The tubular heart differentiates from splanchnic mesoderm and consists of two concentric layers of cells: the outer myocardial layer and the inner endocardial layer, which are separated by cardiac jelly (de Jong et al., 1997). During cardiac looping, the atrial and ventricular myocardia are formed by local differentiation of the myocardium at the outer curvature (Christoffels et al., 2000; Moorman et al., 2000a). Subsequently, the original heart tube becomes septated by fusion of the cardiac cushions, the mesenchymal cap of the primary atrial septum, the spina vestibuli, and the ventricular septum (Wessels et al., 1996, 2000; Webb et al., 1998a). The mesenchymal components of the fused cushions become largely muscularized in later stages of development (de Jong et al., 1997; Mjaatvedt et al., 1999; van den Hoff et al., 1999, 2001; Moorman et al., 2000b). The walls of the pulmonary and caval veins become also muscularized (Jones et al., 1994; Ya et al., 1997; Millino et al., 2000; van den Hoff et al., 2001).
The mechanisms for this process, which can be envisioned to underlie myocardium formation in the intra- and extracardiac mesenchyme, are 1) differentiation (recruitment) of adjacent mesodermal cells into cardiac muscle cells, or 2) growth from the existing myocardium into flanking mesenchyme (myocardialization) (van den Hoff et al., 1999). Muscularization of the cardiac cushions was initially proposed to occur by myocardialization (van den Hoff et al., 1999). However, mesenchyme from the outflow tract (OFT) ridges can be induced to differentiate into cardiomyocytes in vitro (van den Hoff et al., 2001), which suggests that differentiation of mesenchyme into cardiac muscle cells contributes to the muscularization as well. Muscularization of the pulmonary and caval veins has been proposed to be the result of migration from existing atrial cardiomyocytes (Jones et al., 1994; Millino et al., 2000).
As a first attempt to gain more insight into the mechanisms underlying this muscularization in vivo, we analyzed the expression pattern of α-smooth muscle actin (α-Sma) in comparison to the myocardial markers sarcoplasmatic reticulum calcium ATPase (Serca2a), α-myosin heavy chain (Mhc), and β-Mhc during mouse development. The expression of smooth muscle proteins has been shown to precede expression of myocardial markers during the early phases of muscle formation (Ruzicka and Schwartz, 1988; Sugi and Lough, 1992; Miano and Olson, 1996; Ya et al., 1997). Here we report that the expression of α-Sma at the anterior pole of the heart, where it is now accepted that recruitment of mesodermal cells into the myocardial lineage takes place (Virágh and Challice, 1973; de la Cruz and Markwald, 1999; Mjaatvedt et al., 2001; Waldo et al., 2001; Kelly et al., 2001), precedes the expression of myocardial markers. This indicates that α-Sma can be used as an indicator for recruitment of mesodermal cells into the myocardial lineage after the formation of the initial linear heart tube. We then examined the α-Sma expression patterns in the intra- and extracardiac mesenchyme that would become muscularized. The observation that α-Sma precedes the expression of myocardial markers indicates that these mesenchymal cells differentiate into cardiac muscle cells and contribute to the caval and pulmonary myocardia, the smooth-walled atrial myocardium, the myocardial atrioventricular septum, and the myocardial outlet septum.
MATERIALS AND METHODS
FVB mice (Broekman, The Netherlands) were mated and inspected daily for vaginal plugs. The day on which the vaginal plug was found was considered day 1 of gestation. Embryos were fixed in ice-cold modified Amsterdam's fixative (40% methanol : 40% acetone : 20% water) for 4 hr, dehydrated in a graded alcohol series, and embedded in paraplast. Serial 7-μm sections were prepared and mounted onto poly-lysine-coated slides. After deparaffination and hydration in a graded alcohol series, endogenous peroxidase activity was blocked using 3% H2O2 in phosphate-buffered saline (PBS: 150 mmol/L NaCl and 10 mmol/L sodium phosphate, pH 7.4). Following a pretreatment for 30 min in TENG-T (10 mmol/L Tris, 5 mmol/L EDTA, 150 mmol/L NaCl, 0.25% (w/v) gelatin, and 0.05% (v/v) Tween-20, pH 8.0) to reduce nonspecific binding, the sections were incubated overnight with a polyclonal antibody directed against porcine Serca2a (Eggermont et al., 1990), and monoclonal antibodies against α-Sma (IMMH-2; Sigma, St. Louis, MO), myosin light chain 2v (Mlc2v) (Katus et al., 1982) (kindly provided by W. Franz, Lübeck, Germany), and α-Mhc and β-Mhc (Wessels et al., 1991). Antibody binding was visualized using the indirect unconjugated peroxidase-antiperoxidase technique as described previously (Wessels et al., 1990).
Cardiac muscle cell formation after the formation of the linear heart tube was recently described in chickens (van den Hoff et al., 1999; 2001), mice (Kruithof et al., 2003), humans (Lamers et al., 1995; Kim et al., 2001), and partially in rats (Ya et al., 1998). In this study we focus on the pattern of α-Sma expression in intra- and extracardiac mesenchyme, where myocardium formation takes place after formation of the linear heart tube in mice. To this end we performed a comprehensive immunohistochemical study on sections of staged mouse embryos ranging from embryonic day 10 (E10) to E17. Neighboring sections were stained for Serca2a, α-Mhc, and β-Mhc delineating myocardial from nonmyocardial cells, and for α-Sma, respectively. Serca2a is a useful myocardial marker because it stains the entire myocardium at all of the developmental stages analyzed, whereas antibodies directed against α-Mhc, β-Mhc, and Mlc2v stain only part of the myocardium (Fig. 1).
α-Sma Expression Pattern at the Level of the OFT (Figs. 2 and 3)
During the formation of the primary heart tube, the OFT is added anteriorly by recruitment between E9 and E11 of mouse development (Virágh and Challice, 1973; Kelly et al., 2001). At E10, the expression of α-Sma is not only present in the myocardium of the OFT, but also in the adjacent mesodermal cells (Fig. 2a–f) that will be recruited into the myocardial lineage during subsequent development (Virágh and Challice, 1973; Kelly et al., 2001). These data suggest that α-Sma identifies mesodermal cells that during subsequent development will be recruited into the myocardial lineage. At E12, when the recruitment of cardiomyocytes at the arterial pole is complete (Virágh and Challice, 1973; Kelly et al., 2001), α-Sma expression is not observed in the mesodermal cells flanking the OFT myocardium (Fig. 3b). At this stage, α-Sma-positive mesenchymal cells are identified in the aortico-pulmonary septum (APS), which develops from the aortic sac in a caudal direction and begins to fuse with the endocardial cushions (Fig. 3a and b). At E13, when the APS has fused with the endocardial cushions, α-Sma-positive mesenchymal cells are present in the endocardial cushions and reach toward the cardiomyocytes protruding into the mesenchyme (Fig. 3c and d). At E14, α-Sma-positive mesenchymal cells are visible in a large part of the mesenchymal septum, which becomes muscular, as shown by the expression of Serca2a (Fig. 3e and f). These α-Sma-positive mesenchymal cells are aligned to the muscularizing cardiomyocytes, which is not the case with the α-Sma-negative mesenchymal cells (compare the orientation of α-Sma-positive mesenchymal cells with the orientation of the α-Sma-negative mesenchymal cells in Figs. 3e and f). The α-Sma-positive mesenchymal cells have contact with the cardiomyocytes, which are typified by protrusions into the mesenchyme. Short color reactions that are used to visualize the Serca2a and α-Sma antibody binding reveal a gradient in the expression levels of both Serca2a and α-Sma. Myocardial cells protruding into the mesenchyme show a relative low level of Serca2a expression compared to the flanking myocardium (Fig. 3g). This difference in expression level is also seen in Mlc2v- and β-Mhc-stained sections, albeit less prominently (not shown). The cardiomyocytes expressing low levels of Serca2a display a relatively high level of α-Sma expression. This high level of α-Sma expression appears to extend into the flanking mesenchymal cells (Fig. 3h). At E17, the muscularization of the outlet septum is complete (not shown).
α-Sma Expression Pattern During Formation of the Sinus Venosus and Caval Myocardia (Fig. 4)
After formation of the linear heart tube, the myocardium is also added to the inflow tract, forming the sinus venosus. This myocardium formation proceeds in the extracardiac mesenchyme, forming the myocardial caval and pulmonary veins. (The α-Sma expression pattern during the formation of the pulmonary myocardium is described below.) At E11, cells forming the wall of the right and left sinus horns express α-Sma, whereas they show a low, patchy expression pattern of α-Mhc (Fig. 4a and b). At E12, the myocardial sinus venosus is almost completely formed and the myocardium extends to the pericardial border (Fig. 4c), although at the left side the myocardium does not completely reach the pericardial border (arrow in Fig. 4c). The α-Sma expression in the myocardium of the sinus venosus extends beyond the pericardial border in the nonmyocardial cells along the cardinal veins (Fig. 4d). At E14, when the α-Sma expression has become even broader along the caval veins, the first myocardial cells are observed in the wall of the caval veins outside the pericardial cavity (Fig. 4e and f). At E15, the proximal part of the wall of the left cranial caval vein is myocardial (Fig. 4g).
α-Sma Expression Pattern in the Dorsal Mesocardium and the Base of the Spina Vestibuli (Fig. 5)
At the venous pole, the dorsal mesocardium and the base of the spina vestibuli, which is the extension of the mesenchyme of the dorsal mesocardium into the lumen of the atrium, becomes muscularized from E12 to E14. During this muscularization the cardiomyocytes that are present in the dorsal mesocardium display a relatively low level of Serca2a expression (Fig. 5a) and a relative high level of α-Sma expression (Fig. 5b). In the spina vestibuli, α-Sma expression is seen in the mesenchyme beyond the myocardial protrusions (Fig. 5c and d). At E14, the dorsal mesocardium and the neighboring base of the spina vestibuli are both myocardial (Fig. 5e), contributing to the smooth-walled atrial myocardium.
α-Sma Expression Pattern at the Level of the Atrioventricular Canal (AVC) (Fig. 6)
During septation of the AVC the endocardial cushions, the top of the ventricular septum, the mesenchymal cap of the primary atrial septum, and the spina vestibuli fuse, giving rise to the myocardial atrioventricular septum, the mesenchymal mitral and tricuspid valves, the central fibrous body, and the tendon of Todaro in the formed heart. Cardiomyocytes are observed to protrude into the initially mesenchymal atrioventricular septum from the spina vestibuli, the dorsal atrioventricular wall and ventricular septum, the ventral myocardial atrioventricular wall, and the primary atrial septum. This muscularization is most intensive at E14-15. At these stages α-Sma expression is observed in the myocardial protrusions from the primary atrial septum and in the mesenchyme surrounding these protrusions (Fig. 6a, b, g, and h). At E15, this mesenchyme has become myocardial at the ventral side (Fig. 6e), whereas at the dorsal side muscularization continues until E16 (Fig. 6g). The cardiomyocytes that are contiguous with the dorsal aspect of the ventricular septum and protrude into the mesenchyme of the atrioventricular septum express relatively high levels of α-Sma (arrow in Fig. 6d). Close examination shows that α-Sma expression is not limited to the protruding cardiomyocytes but is also observed in immediately adjacent mesenchymal cells (arrowhead in Fig. 6d). At E15, cardiomyocytes that are present in the center of the endocardial cushions below the myocardial cap of the atrial septum are contiguous with the ventral myocardial wall of the AVC. These cardiomyocytes show a high expression of α-Sma (Fig. 6f and h), and a gradient of Serca2a expression that tapers off dorsally (Fig. 6e and g). Muscularization of the AVC is virtually complete at E16, leaving the central fibrous body, the tendon of Todaro, and the tricuspid and mitral valves nonmuscular.
α-Sma Expression Pattern During Formation of the Pulmonary Myocardium (Fig. 7)
At E10, the cells flanking the pulmonary pit show a high expression of α-Sma, whereas the expression of myocardial markers is relative low (Fig. 7a and b). At E12, the pulmonary pit is directed to the left, and the lumen of the pulmonary vein enters from the mesenchymal dorsal mesocardium into the future left atrium. At this stage α-Sma expression can be distinguished in the cells around the pulmonary pit (Fig. 5d) and the forming pulmonary vein (Fig. 7d). The cells of the wall of the pulmonary vein express very little Serca2a (Fig. 5c and 7c). From E14 onward, the cells of the pulmonary vein start to express Serca2a at the entrance into the left atrium. The α-Sma expression extends further distally in the direction of the lungs compared to the Serca2a expression (Fig. 7e and f). At E15, the cells of the wall of the pulmonary vein at the level of the bifurcation into the right and left pulmonary veins express high levels of α-Sma and low levels of Serca2a (Fig. 7g and h). At E17, the pulmonary myocardium extends to at least the fifth bifurcation of the pulmonary veins in the lungs (not shown).
This study shows that the expression of α-Sma precedes the expression of myocardial markers in the intra- and extracardiac mesenchyme, where the myocardium is being formed, which indicates that nonmyocardial cells are recruited into the myocardial lineage.
Recruitment Rather Than Growth From the Existing Myocardium
At the arterial pole, the heart is elongated by differentiation of splanchnic mesoderm into cardiomyocytes (Virágh and Challice, 1973; de la Cruz and Markwald, 1999; Mjaatvedt et al., 2001; Waldo et al., 2001; Kelly et al., 2001). α-Sma expression precedes the expression of myocardial markers at this site of differentiation. This indicates that also in mouse α-Sma accompanies the early phases of myocardial differentiation, and as such can be considered an indicator for recruitment of mesodermal cells into the myocardial lineage in regions of myocardium formation.
At the venous pole of the heart, the newly formed myocardium contributes the caval and pulmonary myocardia and the smooth-walled atrial myocardium (Jones et al., 1994; Webb et al., 1998b; Wessels et al., 2000; Millino et al., 2000; Franco et al., 2000, van den Hoff et al., 2001). The myocardium formation at the sinus venosus is also preceded by α-Sma expression, which suggests that this myocardium formation is achieved by recruitment from an adjacent mesodermal region. Detailed analyses of the expression pattern of α-Mhc mRNA (Jones et al., 1994), and an analysis of transgenic mice, in which the lacZ gene was driven by a truncated Troponin I promoter (Millino et al., 2000), indicate that the origin of the myocardium surrounding the pulmonary veins is the result of a progressive migration of atrial cardiomyocytes into the branching network of pulmonary veins. This hypothesis is based primarily on the observation that the myocardia of the pulmonary wall and the atrium both share expression of the same myocardial markers (Jones et al., 1994; Millino et al., 2000). Franco et al. (2000) showed that the pulmonary myocardium comprises a transcriptional domain different from that of the atrial appendages, which may indicate that the pulmonary myocardium is not derived from the embryonic atrial myocardium. In agreement with that study, our observations that α-Sma expression precedes the expression of myocardial markers during myocardium formation along the pulmonary veins suggest that recruitment underlies the formation of pulmonary myocardium. Similarly, α-Sma was observed in rats to be expressed in the mesenchymal cells adjacent to the pulmonary myocardium, which during subsequent development express α-Mhc (Ya et al., 1997).
The mesenchymal cushion tissue that is involved in septation of the heart becomes largely muscular during subsequent development. Because newly formed myocardial networks are almost always contiguous with the flanking myocardium in vivo and in vitro, migration of existing cardiomyocytes (i.e., myocardialization) was initially thought to underlie muscularization of intracardiac mesenchyme (de Jong et al., 1992; Lamers et al., 1995; van den Hoff et al., 1999). However, in vitro culture experiments showed that mesenchymal cells from the cushions are able to differentiate into cardiomyocytes, provided the proper signals are present (van den Hoff et al., 2001). In agreement with this, at the interface of the myocardial and mesenchymal border in the endocardial cushions, cells are found that display an intermediate phenotype of myocardial and mesenchymal cells (Morris, 1976). Furthermore, a local burst of proliferation of myocardium in the OFT and AVC would be expected with migration of the existing myocardium, which has not been observed (Thompson et al., 1990; Ya et al., 1998; Kubalak et al., 2002). Together with our findings that α-Sma expression precedes the expression of myocardial markers in cushion mesenchyme in vivo, it appears that differentiation of mesenchymal cells (i.e., recruitment) contributes to the muscularization of intracardiac mesenchyme.
Significance of α-Sma Expression
α-Sma expression previously was used to trace the majority of neural crest-derived mesenchymal cells in the endocardial cushions of the OFT (Waller et al., 2000). α-Sma was used as a marker for neural crest-derived cells because α-Sma-positive mesenchymal cells colocalize with lacZ-positive cells in the transgenic pCx43/lacZ mouse. In this transgenic mouse, the lacZ reporter is under the regulation of a truncated connexin-43-promoter, which is preferentially expressed in neural crest cells (Lo et al., 1997; Waldo et al., 1999). However, at E14 in this mouse, lacZ-positive cells were observed around the fusion line of the OFT cushions only, whereas most of the mesenchymal cells present in the outlet septum are α-Sma-positive (Waller et al., 2000) (Fig. 3f). This apparent discrepancy may be due to the fact that at this stage α-Sma expression is not restricted to the neural crest-derived mesenchyme in the OFT, or it may be that the neural crest cells do not express the transgene anymore. The latter explanation seems unlikely because it has been shown (Jiang et al., 2000) that in Wnt1-Cre/R26R embryos (in which the fate of the neural crest cells can be followed), the neural crest contribution is confined to a sub-endocardial layer in the OFT cushion.
Rather than being a marker for neural crest-derived cells in the OFT, α-Sma expression appears to be a feature of migrating mesenchymal cells. The changing of the cell shapes requires reorganization of the cytoskeleton. This is underscored by the observation that at E14 the α-Sma expressing mesenchymal cells become aligned with the myocardial cells of the OFT wall, thereby establishing continuity. This alignment is most prominent in the center of the fused endocardial cushions, which is at the site of the future myocardial septum. The α-Sma-negative mesenchymal cells do not show this alignment. This is in accordance with the previous finding that endocardium-derived mesenchymal cells migrating into the cushions express α-Sma (Nakajima et al., 1997). The mesenchymal cells that are in contact with the cardiomyocytes subsequently up-regulate α-Sma expression, which may be an indication of differentiation toward a myocardial phenotype.
Source of the Cardiomyocyte Precursor
The source of the cardiomyocyte precursors in the endocardial cushions is unknown; however, the endocardium-derived mesenchymal cells appear to be likely candidates, since they are initially present in both the AVC and the OFT cushions. Neural crest cells are present in the OFT and not in the AVC (Kirby, 1999). Epicardium-derived mesenchymal cells populate the atrioventricular cushions only (Gittenberger-de Groot et al., 1998; Männer et al., 2001). Moreover, studies that traced the fate of the pro-epicardium (Dettman et al., 1998; Männer, 1999) and the cardiac neural crest (Waldo et al., 1998; Jiang et al., 2000) did not reveal that these cell populations contribute to the cardiomyocyte lineage. The extracardiac mesenchymal cells that are suggested to differentiate into cardiomyocytes are not endocardium-derived. Nevertheless, the extracardiac and endocardium-derived mesenchymal cells share a common developmental origin, the splanchnic mesoderm (Fig. 8).
Toward a Mechanism
Based on the results described herein, we propose that the following mechanism underlies the formation of myocardium in the endocardial cushions (Fig. 9). Initially, the myocardium that flanks the mesenchyme is smooth. Subsequently, these cardiomyocytes lose their epithelial context and penetrate into the mesenchyme, increasing their contact area with the cushion mesenchyme. This migration process is called myocardialization. However, the cardiomyocytes do not fully migrate into the septum, but remain in contact with the underlying myocardium. The interaction of the penetrating cardiomyocytes and the mesenchymal cells induces differentiation of the mesenchymal cells into cardiomyocytes. A direct interaction between cardiomyocytes and mesenchymal cells seems essential because isolated cardiomyocytes are hardly ever found in the cushion mesenchyme. Prior to the expression of Serca2a and other myocardial markers, these mesenchymal cells up-regulate α-Sma. The newly formed cardiomyocytes likewise induce the adjacent mesenchymal cells to differentiate into cardiomyocytes, until the process of muscularization is complete. Thus both myocardialization and recruitment are necessary for cardiac muscle cell formation in the endocardial cushions, and they are both part of a single muscularization process. Similarly, the muscularization of the pulmonary and caval veins is established by recruitment of the mesenchymal cells lining the veins into the myocardial lineage upon a signal from the neighboring cardiomyocytes. The inducing signal molecules are unknown as yet, although bone morphogenetic proteins and fibroblastic growth factors are possible candidates (Parlow et al., 1991; Schultheiss and Lassar, 1999; Lough and Sugi, 2000). We are currently pursuing this hypothesis.