Cardiopharyngeal mesoderm origins of musculoskeletal and connective tissues in the mammalian pharynx

The pharynx is essential for feeding, breathing and speech, and the pharyngeal arches (PAs), segmental structures in the vertebrate embryo, play a crucial role in pharyngeal development. Two types of mesenchymal progenitor cells are located in the embryonic pharynx and contribute to different elements of the pharyngeal apparatus during morphogenesis.

One is neural crest cells (NCCs) that originate from the dorsal neural tube and reside peripherally in PAs. NCCs have been shown to make a crucial contribution to the vertebrate pharyngeal skeleton and also give rise to connective tissues (CTs) of skeletal and muscular systems (Le Douarin et al., 2004; Noden and Trainor, 2005; Minoux and Rijli, 2010). NCCs further possess patterning information that dictates the morphology of musculoskeletal elements in the head (Schneider and Helms, 2003; Le Douarin et al., 2004; Noden and Trainor, 2005; Rinon et al., 2007; Tokita and Schneider, 2009; Gitton et al., 2010). NCCs are therefore postulated to have a pivotal role in the development and evolution of the vertebrate head and pharyngeal apparatus. Nevertheless, mutant analyses of Hox and Dlx genes that are expressed in NCCs have failed to show a clear homeotic transformation in posterior PA derivatives, unlike jaw and ear elements derived from anterior PAs (Beverdam et al., 2002; Depew et al., 2002; Minoux et al., 2009; Minoux and Rijli, 2010).

The other key component in pharyngeal development is cardiopharyngeal mesoderm (CPM), a subset of head mesoderm in the embryonic pharynx. CPM yields pharyngeal mesoderm in the core of the PAs and second heart field (SHF) cardiac progenitor cells, and contributes to several lineages, including skeletal muscles in the head and neck, SHF-derived cardiac elements and cervical arteries (Kelly et al., 2001; Lescroart et al., 2010, 2015; Gopalakrishnan et al., 2015; Diogo et al., 2015; Wang et al., 2017). Indispensable genes and markers for CPM development include Isl1, and Fgf10 and Mef2c enhancer transgenes (Kelly et al., 2001; Cai et al., 2003; Verzi et al., 2005). The gene encoding the T-box containing transcription factor, Tbx1, is also expressed in CPM. TBX1 is the major gene for 22q11.2 deletion syndrome (22q11.2DS, also known as DiGeorge or Velocardiofacial syndrome), characterized by a spectrum of defects in craniofacial, pharyngeal and cardiovascular structures (Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al., 2001; Yagi et al., 2003). Tbx1 plays an upstream role in the regulatory hierarchy driving branchiomeric but not somitic myogenesis, and mice lacking Tbx1 exhibit severe musculoskeletal defects in the head and neck region (Jerome and Papaioannou, 2001; Lindsay et al., 2001; Merscher et al., 2001; Kelly et al., 2004; Dastjerdi et al., 2007; Sambasivan et al., 2009; Kong et al., 2014; Heude et al., 2018).

The view that NCCs contribute to skeletal elements and CT, while CPM gives rise to musculature during pharyngeal development has been challenged by evidence for a mesodermal origin of components of the pharyngeal skeleton and CT in amniotes (Noden, 1988; Tabler et al., 2017; Heude et al., 2018). In these studies, however, the lateral mesoderm (mesoderm adjacent to the otic vesicle and the first somite) was traced in chicken-quail chimera analysis, and Mesp1-Cre mice, labeling anterior somites, lateral plate mesoderm (LPM) and head mesoderm, were employed in genetic lineage analysis. Therefore, the precise mesodermal origins of the pharyngeal skeleton and CT remain elusive. Given the genetic differences between head and trunk mesoderm, the identification of mesodermal sources of pharyngeal structures is essential to understand the development, pathology and evolution of the pharynx (Buckingham and Vincent, 2009; Sambasivan et al., 2011).

Here, we have mapped the contribution of CPM to pharyngeal structures by genetic lineage tracing using Mef2c gene Anterior Heart Field enhancer (Mef2c-AHF-Cre) and Rosa26^Stop/YFP (yellow fluorescent protein, hereafter called RYFP) transgenic mice (Srinivas et al., 2001; Verzi et al., 2005). Mef2c-AHF-Cre drives Cre expression in CPM including head muscle progenitor cells and the SHF (Verzi et al., 2005). By comparison with genetic lineage analysis of NCC derivatives, and validation with Tbx1-Cre, a second Cre line expressed in CPM (Huynh et al., 2007), we demonstrate that CPM contributes to a broader spectrum of posterior pharyngeal structures than had been previously thought, including cartilage and CT. The latter fate implies a potential role for CPM in muscle patterning, reinforced by the identification of neck muscle patterning defects in mice that are haploinsufficient for the CPM regulatory gene Tbx1. By highlighting the major contribution of CPM, our results bring new clinical and evolutionary insights into development of the mammalian pharyngeal apparatus.

In order to investigate the contribution of the Mef2c-AHF-Cre genetic lineage to pharyngeal structures, we first assessed the specificity of genetic labeling by Mef2c-AHF-Cre at embryonic day (E)15.5. At this stage, musculoskeletal elements are well developed and identifiable in histological sections (Kaufman, 1992). We used immunofluorescence to detect YFP labeling in Mef2c-AHF-Cre;RYFP mice that identifies Cre-expressing cells and their descendants. Consistent with previous results, we observed YFP⁺ cells in PA-derived skeletal muscles (Figs S1 and S2) (Lescroart et al., 2010; Lescroart et al., 2015; Heude et al., 2018). In addition to branchiomeric muscles, YFP⁺ cells were found in PA artery derivatives including the carotid arteries and in the myocardium of the outflow tract and right ventricle (Fig. S2) (Verzi et al., 2005; Wang et al., 2017). In contrast, NCC, LPM and somitic derivatives, as well as ectodermal and endodermal structures, were YFP-negative (see details in Figs S1 and S2) (Köntges and Lumsden, 1996; Chai et al., 2000; Jiang et al., 2000; Huang et al., 2000; Le Douarin et al., 2004; Noden and Trainor, 2005; Matsuoka et al., 2005; Heude et al., 2018). Consistent with these results, activation of Cre or Mef2c-AHF-Cre derivatives has not been observed in NCCs, LPM, somite, ectoderm or endoderm for this transgenic line during early development (from E7.5 to E12.5; this study; Verzi et al., 2005; Lescroart et al., 2015). Collectively, these results confirm the expression of Mef2c-AHF-Cre in CPM and validate the use of this line to genetically trace the contribution of CPM to the pharyngeal region.

Analysis of transverse sections at pharyngeal and shoulder levels revealed that skeletal components were labeled in the Mef2c-AHF-Cre line. Arytenoid, cricoid and tracheal cartilages, as well as the mid-ventral portion of the thyroid cartilage, were YFP⁺ (Fig. 1A-H). However, the dorsal and lateral parts of the thyroid cartilage were devoid of YFP⁺ cells. At the shoulder level, the medial portion of the clavicle and the dorsal end of the sternum were specifically labeled by YFP (Fig. 1E,H), whereas the lateral part of the clavicle and the ventral sternum were YFP^– (Fig. S2F,K).

Close analysis of sections revealed that YFP⁺ cells were distributed in the cell layer surrounding pharyngeal and shoulder skeletons. These perichondrial cells were detected around arytenoid, cricoid and tracheal cartilages (Fig. 1, Fig. S3). The perichondrium of the mid-ventral thyroid cartilage was also YFP⁺, whereas that of the lateral portions of the thyroid cartilage was YFP^–. In addition, the medial perichondrium of the clavicle was marked by YFP. The distribution of labeled CT thus matches the contribution of the Mef2c-AHF-Cre genetic lineage to cartilaginous and bony elements in the pharyngeal and neck region. In addition to skeletal and perichondrial cells, we found Mef2c-AHF-Cre derivatives in medial pharyngeal mesenchyme. YFP⁺ cells were distributed in mesenchymal cells around esophageal and laryngeal endoderm, and also detected adjacent to the thyroid gland (Fig. 1F-H, Fig. S3C-E′). Together, these analyses demonstrate that CPM expressing the Mef2c-AHF-Cre transgene contributes to medial skeletons, skeletal CT and mesenchymal cells in the pharynx and shoulder.

To compare the Mef2c-AHF-Cre genetic lineage with sites of active transgene expression, we next evaluated Cre activity at E15.5 using in situ hybridization. Whereas Cre expression was observed in the outlet region of the heart, where the Mef2c-AHF enhancer was shown previously to be active at fetal stages (Verzi et al., 2005), expression was undetectable in other tissues (Fig. S4). These results are consistent with downregulation of Mef2c-AHF-Cre expression in the right ventricle and venous pole myocardial progenitor cells in the posterior CPM (Verzi et al., 2005; Goddeeris et al., 2008; De Bono et al., 2018). YFP labeling in visceral and shoulder skeletal elements, as well as perichondrial and mesenchymal cells, is thus likely to be the result of recombination events at earlier developmental stages, rather than de novo expression of Cre in these structures at the time of observation.

As NCCs have been proposed to contribute to the pharyngeal skeleton, we further performed a lineage analysis using Wnt1-Cre;RYFP mice, which genetically label NCCs in PAs and their descendants. NCC derivatives were found in the hyoid bone, epiglottis, thyroid cartilage, cranial nerves, smooth muscles of internal carotid artery and mesenchyme around the thyroid and thymus glands, consistent with previous results (Fig. 1I-K, Fig. S5) (Jiang et al., 2000; Matsuoka et al., 2005; Foster et al., 2008; Tabler et al., 2017; Heude et al., 2018). Importantly, medial skeletal components were mostly devoid of Wnt1-Cre lineage-positive cells, including the cricoid and arytenoid cartilages, as well as the mid-ventral thyroid cartilage (Fig. 1I,J). Similarly, Wnt1-Cre derivatives were sparse in the mesenchyme around the pharyngeal endoderm (Fig. 1I,J, Fig. S5). Further observations revealed a minor NCC contribution to the posterior part of the cricoid cartilage near the articulation site with the thyroid cartilage, and no overt Wnt1-Cre derivatives in tracheal cartilages and shoulder skeletons (Fig. 1J,K). Our results support and extend recent analyses of neural crest and mesodermal contributions to the pharyngeal skeleton (Tabler et al., 2017; Heude et al., 2018). In particular, we observed complementary medial and lateral contributions of CPM and NCC to cartilaginous elements in the pharynx and identify CPM as the source of mesodermal pharyngeal skeletal elements.

The observation that CPM contributes to skeletal CT led us to investigate whether CPM also contributes to the CT of pharyngeal muscles. Mef2c-AHF-Cre derivatives were detected in the myofibers of laryngeal muscles (lcary, thary and vm in Fig. 2B), consistent with their branchiomeric as opposed to somitic mesodermal origin (Fig. 2A-C) (Lescroart et al., 2015; Tabler et al., 2017). However, we observed that YFP labeling in these muscles was denser than that of MF20, a myosin heavy chain antibody labeling myofibers, suggesting that YFP⁺ cells contribute to both muscle cells and surrounding muscle CT (MCT) (Fig. 2B-C′). To further investigate this, we used an anti-TCF4 antibody to identify MCT progenitor cells (Kardon et al., 2003). Analysis at single cell resolution revealed co-localization of YFP and MF20 in myofibers, and YFP and TCF4 in surrounding MCT (Fig. 2D). A similar contribution of the Mef2c-AHF-Cre lineage to both myofibers and MCT was also noted in other pharyngeal and laryngeal muscles, and notably YFP⁺ MCT cells were observed in the medial portion of the cricothyroid and sternocleidomastoid muscles (Fig. S6A). Moreover, the attachment sites of pharyngeal muscles to arytenoid and cricoid cartilages were labeled by YFP (Fig. 2B-D, Fig. S6A).

Unexpectedly, we also detected Mef2c-AHF-Cre derivatives in MCT associated with infrahyoid muscles (omh and sthy in Fig. 2B,C). Infrahyoid neck muscles are non-branchiomeric muscles originating in the somites that migrate secondarily into the neck region as part of the hypoglossal cord (Mackenzie et al., 1998; Noden and Trainor, 2005; Heude et al., 2018). Co-immunostaining showed that, although the Mef2c-AHF-Cre genetic lineage does not contribute to MF20⁺ muscle fibers in these muscles, YFP⁺ cells co-localized with TCF4 (Fig. 2E). YFP⁺ MCT was also found in other infrahyoid muscles, and notably at the attachment site of infrahyoid muscles to the body of the hyoid bone (Fig. 2C,E, Fig. S6B). In contrast, somite-derived tongue muscles, including the hyoglossus, that are derived from more anterior regions of the hypoglossal cord, were devoid of YFP⁺ TCF4⁺ MCT (Fig. 2E). As in the case of skeletal elements, in situ hybridization revealed that these sites of Mef2c-AHF-Cre-activated YFP expression are not sites of active Cre expression at E15.5 (Fig. S4). This suggests that YFP labeling in MCT results from genetic lineage tracing of cells expressing Cre at earlier developmental stages.

NCCs are considered to be a major source of MCT in pharyngeal and infrahyoid muscles (Köntges and Lumsden, 1996; Le Douarin et al., 2004; Noden and Trainor, 2005; Matsuoka et al., 2005). Our analysis of Wnt1-Cre;RYFP embryos showed that NCCs contribute to MCT in the lateral, but not medial, component of pharyngeal muscles, such as the cricothyroid muscle, in which MCT is predominantly derived from Wnt1-Cre⁺ cells (Fig. 3A-C,G-I, Fig. S5A-D″). In contrast, MCT in the medial component of these muscles was derived from Mef2c-AHF-Cre-expressing progenitor cells, revealing a strikingly complementary contribution of the two sources of MCT to different regions of individual muscles (Fig. 3D-I). Dual cellular origins of MCT was also observed in somitic infrahyoid muscles, however only very few NCC-derived cells were associated with these muscles, with the exception of the thyrohyoid muscle (Fig. S5C-E″). In summary, CPM gives rise to MCT of both branchiomeric-derived pharyngeal and somitic-derived infrahyoid muscles, and makes a complementary contribution to that of NCC-derived MCT.

The above observations demonstrate that CPM gives rise to medial skeletal elements and CT proximal to pharyngeal endoderm, whereas NCCs contribute to more lateral components in the pharynx (Figs 1–3). Considering that the typical PA configuration is a central CPM core surrounded by peripheral NCCs adjacent to pharyngeal ectoderm and endoderm, the above cellular distribution is unusual, potentially arising by remodeling of PA mesenchyme during development. To clarify this point, we investigated the cellular distribution of CPM, NCC and myogenic progenitor cells in mid-gestation embryos.

At E10.5 at the levels of the 3rd and 4th PA, Mef2c-AHF-Cre-labeled CPM cells were positioned centrally and Wnt1-Cre-labeled NCCs peripherally in PAs. In contrast, at the 6th arch level, CPM-derivatives were positioned medially and NCC-derivatives laterally, similar to the situation at E15.5 (Fig. 4A-B′,D-E′). Mef2c-AHF-Cre derivatives at this level were divided into medial and lateral populations, corresponding to laryngopharyngeal and trapezius/sternocleidomastoid progenitors, respectively. Wnt1-Cre⁺ cells were observed between those two populations, as well as superficially to the lateral Mef2c-AHF-Cre⁺ cells. In contrast, the medial population of Mef2c-AHF-Cre-labeled cells were not surrounded by NCC cells (Fig. 4C,F). The unique distribution of CPM and NCC derivatives in the posterior pharyngeal region thus originates early in development.

We then investigated how CPM myogenic and CT progenitor cells are distributed in mid-gestation embryos. At E10.5, the myogenic regulatory factor MYOD (MYOD1) is expressed in both branchiomeric muscle progenitor cells originating from PA mesoderm, and somite-derived progenitor cells of the tongue and infrahyoid muscles located in the hypoglossal cord (Mackenzie et al., 1998; Kelly et al., 2004; Adachi et al., 2018). In Mef2c-AHF-Cre;RYFP embryos, YFP and MYOD colocalized in 6th PA myoblasts at the boundary of medial Mef2c-AHF-Cre lineage-positive and lateral lineage-negative populations (Fig. 5A-B″). These cells likely correspond to the progenitor cells of pharyngeal muscles with dual CPM and NCC-derived MCT contributions (Figs 2 and 3). YFP^– MYOD⁺ cells in the posterior part of the hypoglossal cord, but not the anterior part, were enclosed by YFP⁺ mesenchymal cells (Fig. 5C-E′). This is consistent with observations that only the anterior part of the hypoglossal cord is surrounded by NCCs at E10.5 (Adachi et al., 2018). Together, these results indicate that the association of CPM with branchiomeric and somite-derived myoblasts in the posterior PA and hypoglossal cord, respectively, is established early during development. Moreover, the relative position of myoblasts and CPM at E10.5 reflects the cellular distribution of myofibers and CTs at later developmental stages, similar to the early association of CPM with NCC CT progenitor cells in anterior PAs (Noden and Trainor, 2005; Grenier et al., 2009).

The early association of NCC cells and branchiomeric muscle progenitors plays an essential role in patterning craniofacial skeletal musculature (Noden and Trainor, 2005; Rinon et al., 2007; Grenier et al., 2009). Our observations that Mef2c-AHF-Cre lineage-labeled cells are closely apposed to myoblasts in the posterior embryonic pharynx and give rise to MCT cells of pharyngeal and infrahyoid muscles raise the hypothesis that CPM may also play a patterning role, analogous to that of NCCs in craniofacial muscles, during pharyngeal and neck muscle development. In order to test this hypothesis we perturbed CPM development using the key CPM regulatory gene, Tbx1. Tbx1 encodes a T-box containing transcription factor and is required for normal activation of the branchiomeric myogenic program and differentiation of esophageal and trapezius muscles (Kelly et al., 2004; Dastjerdi et al., 2007; Grifone et al., 2008; Sambasivan et al., 2009; Theis et al., 2010; Kong et al., 2014; Fuchs et al., 2015; Gopalakrishnan et al., 2015; Lescroart et al., 2015; Heude et al., 2018). However, a role of Tbx1 in subsequent skeletal muscle patterning has not been described. To address this, we first examined the contribution of the Tbx1 genetic lineage to MCT, and then characterized pharyngeal muscle patterning in heterozygous and homozygous Tbx1 mutant embryos.

We genetically traced Tbx1 lineages using Tbx1-Cre;mTmG mice [membrane-targeted tandem dimer Tomato and membrane-targeted green fluorescent protein (GFP)] embryos (Huynh et al., 2007; Muzumdar et al., 2007). Analysis at E15.5 revealed that the Tbx1 genetic lineage gives rise to arytenoid, cricoid, mid-ventral thyroid and tracheal cartilages and their perichondrium. Tbx1-Cre derivatives were also detected in the clavicle and CT at muscle attachment sites on the pharyngeal skeletons, in addition to laryngeal endoderm (Fig. 6A-H). Moreover, the Tbx1-Cre lineage labeled branchiomeric pharyngeal and laryngeal muscles as well as TCF4⁺ MCT cells in branchiomeric and infrahyoid muscles at E15.5 (Fig. 6F-H). In contrast to Mef2c-AHF-Cre, Tbx1 has been reported to be expressed in a subset of NCC derivatives (Funato et al., 2015). However, the Tbx1 and Mef2c-AHF-Cre lineages displayed highly similar labeling of Wnt1-Cre lineage-negative CT and, moreover, we did not observe significant GFP signal in the hyoid bone, the lateral thyroid cartilage or the lateral CT of branchiomeric muscles in which Wnt1-Cre derivatives were detected (Figs 1, 6 and Fig. S5). Analysis at E10.5 revealed that TBX1 was detected in cells surrounding MYOD⁺ cells of the hypoglossal cord (Fig. 6I-J″), in a similar distribution to cells of the Mef2c-AHF-Cre genetic lineage (Fig. 5E,E′). These findings strongly reinforce the conclusion, based on analysis of the Mef2c-AHF-Cre genetic lineage, that these pharyngeal skeletal and CT elements are CPM derivatives. However, in contrast to findings with the Mef2c-AHF-Cre lineage, we observed Tbx1-Cre activated GFP⁺ cells in muscle fibers as well as in MCT of somite-derived infrahyoid muscles (Fig. 6H). Previous evidence has suggested that Tbx1 is activated in hypoglossal cord-derived muscles, including tongue and infrahyoid muscles, at fetal stages, after muscle patterning and differentiation (Okano et al., 2008). In support of late activation of Tbx1 in these muscles, TBX1 is not expressed in MYOD⁺ myogenic progenitor cells in the hypoglossal cord at E10.5 (Fig. 6I-J″).

Haploinsufficiency for TBX1 contributes to 22q11.2DS phenotypes in humans and Tbx1 heterozygous mutant mice display stochastic arch artery anomalies (Scambler, 2010). With the exception of hypoplasia of the levator veli palatini, the branchiomeric myogenic program is normally activated in Tbx1^+/− mice (Kelly et al., 2004; Fuchs et al., 2015). Given the contribution of Mef2c-AHF-Cre- and Tbx1-Cre-labeled CPM to CT in the pharyngeal and neck regions, we investigated skeletal muscle patterning in Tbx1 heterozygous mutant embryos. Comparative analysis of wild-type and heterozygous mutant mice at E15.5 revealed a high frequency of patterning defects in two neck muscles in Tbx1^+/− embryos (Fig. 7, Table S1). The sternocleidomastoid muscle is a branchiomeric muscle derived from the posterior PAs that normally attaches to the mastoid process and extends ventrally to the clavicle and sternum of the shoulder girdle. In Tbx1^+/− embryos, however, the medial part of sternocleidomastoid muscles extended medially and connected ectopically to the greater horn of the hyoid bone (Fig. 7A-C,J,K). Such connections, observed in 8/16 Tbx1^+/− embryos, were not seen in seven wild-type embryos. Similarly, the omohyoid muscle is a somite-derived infrahyoid muscle that attaches to the body of the hyoid bone and extends to the medial side of the scapula in wild-type embryos. Tbx1^+/− embryos displayed ectopic connections of the omohyoid muscles to the greater horn of the hyoid bone that were not observed in wild-type embryos (Fig. 7D-F,J,K). Moreover, in Tbx1^+/− embryos, the omohyoid muscle failed to extend to the normal attachment site on the scapula and instead extended abnormally to the clavicle (Fig. 7G-I,J,K). Ectopic insertions of the sternocleidomastoid and omohyoid muscles were observed both unilaterally and bilaterally, and although ectopic connections of these muscles were not correlated, the two ectopic muscles occasionally fused together when present on the same side (Fig. 7A-F, Table S1). Similarly, stochastic right and left PA artery defects are characteristic of Tbx1^+/− embryos; however, no correlation was observed between lateralized muscle patterning defects and arch artery anomalies (Table S1).

Detailed investigation of Tbx1-Cre;mTmG embryos at E15.5 confirmed that both the sternocleidomastoid and omohyoid muscles contain MCT derived from Tbx1-expressing CPM (Fig. 8). Furthermore, as the Tbx1-Cre allele is a heterozygous null allele, we observed ectopic connections of these two muscles to the greater horn of the hyoid bone, similar to those seen in Tbx1^+/− embryos (Table S1). Immunofluorescence revealed that GFP and TCF4 co-localize in the MCT of these ectopic muscles (Fig. 8D,E). As mentioned above, GFP expression was also observed in muscle fibers of the infrahyoid and tongue muscles owing to the activation of Tbx1 during the fetal period (Fig. 6H, Fig. 8). Although a late role for Tbx1 in patterning these muscles cannot be ruled out, the arrangement of tongue muscles, characterized by NCC-derived CT, in Tbx1 heterozygous embryos was comparable with that in wild-type embryos (Fig. 8, Fig. S7A-E). These results suggest that, subsequent to normal myogenic differentiation, biallelic expression of Tbx1 in CPM-derived MCT is required for correct neck muscle patterning. Moreover, in Tbx1^−/− embryos, infrahyoid muscles were formed, but severely mispatterned, displaying ectopic insertion sites on the lateral part of the clavicle and coalescence of muscle bundles; in contrast, tongue muscle pattern was normal in Tbx1^−/− embryos (Fig. S7). Together, our results reveal that Tbx1 is required for the correct patterning of neck muscles and provide the first evidence implicating CPM-derived CT in muscle patterning.

To further understand the developmental role of Tbx1 in the CT lineage, we crossed Mef2c-AHF-Cre and Tbx1^flox/flox transgenic lines (Xu et al., 2004) and observed the phenotype of Mef2c-AHF-Cre;Tbx1^flox/+ embryos, in which Tbx1 is heterozygous in the Mef2c-AHF-Cre genetic lineage. In control Tbx1^flox/+ embryos, sternocleidomastoid and omohyoid muscles showed the normal pattern of wild-type embryos (Fig. 9A). In contrast, the omohyoid muscle in Mef2c-AHF-Cre;Tbx1^flox/+ embryos exhibited an ectopic connection to the greater horn of hyoid bone and an abnormal extension to the clavicle (n=5/7), phenotypes seen in Tbx1 heterozygous embryos (Fig. 9B, Table S1). No patterning defects were observed in the sternocleidomastoid muscle in Mef2c-AHF-Cre;Tbx1^flox/+ embryos. The different penetrance of the phenotypes between Mef2c-AHF-Cre;Tbx1^flox/+ and Tbx1^+/− embryos may reflect the delayed excision timing of Tbx1 gene in the Mef2c-AHF-Cre genetic lineage and the relative incidence of sternocleidomastoid versus omohyoid patterning defects in Tbx1^+/− embryos (Table S1). Together, our conditional mutant analysis demonstrates that biallelic expression of Tbx1 in CPM-derived CT is required for normal neck muscle patterning.

The pharynx is a crucial organ for the control of food intake, swallowing, respiration and vocalization. However, despite the functional importance of the pharynx, its developmental origins remain controversial. CPM gives rise to branchiomeric skeletal muscles of the head and neck as well as SHF-derived myocardium. Here, using genetic lineage analyses we have shown that CPM has a broader developmental potential than previously recognized and contributes to a spectrum of pharyngeal structures in mice (Fig. 10, Table S2). These include cartilaginous and perichondrial cells in the medial pharyngeal and shoulder skeletons as well as MCT of pharyngeal and somite-derived infrahyoid muscles. Moreover, biallelic expression of the CPM regulatory gene Tbx1 is required to correctly pattern neck muscles with CPM-derived MCT. Together, our results highlight the CPM origins of the mammalian pharynx, and provide new insights into the development, pathology and evolution of the pharyngeal apparatus.

CPM and NCCs contribute in a complementary mediolateral pattern to the skeletons, mesenchyme and MCT of the pharynx. CPM derivatives in the pharyngeal region extend to posterior pharyngeal muscles and MCTs of branchiomeric and infrahyoid muscles, including muscle attachment sites to pharyngeal skeletons. How and when different cell fate choices are made within CPM remains to be defined. However, the contribution of CPM to MCT appears to be established early in the embryonic pharynx, where CPM is closely associated with CPM- and somite-derived muscle progenitor cells. This spatiotemporal distribution of CPM in the pharynx is similar to that of NCC-derived MCT in craniofacial structures that surround branchiomeric muscle progenitor cells and instruct muscle patterning during development (Schneider and Helms, 2003; Le Douarin et al., 2004; Noden and Trainor, 2005; Rinon et al., 2007; Grenier et al., 2009; Tokita and Schneider, 2009; Gitton et al., 2010). Our results thus raise the hypothesis that CPM, like NCCs, may confer patterning information on muscles during morphogenesis of the mammalian pharynx. For example, the differential origins of MCT within individual muscles may contribute to ensuring correct attachment to mesodermal and NCC-derived musculoskeletal elements (Heude et al., 2018). Moreover, as NCC derivatives are implicated in the formation of innervation pattern in craniofacial elements (Köntges and Lumsden, 1996; Begbie and Graham, 2001), the relative contribution of CPM and NCCs to MCT may be linked to muscle innervation in the pharynx. Indeed, muscles innervated by the recurrent branch of the vagus nerve have mostly CPM-derived MCT, whereas the cricothyroid muscle, innervated by the superior laryngeal branch of the vagus nerve, has extensive NCC MCT contributions. Similarly, infrahyoid muscles with extensive CPM-derived MCT have cervical innervation (ansa cervicalis), whereas the thyrohyoid muscle with predominantly NCC-derived MCT is innervated by a branch of the hypoglossal nerve. These observations point to similar developmental roles of CPM and NCC in muscle patterning and innervation.

The T-box transcription factor Tbx1 is a crucial regulator of CPM and pharyngeal development. In support of a CPM origin of infrahyoid MCT, TBX1 is expressed in cells surrounding the hypoglossal cord at midgestation, in a similar distribution to Mef2c-AHF-Cre derivatives. At fetal stages, Tbx1-Cre;mTmG labels infrahyoid muscle myofibers as well as MCT, consistent with later activation of Tbx1 in the myogenic lineage. We found that heterozygous Tbx1 mutant embryos exhibit defective patterning of both branchiomeric and somite-derived neck muscles containing CT derived from Mef2c-AHF-Cre and Tbx1-Cre genetic lineages. In particular, the sternocleidomastoid and omohyoid muscles ectopically connect to the greater horn of the hyoid bone in Tbx1^+/− embryos. Moreover, homozygous Tbx1 mutant embryos show a patterning defect in infrahyoid muscles, with ectopic insertion into the clavicle. Although we cannot exclude that later Tbx1 expression in the muscle lineage contributes to the observed patterning defects, tongue muscles with MCT exclusively derived from NCC are patterned normally in Tbx1 mutants, despite also upregulating Tbx1 expression after differentiation (Okano et al., 2008). Moreover, with the exception of the 4th arch-derived levator veli palatini, which is hypoplastic but correctly patterned, branchiomeric muscles in Tbx1 heterozygous mutants, and even hypoplastic first arch-derived muscles in Tbx1 null embryos, are patterned normally (Kelly et al., 2004; Grifone et al., 2008; Fuchs et al., 2015). Crucially, conditional mutant mice with the deletion of one Tbx1 allele in the Mef2c-AHF-Cre lineage display similar muscle patterning defects to those seen in Tbx1^+/− mice. As the Mef2c-AHF-Cre lineage contributes to MCT but not to myofibers of the omohyoid muscle, this result strongly supports a role for CPM-derived MCT in neck muscle patterning. Together, our data demonstrate that biallelic Tbx1 is required for patterning of a subset of neck muscles and implicate CPM in muscle patterning during pharyngeal development.

We note that several muscles with CPM MCTs retain the normal pattern in heterozygous and conditional heterozygous mutants of Tbx1, possibly because of differential dependency on Tbx1 in different muscles. Tbx1 is downregulated early in medial posterior mesoderm of the embryonic pharynx (De Bono et al., 2018), and other genes may play a patterning role for the medial pharyngeal components such as intrinsic laryngeal muscles. Alternatively, as some muscles have MCT from the NCC lineage, NCC-derived MCT may compensate for impaired CPM-derived MCT in mutant embryos. Although a previous study reported a hyoid bone defect on conditional null mutation of Tbx1 in Wnt1-Cre derivatives, no muscular phenotypes were noted (Funato et al., 2015). Further analyses of lineage-specific Tbx1 mutants, as well as comparison of the gene regulatory networks operating in CPM and NCC-derived CT, will be required to elucidate the cellular basis of the patterning defect. Interestingly, although Mef2c-AHF-Cre is exclusively expressed in CPM, other enhancers from the Mef2c locus have been shown to drive expression in NCCs (Aoto et al., 2015; Hu et al., 2015).

Our identification of muscular patterning defects in the pharynx of Tbx1 heterozygous mutants extends the known haploinsufficient phenotypes in this mouse model. The majority of 22q11.2DS patients, haploinsufficient for a multigene deletion including TBX1, show velopharyngeal insufficiency, facial asymmetry, and feeding and speech problems (Rommel et al., 1999; Eicher et al., 2000; Dyce et al., 2002; McDonald-McGinn et al., 2015). A study further showed that 22q11.2DS patients frequently suffer from dysphagia and feeding difficulties, independent from palatal and cardiac anomalies, and that pharyngeal opening by movement of the hyoid bone is impaired for unknown reasons (Eicher et al., 2000). Given the importance of the hyoid bone for proper function of the pharyngeal apparatus (McFarland, 2009), the patterning defects observed in the sternocleidomastoid and omohyoid muscles of Tbx1 heterozygous mutant mice suggest a potential explanation for pharyngeal dysfunction in 22q11.2DS patients. Indeed, ectopic connections of these muscles to the greater horn of hyoid bone is likely to affect the mobility of the bone. Further characterization of these muscle patterning defects will provide mechanistic insights into the underlying pathology. Moreover, if similar defects underlie feeding and speech difficulties in 22q11.2DS patients, they could be targets for surgical intervention.

The pharynx has undergone various morphological changes in vertebrate history (Romer and Parsons, 1977). Remarkable is the remodeling of gills and the gain of laryngopharyngeal skeletons in the tetrapod lineage as it adopted a terrestrial lifestyle (Coates et al., 2008). As it has been long assumed that skeletons in the tetrapod pharynx are homologous with gill skeletons in fishes (Gegenbaur, 1892; Goodrich, 1930; Romer and Parsons, 1977), the embryonic source of those skeletons, NCCs, have been the major subject of analysis. Many studies have demonstrated the contribution of NCCs to the pharyngeal apparatus in different vertebrate taxa (Schilling and Kimmel, 1994; Köntges and Lumsden, 1996; Olsson and Hanken, 1996; Le Douarin et al., 2004; Matsuoka et al., 2005; Noden and Trainor, 2005; Martin et al., 2009; Kague et al., 2012; Mongera et al., 2013). Among the pharyngeal structures we identify as having CPM origins, such as medial pharyngeal cartilages and MCTs, some have been previously thought to derive from NCCs or LPM (Noden, 1988; Matsuoka et al., 2005). However, genetic lineage tracing using Wnt1-Cre shows a minimal contribution of NCCs to medial pharyngeal cartilages and MCTs (Fig. 1, Fig. S5; Tabler et al., 2017; Heude et al., 2018). In contrast, a mesodermal contribution to the medial pharyngeal skeleton has been suggested in avians and recently shown using Mesp1-Cre lineage tracing in the mouse (Noden, 1988; Tabler et al., 2017). Consistent with heterogeneous origins of pharyngeal skeletal elements, several studies have noted distinct outcomes in the posterior pharyngeal skeleton after genetic perturbations (see discussion in Tabler et al., 2017). However, Mesp1 is expressed in cranial and anterior somitic mesoderm as well as LPM; our results using two different Cre lines now identify CPM as the source of medial skeletons, CT and mesenchymal cells in the mammalian pharynx.

Our comparison of Mef2c-AHF-Cre and Wnt1-Cre genetic lineage contributions shows that NCCs give rise to lateral skeletal elements, CTs and mesenchyme in the pharynx, in a complementary distribution to medially localized CPM derivatives. These findings are consistent with lineage analyses in axolotl, showing that NCCs contribute to the lateral visceral skeletons, whereas the second basibranchial skeleton, the mid-ventral element of the pharynx, originates from the mesoderm associated with pharyngeal muscles and outflow tract (Davidian and Malashichev, 2013; Sefton et al., 2015). Furthermore, based on histological observations in frogs and salamanders, Edgeworth (1935) proposed that mesodermal cells contiguous with the dorsal pericardial wall spread medially around the posterior pharyngeal endoderm to form arytenoid, cricoid and tracheal cartilages, consistent with our genetic evidence that CPM and NCC derivatives distribute in a mediolateral complementary pattern in the pharynx and posterior PA of mouse embryos. The cellular distribution at this level is unique and not found in anterior PAs in mice, nor in fish PAs, suggesting that it is a synapomorphy of the tetrapod lineage (Fig. S8A) (McCauley and Bronner-Fraser, 2006; Abrial et al., 2017; Adachi et al., 2018). Our finding that arch 6 develops differently to anterior arches is consistent with a recent study by Poopalasundaram et al. (2019) showing that posterior PA development in amniotes is markedly distinct from that in other vertebrate clades. However, although Poopalasundaram et al. suggest that myogenesis is suppressed in posterior PAs, we observed MYOD protein accumulation in posterior CPM in the anlagen of branchiomeric muscles derived from arches 4-6 from early developmental stages. The unique cellular distribution of CPM and NCCs in the posterior PA region has important implications for the evolution of the vertebrate pharynx. CPM-derived medial pharyngeal skeletons likely represent evolutionary novelties in the tetrapod lineage, whereas the hyoid bone and lateral thyroid cartilage in mammals are comparable with NCC-derived pharyngeal skeletons in non-tetrapod vertebrates (Fig. S8A,B). Moreover, rearrangement of the distribution of mesenchymal cell populations in the posterior PAs may have contributed to the morphological transition of pharyngeal skeletons during early tetrapod history.

Transgenic lines used in this study have been described previously: Mef2c-AHF-Cre (Verzi et al., 2005), Wnt1-Cre (Danielian et al., 1998), Tbx1-Cre (Huynh et al., 2007), Rosa26-nlacZ (Soriano, 1999), Rosa26-YFP (Srinivas et al., 2001), Rosa26-mTmG (Muzumdar et al., 2007), Tbx1^+/− (Jerome and Papaioannou, 2001) and Tbx1^flox/flox (Xu et al., 2004). Mice were maintained on a mixed background and genotyped using PCR. Noon of the day of the vaginal plug observation was defined as E0.5 and embryos were staged according to Kaufman (1992). Animal experiments were conducted in agreement with National and European laws and approved by the Ethics Committee for Animal Experimentation of Marseille and the French Ministry for National Education, Higher Education and Research.

Embryos were fixed in 4% paraformaldehyde (PFA) and washed in phosphate buffered saline (PBS). Fixed embryos were embedded in paraffin (Paraplast X-TRA, P3808-1KG, Merck Sigma-Aldrich) and sectioned at 10 μm thickness. Sections were stained with hematoxylin and eosin solutions (H&E; Harris Hematoxylin, HHS32-1L and Eosin Y, HT110132-1L, Merck Sigma-Aldrich) and then imaged using an Axio Zoom.V16 microscope with an Axiocam 512 color camera (ZEISS).

Immunofluorescence on paraffin sections was performed as described previously (De Bono et al., 2018). Antigen unmasking solution [10 mM Tris-HCl (pH 9.5), 1 mM EDTA and 0.05% Tween20] was used to reveal epitopes. The following primary and secondary antibodies were used: chicken anti-GFP (1/500, ab13970, Abcam and 1/500, GFP-1020, Aves), mouse anti-MHC1 (1/50, MF20, Developmental Studies Hybridoma Bank), rabbit anti-TBX1 (1/100, Ls-C31179, LSBio), rabbit anti-TCF4 (1/150, C48H11, Cell Signaling Technology), rat anti-MYOD (1/200, 39991, Active Motif), Alexa donkey anti-chicken 488, Alexa donkey anti-mouse 568 and Alexa donkey anti-rabbit 647 (1/500, Thermo Fisher Scientific). The TSA Plus system was employed to detect MYOD (NEL741001KT, PerkinElmer). For paraffin sections of Wnt1-Cre;RYFP and Mef2c-AHF-Cre;RYFP at E10.5, embryos were first incubated in low melting point paraffin and then mounted in paraffin (Avé et al., 1997) (1071501000, Merck Millipore). Immunostained sections were counterstained with Hoechst 33258 (1/1000, 861405, Merck Sigma-Aldrich) and imaged using an Axio Zoom.V16 microscope with an Axiocam 512 color camera and Axio Imager APO Z1 (ZEISS).

The method of in situ hybridization on paraffin sections has been described previously (De Bono et al., 2018). Cre expressions of sections at pharyngeal, shoulder and heart levels were tested on the same microscope slide. The hybridization was performed at 70°C (n=4), 65°C (n=3) and 60°C (n=3). For signal detection, anti-Digoxigenin-AP antibodies (1/2000, 11093274910, Roche) and BM purple (11442074001, Roche) were used, and after color development, sections were counterstained with eosin and imaged using Axio Zoom.V16 with Axiocam 512 color camera (ZEISS).

Embryos were fixed in 4% PFA for 20 min at 4°C and stained in X-gal staining solution (0.1% X-gal, 2 mM MgCl₂, 0.01% deoxycholate, 0.02% Nonidet P40, 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide and PBS) for 2-3 h at 37°C. Embryos were then re-fixed in 4% PFA, washed in PBS, and imaged using Axio Zoom.V16 with Axiocam 512 color camera (ZEISS).

We are grateful to Kelly team members and E. Heude, F. Sugahara, G. Comai, J. Lemberg, J. Yamaguchi, S. Kuroda, W. Takagi and P. Janvier for discussions, and to A. Grimaldi, F. Helmbacher and S. Tajbakhsh for critical reading of the manuscript.

The peer review history is available online at https://dev.biologists.org/lookup/doi/10.1242/dev.185256.reviewer-comments.pdf