The cells that will become the heart are among the first cell lineages to be established in the vertebrate embryo. Significant effort has been directed toward identifying the embryonic and genetic origins of cardiac progenitors. A variety of techniques have been used to localize embryonic heart forming regions and to define the inductive interactions required for recruitment of multipotential cells into the cardiac lineage. With advances in molecular biology, signaling molecules and transcription factors important for cardiogenesis have been identified. The correlation of early cardiac fate maps with gene expression domains should provide important clues into the embryonic origins of the heart. However, the origins of heart cells are still controversial. The resolution of this issue is critical for understanding the molecular and cellular cues that drive a multipotential cell into the cardiac lineage and provides a necessary foundation for therapeutic approaches to generating new sources of cardiomyocytes.
The earliest microscopic study of the embryonic cardiovascular system is attributed to Marcello Malpighi working in the 1600s with chickens developing in ovo (Adelmann, 1966). Subsequent anatomic reports of heart looping and chamber formation were also based largely on avian embryos (Patten, 1951). Beginning early in the past century, the origin of the cardiogenic mesoderm was examined in avian embryos using fate mapping and lineage tracing techniques (Rudnick, 1938; Rawles, 1943; DeHaan, 1963a, 1963b). Avians remain the system of choice for early cardiac lineage and cell fate analysis due to accessibility and manipulability of the embryo. The cellular and molecular events that control early cardiac development are remarkably conserved across vertebrate and even invertebrate embryos (reviewed in Fishman and Chien, 1997; Srivastava and Olson, 2000). Therefore, definition of the heart forming regions of avian embryos has broad application in the study of cardiogenesis across species.
Fate maps depicting the embryonic origins of the developing organs have been generated for the major animal systems used in experimental embryology. These maps result from labeling cells or regions of blastula or gastrula stage embryos and following the marked populations to observe their fate later in development. Avian embryos are particularly well-suited for these studies, because they can be manipulated at early stages of development and then revisited at sequential times in ovo or in whole embryo culture. The heart forming regions of avian embryos have been mapped with inconsistent results using a variety of tracking techniques (DeHaan, 1963a, 1963b; Stalsberg and DeHaan, 1969; Rosenquist, 1970; Garcia-Martinez and Schoenwolf, 1993; Redkar et al., 2001; Waldo et al., 2001). Explantation and culture of cells from different regions of the early embryo have been used to further define the ability of cells from specific regions of the embryo to differentiate into beating cardiac myocytes (Schultheiss et al., 1995; Ehrman and Yutzey, 1999). These approaches have been used to define primary heart fields that form the primitive heart and a secondary heart field in the ventral pharynx that contributes to the conotruncal myocardium.
Mechanistic experiments in several vertebrate and invertebrate embryonic systems have begun to dissect regulatory hierarchies that control early cardiac development (reviewed in Srivastava and Olson, 2000). Genetic analyses in Drosophila, zebrafish, or mouse embryonic systems have identified important molecular regulators of early cardiogenesis (reviewed in Chen and Fishman, 2000). Additional signaling molecules and tissue interactions that promote or inhibit cardiogenesis have been identified through manipulation of the embryo or cell culture environment (reviewed in Lough and Sugi, 2000). The inductive interactions and genetic regulatory mechanisms that together establish the primary heart fields appear to be conserved as additional cardiomyocytes arise in the secondary heart field of the ventral pharynx (Waldo et al., 2001; Mjaatvedt et al., 2001). Thus, the analysis of gene expression patterns for key regulators of cardiogenesis in avian embryos can provide strong support for the localization of cardiac progenitors.
This review is focused on primary and secondary cardiogenesis in avian embryos, because the embryonic origins of the heart can be studied most thoroughly in birds. However, fate mapping and gene expression studies will be compared in mouse and avian systems to provide further support for conservation of cardiac regulatory mechanisms. Although the generalities of early heart formation are not in dispute, there are inconsistencies in the reported localization of heart progenitor populations. The origins of atrial and ventricular precursors in the heart forming region and primitive heart tube have been reported differently (Stalsberg and DeHaan, 1969; de la Cruz et al., 1989; Yutzey et al., 1995; Redkar et al., 2001). Conflicting reports also exist for the timing of the separation of endothelial and myocardial cell lineages (Eisenberg and Bader, 1995; Wei and Mikawa, 2000). Limited information is available regarding the origins of the secondary conotruncal heart field, but lessons of primary cardiomyogenesis seem to be applicable to this cell lineage (Waldo et al., 2001; Mjaatvedt et al., 2001). The purpose of this review is to critically examine conflicting information regarding the localization of cardiac progenitors with consideration of molecular events that control cardiogenesis in an effort to define the embryonic origins of the developing heart.
OVERVIEW OF EARLY HEART DEVELOPMENT
The vertebrate heart is the first organ to form and function during development. In chicken embryos, a beating heart tube is apparent at Hamburger and Hamilton (HH; 1951) stage 10 or approximately 35–40 hr of development (Fig. 1). The cells that will become the heart, or cardiac progenitors, are among the first mesodermal cells to gastrulate through the primitive streak (Garcia-Martinez and Schoenwolf, 1993; Schoenwolf and Garcia-Martinez, 1995). These cells then migrate to an anterior lateral position where they condense to form bilateral heart primordia or primary heart fields (DeHaan, 1965). The primary heart tube, an endothelial tube ensheathed in myocardium, arises as a result of cranial-to-caudal fusion of paired heart primordia. Within the primary heart tube, the cranial regions form the ventricles, whereas atrial precursors are present caudally. Recent studies have identified a second anterior population of cardiac progenitors derived from pharyngeal mesoderm that is incorporated into the cranial outflow tract region (Waldo et al., 2001; Mjaatvedt et al., 2001). The cells of this secondary heart field contribute to the conotruncal myocardium of the looped heart tube. Heart chamber formation is initiated with the morphologic diversification of the atria and ventricles, which occurs at the same time as induction of endocardial cushions at the atrial-ventricular junction and in the outflow tract (reviewed in Fishman and Chien, 1997). This process of early heart formation is remarkably conserved in vertebrate embryos from fish and amphibians to mammals.
FATE MAPPING THE CARDIOGENIC MESODERM IN THE PRIMARY HEART FIELDS
The earliest origins of the cardiac progenitors have been identified before gastrulation in the epiblast cell layer as it separates from the extraembryonic hypoblast. This differentiation occurs within the first few hours of egg incubation at stage XIII (Eyal-Giladi and Kochav, 1976), which precedes the Hamburger Hamilton stages. Cardiac progenitors have been mapped by vital dye injection to the posterior epiblast at stage XII-XIII (Hatada and Stern, 1994). This localization of cardiac progenitors is supported by explantation studies, in which the posterior epiblast has cardiogenic potential and will differentiate into cardiac myocytes in a defined culture system (Yatskievych et al., 1997). During gastrulation, the cells that will become the heart are among the first cells to migrate out of the primitive streak. Mapping and explantation studies of mid-gastrulation stage embryos consistently demonstrate that the heart cells are present in the streak at late stage 3 and have migrated out of the streak at stage 4 (Fig. 1A) (Rudnick, 1938; Rawles, 1943; Garcia-Martinez and Schoenwolf, 1993; Psychoyos and Stern, 1996; Yuan and Schoenwolf, 1999; Wei and Mikawa, 2000).
There is less of a consensus regarding where the heart cells go after they leave the primitive streak, because it is difficult to fate map the cardiac progenitors consistently in their migration route from the primitive streak to the anterior lateral plate. Migratory cardiac progenitors have been mapped in a wide region lateral to the primitive streak (Cohen-Gould and Mikawa, 1996; Redkar et al., 2001). After migration, cardiogenic mesoderm has been localized lateral to the condensing ectodermal neural plate and medial to the area pellucida (Fig. 1B) (DeHaan, 1963a, 1963b; Stalsberg and DeHaan, 1969; Ehrman and Yutzey, 1999). In other reports, more medial positions have been ascribed to the heart forming cells in the stage 5–8 embryo as a result of DiI- or radioisotope-labeling studies (Rosenquist, 1970; Redkar et al., 2001). The more lateral position of the cardiogenic mesoderm is supported by chorioallantoic grafting of lateral plate mesoderm, which demonstrated that “heart-forming potency” extended to the lateral limits of embryo adjacent to the area pellucida (Rawles, 1943). In explantation studies, removal of the far anterior lateral cardiogenic region results in complete loss of cardiac structures on the operated side, but removal of the medial endoderm and mesoderm overlying the condensing neural plate has no effect on cardiogenesis (Fig. 2) (Ehrman and Yutzey, 1999). The confirmation of the laterally placed primary heart forming fields by explantation and cell culture experiments provides strong evidence for a far lateral position of the cardiac progenitors.
The posterior borders of the primary heart forming regions also have been mapped, but the reports are conflicting. Surface marking and time lapse imaging place the posterior border of the heart at approximately the level of the first condensing somite at stage 7 (DeHaan, 1963b, 1965; Ehrman and Yutzey, 1999). However, a more posterior border was reported based on labeling studies with DiI or tritium (Rosenquist, 1970; Redkar et al., 2001). Explantation experiments demonstrate that lateral mesoderm posterior to the first somite does not differentiate into cardiomyocytes in culture and that removal of this tissue does not result in loss of cardiac structures (Ehrman and Yutzey, 1999). In addition, extirpation of the entire heart forming region anterior to the first somite results in loss of all cardiac structures (Mjaatvedt et al., 2001). Thus, the placement of the posterior borders of the primary heart forming region at the level of the first somite is supported by both explantation and cell culture experiments (DeHaan, 1963b; Ehrman and Yutzey, 1999; Mjaatvedt et al., 2001).
TISSUE INTERACTIONS AND SIGNALING MOLECULES THAT INDUCE PRIMARY CARDIOGENESIS
The tissue interactions and signaling molecules that control the earliest stages of cardiogenesis have not yet been fully characterized. Before gastrulation, cells in the posterior, but not anterior, epiblast have cardiogenic potential in culture (Yatskievych et al., 1997; Ladd et al., 1998). Cardiac progenitors explanted at these early stages require added signaling molecules, including Activins, transforming growth factor (TGF)βs, and bone morphogenetic proteins (BMP)s, to differentiate (Yatskievych et al., 1997; Ladd et al., 1998). However, these signaling molecules are not sufficient to recruit noncardiogenic cells to the cardiac lineage and pregastrulation cardiac inductive signals have not been clearly identified in vivo. Cardiac progenitors become specified in the primitive streak or soon thereafter, which is evident in the ability of anterior, but not posterior, lateral mesoderm to differentiate into cardiomyocytes when cultured at clonal density (Gonzalez-Sanchez and Bader, 1990; Antin et al., 1994). Signals important for early cardiac specification are present in the primitive streak, because noncardiogenic cells transplanted to the primitive streak are recruited into the heart (Schoenwolf and Garcia-Martinez, 1995; Schultheiss et al., 1995). However, the regulatory mechanisms that control the initial recruitment of multipotential progenitor populations to the cardiac lineage in vivo have proven to be particularly difficult to define. At present, the nature and source of the initial cardiac specification signals are unknown and molecular markers unique to specified cardiomyocytes have not been identified.
Once the cardiac progenitors have migrated to the anterior lateral plate, there is extensive evidence for a role of anterior lateral endoderm in cardiac lineage determination and differentiation (reviewed in Lough and Sugi, 2000). Coculture with anterior endoderm is sufficient to recruit posterior noncardiogenic mesoderm to the cardiac lineage (Schultheiss et al., 1995). In culture, transient exposure of posterior lateral mesoderm to fibroblast growth factors (FGF)s and BMPs is sufficient to induce cardiogenesis (Barron et al., 2000). The transient nature of the cardiac inductive signal is supported in vivo by differentiation of lateral plate mesoderm from which the endoderm has been removed (Gannon and Bader, 1995). Several FGFs are expressed in the anterior lateral endoderm and FGF signaling is required for cardiomyogenic differentiation (Zhu et al., 1996). BMP-2 also is expressed in the far anterior lateral endoderm and inhibition of BMP signaling by noggin blocks differentiation of cardiogenic mesoderm (Schultheiss et al., 1997; Andree et al., 1998). Local application of BMP-2 is sufficient to induce cardiac gene expression in medial mesoderm, which demonstrates that the anterior medial mesoderm has the potential to enter the cardiac lineage when exposed to the appropriate cell environment (Schultheiss et al., 1997; Andree et al., 1998; Schlange et al., 2000). However, in vivo cardiogenic signals, including BMPs produced by lateral endoderm, apparently establish the primary heart forming regions in a more lateral position.
CARDIAC REGULATORY GENES ARE EXPRESSED IN THE PRIMARY HEART FORMING REGIONS
The regulatory mechanisms that control the initial specification of the cardiac lineages in the primitive streak have not been well characterized. An extracellular signal in the primitive streak appears to be important, because noncardiogenic posterior lateral mesoderm transplanted into the primitive streak is recruited into the cardiac lineage (Schoenwolf and Garcia-Martinez, 1995; Schultheiss et al., 1995). The lack of genetic markers of the heart progenitors as they ingress during gastrulation and migrate laterally contributes to the controversy over their location in the embryo. More information is available regarding the signaling molecules and transcription factors involved in cardiac lineage induction after they reach the anterior lateral plate. Several families of transcription factors have been implicated in cardiac lineage determination and differentiation. Transcriptional regulators of cardiac contractile protein gene expression, including Nkx-2.5, GATA4, and Tbx5, are expressed in the anterior lateral plate mesoderm before formation of the primitive heart tube (Schultheiss et al., 1995; Kostetskii et al., 1999; Srivastava and Olson, 2000; Liberatore et al., 2000; Hiroi et al., 2001). Studies in chick, mouse, Xenopus, zebrafish, and Drosophila have established a role for BMPs in induction of nkx-2.5 gene expression and cardiac differentiation (Frasch, 1995; Schultheiss et al., 1997; Andree et al., 1998; Shi et al., 2000; Liberatore et al., unpublished observations). The localized expression of these molecules can be used to provide support for fate mapping and lineage analysis of the heart progenitors.
Expression of signaling molecules and transcription factors implicated in early cardiogenesis supports the placement of the heart primordia in an anterior lateral position (Schultheiss et al., 1995, 1997; Ehrman and Yutzey, 1999; Colas et al., 2000). At stages 4–5, BMP-2 is expressed in the most anterior lateral regions of the definitive endoderm (Schultheiss et al., 1997; Andree et al., 1998; Ehrman and Yutzey, 1999). By stage 5, nkx-2.5 gene expression is activated in the anterior lateral mesoderm in a portion of the bmp-2 expression domain. The overlapping regions of bmp-2 and nkx-2.5 gene expression are consistent with many of the classic fate mapping studies (Fig. 2) (DeHaan, 1963b; Stalsberg and DeHaan, 1969). By stages 6–7, additional genes involved in cardiogenesis are expressed in the condensing heart primordia. These include transcriptional regulators, such as tbx5 and gata4, as well as early cardiac contractile proteins, including smooth muscle α-actin (smaa) (Kuo et al., 1997; Molkentin et al., 1997; Kostetskii et al., 1999; Horb and Thomsen, 1999; Colas et al., 2000; Liberatore et al., 2000; Hiroi et al., 2001). Placement of the posterior boundary of the heart at the level of the first somite also is consistent with nkx-2.5, tbx5, and smaa expression domains (Schultheiss et al., 1995; Liberatore et al., 2000; Colas et al., 2000). The correlation of fate mapping and gene expression studies defines the heart forming region in the most lateral position in the anterior plate mesoderm adjacent to the extraembryonic cell lineages. Furthermore, the posterior borders of the primary heart forming regions are defined as lateral to the first condensing somite by gene expression and a majority of embryologic analyses (DeHaan, 1963b; Stalsberg and DeHaan, 1969; Ehrman and Yutzey, 1999; Padwardhan et al., 2000; Mjaatvedt et al., 2001).
INHIBITORS OF CARDIOGENESIS ARE EXPRESSED IN THE POSTERIOR LATERAL MESODERM
Fate mapping studies have identified the posterior lateral mesoderm as extraembryonic blood and posterior visceral mesoderm (Schoenwolf et al., 1992; Schultheiss et al., 1995). However, other cell marking and gene expression studies have placed cardiac progenitors in the posterior lateral region (Rosenthal and Xavier-Neto, 2000; Redkar et al., 2001). Recently, regulatory molecules that inhibit cardiogenesis have been identified, including Wnts and upstream regulators of Wnt gene expression (Marvin et al., 2001; Ehrman and Yutzey, 2001). Wnt8c and its upstream regulator cCdx-B are expressed in the posterior lateral mesoderm consistent with regions fated to be incorporated into noncardiac lineages (Marvin et al., 2001; Ehrman and Yutzey, 2001). They are not expressed in the anterior heart forming region, and ectopic anterior expression of Wnt8c inhibits cardiac differentiation (Marvin et al., 2001). During normal development, the Wnt inhibitor crescent is expressed in anterior cardiogenic regions of the embryo and ensures that Wnt signaling will be low in the heart forming region. Therefore, regulatory pathways that promote cardiogenesis are present in the anterior lateral plate, and anti-cardiogenic regulatory pathways act in the posterior. Taken together, the majority of fate mapping, explantation, and molecular studies demonstrate that regions of the embryo expressing repressors of cardiogenesis such as Wnt8c are not included in the primary cardiac fields.
ORIGINS OF ATRIAL AND VENTRICULAR CARDIOMYOCYTES
The regions of the primitive heart tube that will develop into the different cardiac chambers are initially arranged sequentially along the anteroposterior/craniocaudal axis of the embryo with the ventricles more anterior and the atria more posterior. Fate mapping studies of the primitive streak demonstrate that the anteroposterior position at the time of gastrulation is reflected in anteroposterior position in the primitive heart (Garcia-Martinez and Schoenwolf, 1993). Thus, cardiac progenitors in the anterior streak are incorporated into the anterior ventricular segments of the primitive heart, whereas cells in the posterior streak are incorporated in the sinoatrial segments. The positional identity of transplanted cells can be respecified within the streak, because cardiac progenitors migrate and differentiate consistent with their final position during gastrulation (Garcia-Martinez and Schoenwolf, 1992; Schoenwolf and Garcia-Martinez, 1995; Schultheiss et al., 1995). Together, these studies suggest that positional restriction of atrial and ventricular precursors occurs during the earliest stages of cardiogenesis. However, molecular regulators of chamber-specific cell fates have not been identified in gastrulating cardiac progenitors.
The embryonic locations of atrial and ventricular progenitors in the primary heart fields have been mapped differently in conflicting reports. DeHaan and collaborators reported positional restriction of ventricular and atrial precursors in the primary heart forming fields (DeHaan, 1963b, 1965; Stalsberg and DeHaan, 1969). However, other marking studies concluded that there is no prepatterning of the heart forming region with regard to atrial and ventricular precursors (Redkar et al., 2001). The inclusion of posterior atrial progenitors in the primary heart field anterior to the somites also has been disputed (Rosenthal and Xavier-Neto, 2000; Redkar et al., 2001). Evidence for the presence of atrial progenitors in the primary heart forming region is provided by explantation experiments. Cardiac progenitors removed from the posterior third of the heart forming region anterior to the somites express atrial myosin heavy chain in culture (Fig. 3A) (Yutzey et al., 1994, 1995). These explantation and gene expression studies indicate that atrial precursors are present caudally within the primary heart forming region.
There is controversy over which heart segments are included in the primitive heart tube before looping. Surface marking studies have concluded that the primitive heart consists only of ventricular segments at the time beating is initiated (HH stage 10–11) and that the atrial-ventricular junction, atria, and sinus venosus segments are incorporated later in the looping heart tube (de la Cruz et al., 1989; de la Cruz and Markwald, 1998). However, other marking studies of the linear heart tube demonstrated that the atrial progenitors are included in the primitive heart tube before looping and are derived from the primary heart fields (Stalsberg and DeHaan, 1969; Waldo et al., 2001). The discrepancies in these studies may arise from single point analysis in cell marking and torsion of the primitive heart tube. In addition, the atrial and ventricular segments of the heart forming region may not be in an absolutely linear arrangement. Support for the presence of atrial progenitors in the heart before looping is provided by gene expression studies. Restricted expression of an atrial myosin heavy chain gene (AMHC1) is apparent in the posterior segments of the primitive heart tube before looping (Fig. 3B) (Yutzey et al., 1994; Liberatore et al., 2000). This atrial gene expression pattern, supported by a subset of the fate mapping studies, strongly suggests that the primitive heart tube before looping includes both atrial and ventricular myocytes (Stalsberg and DeHaan, 1969; Yutzey et al., 1994).
SEPARATION OF CARDIOMYOGENIC AND ENDOTHELIAL CELL LINEAGES
Fate mapping studies and cell lineage analysis of the endocardial endothelial progenitors demonstrate that they arise from the same regions of the embryo as the cardiomyocyte precursors (Stalsberg and DeHaan, 1969; Cohen-Gould and Mikawa, 1996; Wei and Mikawa, 2000). By neurulation stages, distinct endothelial cells can be marked with QH-1 antibody staining in broad lateral and extraembryonic regions that include cardiac and vascular endothelial precursors (Coffin and Poole, 1988; Linask and Lash, 1993). The rudiments of the endocardium are apparent within the condensing heart promordia, and a continuous endocardial endothelium lines the primitive heart tube (Coffin and Poole, 1988; Viragh et al., 1989). Retroviral cell lineage analyses demonstrate that cells in the primitive streak at stage 3 or lateral mesoderm at stage 4 contribute to either the myocardial or endothelial cell layers of the tubular heart (Cohen-Gould and Mikawa, 1996; Wei and Mikawa, 2000). Neither study found a common progenitor for endothelial and myocardial cells even though cardiomyogenic and endothelial progenitors are intermixed during gastrulation and migration. Because no common stem cell population contributing to both cell lineages was observed, it was concluded that the endothelial and myocardial lineages are distinct from each other at or soon after gastrulation. Evidence for a common progenitor is provided by a cell line derived from quail embryo anterior mesoderm that can express either myocardial or endothelial cell markers, depending on the cell culture environment (Eisenberg and Bader, 1995). Lineage-specific genes for endothelial and myocardial precursors have not been identified at these early stages; however, in vivo cell lineage analysis demonstrates that most, if not all, cardiomyocytes and endothelial cells arise from separate progenitor populations in the primitive streak and anterior lateral plate. These endothelial precursors may not represent the entire endocardium, because outflow tract endothelial cells have been mapped to head mesoderm adjacent to the hindbrain (Noden et al., 1995). This cranial endothelial precursor population may correspond to a secondary cardiogenic field but further studies are necessary to unequivocally define the origins of outflow tract myocardial and endothelial precursors.
CONSERVATION OF PRIMARY CARDIOMYOGENESIS IN CHICKEN AND MOUSE EMBRYOS
Fate mapping and gene expression studies show that early heart development is highly conserved in mammalian and avian embryos. In mice, the heart cells are among the first to gastrulate at early to mid-streak stages (Tam et al., 1997). Before gastrulation, mouse epiblast has cardiogenic potential and the cardiac lineage is specified at the late streak stage (Auda-Boucher et al., 2000). The cardiac progenitors migrate from the primitive streak to the most anterior lateral position in the embryo where they initiate early cardiac expression of nkx-2.5 and tbx5 (Lints et al., 1993; Liberatore et al., 2000). The cardiogenic region of the mouse has traditionally been described as a crescent (Kaufman and Navaratnam, 1981). However, analysis of early nkx-2.5 and tbx5 expression in the late streak stage embryo clearly shows paired heart primordia similar in position and gene expression to the heart forming fields in avian embryos (Fig. 4) (Searcy et al., 1998; Liberatore et al., 2000). As in the chick, the presence of endoderm promotes myogenesis in late gastrulation stage mouse embryos (Arai et al., 1997; Auda-Boucher et al., 2000). Visceral endoderm, adjacent to the most lateral mesoderm in late streak stage embryos, is required for cardiac differentiation and heart formation, which supports a far lateral localization of the primary heart fields (Arai et al., 1997; Narita et al., 1997). BMP-2 can promote cardiogenesis in cultured mouse embryos and potential target sites for BMP-mediated gene activation have been identified in early cardiac regulatory sequences of the mouse nkx-2.5 gene (Searcy et al., 1998; Liberatore et al., 2002). Thus, the timing, tissue interactions, and molecular regulatory mechanisms that establish the primary cardiogenic field are highly conserved between avian and murine embryos.
Fate mapping of the atrial and ventricular lineages in the mouse primitive heart tube has not been reported. This has led to some confusion regarding the specific nature of defects in heart chamber formation, resulting from targeted mutagenesis of cardiac regulatory genes (reviewed in Srivastava and Olson, 2000). Evidence for the positional restriction of atrial and ventricular lineages in the primitive heart is provided by transgenic mice that express an atrial-specific reporter gene (Xavier-Neto et al., 1999). Expression of the transgene in the posterior segments of the primitive heart tube indicates that the atrial progenitors are included in the primitive heart tube at stages before looping. Similarly, targeted mutation of tbx5, a transcription factor associated with atrial chamber formation and gene expression, inhibits cardiomyogenesis in the posterior primitive heart tube (Bruneau et al., 2001). Ventricular-specific expression of mlc2v also is observed in the anterior segments of the primitive heart tube at this stage (O'Brien et al., 1993). Together, these studies provide evidence for positionally restricted ventricular and atrial lineages in the primitive heart tube before looping in mammalian as in avian embryos.
IDENTIFICATION OF THE SECONDARY HEART FIELD
Although fate-mapping studies assumed that all of the myocardium of the mature heart was present in the initial or primary heart tube, marking studies in chick and molecular studies in both chick and mouse suggested that the myocardium of the definitive outflow tract is added to the primary heart tube during looping (reviewed in de la Cruz and Markwald, 1998). The potential explanation for this misassumption is that the original cardiac fate-mapping studies were done in New culture in which whole chick embryos at early stages are explanted ventral side up onto a bed of agar-albumen. The New culture technology that has been available until very recently allowed growth of the embryos in excellent condition until approximately stages 10–12. At stage 10, the primary heart tube has formed and, by stage 12, it is in the early process of looping. Most investigators assumed that all of the components of the mature heart were contained in the tube at this stage. Thus, it was thought that the myocardium in the primary or straight heart tube composed all of the myocardium that later forms the mature heart (Rosenquist and DeHaan, 1966; Garcia-Martinez and Schoenwolf, 1993).
The straight heart tube apparent at stage 10 begins to loop at stage 12, a process that continues over several days. During looping, the myocardial tube is composed of a descending, or inflow limb, and an ascending, or outflow limb. The outflow tract is the final portion of the ascending limb, and it attaches to the ventral pharynx. Before septation of the outflow tract into aortic and pulmonary channels can occur, the outflow tract achieves a position over the prospective ventricles such that blood can flow into it from both the presumptive left ventricle located in the descending limb of the loop, and the presumptive right ventricle in the ascending limb of the loop. The outflow tract lengthens in the final stages of looping to accommodate this torsion. Because the circulation has been established before the tube lengthens, enough myocardial cells must be available to reinforce the wall of the lengthening tube to ensure normal cardiac output for the embryo (Clark, 1984; Leatherbury et al., 1991). In addition myocardial cells of the outflow tract will later be needed for myocardialization of the conal ridges, an important event in closure of the ventricular septum (Van den Hoff et al., 1999). With incomplete or absent myocardialization of the conus, the subarterial ventricular septum cannot close, leaving a subaortic or subpulmonary ventricular septal defect.
Early morphologic evidence that the conotruncal segments of the heart are added during looping was provided by in ovo marking studies of chicken embryos. Studies by de la Cruz and colleagues in the 1960s showed that a mark placed at the junction of the outflow tract with the body wall during early looping (stage 14) was incorporated into the body of the mature right ventricle. Only at stage 22, 2 days later, did a mark placed at the distal end of the outflow tract stay at the ventriculoarterial junction (Arguello et al., 1975; de la Cruz et al., 1977, 1987, 1989, 1991). This finding indicates that neither the conus nor the truncus is present in the primary heart tube but that these segments are added to the outflow tract between stages 12 and 22 (approximately 2 to 4 days of incubation). However, the source of the conotruncal myocardium was not identified in these experiments.
Molecular studies in chick and mouse embryos also indicated that the outflow myocardium is molecularly distinct from myocardium in the rest of the heart tube. In the chick, smooth muscle α-actin, (Ruzicka and Schwartz, 1988), frizzled 2 (VanGijn et al., 2001), and an adenovirally introduced cytomegalovirus promoter (Watanabe et al., 1998) show differential expression in outflow tract myocardium. In mice, transgenic expression of lacZ reporter genes containing regulatory sequences of nkx-2.5, gata6, or mlc2v genes also showed specific gene expression in the embryonic outflow tract (Ross et al., 1996; Searcy et al., 1998; Molkentin et al., 2000; Davis et al., 2000). Recent gene expression studies also have identified discrete myocardial molecular compartments, supporting the idea that the myocardium of the primary heart tube does not represent all of the myocardium of the mature heart (Christoffels et al., 2000) and that the origin of the outflow myocardium is heterogeneous. Although this specialized compartment of gene expression was recognized before the discovery of the secondary heart field, its significance was not evident.
The derivation of the myocardial cells for the conotruncal segments of the heart from a “secondary” or “anterior” heart field that is independent from the primary heart fields has been described recently (Fig. 5) (Waldo et al., 2001; Mjaatvedt et al., 2001; Kelly et al., 2001). Fate mapping studies in avian embryos demonstrate that the conotruncal myocardium is derived from pharyngeal splanchnic mesoderm cranial to the primitive heart tube (Waldo et al., 2001; Mjaatvedt et al., 2001). Further support for existence of the secondary heart field is provided by normal development of the conotruncus in embryos with complete ablation of the primary heart fields (Mjaatvedt et al., 2001). Similarly, in mice, DiI labeling placed in the pharynx of cultured embryos at embryonic day 8.25 could be seen in the outflow tract at embryonic day 10.5 (Kelly et al., 2001). Thus, there is increasing molecular and embryologic evidence for a secondary heart field in the ventral pharynx that contributes to the conotruncal myocardium.
SECONDARY INDUCTION OF CARDIOGENESIS IN THE CONOTRUNCAL HEART FIELD
Because the secondary heart field has only been described recently, very little is known about it. Studies by Waldo and colleagues and Mjaatvedt and colleagues showed that the field appears in the ventral midline splanchnic mesoderm underlying the endoderm of the caudal pharynx as the primary heart fields have finished adding myocardium to the cardiac inflow tract (Waldo et al., 2001; Mjaatvedt et al., 2001). Culture of the secondary heart field with serum results in myocardial differentiation, which is absent in serum-free medium (Mjaatvedt et al., 2001). However, the serum can be replaced by coculture of the secondary heart field with the distal rim of outflow myocardium. That the distal outflow myocardium is inductive has also been shown in vivo by using quail-chick chimeras. When the secondary heart field is transplanted from a quail embryo to the floor of the chick pharynx caudal to the outflow tract, it differentiates as a vesicle of myocardium (Waldo et al., 2001). When secondary heart field is transplanted to the lateral pharynx, no myocardial markers appear. This finding suggests that the outflow myocardium is inductive or potentiates the differentiation of the myocardial phenotype from the secondary heart field.
Both FGF and BMP family members are present in the caudal pharynx and outflow tract during the time that myocardium is added to the heart tube from the secondary heart field (Waldo et al., 2001). Together these signaling molecules are important for induction of cardiogenesis in the primary heart fields and appear to have a role in secondary cardiomyogenesis (Fig. 5). Inhibition of Wnt signaling also has been implication in primary cardiogenesis but a role for Wnt antagonists in the secondary heart field has not been established (Marvin et al., 2001). In the avian primary heart field, signals from the anterior endoderm, including BMPs, are required for induction of cardiogenesis (see previous section; Schultheiss et al., 1997; Schlange et al., 2000). Expression of nkx-2.5 and gata4 in the primary heart field is apparent before myocardial differentiation and can be induced by BMPs and FGFs (Schultheiss et al., 1995; Jiang et al., 1998; Reifers et al., 2000; Schlange et al., 2000). Although the origin of presumptive myocardial cells in the secondary heart field is separate from the primary heart fields, secondary cardiomyocytes also express nkx-2.5 and gata4 during the early stages of determination and differentiation (Schultheiss et al., 1995; Jiang et al., 1998; Waldo et al., 2001). Thus, the expression of FGFs and BMPs together with early cardiac transcription factors in the secondary heart field is consistent with a recapitulation of the inductive events of primary cardiogenesis in the conotruncal myocardium.
FGFs are present in the mesoderm and endoderm of the pharynx at critical stages of induction and differentiation of the secondary heart field of the conotruncal myocardium (Waldo et al., 2001; Kelly et al., 2001). In chicken embryos, FGF-8 is present in the endoderm and ectoderm of the lateral pharynx at stage 14 when the myocardium from the secondary heart field begins to differentiate (Wendling et al., 2000; Waldo et al., 2001). This expression is extinguished by stages 22–24 when outflow lengthening is finished (Waldo et al., 2001). Coculture studies have shown that the presence of lateral pharynx, which expresses fgf-8, causes a remarkable increase in proliferation of cells in the secondary heart field (Farrell et al., 2001). In mice, fgf-8 and fgf-10 are both expressed in an area medial to the primary heart forming regions bilaterally (Kelly et al., 2001). Fgf-10 expression continues in the right ventricle and outflow myocardium and in the splanchnic mesoderm dorsal and caudal to the attachment of the outflow tract to the ventral pharynx (Kelly et al., 2001). Transgenic mice generated with fgf-10 regulatory elements linked to a lacZ reporter gene express β-gal in a region comparable to the avian secondary heart field shown by Waldo et al. (Kelly et al., 2001). Thus, FGFs are expressed in endoderm and mesoderm of the ventral pharynx consistent with the temporal and spatial regulation of secondary cardomyogenesis.
BMPs also are expressed in the secondary heart field and myocardium of the outflow tract in chicken and mouse embryos in a region consistent with an inductive role in the secondary heart field. (Jones et al., 1991; Waldo et al., 2001). The addition of Noggin, a BMP inhibitor, delays differentiation of the myocardial cell phenotype in explanted tissue from the secondary heart field (Waldo et al., 2001). Thus, BMP signaling is required for the initial differentiation of the conotruncal myocardium. Further support for BMP-mediated induction of outflow tract myocardial gene expression is provided by transgenic mice generated with nkx-2.5 regulatory sequences. Deletion of a Smad consensus region from a nkx-2.5 early cardiac regulatory element results in delay of transgene expression not only in the primary heart fields but also in the outflow tract myocardium (Liberatore et al., 2002). Thus, BMP signaling mediated by Smad proteins may directly activate nkx-2.5 gene expression in both primary and secondary heart fields. It is likely that BMPs act in concert with FGFs to control secondary cardiomyogenesis and studies by Waldo et al. (2001) suggest that BMP-2 and FGF-8 are the factors that balance proliferation and differentiation of secondary myocardium in avian embryos.
SUMMARY AND PERSPECTIVES
Significant effort has been devoted toward identifying the origins of the developing heart and the regulatory mechanisms that control cardiogenesis. This has led to conflicting reports as to the locations of the heart forming fields in the embryo. Although it is not possible to reconcile the conflicting fate maps retrospectively, it is possible to weigh the discrepancies in terms of new molecular and genetic techniques. At best, this will provide an approximation of the heart forming fields, and there may still be surprises ahead as to sources of heart cells in the embryo. Recently, a secondary source of cardiomyocytes has been identified that contributes to the conotruncal myocardium of the outflow tract. The regulatory mechanisms that control primary cardiogenesis, including inductive interactions with endoderm and activation of early cardiac transcription factors, appear to be conserved in the secondary heart field. These tissue interactions and molecular regulatory mechanisms are not yet fully characterized for the secondary heart field, but extensive analysis of primary cardiogenesis has greatly facilitated these studies. Thus, the lessons of early cardiogenesis learned in the study of early stage avian embryos can be applied not only across species but also to novel sources of cardiomyocytes in the embryo.
Important gaps in our understanding of cardiac lineage development remain. The earliest events that control cardiac specification have proven to be particularly elusive. Many investigators working in a variety of developmental systems have been unsuccessful in identifying “factor X,” which confers cardiac potential on heart progenitor cells (Schultheiss et al., 1997). It is also very likely that critical regulators of cardiomyocyte lineage commitment and differentiation have not yet been found. Certainly our knowledge of the molecular mechanisms that control endothelial lineage development are incomplete and the initiators of atrial vs. ventricular lineage determination are not well characterized. Clues into each of these processes have been uncovered; however, there is still significant work to be done before the development of unspecified mesoderm into a beating heart is completely understood. This lack of information is obvious in our present inability to convert multipotential stem cells to a cardiac phenotype in a controlled and reliable manner.
We thank Lisa Ehrman for figure preparation and Melissa Colbert, Paul Bushdid, and Lisa Ehrman for helpful comments. K.E.Y. received an American Heart Association Established Investigator Award.