-distal boundary of expression of Barx-1 is not effected by Noggin beads at this stage (I). At E11 onward, the expression of Msx-1 expands proximally to include the condensing mesenchyme immediately underneath the developing incisor and molar tooth germs (2) (J). At this stage, Noggin beads are still able to inhibit expression, as can be seen by the loss of Msx-1 under the incisor bud that is nearest to the beads (arrow). Scale bars, 500 μm.

Table 1

Transformation of incisors to molars. Incisors do not develop as well as molar tooth germs in kidney capsules, despite being cultured as pairs, with two out of three (70%) forming cysts at E10 and E11. After E11, incisor tooth germs develop well in kidney capsules. Some secondary molars also developed alongside primary molars in the Noggin/BSA presumptive molar regions.

View this table:

The distinct expression boundaries of homeobox genes in the mandibular ectomesenchyme that are found before the tooth-germ initiation resulted in the suggestion that these domains provide the positional information for tooth patterning (13). This suggestion was supported by mutations in the Dlx-1 andDlx-2 genes, which resulted in a restricted loss ofBarx-1 expression and the specific arrest of maxillary molar development (14). We found that ectopic expression ofBarx-1 in the distal, presumptive incisor tooth ectomesenchyme results in a change in morphogenesis such that molar teeth develop in place of incisors. The observation that this change of tooth type is not always complete, with molar-incisor hybrid teeth also being formed occasionally, suggests that a transformation of incisor tooth germs to molar tooth germs has taken place. Although the observed phenotype is entirely consistent with the proposed role ofBarx-1 in molar development, we cannot exclude the possibility that ectopic expression of Barx-1 alone is not sufficient for transformation and may need to be accompanied by other molecular changes to the ectomesenchyme cells (such as a loss ofMsx-1 expression), which also occur after the inhibition of BMP signaling. A loss of Msx-1 alone, however, is not sufficient to produce a transformation of tooth type because a loss ofMsx-1 expression at E11, which is not accompanied by a change in Barx-1 expression, does not result in a transformation of tooth identity (Fig. 3, I and J). A loss of the function of genes that are involved in tooth patterning (such asDlx1/2 and Activin βA knockout mice) results in the early arrest of tooth development rather than in a transformation of tooth type, which appears to require misexpression (11, 14).

At E9.5, Fgf8 beads can induce ectopic Barx-1 expression in distal regions of the mandible, indicating that all neural crest–derived ectomesenchymal cells of the mandibular arch are equally responsive to epithelial signals. This implies that the neural crest cells that populate the mandibular arch are not prespecified but are patterned by contact with epithelial signals. The evidence for a prepatterning of cranial neural crest cells is limited to the proximal cells of the mandibular arch; our results are consistent with data showing that distal mandibular arch cells have a different axial origin than proximal cells (15).

The antagonistic effects of Fgf8 and Bmp4, which act to establish the distal boundary of Barx-1 expression, are similar to those reported for Pax-9 (16).Barx-1 is, however, expressed earlier than Pax-9and, unlike Pax-9, is not induced at localized sites that underlie all developing teeth. This antagonistic signaling interaction thus has at least two roles in the specification of mandibular mesenchymal fates: an early role in proximal-distal axis specification, which determines whether mesenchymal cells have presumptive molar or incisor fates, and a later role in determining the sites of tooth bud initiation.

  • * To whom correspondence should be addressed. E-mail: psharpe{at}


Stay Connected to Science

Navigate This Article