The past two decades have witnessed a sea of change in our understanding of the development of the heart. In the past, our knowledge of cardiac development was largely culled from investigations of serially sectioned embryos, using reconstructions to clarify the arrangement of regions difficult to interpret from the sections themselves. Using these techniques, it was not possible to trace the fate of given cells or areas of cells, nor to distinguish with accuracy the differentiation, if any, of the building blocks of the developing cardiac chambers. The development of molecular techniques has changed all that. Clonal analysis in the mammalian heart, using transgenic technology, permits tracking of the fate of daughters of any given cell (Meilhac et al., 2004). Use of Cre-recombinase, again when combined with transgenic technology, allows the tracking of a population of cells known to express a particular gene (Cai et al., 2003). The use of these techniques now has shown that the concept of cardiac development based on “segmentation” is far from accurate. Instead, the various parts of the developing heart are added sequentially to a linear tube derived from a primary cardiac crescent, with the initial tube itself providing the basis of the left ventricle in the formed heart (Cai et al., 2003; Meilhac et al., 2004). The studies that have made us rethink our overall concepts of development, leading to a model based on “ballooning” rather than “segmentation” (Moorman and Christoffels, 2003; Christoffels et al., 2004), are equally pertinent to concepts for development of the venous pole. In this review, therefore, we show the relevance of the new technologies in adding further clarity to our knowledge of the mechanism involved in formation of the systemic venous component of the right atrium, and the pulmonary venous component of the left atrium.
ORIGIN OF THE HEART
At the stage at which the embryo has become a disc, as seen in the chick at Hamburger and Hamilton stage 3, in the human at Carnegie stage 7 (approximately 16 days of development), and in the mouse as a cup at approximately embryonic day 6.5, the primitive streak is well established, with the node at its cranial extent (Fig. 1a). At this stage, as part of gastrulation, populations of cells migrate from the sides of the streak and occupy areas on the right and left sides of the mid-portion of the disc. These areas are commonly known as the heart-forming regions (Fig. 1b). The data derived using clonal analysis (Meilhac et al., 2004) have been interpreted to show that there are two waves of migrations from these regions (Fig. 1c). Cells migrating as the first lineage form the primary cardiac crescent, the part previously presumed to give rise to the entirety of the heart tube. The new data show that a second migration of cells forms an additional cardiogenic area, the secondary heart lineage. Evidence provided from the clonal analysis (Meilhac et al., 2004) and the Cre-recombinase experiments (Cai et al., 2003), shows that the primary cardiac crescent provides the precursor cells for the left ventricle and that cells derived from both lineages give rise to the atrial chambers and the right ventricle, albeit with most coming from the second lineage. The outflow tract is derived exclusively from the second lineage. How is all this new information pertinent to the development of the venous pole?
WHAT IS THE VENOUS POLE?
In lower vertebrates such as fishes, oxygenation takes place in the gills that are part of the systemic circulation. Other than in species such as the lungfish, there are no lungs found in these animals. The only venous structures, therefore, are the systemic veins. They enter a discrete chamber of the heart that is called the sinus venosus, defined as “a thin-walled sac, with little muscular tissue, essentially a place for the collection of venous blood….” (Romer and Parsons, 1986). In mammals, of course, the heart has become a four-chambered organ, with oxygenation taking place in the lungs and with separate formation of pulmonary and systemic veins. In normal mammalian hearts, the systemic venous tributaries all terminate in the morphologically right atrium. In the mouse, anatomic structures known as the venous valves demarcate an obvious junction between the component of the right atrium receiving the systemic veins and the remainder of the chamber (Fig. 2). In the left atrium, there are no discrete anatomic boundaries marking the pulmonary venoatrial junctions. In the mouse, the pulmonary veins open through a solitary orifice located in the left atrium adjacent to the atrioventricular junction (Fig. 2). In man, in contrast, the pulmonary veins have four or more discrete openings located at the corners of the atrial roof, albeit still without discrete anatomic junctions between the venous orifices and the body of the left atrium (Fig. 3).
During the early stages of development of the mammalian heart, specifically at day 9.5 in the mouse or day 20 in humans, there are no structures such as the venous valves in either man or mouse that mark any of the venoatrial junctions. When first formed, the inlet of the heart tube is relatively symmetrical, with two conduits returning the blood from the developing embryo (Fig. 4). At this stage, during the ninth day in mouse and before Carnegie stage 12 (approximately 4 weeks) in the human, the lungs have still to appear; hence, there are no pulmonary venous structures.
All of this is pertinent to the further development of the venous pole and the very nature of the embryonic sinus venosus. If we turn to well-recognized textbooks of embryology, we find that “the sinus venosus consists of the partially confluent left and right sinus horns….” (Larsen, 2005). Recent findings have shown, however, that the channels that, at this stage, drain the blood from the embryo to the developing heart will not persist as the venous tributaries, be they systemic or pulmonary (Soufan et al., 2004). In reality, the initial linear heart tube never possesses compartments that are analogous to the segments found in the body plan of lower animals or to the somites found within the developing embryos of all vertebrates. The areas that, at first sight, seem destined to become the systemic venous tributaries and that are usually recognized as the embryonic sinus venosus (Larsen, 2005) become incorporated at the caudal end of the primary tube as the primordium of the atrial component of the heart.
STRUCTURE OF THE PRIMARY HEART TUBE
The myocardial tissue forming the primary cardiac crescent becomes positioned in the cervical region of the developing embryo (Fig. 1). In the developing mouse, by the nine-somite stage, corresponding to the eighth day of development, the cells within this crescent have coalesced to form a tube. When initially formed, the tube is attached throughout its length to the body of the embryo through the so-called dorsal mesocardium, the attachment acting as a point of reference with which to divide the tube longitudinally into right and left halves (Fig. 5). At this initial stage, as already discussed, the junctions between the newly formed tube and the conduits draining blood from the embryo are relatively symmetrical and join the right and left halves of the inlet to the tube (Fig. 4). Because additional components are being added to the tube, this process occurring at its inlet and outlet extremities, the tube elongates. While it lengthens, it loses this initial symmetry in the process described as looping. As the tube bends within the pericardial cavity, the dorsal mesocardium disappears in the developing ventricular regions. Further development of the atrial and ventricular parts is then dependent on the process suitably described as ballooning. In the mouse, this process is seen during day 8.5 for the ventricular part and at day 9.5 for the atrial part (Christoffels et al., 2004).
Ballooning is the expansion of pouches from the cavity of the primary heart tube. The pouches balloon out in parallel from the newly formed atrial part of the tube at the inlet but in sequence from two areas of the developing ventricular part (Fig. 6). At the inlet of the tube, the incorporation of the draining channels forms the primordium for formation of both definitive atrial chambers. The atrial pouches then balloon anteriorly from both sides of this part of the tube. In the ventricular loop, in contrast, the pouches excavate in series from the outer curvature of the ventricular loop, leaving the cavity of the tube itself as the primary interventricular foramen, with the ballooning producing the primary ventricular septum between the pouches.
The myocardium that initially formed the walls of the heart tube is known to be slowly conducting and can be designated as primary myocardium (Moorman et al., 1998). The myocardium that forms the walls of the pouches, in contrast, conducts the cardiac impulse more rapidly. This formation along the heart tube of areas of slowly and rapidly conducting myocardium coincides with the ability to record an adult type of electrocardiogram from the developing embryo, this event occurring long before it is possible to distinguish an anatomically discrete conduction system (Moorman et al., 1998). In molecular terms, the primary myocardium can be distinguished from the myocardium that forms the walls of the pouches by the expression of connexin40 and natriuretic precursor peptide A, the latter also known as atrial natriuretic peptide (Delorme et al., 1997; Christoffels et al., 2000). The primary myocardium does not express either of these markers, whereas the chamber myocardium expresses both markers (Fig. 7).
These changes occurring at the inlet of the heart tube set the scene for the formation of the definitive right and left atrial chambers. At the earliest stage, as discussed, it seems that the systemic venous tributaries join directly with the heart tube. As also explained, however, the straight part of the primary tube seen at this early stage is destined to form little more than the definitive left ventricle. The parts initially seeming to be the venous tributaries become integrated into the heart as the primordium of a common atrium, with its right and left halves determined by the persisting connection to the body of the embryo through the dorsal mesocardium (Fig. 5). The atrial pouches balloon from this common atrium, one to the right and the other to the left of the developing arterial pole. Like the apical parts of the developing ventricles, these atrial pouches are made of secondary myocardium, which expresses both connexin40 and atrial natriuretic peptide (Fig. 7). Staining for the marker pitx2 (Franco et al., 2000) shows that it is only the pouch ballooned from the left side of the tube that expresses the genes determining morphological leftness. Eventually, this pouch will become the left atrial appendage, this being the part that also permits morphological distinction of the left atrium in postnatal life (Uemura et al., 1995). The pouch, ballooned from the right side of the primary atrium, in contrast, does not express pitx2, and is destined to become the morphologically right appendage. Because of their origin by ballooning from the same part of the primary tube, the atrial appendages, under suitable genetic control, can develop in an isomeric manner (Uemura et al., 1995; Brown and Anderson, 1999). Because the ventricular pouches develop in sequence and contain ventricular myocardium from the primary tube that is molecularly right and left, they do not respond in like manner to the genetic control (Brown and Anderson, 1999); therefore, morphological isomerism is extremely rare within the ventricles (Rinne et al., 2000).
FURTHER DEVELOPMENT OF THE ATRIAL CHAMBERS
The atrial appendages represent no more than ballooning from the common atrium formed from the primary heart tube. The definitive right and left atrial chambers of the mammalian heart, however, are also each connected separately to its own ventricle, and each possesses a discrete venous component. Furthermore, the two atria are separated from one another by the atrial septum. The connection to the ventricles is achieved by incorporating into the developing atrial chambers that part of the primary heart tube recognized initially as the atrioventricular canal. This component is made of primary myocardium and stains negatively for both connexin40 and atrial natriuretic factor (Christoffels et al., 2000; Soufan et al., 2004). Subsequent to division of the canal by fusion of the atrioventricular endocardial cushions and formation of the insulating tissues of the atrioventricular junctions, this atrioventricular canal myocardium becomes sequestered in the atrial chambers as the vestibules of the newly formed tricuspid and mitral valves (Wessels et al., 1996).
Formation of the venous components of the atrial chambers, in contrast, requires remodeling of the entire inlet of the primary tube, as illustrated in the reconstructions and schemes of the atrial part of the mouse heart (Fig. 8). At the stage of initial ballooning of the appendages from the primary atrial component, at day 9.5 in the mouse, the conduits returning the systemic venous blood to the heart were relatively symmetrical, albeit with a slight reduction seen even at this stage in the size of the left-sided venous channel (Figs. 4, 8). The myocardium forming the walls of these systemic venous channels is primary in nature, staining negatively for both connexin40 and atrial natriuretic factor (Soufan et al., 2004). Although the primary atrial component of the tube is itself joined by means of the dorsal mesocardial connection to the developing body wall at this stage (Fig. 5), there is no lumen extending into the mediastinum. This finding is equivalent to 4 weeks of human development, because the lungs have yet to form within the developing thorax (Blom et al., 2001). In the mouse at this stage, however, myocardium at the site of the dorsal mesocardium is already distinct from both the primary myocardium of the tube itself and the secondary myocardium of the appendages, because it stains positively for connexin40 but negatively for atrial natriuretic peptide (Soufan et al., 2004; Fig. 7). It is the nature of this myocardium, which can be termed mediastinal myocardium (Franco et al., 2000), and the temporal delay in appearance of a pulmonary vein compared with the systemic venous tributaries that help us determine the origin of the venous component of the morphologically left atrium.
ORIGIN OF THE PULMONARY VEIN
The origin of the pulmonary vein remains controversial. Although all now agree that the vein itself canalizes from a mid-pharyngeal strand at approximately 6 weeks of development in the human (Blom et al., 2001) and opens into the atrium between the ridges marking the site of the dorsal mesocardial connection, arguments continue as to whether or not this part of the developing atrium should be considered part of the sinus venosus (Hochstetter, 1908; Auër, 1941). Some argue that the area should be considered as part of the sinus venosus because the tissues surrounding the dorsal mesocardium, along with those surrounding the systemic venous tributaries, stain positively for an antibody to human natural killer cells (deRuiter et al., 1995). The staining properties of the antibody to the human killer cells, however, are also cited as evidence that cells are derived from the neural crest (Verberne et al., 1998) or that they are primordia of conducting tissues (Blom et al., 1999). In reality, these antibodies stain migrating populations of cells (Tucker et al., 1984; Kuratani and Kirby, 1991), including those derived from the neural crest, and those entering the heart to form the mediastinal myocardium. The neural crest, furthermore, also produces the parasympathetic neural input to the heart, which also enters the heart primarily through the venous pole. Positive staining of antibodies to human natural killer cells, therefore, does not prove that the myocardium surrounding the pulmonary vein is derived from the sinus venosus. In fact, our own findings using molecular markers show that the myocardium surrounding the pulmonary venous orifice, from the time of its first appearance within the developing atrium, stains positively for connexin40. Thus, it is readily distinguished from the primary myocardium surrounding the orifices of the systemic venous tributaries. This evidence, coupled with the timing of appearance of the pulmonary venous channel, at approximately Carnegie stage 12 in the human and during the ninth day in the mouse, shows that the pulmonary vein is a new structure. It does not take its origin from the sinus venosus, the latter never existing as a discrete compartment of the developing mammalian heart.
PULMONARY VEIN IS A NEW DEVELOPMENT
Once it has canalized in the mediastinum and gained its access to the heart, the pulmonary vein opens initially as a solitary orifice adjacent to the atrioventricular junctions. This is the arrangement not only in the mouse heart (Fig. 2) but also in the developing human heart (Fig. 9). The myocardium surrounding this area stains negatively for atrial natriuretic factor but positively for connexin40, indicating that it is fast-conducting and, hence, not part of the primary heart tube (Fig. 7). This mediastinal myocardium becomes integrated into the developing atrial chambers concomitant with the remodeling of the atrioventricular canal and also with remodeling of the systemic venous tributaries (Cai et al., 2003; Soufan et al., 2004; Fig. 8). It is the overall remodeling that sets the scene for atrial septation, with the primary atrial septum also taking its origin from the mediastinal myocardium, which then forms the bodies of both the definitive right and left atrial chambers. As part of the remodeling, with growth of the embryo, the channel draining the veins from the left side of the body decreases significantly in size. It becomes incorporated into the developing left atrioventricular groove as the left superior caval vein in the mouse (Fig. 10) and eventually as the coronary sinus in man (Fig. 9). As it becomes an integral part of the left atrioventricular groove, the venous channel retains its own discrete walls (Figs. 9, 10) but comes to open within the confines of the systemic venous component of the right atrium (Knauth et al., 2002). It is only by the time that it has shifted to open exclusively within the right atrium that the junction between the systemic venous tributaries and the developing right atrium become anatomically discrete, this being heralded by the appearance of the so-called venous valves (Fig. 2). By this time, when judged relative to the extent of the secondary and mediastinal myocardium, the primary myocardium of the atrial component of the tube has decreased markedly in size (Soufan et al., 2004). It now forms a corridor of tissue in the inferior and posterior wall of the developing right atrium, running from the atrioventricular canal to the so-called septum spurium. The rightward shift of the primary myocardial corridor, along with the incorporation of the atrioventricular canal, takes place as the mediastinal myocardium expands to form the posterior walls and roofs of the newly forming right and left atrial chambers. The primary atrial septum, known to be mediastinal myocardium because it stains positively for connexin40 but negatively for atrial natriuretic peptide, then grows toward the atrioventricular cushions to divide the two atrial chambers. The primary septum, along with the strip of tissue forming the posterior wall of the right atrium to the level of the left venous valve, also stains positively for pitx2, showing that it is molecularly left (Franco et al., 2000). The primary atrial septum, having grown to the right side of the pulmonary venous orifice, places the newly developed pulmonary venous component within the morphologically left atrium (Webb et al., 1998, 2000). The mediastinal myocardium, therefore, is always distinct from the sinus venosus. When the openings of the systemic venous tributaries are first recognized as discrete structures, subsequent to formation of the venous valves, they are exclusively part of the morphologically right atrium. The myocardium of the pulmonary venous component stains positively for connexin40, showing that, unlike the mouths of the systemic venous tributaries, it has never been part of the primary heart tube.
In the mouse heart, the pulmonary vein retains its position as a solitary orifice adjacent to the left atrioventricular junction (Fig. 2). In the human heart, in contrast, subsequent to the division of the primary atrial chamber into its right and left parts by growth of the primary atrial septum, the pulmonary veins remodel so as to gain separate entrances at the four corners of the roof of the body of the left atrium (Fig. 3). It is only as the right pulmonary veins gain their entrance to the atrial roof that, in the human heart, we find infolding of the wall between their mouths and the orifices of the systemic venous sinus entering the right atrium (Webb et al., 2001). This infolding produces the so-called septum secundum, in reality the superior interatrial fold (Röse, 1899; Anderson et al., 2002). In the mouse heart, this so-called septum is an insignificant structure, because the pulmonary veins open into the left atrium through a solitary orifice adjacent to the atrioventricular junction (Fig. 2). In humans, there is a specific anomaly involving the right upper pulmonary vein that is known as a sinus venosus defect. In this lesion, a hole is found joining together the atrial chambers but outside the confines of the atrial septum (Al Zaghal et al., 1997). The hole cannot exist until the right pulmonary veins have achieved their definitive position on the dome of the left atrium subsequent to the completion of atrial septation. It is spurious, therefore, to argue that evidence from the so-called sinus venosus defect lends credence to the notion that the pulmonary vein originates from the embryonic systemic venous sinus (Blom et al., 2001). Any explanation of formation of the so-called sinus venosus defect (Anderson et al., 2004) must take account of the mechanisms of incorporation of the mediastinal myocardium to form the bodies of both atrial chambers, along with the primary atrial septum and the pulmonary component of the morphologically left atrium.
The evidence accruing over the past decade has mandated a re-thinking of concepts of development of the heart tube, in particular its venous pole. The heart tube never contains the so-called segments described in classic accounts. Instead, the primary myocardium derived from the cardiac crescent is reinforced by other contributions of primary myocardium from the anterior, or secondary, heart field, these forming the right ventricle and outflow tract at the outlet of the tube, and the atrioventricular canal, atrial component, and systemic venous tributaries at the inlet to the tube. The myocardial components forming the parts of the chambers that permit their unequivocal recognition in postnatal life, namely the appendages, are derived by ballooning from the heart tube. For their completion, however, the atrial chambers require incorporation of the atrioventricular canal to form the vestibules of the atrioventricular valves, along with incorporation of mediastinal myocardium through the dorsal mesocardium. It is this mediastinal myocardium that provides the site of formation of the bodies of both atrial chambers, the growth of the primary atrial septum, and, equally significantly, the site of access for the pulmonary vein to canalize and grow into the morphologically left atrium. The pulmonary vein, therefore, has never been part of the so-called sinus venosus, nor does it take its origin from the systemic venous tributaries.
We thank Dr. V.M. Christoffels for critical discussions and providing the photographs for Figure 7; Mrs. A.T. Soufan, P.A.J. de Boer, and J. Hagoort for making the reconstructions presented in Figure 8; and Mrs. D. Berkhof for expert secretarial assistance. R.H.A. and N.A.B. were funded by the British Heart Foundation, R.H.A. was funded by the Joseph Levy Foundation and the Great Ormond Street Hospital, and A.F.M.M. was funded by The Netherlands Heart Foundation.