Head muscle progenitors in pharyngeal mesoderm are present in close proximity to cells of the second heart field and show overlapping patterns of gene expression. However, it is not clear whether a single progenitor cell gives rise to both heart and head muscles. We now show that this is the case, using a retrospective clonal analysis in which an nlaacZ sequence, converted to functional nlacZ after a rare intragenic recombination event, is targeted to the αc-actin gene, expressed in all developing skeletal and cardiac muscle. We distinguish two branchiomeric head muscle lineages, which segregate early, both of which also contribute to myocardium. The first gives rise to the temporalis and masseter muscles, which derive from the first branchial arch, and also to the extraocular muscles, thus demonstrating a contribution from paraxial as well as prechordal mesoderm to this anterior muscle group. Unexpectedly, this first lineage also contributes to myocardium of the right ventricle. The second lineage gives rise to muscles of facial expression, which derive from mesoderm of the second branchial arch. It also contributes to outflow tract myocardium at the base of the arteries. Further sublineages distinguish myocardium at the base of the aorta or pulmonary trunk, with a clonal relationship to right or left head muscles, respectively. We thus establish a lineage tree, which we correlate with genetic regulation, and demonstrate a clonal relationship linking groups of head muscles to different parts of the heart, reflecting the posterior movement of the arterial pole during pharyngeal morphogenesis.
All skeletal muscles are not identical. Notably, trunk and head muscles differ in a number of important respects. They derive from different embryonic regions: trunk muscles form from paraxial mesoderm of the somites, whereas most head muscles are formed from unsegmented cranial paraxial mesoderm (Noden and Francis-West, 2006). Trunk and head muscles also have distinct gene regulatory programmes such that Pax3, which is an important upstream regulator of trunk myogenesis (Buckingham and Relaix, 2007), is not expressed in head mesoderm, whereas other genes such as Tbx1 or Pitx2 (Grifone and Kelly, 2007) are upstream regulators of skeletal myogenesis in the head, but not the trunk. Differences can already be observed at the onset of gastrulation when cells that give rise to cranial paraxial mesoderm ingress through the streak before cells that give rise to somitic paraxial mesoderm (Kinder et al., 1999; Parameswaran and Tam, 1995; Tam et al., 1997).
Craniofacial muscles can be classified into different groups: somite-derived neck and tongue muscles, branchiomeric muscles that are involved in mastication, facial expression and function of the larynx and pharynx, and extraocular muscles that control eye movement. Branchiomeric muscles are derived from pharyngeal mesoderm, which includes both lateral splanchnic and paraxial mesoderm. This forms the mesodermal core of the branchial arches and is then repositioned within the head during the morphogenetic movements that accompany craniofacial development. Caudal branchial arches give rise to laryngeal and pharyngeal muscles and, in mammals, the first and second arches give rise to progenitors of jaw and facial expression muscles, respectively (Larsen et al., 2009). Extraocular muscles derive mainly from prechordal mesoderm, although it is not clear whether there is also a contribution from paraxial mesoderm (Noden and Francis-West, 2006). However, extraocular muscles are clearly subject to different genetic regulation from branchiomeric muscles. Transcription factors such as Tbx1, MyoR or capsulin, required for branchiomeric myogenesis, do not play a role in their development (Kelly et al., 2004; Lu et al., 2002). Furthermore, the hierarchy of myogenic determination factors of the MyoD family differs between extraocular and branchiomeric muscles (Sambasivan et al., 2009). However, Pitx2 function is required for both extraocular and first-branchial-arch-derived muscles (Dong et al., 2006).
Classic fate mapping and lineage tracing experiments had indicated a close relationship between progenitors for cranial paraxial mesoderm and mesoderm that will form the heart, which ingress through the streak at the same stage (Kinder et al., 1999). In addition, grafting experiments showed that cardiac and cranial paraxial mesoderm progenitors are present in the same region (Parameswaran and Tam, 1995; Tam et al., 1997). In single-cell labelling experiments in the epiblast, 60% of clones contributed to more than one structure, including cranial paraxial mesoderm and lateral mesoderm from which the heart is derived (Buckingham et al., 1997; Lawson and Pedersen, 1992). The first heart field arises from lateral splanchnic mesoderm and forms the primitive heart tube. It is now established, from experiments in chick and mouse embryos, that pharyngeal splanchnic mesoderm, which constitutes the second heart field (SHF), contributes to the growth of the developing cardiac tube (Buckingham et al., 2005). In the mouse embryo, cells from this field form the outflow tract myocardium and also contribute to the right ventricle, as well as to the venous pole of the heart. The SHF is regulated by a genetic network that includes genes such as Islet1, the expression of which marks these cardiac progenitor cells (Cai et al., 2003). Pharyngeal mesoderm, contributing to the outflow region of the heart, is contiguous with that contributing to the branchiomeric muscles, as shown by dye labelling experiments of the mesodermal core of the first two branchial arches in mouse and chick embryos (Kelly et al., 2001; Nathan et al., 2008). Furthermore, common gene expression profiles are observed in these cells (Bothe and Dietrich, 2006; Grifone and Kelly, 2007; Tzahor, 2009), with a proximal-distal gradient of gene expression within the mesodermal core, corresponding to markers associated with branchiomeric rather than SHF progenitors (Nathan et al., 2008). Tbx1 provides an example of a regulatory gene that is implicated in branchiomeric myogenesis (Kelly et al., 2004) and also plays an important role in the formation of the cardiac outflow tract (Xu and Baldini, 2007). Genetic tracing experiments with Islet1-Cre or Mesp1-Cre activated in precardiac mesoderm, show expression of the conditional Rosa26 reporter in branchiomeric muscles as well as in the heart, indicating that these skeletal muscles also derive from cells that had expressed Islet1 or Mesp1. Dye labelling and manipulation of signalling pathways in explants of cranial mesoderm at earlier stages in the chick embryo also show overlapping cardiac and skeletal muscle potential (Tirosh-Finkel et al., 2006).
Mesodermal cells in the pharyngeal region can thus contribute to heart and head muscle, and branchiomeric muscle progenitors express genes that characterize the SHF, prior to entering the myogenic programme. However, it is not clear whether a single progenitor cell gives rise to descendants in both types of striated muscle or whether progenitors are initially intermingled. Dye labelling of populations of cells and genetic tracing experiments do not distinguish between these possibilities. We have used retrospective clonal analysis to investigate lineage relationships between head and heart muscle. We show that there are two branchiomeric muscle lineages, the first of which also contributes to extraocular muscles. The first branchiomeric muscle lineage gives rise to the temporalis and masseter muscles, which are first-arch derivatives (Larsen et al., 2009), and also unexpectedly contributes myocardial cells to the right ventricle. The second branchiomeric muscle lineage gives rise to muscles of facial expression that are second-arch derivatives and also contributes to myocardium at the arterial pole of the heart. Within this second lineage, a further subdivision is observed between myocardium at the base of the pulmonary trunk and the aorta. These sublineages contribute to left or right muscles of facial expression, respectively. There is therefore a clonal relationship between head and heart muscle progenitors, with further sublineages that form a lineage tree, as discussed here.
MATERIALS AND METHODS
The Mlc1v-nlacZ-24 transgenic line (Kelly et al., 2001), the T4 transgenic line (Biben et al., 1996) and the αc-actinnlaacZ1.1/+ mouse line (Meilhac et al., 2003) have been described previously. Mef2c-AHF-enhancer-Cre males (Verzi et al., 2005) were crossed to the Rosa26R-nlacZ reporter line (J.-F.N., E. Tzouanacou and V. Wilson, unpublished).
X-gal staining, immunochemistry and histology
Mlc1v-nlacZ-24 and Mef2c-AHF-enhancer-Cre;Rosa26R-nlacZ embryos were sectioned using a cryostat. Immunochemistry was performed with MyoD (Dako) and β-galactosidase (J.-F.N.) antibodies.
Retrospective clonal analysis
A total of 627 embryos at embryonic day (E) 14.5 had been collected in a previous study (Bajolle et al., 2008) and 1596 additional embryos were collected here. Statistical analyses were carried out on the newly collected embryos only, as some hearts of the first series had been sectioned for other purposes such that the collection was no longer complete and therefore no longer fulfilled the random criterion required for statistical analysis. Most of the E14.5 αc-actinnlaacZ1.1/+ embryos present multiple clusters of β-galactosidase-positive cells or fibres and it is therefore essential to establish by statistical analysis whether labelled cells derive from a single recombination event and are thus clonally related. Because recombination is random, there is a low probability that such an event occurs in the same location a second time and, therefore, a cluster probably contains clonally related cells (Meilhac et al., 2004; Meilhac et al., 2003). We distinguished between large and small clusters of labelled cells and fibres as large clusters are derived from an earlier recombination event than small clusters and are therefore more interesting for our study. Large clusters in skeletal muscles or in myocardium were defined as clusters with more than 10 fibres or 10 cells labelled, respectively. Such large clusters in head muscles were seen in 1.8% of embryos, whereas small clusters occurred at a much higher frequency (17.5%).
We estimated the expected frequency of double recombination events in two different regions, which, according to the law of independent probabilities, is equal to the product of the frequency of labelling in each region (Tables 1, 2). In order to decide whether the observed frequency of common labelling in two distinct regions, for instance branchiomeric muscles and heart myocardium, was consistent with the expected frequency, we performed a statistical test. We have used the non-parametric Fisher's exact test that allows us to work with small numbers of labelled embryos. The null hypothesis is that the labelling in both regions results from two independent events. When the P-value is less than 0.05, the null hypothesis can be confidently rejected, leading to the conclusion that the labelling probably derives from a single recombination event.
Expression of the Mlc1v-nlacZ-24 transgene reveals early continuity between splanchnic mesoderm and the mesodermal core of the first and second branchial arches
In the Mlc1v-nlacZ-24 transgenic line, in which reporter gene expression is driven by Fgf10 regulatory elements (Kelly et al., 2001), the outflow tract and part of the right ventricle of the developing heart are β-galactosidase-positive (Fig. 1A-C). At E8.5, X-gal staining was seen in the mesodermal core of the developing arches (Fig. 1A,B), where Fgf10 is also expressed (Kelly et al., 2001). This expression domain is contiguous with β-galactosidase-positive cells in the splanchnic mesoderm of the SHF and its myocardial derivatives at the arterial pole of the heart. This continuity was maintained at E9.5 when transgene expression was observed in the mesodermal core of the arches where the branchiomeric skeletal muscle programme initiates (Fig. 1C). Mlc1v-nlacZ-24 expression thus illustrates the continuity between myocardial and skeletal muscle progenitor cells in its expression domain in pharyngeal mesoderm. As shown in Fig. 1D, the Mlc1v-nlacZ-24 transgene continued to be expressed in branchiomeric head muscles at later stages.
A retrospective clonal analysis at E14.5
In order to examine a possible lineage relationship between myocardium at the arterial pole of the heart and branchiomeric head muscles, we carried out a retrospective clonal analysis (Bonnerot and Nicolas, 1993) using the αc-actinnlaacZ1.1/+ line (Meilhac et al., 2003). In this line, the reporter has been introduced into an allele of the α-cardiac actin gene, which is expressed throughout the myocardium (Fig. 1E) and also in all developing skeletal muscles (Fig. 1F) (Biben et al., 1996; Sassoon et al., 1988). This gene therefore provides an appropriate endpoint for clonal analysis of these tissues. The retrospective clonal approach avoids preconceived ideas about lineage relationships and is based on the analysis of a collection of embryos rather than isolated examples. It employs an nlaacZ reporter in which a duplication introduces a stop codon into the β-galactosidase coding sequence, rendering it non-functional. A rare intragenic recombination event results in random removal of the duplication, independently of gene expression, so that the reporter now makes functional β-galactosidase when the gene into which it is integrated is expressed. Cells that are descended from a progenitor that has undergone such a recombination event give rise to labelled cells that are clonally related (see Fig. S1 in the supplementary material). In the case of skeletal muscle, where cells fuse to form fibres, this is assessed as fibres with labelled nuclei.
At E14.5, head muscle primordia have formed and αc-actin is strongly expressed in these skeletal muscles, as well as in the heart (Fig. 1E,F). This timepoint has therefore been selected for the retrospective analysis using the αc-actinnlaacZ1.1/+ line. At this stage, all embryos scored contained β-galactosidase-positive cells. This frequency means that multiple recombination events have probably taken place per embryo to convert the nlaacZ sequence to a functional nlacZ reporter. Although multiple clusters of β-galactosidase-positive fibres were present in body muscles, labelling was much more rare in head muscles (19.3%), which represent a small fraction of the total musculature. Most of the embryos with such labelling in head muscles (77.7%) only had one or two β-galactosidase-positive fibres. As small clusters of labelled fibres probably correspond to more recent recombination events, which are less informative about the progenitor cell pool, we subsequently focused our analysis on clusters containing more than 10 labelled fibres. Only 39 out of the 2223 embryos (1.8%) had clusters of more than 10 β-galactosidase-positive fibres in head muscles (Table 1). Therefore, the probability that more than 10 labelled fibres per embryo derive from multiple recombination events is very low (see Fig. S2 in the supplementary material). The calculation of probability is an essential feature of this clonal approach; the statistical demonstration, based on observation of many embryos, that a double recombination event is very improbable leads to the conclusion that labelled cells descend from a single recombined progenitor and are therefore clonally related.
Two subpopulations of progenitor cells contribute to muscles derived from mesoderm of the first two branchial arches
Examples of αc-actinnlaacZ1.1/+ embryos with β-galactosidase activity in head muscles are shown in Fig. 2. Two distinct distributions of labelled fibres were observed in subsets of branchiomeric muscles, as illustrated in Fig. 2A,B compared with Fig. 2C,D. These results are summarized in Fig. 2G,H. Fourteen embryos had β-galactosidase-positive fibres that were restricted to the temporalis and masseter muscles (category I). In 21 embryos, these muscles were not labelled but β-galactosidase-positive fibres were observed in muscles of facial expression, as indicated (category II). Only four embryos had β-galactosidase-positive fibres in both categories of muscles. This labelling is highly unlikely to result from two independent recombination events as the expected frequency of double recombination events in category I and category II muscles is very low (Table 2) and labelled cells are therefore clonally related. The number of muscles labelled increases with the number of β-galactosidase-positive fibres, in particular for category II, which includes a greater number of distinct muscles and involves widespread migration of branchial-arch-derived cells. In addition to the number of muscles and their size, which affects the extent of labelling, the number of labelled cells in the clone reflects the number of divisions and hence timing of the recombination event in the progenitor cell. In embryo number 1304, for example, recombination probably occurred in a more recent progenitor that gave rise only to the masseter muscle, whereas in most cases a common progenitor gives rise to both category I type muscles (masseter and temporalis). In the case of category II muscles, there is some indication that the zygomaticus, buccinator and auricularis muscles are more frequently labelled; however, these muscles are larger than other category II muscles. In general, the distribution of β-galactosidase-positive fibres between muscles indicates a dispersion of cells after the recombination event. The four embryos that had labelling in both categories (Fig. 2E,F,H) had large numbers (>40) of β-galactosidase-positive fibres, with two of the embryos (2880, 1779) also having a very large number of β-galactosidase-positive fibres throughout the body, indicative of a very early recombination event. The presence of two distinct labelling patterns and the fact that both were seen only in embryos with many labelled cells (+ or ++), indicates that category I and II derive from early common progenitor cells that segregated into distinct lineages.
Most embryos showed labelling in muscles on either the left or right side of the head (Fig. 2H), with no particular bias. Some embryos had head muscle labelling on both sides. In four cases (385, 2084, 284, 1436) this was restricted to category II muscles. In two other examples (2080, 2880), one side of the embryo had extensive labelling in category II or both category I and II muscles, whereas only category I muscles were labelled on the other side (Fig. 2F). In these examples, a common progenitor for both lineages must have given rise to asymmetrically distributed descendants that were only of category I type on the right side. The two very extensively labelled embryos (1779, 2688) had complete labelling on both sides (Fig. 2E). Bilateral labelling indicates that the recombination event preceded the onset of gastrulation, whereas monolateral labelling can arise before or after gastrulation and cannot be used as a criterion to date the clones (Lawson et al., 1991; Selleck and Stern, 1991).
Extraocular muscles, which lie in close proximity to the eye, are thought to derive from both prechordal and paraxial mesoderm (Evans and Noden, 2006; Noden and Francis-West, 2006). We have observed ten embryos (out of 2223), which had β-galactosidase-positive fibres in extraocular muscles. This includes three embryos that also show labelling in both category I and II muscles (2880, 1779, 2688). In three other cases (3243, 394, 2901), there was labelling only in category I branchiomeric muscles (Fig. 3A-C). Given the high number of embryos scored and the very low number of embryos with labelling in these small muscles, it is very improbable that more than one recombination event had occurred, leading to the conclusion that there is a clonal relationship between these muscle groups (Fig. 3D; P=0.0004). Embryos 3243 and 2880 showed labelling only in a subset of extraocular muscles (dorsal rectus muscles for 3243 and dorsal rectus and lateral rectus muscles for 2880); however, the four other embryos with labelling in category I muscles showed labelling in all six extraocular muscles as shown in Fig. 3B,C. No link was observed with category II muscles (P=0.96).
Other branchiomeric muscles situated deep within the embryo, including those of the pharynx and larynx derived from posterior branchial arches, were not analysed in detail in our analysis, which concentrated on the muscles presented in Fig. 2H. However, we also scored additional muscles located underneath the mandible in the context of category I and category II labelling (see Fig. S3 in the supplementary material).
We also examined whether embryos with β-galactosidase-positive fibres in branchiomeric head muscles also showed labelling in muscles of the trunk and limbs that are derived from somites (Table 1). We conclude that there is an early segregation of branchiomeric and somitic muscle progenitor cells (Table 2; see Fig. S4 in the supplementary material).
Branchiomeric head muscles share common progenitors with right ventricular and arterial pole myocardium
We next examined how many of the embryos with category I or II labelling in head muscles also showed labelling in myocardial derivatives of the anterior SHF, namely myocardium of the right ventricle and at the base of the pulmonary trunk and aorta, described as the arterial pole of the heart. The frequency of embryos with labelling in this myocardium is presented in Table 1. The left ventricle, which does not derive from the SHF (Meilhac et al., 2004), is presented as a negative control.
None of the embryos in category I had any β-galactosidase-positive cells in the arterial pole of the heart; however, seven of them had labelled clusters (>10 cells) in the right ventricle. A summary for all embryos with head muscle labelling is given in Fig. 4A and an example is shown in Fig. 4B,B′. The low probability of double recombination events in the two regions reported in Table 2 suggests that labelling in category I muscles and right ventricular myocardium probably results from a single event. Furthermore, statistical analysis, using the Fisher's exact test and based on the figures shown in Fig. 4C, strongly supports the conclusion that the β-galactosidase-positive cells in the right ventricle and fibres in category I skeletal muscles arise from a common progenitor (P=3×10−5).
We then examined embryos with category II head muscle labelling and found in this case that many of them also had β-galactosidase-positive cells in the arterial pole of the heart (12/21). The results are summarized in Fig. 4A and an example is shown in Fig. 4E,E′. Statistical analysis, using the Fisher's exact test and based on the numbers shown in Fig. 4F, supports the conclusion that the β-galactosidase-positive cells in arterial pole myocardium and fibres in category II skeletal muscles arise from a common progenitor (P=9×10−5). In some cases, labelling in the arterial pole extended into the right ventricle, but no labelling of only the right ventricle was observed in category II embryos.
Category II labelling in left or right head muscles correlates with labelling in myocardium at the base of the pulmonary trunk or aorta
By E14.5, myocardium at the arterial pole of the heart was located at the base of the pulmonary trunk or aorta. When we looked more closely at embryos with β-galactosidase-positive cells in this region, as well as labelled fibres in the head, we distinguished labelling in one or both of these arteries (Fig. 5A″,B″,C″). Examination of skeletal muscle labelling had indicated that most embryos had β-galactosidase-positive fibres on either the left or right side of the head. We observed a striking correlation between left or right labelling of category II skeletal muscles and labelling of myocardium at the base of the pulmonary trunk or aorta, respectively. In cases where both arteries had β-galactosidase-positive cells, labelling was predominantly seen on both sides of the head (Fig. 5A-D). This was observed in embryos with large numbers of labelled fibres in head muscles. The significance of this observation was examined and validated by phylogenetic tools to generate the tree shown in Fig. S5 in the supplementary material. This result indicates that there is a common progenitor for pulmonary trunk myocardium and category II muscles on the left side of the head, whereas those on the right side of the head share a common progenitor with myocardium at the base of the aorta.
Mef2c-AHF-Cre genetic tracing shows differences in the two head muscle lineages
We next examined the descendants of progenitors in which the Mef2c-AHF-enhancer had been activated. This enhancer marks progenitors of the outflow tract and right ventricular myocardium (Verzi et al., 2005), as well as some branchial-arch-derived muscles (Dong et al., 2006).
We crossed the Mef2c-AHF-enhancer-Cre line with a Rosa26R-nlacZ reporter line. At E10.5, labelled cells were found in the mesodermal core of the first and second branchial arches, but also in the outflow tract, right ventricular myocardium and mesoderm behind the heart tube (Fig. 6A-B′). However, within the branchial arches, labelled cells were found only in the more distal part of the second branchial arch (Fig. 6A). Sections show that although in the first branchial arch there was overlapping expression of β-galactosidase and MyoD, β-galactosidase-positive cells were mainly negative for MyoD in the second branchial arch (Fig. 6B,B′). This indicates that positive cells in the second branchial arch have not adopted a myogenic fate. In keeping with this, at E14.5, labelled cells were found only in category I head muscles (Fig. 6C), as well as in the right ventricular and outflow tract myocardium as expected. We propose that first branchial arch head and heart derivatives are marked by the activation of this enhancer, whereas in the second arch, activation is restricted to cells giving rise to cardiac derivatives. This result reveals molecular differences in the regulation of skeletal myogenic progenitor cells in the first and second arch.
We have shown the lineage relationships between head muscles and second heart field derivatives, as summarized in Fig. 6D. This retrospective clonal analysis demonstrates that branchiomeric skeletal muscle and SHF-derived myocardium derive from common progenitors and that there are distinct lineage and sublineage relationships within this framework. Notably, we show clonality between skeletal muscles derived from the first branchial arch and right ventricular myocardium. This unexpected finding, together with clonality between outflow-tract-derived myocardium at the base of the great arteries and second-branchial-arch-derived muscles, concords with the posterior movement of the arterial pole of the heart during pharyngeal morphogenesis.
Two non-somitic head muscle lineages
Skeletal muscles of the head fall into two categories, both clonally distinct from somitic muscles, indicating early segregation of these different myogenic lineages, as expected from fate mapping experiments (Parameswaran and Tam, 1995; Tam et al., 1997). First, there are progenitors that contribute to temporalis and masseter muscles, derived from the first branchial arch (Larsen et al., 2009). We also detected a significant clonal relationship with extraocular muscles. This comprised all the extraocular muscles, not just the dorsal oblique and lateral rectus, previously proposed to be derived from mesoderm at the level of the arches (Noden and Francis-West, 2006). We observed two clones in which only a subset of extraocular muscles are labelled (dorsal rectus or dorsal and lateral rectus), suggesting differences in the timing of progenitor segregation. The clonality between extraocular and first branchial arch muscles indicates that paraxial mesoderm, as well as prechordal mesoderm, contributes to all these muscles. This link between first branchial arch muscle derivatives and extraocular muscles is also observed at the level of Pitx2 function (Dong et al., 2006). The clonal relationship with right ventricular myocardial cells indicates a contribution from a progenitor for lateral splanchnic, as well as paraxial pharyngeal, mesoderm. Pitx2 also marks cardiac progenitors; however, its dynamic expression within the myocardium complicates the interpretation of genetic tracing experiments (Franco and Campione, 2003). We also identified a second category of muscles, mainly involved in facial expression, that arise from the second branchial arch (Larsen et al., 2009). First and second branchial arch muscles are therefore derived from distinct lineages.
The progenitor cells that give rise to right or left head muscles reflect right-left segregation. Clones that contribute to both right and left head muscles must derive from a progenitor that precedes bilateralisation of the mesoderm prior to gastrulation (Lawson et al., 1991). A spatial boundary between mesoderm of the first and second branchial arches has been demonstrated at E8.5 by orthotopic grafts of cranial paraxial mesoderm (Trainor et al., 1994). We now propose that the cell lineage segregation between the mesoderm of the first two arches has already taken place by the time of gastrulation.
Branchiomeric muscles share common progenitors with right ventricular or arterial pole myocardium
Our data reveal a clonal relationship between branchiomeric craniofacial muscles and myocardial derivatives of pharyngeal mesoderm in the SHF. From previous retrospective clonal analyses, we know that the anterior SHF contributes to outflow tract and right ventricular myocardium (Meilhac et al., 2004). The Mlc1v-nlacZ-24 transgenic line shows early continuity between Fgf10-expressing cells in the SHF and the mesodermal core of the first and second branchial arches. We now report that category I, first-branchial-arch-derived head muscles show a clonal relationship with cells in the right ventricle. The anterior boundary of the linear heart tube is positioned on the anterior-posterior axis at the level of the first branchial arch (Waldo et al., 2001). Mesoderm in the core of this arch may thus be continuous with the arterial pole of the heart at the time cells migrate into the early cardiac tube to contribute to the right ventricle. This result differs from the conclusions drawn for the chick embryo, where dye labelling experiments suggested that a population of cells contributing to the mesoderm of the first branchial arch, and subsequently to the masseter muscle, also contribute to the outflow tract of the heart (Tirosh-Finkel et al., 2006). This may reflect a difference in developmental timing between birds and mammals, but may also be due to dye labelling of a population of cells such that outflow tract progenitors are labelled at the same time as the mesodermal core of both branchial arches. Category II, second-branchial-arch-derived head muscles share a clonal relationship with myocardium at the base of the aorta and pulmonary trunk, which form by septation of the outflow tract. Consistent with these findings, as development proceeds, the heart tube moves posteriorly relative to the arches so that at a later stage its anterior boundary is aligned with the second branchial arch. Indeed, dye labelling of second branchial arch mesoderm in the mouse embryo showed subsequent localisation of labelled cells in outflow tract myocardium (Kelly et al., 2001).
Different head muscles, derived from the first branchial arch, did not show any distinction with respect to their clonal relationship to myocardium and, similarly, no such clonal distinction was observed for second branchial arch muscles.
Many large clones in the right ventricle and also in arterial pole myocardium do not show labelling in head muscles, suggesting that only a subset of myocardial progenitor cells share a clonal relationship with head muscles. There is no evident regionalization of such clones within the right ventricle. As expected, within the large clones that also contribute to head muscles, we see some that colonise both arterial pole and right ventricular myocardium, consistent with the contribution of the second myocardial cell lineage (Meilhac et al., 2004).
Estimation of the date of lineage segregation
The retrospective clonal analysis gives information about the age of a common progenitor cell, as well as the location of its descendants. To estimate the date of segregation of myocardium and head muscles, we have compared E14.5 embryos showing this distribution with the pattern of labelling in the heart of E8.5 αc-actinnlaacZ1.1/+ embryos (Meilhac et al., 2004). Most of these E14.5 αc-actinnlaacZ1.1/+ embryos have labelling that is restricted to the right ventricular (category I) or arterial pole (category II) myocardium. At E8.5, αc-actinnlaacZ1.1/+ embryos, which also show labelling that is restricted to the right ventricle or outflow region, have 3-10 β-galactosidase-positive cells, indicative of a recent recombination event. Therefore, we can propose that the lineage for muscles and for myocardium segregates late in the context of heart development. Embryos with labelling in head muscles derived from both first and second branchial arches showed a more extensive pattern of labelling in the heart, with more chambers labelled, including the left ventricle, suggesting that the recombination event predates the segregation of the first and second myocardial lineages (Meilhac et al., 2004).
Clonal distribution in the pulmonary trunk or aorta correlates with head muscle laterality
We show that clones in head muscles derived from the second branchial arch have β-galactosidase-positive cells in myocardium of the pulmonary trunk or the aorta. Segregation between the base of the aorta and pulmonary trunk was previously reported for clones in the hearts of αc-actinnlaacZ/+ embryos at E14.5, with regionalisation of clonal distribution noted in outflow tract myocardium at E10.5 (Bajolle et al., 2008). Gene and transgene expression patterns also show similar regionalisation, which extends to a subpopulation of cells in the SHF, indicating differences in transcriptional regulation in these myocardial subpopulations (Bajolle et al., 2008; Rochais et al., 2009; Theveniau-Ruissy et al., 2008). Strikingly, clones that colonised the pulmonary trunk contributed to second-branchial-arch-derived head muscles on the left side of the embryo, whereas clones in the aorta contributed to muscles on the right side of the head. Dye injections in the chick embryo have shown, however, that when right pharyngeal mesoderm is marked (Hamburger and Hamilton stage 13), this subsequently labels pulmonary trunk myocardium and a timecourse on labelled cells shows a spiralling movement of this mesoderm as it migrates into the outflow tract of the heart (Ward et al., 2005). In mouse embryos, dye injection into the outflow tract shows that it undergoes rotation between E9.5 and E10.5 (Bajolle et al., 2006). These observations might lead one to expect that there would be a correlation between clones in right head muscles and the pulmonary trunk. The converse correlation that we observe leads to a re-evaluation of the timing of addition of cells to the outflow tract. Previous experiments in the mouse embryo have shown that cells from the second branchial arch, labelled at E9.5, are found within the outflow tract at E10.5 (Kelly et al., 2001), suggesting that cells from the mesodermal core of the second arch are added to the arterial pole of the heart after E9.5 and therefore probably after the initiation of rotation. This late addition is consistent with dye labelling and observations on cardiac gene expression within the core mesoderm of the arches in the chick embryo, indicating that cells located within the arch contribute to the outflow tract (Nathan et al., 2008; Tirosh-Finkel et al., 2006).
Correlation with genetic tracing experiments using Cre lines and the Rosa26 reporter
Mesp1 is expressed early in the common progenitor of extraocular and branchiomeric muscles, as well as the myocardium (Harel et al., 2009; Saga et al., 2000). Islet1-Cre and Nkx2.5-Cre lines with the Rosa26R reporter also showed labelled head muscles and heart myocardium, but did not mark extraocular muscles (Harel et al., 2009). The onset of Islet1 (and Nkx2.5) expression is therefore too late or too limited to mark common progenitors of extraocular and first arch muscles that are revealed by retrospective clonal analysis. Differences between first branchial arch muscle derivatives might have been predicted based on genetic tracing with the Islet1-Cre (Nathan et al., 2008); however, this was not observed in the clonal analysis. Tbx1 also plays a crucial role in the two first branchial arch lineages. Tbx1 has been shown to be required for branchiomeric myogenesis (Grifone et al., 2008; Kelly et al., 2004) and is involved in the development of right ventricular and arterial pole myocardium (Xu and Baldini, 2007; Xu et al., 2005). Mutants for Tbx1 show defects in branchiomeric muscles and in arterial pole myocardium. We show here that the Mef2c-AHF-enhancer, activated in progenitors of the core mesoderm of the branchial arches and of the outflow tract and right ventricular myocardium (Dong et al., 2006; Verzi et al., 2005), distinguishes the first- and second-branchial-arch-derived muscle lineages. The Mef2c-AHF-enhancer-Cre is activated in the common progenitor of right ventricular myocardium and first-branchial-arch-derived head muscles, but marks only arterial pole myocardium and not head muscles derived from the second branchial arch, presumably reflecting a later activation of the enhancer in this lineage. This may correspond to the retinoic-acid-sensitive population that contributes later to the outflow tract described by Li et al. (Li et al., 2010).
Correlation of lineage segregation obtained by clonal analysis with the expression or role of transcriptional regulators (shown schematically in Fig. 6D) is delicate as genetic tracing experiments can be misleading and may only give a partial picture. Signalling pathways also classically affect cell fate choices. This is clearly demonstrated for the role of bone morphogenetic protein in promoting myocardial versus myogenic differentiation (Tirosh-Finkel et al., 2006); however, the spatiotemporal complexity of signalling inputs makes it difficult to extend this analysis to lineage domains.
In conclusion, this clonal analysis provides novel insights into the lineage relationships between head muscles and the heart, both in terms of contributions to different muscle structures or myocardial compartments and of the suggested timing of lineage segregation during early development (summarised in Fig. 6).
We thank C. Bodin for technical help. The work in M.B.'s laboratory was supported by the Pasteur Institute and the CNRS, with grants from the E.U. Integrated Projects ‘Heart Repair’ [LH SM-CT2005-018630 (also to R.G.K.)] and ‘CardioCell’ (LT2009-223372), which have funded J.-F.L.G. M.B. and R.G.K. also acknowledge the support of the Association Française contre les Myopathies. S.M.M. and R.G.K. are INSERM research scientists. F.L. benefits from a doctoral fellowship from the Ile de France region.
Note added in proof
A recent paper on the ascidian Ciona intestinals demonstrates a common precursor for head and heart muscles. (Stolfi et al., 2010).
Competing interests statement
The authors declare no competing financial interests.
Supplementary material for this article is available at http://dev.biologists.org/lookup/suppl/doi:10.1242/dev.050674/-/DC1
- Accepted June 16, 2010.
- © 2010.