Activin and Bmp4 Signaling Converge on Wnt Activation during Odontogenesis

Abstract

Previous studies show that both activin and Bmp4 act as crucial mesenchymal odontogenic signals during early tooth development. Remarkably, mice lacking activin-βA (Inhba^−/−) and mice with neural crest–specific inactivation of Bmp4 (Bmp4^ncko/ncko) both exhibit bud-stage developmental arrest of the mandibular molar tooth germs while their maxillary molar tooth germs completed morphogenesis. In this study, we found that, whereas expression of Inhba and Bmp4 in the developing tooth mesenchyme is independent of each other, Bmp4^ncko/nckoInhba^−/− compound mutant mice exhibit early developmental arrest of all tooth germs. Moreover, genetic inactivation of Osr2, a negative regulator of the odontogenic function of the Bmp4-Msx1 signaling pathway, rescues mandibular molar morphogenesis in Inhba^−/− embryos. We recently reported that Osr2 and the Bmp4-Msx1 pathway control the bud-to-cap transition of tooth morphogenesis through antagonistic regulation of expression of secreted Wnt antagonists, including Dkk2 and Sfrp2, in the developing tooth mesenchyme. We show here that expression of Dkk2 messenger RNAs was significantly upregulated and expanded into the tooth bud mesenchyme in Inhba^−/− embryos in comparison with wild-type littermates. Furthermore, in utero treatment with either lithium chloride, an agonist of canonical Wnt signaling, or the DKK inhibitor IIIC3a rescued mandibular molar tooth morphogenesis in Inhba^−/− embryos. Together with our finding that the developing mandibular molar tooth bud mesenchyme expresses significantly higher levels of Dkk2 than the developing maxillary molar tooth mesenchyme, these data indicate that Bmp4 and activin signaling pathways converge on activation of the Wnt signaling pathway to promote tooth morphogenesis through the bud-to-cap transition and that the differential effects of loss of activin or Bmp4 signaling on maxillary and mandibular molar tooth morphogenesis are mainly due to the differential expression of Wnt antagonists, particularly Dkk2, in the maxillary and mandibular tooth mesenchyme.

Introduction

Tooth development is controlled by reciprocal signaling interactions between epithelial and mesenchymal tissues and has long been used as a model for understanding cell-cell signaling interactions that regulate mammalian organogenesis (Thesleff et al. 1995; Jernvall and Thesleff 2000). Tissue explant culture assays have shown that the odontogenic potential initially arises in oral ectoderm, where dental placodes form, and shifts to the prospective tooth mesenchyme at early bud stage, with tooth mesenchyme from bud and later stages capable of inducing tooth organogenesis even when recombined with embryonic nondental epithelium (Kollar and Baird 1970a, 1970b; Ruch et al. 1973; Mina and Kollar 1987; Lumsden 1988; Jernvall and Thesleff 2000). At bud stage, mesenchymal signals induce formation of the primary enamel knot (PEK), a signaling center consisting of nonproliferating epithelial cells in the distal tooth bud region that secretes many growth factors, including members of Bmp, Fgf, and Wnt families and Shh, which drives tooth morphogenesis through bud-to-cap transition and beyond (reviewed by Jernvall and Thesleff 2000; Thesleff 2003; Tucker and Sharpe 2004; Zhang et al. 2005; Lan et al. 2014).

Among signaling molecules expressed in the early dental epithelium and mesenchyme, Bmp4 exhibits an expression pattern that coincides with the shift of odontogenic potential from epithelium to mesenchyme during tooth bud formation (Vainio et al. 1993). Bmp signaling activates expression of Msx1, encoding a homeodomain-containing DNA-binding transcription factor, in developing tooth mesenchyme (Vainio et al. 1993; Chen et al. 1996; Tucker et al. 1998). Mice lacking Msx1 exhibit tooth developmental arrest at bud stage, with significantly reduced Bmp4 messenger RNA (mRNA) expression in the tooth mesenchyme (Satokata and Mass 1994; Chen et al. 1996; Tucker et al. 1998). Remarkably, addition of recombinant Bmp4 protein rescued morphogenesis of Msx1^−/− molar tooth germs to bell stage in explant cultures (Bei et al. 2000). However, in contrast to the developmental arrest of all tooth germs in Msx1^−/− mice, only mandibular molar development was arrested at bud stage in mice lacking Bmp4 expression in the dental mesenchyme (Jia et al. 2013; Jia et al. 2016), suggesting that another mesenchymal factor partly complemented Bmp4 function in maxillary molar and incisor development.

Wnt signaling is another key molecular pathway driving tooth organogenesis. In the absence of Wnt activation, β-catenin is phosphorylated in the cytoplasm by GSK3β and targeted for degradation. Wnt activation of the frizzled and Lrp5/6 co-receptors leads to inhibition of GSK3β, resulting in stabilization and translocation of β-catenin into the nucleus, where it activates target gene expression (MacDonald et al. 2009). Tissue-specific inactivation of β-catenin in either the embryonic oral epithelium or early tooth mesenchyme caused tooth developmental arrest at bud stage (Andl et al. 2002; Liu et al. 2008; Chen et al. 2009). On the other hand, constitutive stabilization of β-catenin in the mouse embryonic oral epithelium caused supernumerary tooth formation (Järvinen et al. 2006; Wang et al. 2009). However, constitutive stabilization of β-catenin could not induce tooth formation in the absence of BMP receptor-Ia in the dental epithelium (O’Connell et al. 2012), indicating that BMP and Wnt signaling acts synergistically to induce tooth organogenesis.

Recent studies show that the Bmp4-Msx1 pathway regulates expression of secreted Wnt antagonists during early tooth development. Multiple secreted Wnt inhibitors, including Dickkopf family members Dkk2 and Dkk3, as well as secreted frizzled-related proteins Sfrp1, Sfrp2, Sfrp3, are expressed in developing tooth mesenchyme (Sarkar and Sharpe 1999; Fjeld et al. 2005; Jia et al. 2013; Jia et al. 2016). Inactivation of Bmp4 in the neural crest lineage (Bmp4^ncko) caused increased expression of Dkk2 mRNAs in the developing tooth mesenchyme (Jia et al. 2013). Remarkably, in utero treatment with a small-molecule inhibitor of DKKs, IIIC3a (Li et al. 2012), rescued mandibular molar morphogenesis in Bmp4^ncko/ncko embryos (Jia et al. 2016). Further studies showed that expression of Dkk2 and Sfrp2 in the developing tooth mesenchyme depended on the function of Osr2, a zinc finger transcription factor that interacts with Msx1 and antagonizes its function in tooth development (Zhang et al. 2009; Zhou et al. 2011; Jia et al. 2016). Whereas Msx1^−/− embryos exhibit increased expression of Dkk2, Sfrp1, and Sfrp2 mRNAs in the tooth bud mesenchyme, genetic inactivation of Sfrp2 and Sfrp3 in combination with IIIC3a treatment partly rescued molar tooth morphogenesis in Msx1^−/− embryos (Jia et al. 2016).

In addition to Bmp4, activin-A has been identified as an important mesenchymal odontogenic signal. Mice lacking activin-βA, encoded by the Inhba gene, exhibit mandibular molar arrest at bud stage, but their maxillary molar teeth develop normally (Ferguson et al. 1998). Expression of follistatin (Fst) and Irx1, 2 downstream target genes of activin signaling, were significantly downregulated in both maxillary and mandibular molar tooth germs in Inhba^−/− embryos, which led to the hypothesis that maxillary and mandibular molar tooth morphogenesis involves distinct genetic pathways (Ferguson et al. 1998; Ferguson et al. 2001). In this study, we show that Bmp4^ncko/nckoInhba^−/− compound mutant mice exhibit early bud developmental arrest of all tooth germs, whereas genetic inactivation of Osr2 rescues mandibular molar morphogenesis in Inhba^−/− embryos. Moreover, we demonstrate that Inhba^−/− embryos exhibit significantly increased expression of Dkk2 in the tooth bud mesenchyme and that in utero treatment with IIIC3a also rescues mandibular molar morphogenesis in Inhba^−/− embryos. These data indicate that the activin and Bmp4 signaling pathways converge on the regulation of Wnt pathway activation in both maxillary and mandibular molar tooth germs to control early tooth morphogenesis.

Materials and Methods

Animals

Bmp4^f/f, Inhba^+/−, Osr2^+/−, and Wnt1Cre transgenic mice and generation of Bmp4^f/f;Wnt1Cre (Bmp4^ncko/ncko) compound mutant mice were previously described (Matzuk et al. 1995; Danielian et al. 1998; Lan et al. 2004; Jia et al. 2013). These mice were maintained in a CD1 mixed background. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the Cincinnati Children’s Hospital Medical Center. This study conformed with ARRIVE guidelines for preclinical animal studies.

Laser Capture Microdissection and Quantitative Real-Time Reverse Transcription Polymerase Chain Reaction

Embryo heads microdissected at E13.5 were frozen in Tissue-Tek OCT compound. Specimens were cryosectioned, and molar tooth mesenchyme tissues were isolated by laser capture microdissection (LCM) (Appendix Fig. 1) as described previously (Jia et al. 2016). Tissues were pooled from 3 embryos of the same genotype, and total RNAs were extracted by using the RNeasy Micro Kit (Qiagen). Complementary DNAs were analyzed using real-time polymerase chain reaction (PCR) using the C1000 Touch Thermal Cycler (Bio-Rad) and Advanced SYBR Green Supermix (Bio-Rad). PCR experiments were carried out in duplicate. Student’s t test was used to analyze pairwise differential expression, and P values <0.05 were considered statistically significant.

Histology, In Situ Hybridization, and Immunofluorescence Assays

Embryos were collected from timed pregnant females, fixed in 4% paraformaldehyde, dehydrated through graded ethanol series, embedded in paraffin, and sectioned at a 7-µm thickness. For histological analysis, sections were stained with hematoxylin and eosin as described previously (Chen et al. 2009). Section in situ hybridization was performed as described in Zhang et al. (1999 ). Cryosectioned embryo heads were used for immunofluorescence staining using a rabbit monoclonal antibody for pSmad1/5/9 (D5B10; Cell Signaling Technology) as previously described (Zhou et al. 2011).

Lithium Chloride and IIIC3a Treatments

Lithium chloride (LiCl) was injected intraperitoneally into pregnant mice (6–12 µmol/g body weight) once every 24 h from gestational days 11 to 14. NaCl solution was used as negative control for LiCl treatment. IIIC3a (Millipore) was dissolved in dimethyl sulfoxide (DMSO) and injected intraperitoneally into pregnant mice (10–25 µg/g body weight) once every 24 h from gestational days 11 to 13 or 14. Control mice were injected with the same concentration of DMSO. Embryos were harvested at E13.5 or E18.5 for in situ hybridization and histological analyses, respectively.

Results

Bmp4 and Inhba Are Independently Expressed in the Developing Tooth Mesenchyme and Act Synergistically to Drive Tooth Morphogenesis

To investigate possible interactions between Bmp4 and activin signaling during tooth development, we first compared Bmp4 expression during tooth development in Inhba^−/− embryos and control littermates. Bmp4 mRNAs were similarly expressed in tooth bud mesenchyme in control and Inhba^−/− embryos (Fig. 1A, B). We also compared the expression of Inhba mRNAs in Bmp4^ncko/ncko embryos and control littermates and did not find any significant difference (Fig. 1C, D). Moreover, immunofluorescence staining for pSmad1/5/9 showed that BMP signaling activity was not altered in Inhba^−/− tooth germs (Appendix Fig. 2). These data indicate that Bmp4 and activin-A do not regulate each other’s expression during early tooth development.

Figure 1.

Comparison of Bmp4 and Inhba messenger RNA (mRNA) expression in the developing tooth bud mesenchyme in control, Inhba^−/−, and Bmp4^ncko/ncko embryos. (A) Quantitative real-time reverse transcription polymerase chain reaction (RT-PCR) analysis of Bmp4 mRNAs in E13.25 maxillary (Max) and mandibular (Man) molar mesenchyme in the control and Inhba^−/− embryos. (B) In situ hybridization analysis of Bmp4 mRNA expression in frontal sections through the first molar tooth germs in E13.25 control and Inhba^−/− embryos. (C) Quantitative real-time RT-PCR analysis of Inhba mRNAs in E13.5 maxillary (Max) and mandibular (Man) molar mesenchyme in control and Bmp4^ncko/ncko embryos. (D) In situ hybridization analysis of Inhba mRNA expression in frontal sections through the first molar tooth germs in E13.5 control and Bmp4^ncko/ncko embryos. Red dashed lines mark the boundary between the tooth bud epithelium and mesenchyme; mandibular region is shown below the maxillary region; lingual side is to the left in panels B and D. Error bars represent standard error of the mean in panels A and C. Panels B and D are representative results of each group (n = 8 for each group).

To gain better understanding of the molecular mechanisms mediating activin signaling regulation of tooth development, we generated and analyzed tooth development in Bmp4/Inhba compound mutants. Whereas Bmp4^ncko/ncko and Bmp4^ncko/+ Inhba^−/− embryos showed continued maxillary molar morphogenesis beyond bud stage (Fig. 2B, C, F, G, J, K), all Bmp4^ncko/ncko Inhba^−/− mutants (n = 8) exhibited bud-stage arrest of maxillary molars (Fig. 2D, H, L) and all incisor tooth germs (Appendix Fig. 3), with their mandibular molar tooth germs arrested at early bud formation (Fig. 2D, H, L). These data indicate that Bmp4 and activin signaling act synergistically to regulate organogenesis of both the maxillary and mandibular teeth.

Figure 2.

Bmp4 and activin-A act synergistically to drive tooth germ morphogenesis. (A–L) Hematoxylin and eosin–stained frontal sections through the first molar tooth germs from E13.5 to E16.5 (n = 8 for each group). (A, E, I) Control embryos show normal bud, cap, and bell stages of tooth development at E13.5, E14.5, and E16.5, respectively. (B, F, J) Bmp4^ncko/ncko embryos show bud-stage arrest of the mandibular molar tooth germs (arrowheads). (C, G, K) Bmp4^ncko/+Inhba^−/− embryos show bud-stage arrest of the mandibular molar tooth germs (arrowheads). (D, H, L) Bmp4^ncko/nckoInhba^−/− compound mutants show bud-stage arrest of the maxillary molar tooth germs and even earlier developmental arrest of the mandibular molar tooth germs (arrowheads).

Genetic Inactivation of Osr2 Rescued Mandibular Molar Tooth Morphogenesis in Inhba^−/− Mice

The Osr2 transcription factor is known to suppress Msx1-mediated propagation of Bmp4 expression in the developing tooth mesenchyme (Zhang et al. 2009; Zhou et al. 2011). We investigated whether inactivating Osr2 could affect mandibular molar development in Inhba^−/− mice. At E16.5, whereas Inhba^−/− embryos exhibit bud-stage arrest of mandibular molar tooth germs compared to wild-type littermates (Fig. 3A, B), mandibular first molar tooth germs in Inhba^−/−Osr2^+/− (n = 6) and Inhba^−/−Osr2^−/− (n = 4) embryos progressed to cap and early bell stages, respectively (Fig. 3C, D). By E18.5, whereas mandibular first molar tooth buds were still developmentally arrested in Inhba^−/− embryos (n = 8) (Fig. 3F), all Inhba^−/−Osr2^+/− (n = 10) and Inhba^−/−Osr2^−/− (n = 6) embryos showed bell-stage maxillary and mandibular first molar tooth germs (Fig. 3G, H). However, the mandibular molar tooth germs in Inhba^−/−Osr2^+/− and Inhba^−/−Osr2^−/−embryos showed abnormal cusp patterning (compare Fig. 3G and H with Fig. 3E). Moreover, expression of Fst and Irx1 mRNAs in the molar tooth epithelium was similarly reduced in E13.5 Inhba^−/− and Inhba^−/−Osr2^−/− embryos compared to control littermates (Appendix Fig. 4). These results indicate that reducing Osr2 function could rescue mandibular molar development through the bud-to-cap transition in Inhba^−/− embryos, but activin signaling has an Osr2-independent role in dental epithelial differentiation and cusp patterning.

Figure 3.

Genetic inactivation of Osr2 rescued mandibular molar morphogenesis in Inhba^−/− embryos. (A–H) Hematoxylin and eosin–stained frontal sections through the maxillary and mandibular first molar tooth germs at E16.5 and E18.5, respectively. At E16.5, wild-type embryos show both maxillary and mandibular first molar tooth germs at early bell stage (A). Inhba^−/− embryos show bud-stage developmental arrest of the mandibular molar (n = 6) (B). In Inhba^−/−Osr2^+/− (n = 6/6) and Inhba^−/−Osr2^−/− (n = 4/4) embryos, the mandibular first molar tooth germs progressed to late cap and early bell stages, respectively (C, D). At E18.5, wild-type embryos show both maxillary and mandibular first molar at bell stage (E). Inhba^−/− embryos show bud-stage developmental arrest of the mandibular first molar tooth germs (n = 8) (F). Both Inhba^−/−Osr2^+/− (n = 10/10) and Inhba^−/−Osr2^−/− (n = 6/6) embryos show mandibular first molar tooth germs at bell stage (G, H). Arrowheads point to the mandibular first molar tooth germs. A white line separates the pictures for maxillary and mandibular molar tooth germs in panels D–H because the 2 pictures are taken from distinct sections of a series from the same embryo. Scale bars, 200 µm.

Expression of Dkk2 Is Significantly Upregulated in the Tooth Bud Mesenchyme in Inhba^−/− Embryos

We recently reported that Dkk2 is expressed at significantly higher levels in the mandibular molar mesenchyme than in the maxillary molar mesenchyme in wild-type embryos and that its expression is significantly upregulated in both maxillary and mandibular molar tooth bud mesenchyme in Bmp4^ncko/ncko embryos (Jia et al. 2013). In addition, we showed that Msx1 and Osr2 have opposing effects on expression of Dkk2, Sfrp1, and Sfrp2 in the developing tooth bud mesenchyme, with expression of each of these 3 Wnt antagonist genes significantly upregulated in the tooth bud mesenchyme in Msx1^−/− embryos (Jia et al. 2016). To investigate whether activin signaling also affects expression of these Wnt antagonists in the developing tooth mesenchyme, we isolated E13.5 maxillary and mandibular molar mesenchyme, respectively, from Inhba^−/− and control (wild-type and heterozygous) littermates using LCM and carried out quantitative real-time reverse transcription (RT)–PCR analyses. We confirmed that expression of Dkk2 mRNAs in the mandibular molar tooth bud mesenchyme is more than 2-fold of that in the maxillary tooth mesenchyme in control embryos (Fig. 4A). Moreover, expression of Dkk2 mRNAs is significantly increased in both maxillary and mandibular tooth bud mesenchyme in Inhba^−/− embryos (Fig. 4A). Expression of Sfrp1 and Sfrp2 mRNAs, respectively, is not significantly different in the maxillary and mandibular molar mesenchyme in control embryos, with expression of Sfrp2 mRNAs increased about 2-fold in the mandibular molar mesenchyme in Inhba^−/− embryos compared to control littermates (Fig. 4A). In situ hybridization analysis showed that Dkk2 mRNAs were preferentially expressed in the oral mesenchyme lingual to developing tooth buds and excluded in the condensing tooth mesenchyme surrounding the distal tooth bud epithelium in control embryos (Fig. 4B). Expression of Dkk2 mRNAs expanded into the distal tooth mesenchyme in both maxillary and mandibular tooth germs in Inhba^−/− embryos (Fig. 4C).

Figure 4.

Comparison of Dkk2, Sfrp1, and Sfrp2 messenger RNA (mRNA) expression in the developing tooth mesenchyme in E13.5 Inhba^−/− mutants and their control littermates. (A) Quantitative real-time reverse transcription polymerase chain reaction (RT-PCR) analysis of Dkk2, Sfrp1, and Sfrp2 mRNAs in the maxillary (Max) and mandibular (Man) molar mesenchyme. Error bars represent standard error of the mean. **P < 0.05. (B, C) Frontal sections showing expression patterns of Dkk2 mRNAs in and around the maxillary (upper) and mandibular (lower) molar tooth mesenchyme in the control (B) and Inhba^−/− (C) embryos (n = 6). Red dashed lines mark the boundary between the tooth bud epithelium and mesenchyme; lingual side is to the left; arrowheads point to the first molar tooth germs in panels B and C.

Pharmacological Activation of Canonical Wnt Signaling Pathway or Inhibition of DKKs Rescued Mandibular Molar Development in Inhba^−/− Mice

To investigate the possibility that mandibular molar developmental arrest in Inhba^−/− embryos is due to inhibition of Wnt signaling, we tested whether activation of canonical Wnt signaling could rescue mandibular molar tooth morphogenesis in Inhba^−/− mice. We treated pregnant mice bearing Inhba^−/− embryos with intraperitoneal injection of LiCl, a simple salt that inhibits the kinase activity of Gsk3β and thus increases intracellular β-catenin (Klein and Melton 1996; Song et al. 2009; Zeilbeck et al. 2014), from gestational days 11 to 14. As control, NaCl treatment did not significantly affect development of the embryonic tooth germs in control and Inhba^−/− embryos (Appendix Fig. 5A, B). Whereas LiCl treatment did not significantly affect tooth morphogenesis in control embryos (Appendix Fig. 5C), both maxillary and mandibular first molar tooth germs progressed to bell stage by E18.5 in 7 of 15 LiCl-treated Inhba^−/− embryos (n = 14/30 mandibular first molars) (Appendix Fig. 5D).

We then investigated whether inhibiting DKK activity directly could rescue mandibular molar morphogenesis in Inhba^−/− embryos by treating pregnant Inhba^+/− mice with intraperitoneal injection of IIIC3a from gestational days 11 to 14. Remarkably, mandibular molar tooth germs progressed to bell stage by E18.5 in 4 of 8 IIIC3a-treated Inhba^−/− embryos (Fig. 5C, D), with restored expression of Wnt target genes Lef1 and Axin2 in the mandibular molar mesenchyme at E13.5 (Fig. 5E–L). These data indicate that an important role for activin signaling in early tooth development is to regulate expression of Wnt antagonists, particularly Dkk2, in the developing tooth mesenchyme.

Figure 5.

IIIC3a treatment in utero rescued mandibular molar tooth morphogenesis and Wnt signaling activity in Inhba^−/− embryos. (A–L) Frontal sections through the maxillary and mandibular first molar tooth germs in Inhba^+/− control (A, C, E, G, I, K) and Inhba^−/− (B, D, F, H, J, L) embryos treated with DMSO (A, B, E, F, I, J) or IIIC3a (C, D, G, H, K, L). (A–D) Hematoxylin and eosin (H-E) staining at E18.5. n = 6 for Inhba^+/− embryos treated with DMSO. Rescue rate is 7/16 with IIIC3a treatment. A white line separates the pictures for maxillary and mandibular molar tooth germs in panel A because the 2 pictures are taken from distinct sections of a series from the same embryo. (E–L) Messenger RNA (mRNA) expression of Lef1 (E–H) and Axin2 (I–L) (n = 2 embryos for each genotype for each probe). Red dashed lines mark the boundary between the tooth bud epithelium and mesenchyme. Arrowheads point to the mandibular first molar tooth germ in Inhba^−/− mutants; lingual side is to the left; scale bars, 200 µm.

Discussion

Previous studies of Inhba^−/− mice led to the hypothesis that development of maxillary and mandibular molar teeth involves fundamentally distinct molecular mechanisms (Ferguson et al. 1998; 2001). Inhba mRNAs are expressed strongly in the early developing tooth mesenchyme of all tooth germs, and expression of several downstream target genes, including Fst and Irx1, was dramatically reduced in both maxillary and mandibular molar tooth germs in Inhba^−/− embryos (Ferguson et al. 1998, 2001). Further studies ruled out the possibility that continued maxillary molar morphogenesis in Inhba^−/− embryos was due to functional complementation by another activin-like ligand (Ferguson et al. 2001). However, comparative expression analysis showed normal expression patterns of a large number of important tooth developmental genes in Inhba^−/− embryos (Ferguson et al. 1998, 2001), and the molecular mechanism mediating the differential requirement for activin signaling in maxillary and mandibular molar morphogenesis has remained a mystery.

We recently reported that Bmp4^ncko/ncko mice exhibit bud-stage developmental arrest of mandibular molars while their maxillary molars continued to develop to mineralized teeth (Jia et al. 2013). Through RNA-seq analysis of LCM-isolated E13.5 tooth mesenchyme, we found that the mandibular molar mesenchyme expresses significantly higher levels of Dkk2 mRNAs than the maxillary molar mesenchyme in wild-type embryos and that Dkk2 expression is significantly increased in both maxillary and mandibular molar mesenchyme in Bmp4^ncko/ncko embryos (Jia et al. 2013). Subsequently, we demonstrated that pharmacological activation of canonical Wnt signaling by in utero treatment with LiCl or IIIC3a partly rescued mandibular molar morphogenesis in Bmp4^ncko/ncko mice (Jia et al. 2016). These results indicate that higher levels of Wnt antagonists in the mandibular molar mesenchyme could account for the differential effects of loss of mesenchymal Bmp4 on maxillary and mandibular molar organogenesis. In this study, we found that, although expression of Bmp4 in the developing tooth mesenchyme is unaffected in Inhba^−/− embryos, Dkk2 expression was significantly increased in both maxillary and mandibular molar mesenchyme in Inhba^−/− embryos. Moreover, we found that expression of Inhba in the tooth bud mesenchyme is unaffected in Bmp4^ncko/ncko embryos. These results suggest that activin and Bmp4 signaling regulate Dkk2 expression in the developing tooth mesenchyme independently. However, in contrast to Inhba^−/− and Bmp4^ncko/ncko mutants, which exhibit continued morphogenesis of maxillary molars, all Bmp4^ncko/nckoInhba^−/− embryos exhibit bud-stage arrest of maxillary molar tooth germs and even earlier developmental arrest of mandibular molar tooth germs. These results indicate that Bmp4 and activin signaling act synergistically and partly redundantly to regulate early morphogenesis of both maxillary and mandibular molar tooth germs.

Bmp4 is known to act in a positive feedback loop with Msx1 (Chen et al. 1996; Bei et al. 2000). We recently showed that the Msx1 and Osr2 transcription factors interact and act antagonistically to regulate Dkk2 and Sfrp2 expression (Zhou et al. 2011; Jia et al. 2016). In this study, we found that genetic inactivation of 1 Osr2 allele is sufficient to rescue mandibular molar morphogenesis through the bud-to-cap transition in Inhba^−/− embryos, and mandibular molar morphogenesis progressed to bell stage by E18.5 in Inhba^−/−Osr2^−/− compound mutants. Since Osr2^−/− embryos exhibit increased expression of Bmp4 in the developing tooth mesenchyme (Zhang et al. 2009), one possibility is that the increase in Bmp4 expression is responsible for driving mandibular molar morphogenesis through the bud-to-cap transition in Inhba^−/−Osr2^+/− and Inhba^−/−Osr2^−/− embryos. Alternatively, Osr2 might directly interact with components of the activin signaling pathway to regulate Dkk2 expression. Further studies are necessary to elucidate the molecular mechanisms acting downstream of activin signaling to regulate Dkk2 expression during tooth development.

Our findings that in utero treatment with the DKK inhibitor IIIC3a rescued mandibular molar morphogenesis in both Inhba^−/− and Bmp4^ncko/ncko mice (Jia et al. 2016 and this study), indicate that the differences in maxillary and mandibular molar tooth phenotypes in these mutant mice, respectively, are due, at least in part, to intrinsic differences in the levels of Wnt antagonists, particularly Dkk2, expressed in the developing tooth mesenchyme. Whereas the molecular mechanism underlying the differential levels of Dkk2 expression in the maxillary and mandibular molar mesenchyme remains to be resolved, our results do not support the hypothesis that development of maxillary and mandibular molar teeth involves fundamentally distinct genetic pathways. Rather, the odontogenic activity of either activin or Bmp4 is sufficient to overcome the inhibitory effects of relatively low levels of Wnt antagonists in the maxillary molar mesenchyme, but both pathways are required to overcome high levels of Wnt inhibition in the mandibular molar tooth germs to drive their morphogenesis through bud-to-cap transition. Although we did not directly measure Wnt signaling activity in the tooth germs in Bmp4^ncko/nckoInhba^−/− embryos, it is most likely that combined loss of both activin and Bmp4 signaling causes more dramatic inhibition of Wnt signaling in all tooth germs than that in the single mutants. Indeed, the early tooth developmental arrest phenotype of the Bmp4^ncko/nckoInhba^−/− embryos is remarkably similar to those of mouse embryos with either overexpression of Dkk1 or tissue-specific inactivation of β-catenin in the early oral epithelium (Andl et al. 2002; Liu et al. 2008). Together, these results indicate that activin-A and Bmp4 act partly redundantly and converge on the regulation of expression of secreted Wnt antagonists during early tooth development. Furthermore, rescue of mandibular molar morphogenesis in both Inhba^−/− and Bmp4^ncko/ncko mice by IIIC3a treatment suggests that small-molecule inhibitors of Wnt antagonists could be useful for treatment of congenital tooth agenesis in humans.

Author Contributions

H.-J.E. Kwon, R. Jiang, contributed to design, data acquisition, analysis, and interpretation, drafted and critically revised the manuscript; S. Jia, Y. Lan, contributed to design, data acquisition, analysis, and interpretation, critically revised the manuscript; H. Liu, contributed to data acquisition and analysis, critically revised the manuscript. All authors gave final approval and agree to be accountable for all aspects of the work.