Jaw muscularization requires Dlx expression by cranial neural crest cells

Histo-anatomical analysis of embryonic day 18.5 (E18.5) EdnRA^−/− and Dlx5/6^−/− mutant heads shows that the skeletal malformations are associated with severe muscle defects (Fig. 1, Fig S2, and Table S1). In Dlx5/6^−/− double-mutant mice, all PA1-derived masticatory muscles and PA2-derived muscles are absent and replaced by undifferentiated reticular connective tissue (Fig. 1 C and C′ and Fig. S2G). This observation contrasts sharply with the persistence of transformed jaw skeletal elements that are therefore devoid of any muscular connection. Oculo-motor muscles are present, but their morphology is altered in parallel with eye malformations (Fig. 1C). More posterior muscles are also present, but their morphology is altered and individual muscle masses are not identifiable (Fig. 1 D and F).

In EdnRA^−/− mutants, masticatory muscles are present, but their morphology is perturbed (Fig. 1 B, B′, and E; Fig. S2D). However, they can often be unequivocally identified through their attachment to specific skeletal elements. In wild-type fetuses, the superficial and lateral masseter is inserted in the lateral aspect of the mandible (Fig. 1D), whereas the deep masseter passes through the infraorbital foramen to insert into the medial aspect of the mandible. In EdnRA^−/−, the situation is different; all components of the masseter pass through the infraorbital foramen of the transformed lower jaw (see black arrowhead Fig. 1B) and have a large insertion in the medial aspect of this bone. In EdnRA^−/−, the pterygoid and temporal muscles are indistinguishable and form a single muscular mass (Fig. 1E). PA2-derived muscles are reduced whereas oculo-motor muscles are normal.

In both EdnRA^−/− and Dlx5/6^−/− mice, intrinsic muscles of the tongue and sublingual muscles are severely affected: the genioglossus and the geniohyoid are absent, and other intrinsic muscles of the tongue are reduced and disorganized, thus making the tongue a vestigial medially located structure (Fig. 1 B, B′′, C, C′′, E′, and F′ and Fig. S2 B, E, and H). The mylohyoid and digastric muscles are absent and are replaced by reticular connective tissue with only a few identifiable muscles fibers (Fig. 1 B′′′ and C′′′ and Fig. S2 C, F, and I). To understand the ontogeny of these muscular defects, we then analyzed the myogenic developmental process in EdnRA^−/− and Dlx5/6^−/− mutant embryos.

Craniofacial muscle formation initiates with the comigration of CNCCs and the myogenic mesoderm toward PA1 and PA2 (26). The initial stage of specification is characterized by the activation of Tbx1 and Capsulin specifically in head myogenic lineages (15, 27). In contrast, tongue muscles are generated from the rostral migration of myoblasts deriving from the first five somites. The first stages of craniofacial myogenesis are not affected in Dlx5/6^−/− embryos. Although the shape of Dlx5/6^−/− PA1 is already slightly altered at E9.5, the migration and initial specification of head myogenic mesoderm is taking place, as shown by the mesodermal activation of Tbx1 and Capsulin at E9.5 in PA1 and PA2 (Fig. 2 A–D). Furthermore, migration of somitic tongue muscle precursors along the hypoglossal cord is not altered, as shown by the expression pattern of Pax3 at E10.5 (Fig. 2 E–F).

The initial stages of craniofacial muscle determination are characterized by expression of Myf5 (28). To follow the onset of myogenic determination, we used Myf5^nLacZ/+ reporter embryos (29). At E10.5, Myf5-LacZ is expressed in the extraocular premuscle mass (peom) and in the mandibular arch myoblasts, which later develop into masticatory premuscle masses (pmm) at E11.5–E12.5 (Fig. 3 A, C, and E). In E10.5–E11.5 Dlx5/6^−/−;Myf5^nLacZ/+ embryos, Myf5-LacZ expression is strongly reduced in the ocular and masticatory region (Fig. 3 B and D) and is barely detectable at E12.5 (Fig. 3F), indicating a defect in myogenic determination.

The myogenic determination factor MyoD is under the control of both the Myf5 and Tbx1 (28). In E11.5 wild-type embryos, MyoD is expressed in the extraocular and masticatory premuscle masses; this territory of MyoD expression is preserved, but reduced in both EdnRA^−/− and Dlx5/6^−/− embryos (Fig. 4 A–C).

At later stages, in wild-type fetuses, the masticatory premuscle mass at E12.5 and the masseter muscle at E14.5 and E18.5 are MyoD (Fig. 4 D and G), Desmin (Figs. S3 A and D and S4A), and fast myosin heavy chain (fMHC) (Fig. S4A′) positive, indicating that myogenic differentiation is taking place. In EdnRA^−/− mutants, the presumptive masseter is present and positive for the above-mentioned myogenic markers, but is clearly reduced at E12.5 and E14.5 (Fig. 4 E and H and Fig. S3 B and E) and shows a different morphology and orientation at E18.5 (Fig. S4 B and B′). In the masseter region of Dlx5/6^−/− mutants, only a few cells express MyoD, Desmin, and fMHC; the few positive myoblasts do not fuse into myotubes, resulting in the absence of masticatory muscle masses (Fig. 4 F and I and Figs. S3 C and F and S4 C and C′). In EdnRA^−/− and in Dlx5/6^−/− mutants, we find expression of muscle determination and differentiation markers in the vestigial tongue from E12.5 to E18.5 (Fig. 4 J–O and Figs. S3 G–L and S4 D–F and D′–F′), the genioglossus muscle is never present, and sublingual muscles are barely differentiated (Fig. 4 K, L, N, and O and Fig. S3 H, I, K, and L).

The presence of masticatory muscles in EdnRA^−/− demonstrates that they can form even in the presence of the lower-to-upper jaw transformation that occurs in both EdnRA^−/− and Dlx5/6^−/− mutants. An explanation for the muscle development in EdnRA^−/− embryos can be found by analyzing the Edn1-independent expression of Dlx5/6. Inactivation of EdnRA completely prevents the activation of Dlx5 and Dlx6 in E9.5 PA1 during CNCC specification. However, in the same mutant at E10.5, Dlx5/6 are expressed in an Edn1-independent territory in the proximal part of PA1 (25). Similarly, in E11.5 EdnRA^−/− mutants, distal expression of Dlx5 and Dlx6 is lost in the arches, but is maintained in the proximal part of the mandibular and hyoid arches and in the maxillary bud (Fig. 5). Reflecting the mirror transformation of the lower into an upper jaw, at E11.5 the expression of Dlx5 and Dlx6 also displays mirror-like symmetry (Fig. 5 B and D). Remarkably, in EdnRA^−/− mutants, the myogenic territory at the origin of masticatory muscles [defined by the expression of Myf5 (Fig. 3C) and MyoD (Fig. 4A)] is included within the Edn1-independent Dlx5/6-positive region, suggesting that the presence of Dlx5 and Dlx6 is at the origin of the myogenic differentiation observed in these mutants. Similarly, PA2-derived muscles present milder defects in EdnRA^−/− than in Dlx5/6^−/− mutants (Table S1). This might reflect the presence of an Edn1-independent Dlx5/6-positive region also within PA2 (Fig. 5). The masticatory muscular defects observed in EdnRA^−/− and in Dlx5/6^−/− mutants are summarized in Fig. 6.

In summary, our data show that Dlx5 and Dlx6 expression by CNCCs is necessary to maintain the myogenic program in the craniofacial mesoderm. In the absence of Dlx5/6, the myoblasts cannot initiate their differentiation. We reasoned that a possible mechanism could involve the derepression of BMP and Wnt signaling, which are known to prevent craniofacial myogenesis (17). Indeed, in a systematic quantitative PCR comparative analysis of microdissected E10.5 wild type and Dlx5/6^−/− PA1s, we have shown that BMP7 and Wnt5a expression is increased two- and threefold, respectively, in the double-mutant mouse (30).

EdnRA and Dlx5;Dlx6 mutant mice were genotyped by PCR using allele-specific primers as reported (20, 36). Embryos were fixed in PBS (Sigma) 4% paraformaldehyde (E9.5–E12.5) for 1 d or in Bouin's fixative solution (75% saturated picric acid, 20% formol 40%, 5% acetic acid) for 5 d (E14.5 and E18.5). E18.5 fetuses were decalcified for 2 d with Jenkin's solution (50 mL, 2 d, three changes, 4% hydrochloric acid 37%, 3% acetic acid, 10% chloroform, 10% dH2O, 73% absolute ethanol). Fixed embryos and fetuses were embedded in paraffin and sectioned at 8–12 μm. Serial sections of E18.5 fetuses were subjected to Mallory trichromic staining (22) for anatomical analysis.

Dlx5/6^+/−;Myf5^nLacZ/+ heterozygous parents, obtained by crossing Myf5^nLacZ/+ transgenic males (29) with heterozygous Dlx5/6^+/− females, were crossed with Dlx5/6^+/− mice to obtain Myf5^nLacZ/+, Dlx5/6^+/−;Myf5^nLacZ/+, and Dlx5/6^−/−;Myf5^nLacZ/+ embryos, which were then fixed in PBS 4% PFA and stained to reveal LacZ expression.

Immunohistochemistry was performed on deparaffinized sections using standard protocols of the Dako Envision kit and the Dako ARK kit. Polyclonal rabbit primary antibodies (anti-MyoD; Santa Cruz C-20:sc-304, 1/1200) and anti-Desmin (Abcam AB15200, 1/800) or mouse monoclonal primary antibody (anti-fMHC; Sigma MY-32, 1/100) were diluted in PBS, 1% BSA (Sigma).

Whole-mount in situ hybridization was performed with digoxigenin-labeled RNA probes corresponding to the antisense sequence of murine Tbx1, Capsulin, Pax3, MyoD, Dlx5, and Dlx6 (all previously reported in refs. 29, 37, and 38), on E9.5–E11.5 embryos adapting the procedure described by Tajbakhsh (29).