The first and second branchial arch mesoderm cells contribute to the jaw muscles – the temporalis, pterygoid, masseter, mylohyoid and anterior digastric, and to the facial expression muscles – the auricularis, buccinator, posterior digastric, frontalis, orbicularis oculi, quadratus labii, zygomaticus and others, respectively (Figure 1M) (Lescroart et al., 2010). To examine the effect of exogenous RA signaling on branchiomeric muscle development, we performed in situ hybridization using Myogenin as a myogenic determination marker (Hasty et al., 1993) in heads of E13.5 RA-treated embryos (Figures 1A–L). Muscles derived from BA1 displayed specific and consistent abnormalities in association with elevated RA signaling. The temporalis and masseter muscles failed to form (Figures 1B,H,J,L), and the pterygoid, mylohyoid and anterior digastric muscles were present, but were reduced in size and thickness, and the buccinator muscles were fragmented, as evidenced by the expression of Myogenin (Figures 1D,H,J,L). In the case of muscles derived from BA2, the frontalis and orbicularis oculi muscles were missing, whereas the others were hypoplastic and disorganized compared to controls (Figures 1D,F). The phenotypes of individual muscles derived from BA1 and BA2 are summarized in Figure 1N. Both intrinsic and extrinsic tongue muscles which originate from the occipital somite (Noden and Francis-West, 2006), formed normally and did not show any noticeable structural abnormality in RA-treated heads at E13.5 (Figures 1G–L and Supplementary Figure 1). Together, these data showed that excessive RA signaling results in defects in the development of BA1- and BA2-derived cranial muscles. In order to understand the cellular mechanism underlying this phenotype, we further analyzed the myogenic developmental process in RA-treated embryos.

Conversely, we also assessed the effect of reduced RA signaling on developing craniofacial muscle using conditional Rdh10 knock out embryos (CreErt2;Rdh10^flox/flox), which exhibit a severe reduction of RA signaling in the craniofacial region (Kurosaka et al., 2017). Interestingly, most of the cranial muscles showed subtle differences between control and CreErt2;Rdh10^flox/flox embryos (Supplementary Figure 2).

We examined the expression patterns of Pitx2 and Tbx1, which mark muscle precursor cells in BA1 and BA2. In the control embryos, Pitx2 was expressed at E10.5 in the mesodermal core of BA1 and BA2, while in the RA-treated embryos Pitx2 expression was substantially reduced (Figures 2A,B). In contrast, increased Pitx2 expression could be observed in the epithelium of BA1 (Figures 2A,B). Tbx1 expression was also reduced in BA1 and BA2 of E10.5 RA-treated embryos compared to controls (Figures 2C,D). These results indicated that excessive RA signaling interferes with the expression of Pitx2 and Tbx1, which are essential for differentiation of BA1 and BA2 myogenic mesoderm.

The myogenic determination factor MyoD1 is under the control of both Pitx2 and Tbx1 (Braun and Gautel, 2011). In E10.5 control embryos, MyoD1 was expressed in BA1 and BA2 myoblasts, which later develop into the masticatory and facial premuscle masses at E11.5. In RA-treated embryos, MyoD1 expression was a substantially reduced in BA1 and BA2, which is indicative of a defect in myogenic determination (and later differentiation). Furthermore, this is likely an effect of perturbed specification due to reduced Pitx2 and Tbx1 expression in BA1 and BA2 (Figures 2E,F,I,J). In contrast, MyoD1 expression in the hypoglossal cord, from which the tongue myoblasts migrate (Huang et al., 1999), remained unchanged in the RA-treated group (Figures 2G,H).

Collectively, these findings indicate that excessive RA perturbs BA1 and BA2 muscle precursor specification and determination. We also analyzed the expression of Aldh1a2 and Aldh1a3, which encode indispensable enzymes for synthesizing RA, in RA-treated and RA loss-of-function, CreErt2;Rdh10^flox/flox embryos. We detected subtle differences in the expression of these genes in the developing craniofacial region, between control and RA-treated embryos. This was also true for the expression of Aldh1a2 in Rdh10 mutant embryos, whereas in contrast Aldh1a3 expression was substantially reduced in the head (Supplementary Figure 3).

Pitx2 and Tbx1 have previously been implicated in proliferation and survival of muscle progenitor cells in the branchiomeric muscles (Kelly et al., 2004; Dong et al., 2006; Shih et al., 2007) and we observed altered Pitx2 and Tbx1 expression in E10.5 RA-treated embryos. We further investigated the behavior of BA1 and BA2 muscle progenitor cells via staining for Islet1 (Nathan et al., 2008) in conjunction with phosphorylated Histone H3 (pHH3) and TUNEL to assess cell proliferation and cell death, respectively. Cell proliferation was unchanged in the muscle progenitor cells of both BA1 and BA2 in E10.5 RA-treated embryos compared to controls (Figures 3A,B and Table 1). In contrast, increased cell death was detected in the muscle progenitor cells of both BA1 and BA2 in E10.5 RA-treated embryos compared to controls (Figures 3C,D and Table 1). These results demonstrate that excessive RA signaling results in cell death in muscle progenitor cells in both BA1 and BA2.

CNCC-mesoderm interactions are crucial for proper branchiomeric myogenesis (Grenier et al., 2009) and we have previously shown that excessive RA signaling at E8.5 affects CNCC development (Wang et al., 2019). Therefore, we hypothesized that a defect in CNCC development could underly the etiology of RA-induced BA1- and BA2- derived muscle malformation. To test this idea, we assessed the activity of Dlx5 and Dlx6, which are expressed by CNCCs and are required for the myogenic differentiation and patterning of craniofacial muscles (Heude et al., 2010). Dlx5 and Dlx6 expression was down-regulated in the proximal but not regions of distal branchial arches in E10.5 RA-treated embryos. The reduction was more pronounced in BA1 than BA2 (Figures 4A–D). Therefore, the BA1-derived muscle anomalies observed in the RA-treated embryos likely manifest as a result of perturbed Pitx2 and Tbx1 expression in the progenitor branchial arch mesoderm, together with loss of Dlx5 and Dlx6 expression by CNCCs in the proximal regions BA1, which impacts proximal-distal patterning. We also detected bilateral fusion of the upper and lower jaw in RA-treated embryos (Figures 4E–H). To evaluate the phenotype in detail, we dissected the mandibular structures from skeletal preparations of E18.5 embryos. In the control embryos, the main features – the coronoid process, the condylar process, the angular process and the molar alveolus – could be recognized (Figure 4G, box). In contrast to the controls, the position around the coronoid process and dentary bone in the mandible of RA-treated embryos was fused to the posterior/lateral position of maxilla (Figure 4H, box). Additionally, the condylar process was not detectable and the angular process was reduced in size (Figure 4H, box). The altered expression of Dlx5 and Dlx6 in the BAs could also have contributed to causing these defects (Vieux-Rochas et al., 2007; Vieux-Rochas et al., 2010).

RA signaling is mediated by a nuclear receptor superfamily consisting of multiple RARs (RARα, RARβ, and RARγ) and their heterodimeric binding partner RXRs (RXRα, RXRβ, and RXRγ). In order to further analyze the effects of enhanced and reduced RA signaling during craniofacial muscle development, the expression of RARs and RXRs was analyzed in both RA-treated and CreErt2;Rdh10^flox/flox embryos whose dams were administered tamoxifen at E7.0. Overall, we detected subtle differences in the expression levels and patterns of both RARs and RXRs between the control group and RA-treated embryos at E11.0 except that Rxra showed reduced expression in the developing frontonasal process (Supplementary Figure 4). In contrast, the expression of RARa, RARb, RXRa, RXRb, and RXRg was substantially reduced in the developing craniofacial region of E11.5 CreErt2;Rdh10^flox/flox embryos compared to controls(Supplementary Figure 4).

Numerous studies have attempted to elucidate the effects of stage- and/or dose- dependent effects of RA signaling on the development of various different muscles. It has been reported that RA signaling exhibits different effects on the differentiation of embryonic stem cells into cardiomyocytes depending on the timing of RA supplementation (Edwards and McBurney, 1983; Wobus et al., 1997; Nakajima, 2019). Compared with the levels of RA that enhance cardiomyogenesis, lower levels of RA signaling are known to enhance skeletal myogenesis (Edwards and McBurney, 1983; Halevy and Lerman, 1993). Previous studies have shown that both facial and tongue muscles are severely affected after RA-treatment (200 mg/kg) at E8.0 in the mouse fetus (Padmanabhan and Ahmed, 1997). We examined the effects of dose and timing of RA administration in our previous work on the effect of exaggerated RA signaling midgestational development. As a result, we found that daily administration of 25 mg/kg RA from E8.5 to E10.5 had the effect of disrupting multiple craniofacial structures (Wang et al., 2019). Also, in the present study, we detected BA1- and BA2-derived muscle malformations but minimal effects of RA-treatment on tongue muscle development. Our results indicate that proper RA signaling is essential for survival of muscle precursor cells in the branchial arch mesoderm. Branchiomeric muscle development is mediated by signaling and genetic pathways that are distinct from those of extraocular, tongue and laryngeal muscles in the head (Braun and Gautel, 2011). Interestingly, our study showed that in contrast to the induction of branchiomeric muscle defects, excess RA signaling did not cause defects in MyoD1 expression in the hypoglossal cord before E13.5. Altogether, these results indicate that muscle precursor cells react to RA signaling in different ways depending on the tissue type, developmental timing and dosage of RA. Interestingly, reduced RA signaling did not show as strong a disruption of BA1- and BA2-derived muscle development as elevated RA signaling. These results indicate that craniofacial muscle progenitor cells respond differently to elevated and reduced RA signaling.

Pitx2 mouse mutants exhibit a failure of BA1-derived muscle development in association with altered MyoD expression and elevated apoptosis of undifferentiated muscle progenitor cells (Dong et al., 2006). RA signaling has also been implicated in directly regulating Pitx2 in embryonic eye development (Kumar and Duester, 2010) and it has been also shown that RA signaling and Pitx2 regulate CNCCs during zebrafish development (Chawla et al., 2016). Exogenous RA (bead implantation) in chick embryos impacts the expression of the myogenic markers Pitx2 and Tbx1 (Bothe et al., 2011). Furthermore, Retinoid X receptor (RXR) co-localizes with PITX2 in extraocular muscle cells (Hebert et al., 2017). Together with our results, this clearly shows that perturbed RA signaling results in downregulation of Pitx2 in BA1 and BA2, which in turn results in branchiomeric muscle defects. However, although muscles derived from BA1 are absent in Pitx2 mutants, the muscles derived from BA2 are merely distorted (Shih et al., 2007). To explain the severe defects in muscles derived from BA2 in our study in association with elevated RA signaling, we examined the premyoblast specification marker, Tbx1. Tbx1 mutants exhibit severe perturbation or absence of both BA1 and BA2 derived muscles (Kelly et al., 2004). Furthermore, Pitx2 and Tbx1 are known to be molecular partners that regulate parallel pathways of common target genes in craniofacial muscle development (Dong et al., 2006; Bothe et al., 2011; Braun and Gautel, 2011). Additionally, interactions between Tbx1 and RA signaling have been demonstrated in a mouse model of DiGeorge syndrome (Ryckebusch et al., 2010). Consistent with these findings, we observed a substantial reduction of Tbx1 expression in the mesodermal core of BA1 and BA2 in RA-treated embryos. Taken together, our results further support the conclusion that elevated RA signaling results in defects in the specification of undifferentiated muscle progenitor cells in BA1 and BA2, and thus contributes to the etiology of branchiomeric muscle malformation (Figure 5).

Previous in vitro and in vivo studies showed that proper RA signaling is essential for normal NCC development (Lee et al., 1995; Niederreither et al., 2003; Wang et al., 2019). Furthermore, CNCCs surround mesoderm cells in BAs that give rise to myogenic progenitors (Trainor et al., 1994; Trainor and Tam, 1995), and these cellular interactions are crucial for branchiomeric muscle formation (Noden, 1983, 1986; Chai et al., 2000; Sambasivan et al., 2011). For instance, Dlx5-positive CNCCs have been reported to guide the migration of muscle progenitors derived from BAs (Sugii et al., 2017). Additionally, CNCCs induce cranial myogenic formation by secreting inhibitory factors of the BMP and WNT signaling pathways in the BAs of chick embryos (Tzahor et al., 2003). In the present study, Dlx5/6 expression was decreased in the proximal region of BA1 and BA2. The loss of Dlx5/6 is known to impact muscle patterning and differentiation during the later stages of branchiomeric muscle development, however, Dlx5/6 do not affect the expression of the myogenic specification markers (Heude et al., 2010). Taken together, these facts suggest that several mechanisms contribute to the branchiomeric muscle defects observed in the present study. Firstly, early branchiomeric muscle formation is inhibited by defects in CNCC development (Heude et al., 2010) and subsequent differentiation and patterning are affected by altered Dlx5/6 expression in BAs as a result of disturbed RA signaling. Failure of Dlx5/6 expression leads to intrinsic tongue and sublingual muscle defects and masticatory muscle defects (Heude et al., 2010). Interestingly, in our study, reduced proximal Dlx5/6 expression led to masticatory muscle progenitor cell defects, but intrinsic tongue and mylohyoid muscles, which derive from a different mesodermal source, were present. We speculate that elevated RA has a small effect on mesodermal precursors of those muscles at the stages and under the conditions we have investigated. Additionally, since muscle attachments to bone are critical for shaping bone during development, it is possible that absence of the muscles of mastication is one reason for retarded condylar and angular processes development in the mandibles RA-treated embryos.

Retinoic acid receptors mediate RA signaling and thus play important roles in transducing RA signaling during embryonic development (Rhinn and Dolle, 2012). In the present study, we detected subtle differences in the expression of most RARs and RXRs between E11.0 control and RA-treated embryos, with the exception of RXRa, which showed reduced expression in the RA-treated group. These results indicated that daily administration of 25 mg/kg RA from E8.5 to E10.5 does not critically influence the expression of either RARs or RXRs during craniofacial development. In contrast, we detected a substantial diminishment in retinoic acid receptor expression in embryos in which RA signaling was reduced. There is a large degree of functional redundancy between retinoic acid receptors (Mark et al., 2009), hence, further investigations including genetic complementation experiments will be required to reveal the effects of altered expression of retinoic acid receptors on craniofacial muscle development.

Pregnant female Institute of Cancer Research (ICR) mice (CLEA, Japan) were administered all-trans RA (25 mg/kg body weight) (Sigma-Aldrich) by oral gavage. All of the mice were housed with a 12 h dark-light cycle in which the light phase started from 8 a.m. RA (25 mg/ml in dimethylsulfoxide) was diluted 1/10 in corn oil just before use. Control animals were given the equivalent volume of the carrier. Oral gavage was performed once per day at gestation stages (from E8.5 to E10.5). Embryonic stage E0.5 was defined on the morning of vaginal plug confirmation. The approximate somite number in ICR mice at E8.5 was seven pairs.

Rdh10^flox/flox and Cre-ERT2 mice were maintained and used as previously described (Sandell et al., 2007, 2012; Kurosaka et al., 2017). In order to eliminate RDH10 from developing embryos, Rdh10^flox/flox female mice were crossed with Cre-ERT2:Rdh10^flox/flox male mice followed by administration of tamoxifen at E7.0 as previously reported (Kurosaka et al., 2017). Cre-ERT2 mice carry the gene for Cre recombinase fused to the estrogen receptor T2 cassette inserted into the Rosa 26 locus (Ventura et al., 2007). Consequently, recombination takes place in a ubiquitous manner after administration of tamoxifen (Hayashi and McMahon, 2002). Rdh10^flox/flox embryos were used as control samples throughout the study.

All animal experiments were performed in accordance with the guidelines of the Animal Care and Use Committee of the Osaka University Graduate School of Dentistry, Osaka, Japan. The committee on the ethics of animal experiments of the same university approved the study protocol (permit number: 26-028-0, 26-020-0).

We used the R_ mCT2 system (Rigaku) with scanning parameters 50 kV, 200 μA to perform micro-CT. Scans were reconstructed and analyzed using 3D viewer and Volume Rendering Control software (Rigaku), according to standardized protocols.

E18.5 embryos were skinned and eviscerated. The embryos were fixed in 100% ethanol overnight and then stained for 24 h with Alcian Blue (150 μg/ml in 20 ml of glacial acetic acid and 80 ml of 95% ethanol). After washing in 100% ethanol, soft tissues were dissolved in 2% KOH overnight and stained with Alizarin Red (50 μg/ml in 1% KOH) overnight. Stained embryos were kept in 20% glycerol/1% KOH until skeletons became clearly visible. Embryos were stored in 50% glycerol/50% water.

Whole-mount and sectional in situ hybridization was performed as described with minor modifications using digoxigenin (DIG)-UTP (Roche)-labeled antisense RNA probes corresponding to the sequences of Myogenin, Pitx2, Tbx1, Myod1, Dlx5/6, Aldh1a2, Aldh1a3, Rara, Rarb, Rarg, Rxra, Rxrb, and Rxrg. Sequences used were previously reported in the Allen Brain Atlas¹. For all in situ hybridization analyses, a minimum of three embryos of each sample were examined per probe.

Analyses of apoptotic cells were performed using an in situ cell death detection kit (Roche) following the manufacturer’s instructions. For analyses of proliferation, samples were incubated with a mouse anti-pHH3 antibody (1:200, Millipore) at 4°C overnight followed by secondary Alexa-Fluor-488 donkey anti-mouse IgG (1:200, Invitrogen) for 6 h at room temperature for sections and overnight at 4°C for whole embryos. To label muscle progenitor cells in BA1 and BA2, sections were counterstained with a goat anti-Islet1 antibody (5 μg/ml, Abcam) at 4°C overnight, followed by secondary antibody (Alexa-Fluor-546 donkey anti-goat IgG, 1:200, Molecular Probes). Cells in at least five adjacent sections were counted in each assay. Statistical significance was assessed using Fisher’s exact test.