Much has been learnt since the second edition of the book was published. It remains a fact that, as was emphasised at the start of this chapter in the second edition, from the functional point of view the heart is simply a specialised part of the vascular system. The development of the heart as a specialised pump, nonetheless, is obviously of great significance, as is the formation of a coelomic cavity around the developing organ so as to aid its action. We have learnt a great deal over the past decade regarding the origin of the muscular parts of this pumping organ. In the previous edition, emphasis was placed on the so-called segments of the developing heart tube, since it was believed that the initial linear tube contained the precursors of the components as seen in the postnatal heart. We now know that this is not the case, and that tissue is continually added to the heart tube as it grows and loops. The initial straight part of the tube ( Fig. 3-1 ) eventually forms little more than the left ventricle. This knowledge now permits us better to interpret the morphogenesis of many congenital cardiac malformations. Embryology, therefore, is no longer a hindrance in this regard, as one of us stated somewhat controversially over 2 decades ago. 1 It is the new evidence that has emerged concerning the appearance of the cardiac components that we will emphasise in the opening sections of our chapter. Thereafter, we will revert to providing an account of the various cardiac segments, as was done in the second edition. 2 In this respect, we should emphasise, again as was done in the second edition, that we do not use segment in its biological sense when describing the development of the cardiac components. Thus, we do not imply that each purported segment is identical to the others, as is seen in invertebrates such as annelids. As explained in Chapter 1 , however, the so-called segmental approach is now the preferred means of describing the cascade of information acquired in interrogating cardiac structure during the diagnostic process. As in the previous edition, therefore, we will continue to describe the cardiac components as segments, hoping to provide the necessary background to understand the anomalous development that leads to congenitally malformed hearts with abnormal connections between them. It is also the case that there are discrepancies between the terms used by biologists to describe developing heart and the attitudinally appropriate terms used by clinicians when describing the formed organ. Biologists and embryologists use the term anterior to describe structures that are towards the head, and posterior for those towards the feet. We circumvent these problems by describing cranial and caudal structures. So as to avoid confusion, we also need to avoid use of these terms when describing those structures located towards the spine and sternum, as is the wont of clinicians. For this purpose, we will use the adjectives dorsal and ventral. Right and left, of course, retain their time-honoured usage.
ORIGIN OF THE HEART TUBE
Recent molecular studies 3–5 have validated the time-honoured concept 6–8 that, after formation of the linear heart tube (see Fig. 3-1 ), cells are continuously added at both its venous and arterial poles. The source of this new material has been called the second heart field, albeit that the difference between this purported second field and the presumptive primary field has not adequately been defined. 9 Despite this lack of adequate definition, the rediscovered information has revolutionised our understanding of the building plan of the heart. Acceptance of this concept of temporal addition of new material to the heart has also helped our understanding of cardiac morphogenesis, but new questions arise. Are there two developmental fields as opposed to just one, or three, or perhaps more? How are the alleged fields distinguished from the surrounding tissues, and from each other? Do the fields represent territories of gene expression, or of morphogenetic signaling? Most importantly, do the alleged first and second fields represent distinct cellular lineages, the descendants of which form distinct compartments within the definitive heart? An understanding of the mechanics of formation of the cardiac tube can provide some answers to these questions. 9
Subsequent to gastrulation, the embryo possesses three germ layers. These are the ectoderm, which faces the amniotic cavity, the endoderm, which faces the yolk sac, and the intermediate mesodermal layer ( Fig. 3-2 ). At this stage, we can recognise the so-called embryonic disc, marked by the junction of the developing embryo itself with the extra-embryonic tissues formed by the amnion and yolk sac. It is subsequent folding of this disc, concomitant with extensive growth, that gives the embryo its characteristic shape. Eventually, the original junction between the disc and the extraembryonic tissues becomes the navel of the embryo. As part of this process of remodelling, the parts of the disc initially positioned peripherally attain a ventral location within the embryo. At the initial stage, the region where ectoderm and mesoderm face one another without interposing mesoderm, is the stomato-pharyngeal membrane, which closes the orifice of the developing mouth. This membrane is flanked centrally by pharyngeal mesoderm that, in turn, is bordered peripherally by the cardiogenic mesoderm. The cardiac area is itself contiguous with the mesoderm of the transverse septum, in which will develop the liver. This transverse septum is the most peripheral part of the mesodermal layer of the embryonic disc and, after the completion of folding, it is located cranial to the navel.
During the process of gastrulation, the cells that will form the heart migrate from the anterior part of the primitive streak. Subsequent to their migration, they give rise to two heart-forming regions in the mesodermal germ layer. These areas are positioned on either side of the midline. 10–12 With continuing development, they join across the midline to form a crescent-shaped area of epithelium ( Fig. 3-3 ). It is this tissue which subsequently provides the material for the heart tube, albeit in complex fashion. Initially, the crescent becomes a trough, which starts to close dorsally along the margins closest to the developing spine. Starting at the level of the part of the tube that will eventually become the left ventricle, closure proceeds by a process of zipping, the edges of the trough coming together in both cranial and caudal directions ( Fig. 3-4 ). After the process of folding, the part of the crescent that was initially positioned centrally retains this position, becoming positioned medio-dorsally in the formed body. The peripheral part of the cresent, in contrast, becomes the ventral part of the definitive heart tube. It is the centro-medial part of the initial heart-forming area that several investigators have described as the secondary, or anterior, heart field. 3–5 Its location permits it to contribute to those parts of the heart that develop at the arterial pole, specifically the right ventricle and outflow tract, and also to those which will form at the venous pole. At the venous pole, there is formation of the so-called mediastinal myocardium, 13 this area eventually providing the site of entry for the pulmonary veins. 14,15 The peripheral part of the initial crescent flanks the transverse septum, which is the site of formation of the liver, and the location of the termination of the systemic venous tributaries.
It follows from the description given above that the tissues directly adjacent to the cardiac crescent are made of pharyngeal mesoderm. From the outset, therefore, the tissues that will form the right ventricle and outflow tract at the arterial pole are directly adjacent to the developing pharyngeal arches. This pharyngeal mesoderm, also containing cells derived from the so-called second heart field, eventually provides the tissues not only for the myocardial components of the right ventricle and outflow tract, but also for the non-myocardial intrapericardial arterial trunks and their valves and sinuses.
It has been analyses of molecular lineage that have demonstrated in unambiguous fashion that myocardium is added to the already functioning myocardial heart tube. 16,17 Thus, use of these techniques revealed a similar programme to that observed in the differentiation of the primary heart field, involving the transcription factors Nkx2-5, Gata4 and Mef2c, as well as fibroblast growth factors and bone morphogenetic proteins. 5 Differences in opinion remain regarding the structures formed from the so-called second field. One group has argued that this field forms only the outflow tract, 4 but another group claims that, as well as the outflow tract, the field also gives rise to the non-myocardial but intrapericardial portion of the arterial pole. 5 Such differences may be more apparent than real, since boundaries between morphological regions are not necessarily formed at subsequent stages by the same cells as were present initially. Interpretation has also been hampered by assumptions that, in avian hearts, the right ventricular precursors are derived from the initial linear heart tube, which contains the precursors of only the left ventricle in mammals. 17 It is now known that, in the developing avian heart, as in the mammalian heart, if the straight tube is labelled at the pericardial reflection of the arterial pole, 18 the entirety of the right ventricle is seen to be added to the linear heart tube later in development, consistent with the data derived in the mouse. 3
Some authors 19 have already suggested that there is complex patterning in the primary heart field, rather than the existence of multiple fields. Based on our interpretation of the processes of folding that lead to the formation of the linear heart tube (see Fig. 3-4 ), we endorse this notion. 9 Acceptance of the notion, nonetheless, carries with it the implication that the cells within this solitary field initially have the capacity to form all parts of the heart, depending on their position in the field. This, in turn, is in keeping with current views on cellular diversity, since differences in the concentration of diffusing morphogens can create a number of different fates for a given cell, promoting diversity within a field that was initially homogeneous.
Even if we accept that the material from which the heart is formed is derived from the same basic field, there is an obvious temporal order in the differentiation of the alleged first and second cardiac heart-forming regions. This order does no more than reflect the evolutionary development of the cardiovascular system. When initially developed during evolution of the animal kingdom, the heart contained no more than the components of the systemic circulation, namely an atrium with left and right appendages, a ventricle, and a myocardial outflow tract. The pulmonary circulation, represented by the right ventricle, and the dorsal atrial wall, including the atrial septum, appears appreciably later in evolutionary development. Within the evolutionary tree, it is in the lungfish that the pulmonary veins are first seen to unite, and drain directly to the left atrium. This sets the scene for a separate pulmonary circulation, and also for the appearance of the atrial septum. It cannot be coincidental, therefore, that the atrial septum in mammals, along with the dorsal atrial wall, is formed from mediastinal myocardium. 13 This mediastinal myocardium, along with the right ventricle, is added to the heart relatively late in its development. The evolutionary considerations suggest strongly that novel patterning, with different temporal sequences, but within a single heart field, is sufficient to provide all the material needed to construct the four-chambered hearts of birds and mammals, albeit that not all precursors are present in the linear heart tube when it first appears (see Fig. 3-1 ).
FORMATION OF THE CARDIAC LOOP
When first seen, the developing heart is more or less a straight tube (see Fig. 3-1 ). Very soon, it becomes S-shaped ( Fig. 3-5 ). The changes involved in producing the bends are described as looping. It had been thought that the curvatures produced were the consequence of rapid growth of the tube within a pericardial cavity that expanded much more slowly. 20 Experiments showed that the tube continues to loop even when deprived of its normal arterial and venous attachments, 21 and also loops when no longer beating, ruling out the role of haemodynamics as a morphogenetic factor. 22 Looping, therefore, is an intrinsic feature of the heart itself, albeit that the exact cause has still to be determined. Be that as it may, the tube usually curves to the right. This rightward turning is independent of the overall left–right asymmetry of the developing embryo. It is often said that rightward looping is the first sign of breaking of cardiac symmetry. This is incorrect. Asymmetry is evident in the structure of the atrioventricular canal when it is first seen (see Fig. 3-1 ), 23 but asymmetry can even be seen in the morphology and extent of the heart-forming fields. 13
THE CARDIAC COMPONENTS
Only after looping of the heart tube has taken place is it possible to recognise the appearance of the building blocks of the cardiac chambers, along with the primordiums of the arterial trunks, and the venous tributaries. In the second edition of this book, it was suggested that five segments could be seen in the developing tube, on the basis of constrictions between the various parts, and that these constrictions played important roles in further development. As has been explained above, we now know that this is not the case.
The development of the cardiac chambers depends on ballooning of their cavities from the lumen of the primary heart tube. Addition of new material from the heart-forming area produces the primordiums of the right ventricle and outflow tract at the arterial pole, whilst addition of material at the venous pole produces initially the atrioventricular canal, followed by the atrial primordium, to which drain the systemic venous tributaries. The addition of this material is part and parcel of the appearance of the looped tube. Subsequent to looping, the cells making up the larger part of the tube are negative for both connexin40 and atrial natriuretic factor, permitting them to be labelled as primary myocardium ( Fig. 3-6 ).
As the parts of the cavity begin to balloon out from both the atrial and ventricular components of the tube, so does the myocardium forming the walls of the ballooning components change its molecular nature, being positive for both connexin40 and atrial natriuretic peptide ( Fig. 3-7 ). This myocardium is called chamber, or secondary, myocardium. The parts ballooning from the atrial component do so symmetrically with the newly formed pouches appearing to either side of the outflow tract. The pouches will eventually become the atrial appendages. Examination of the atrial component of the heart at this early stage, however, reveals the presence of a third population of cells. These cells are positive for connexin40, but negative for atrial natriuretic peptide. They make up the part of the tube that retains its connection with the developing mediastinum through the dorsal mesocardium, and hence are described as being mediastinal myocardium. 13 These cells will eventually form the dorsal wall of the left atrium, or pulmonary venous component, a small part of the dorsal wall of the right atrium bordered at the right side by the left venous valve and at the left side by the primary atrial septum, which also is derived from the mediastinal component ( Fig. 3-8 ). From the stance of lineage, the cells are derived from the mesenchyme of the mediastinum that surrounds the developing lung buds. 24 There is also ballooning of cavities from the ventricular part of the heart tube. Unlike the situation in the atrial component, where the appendages of both definitive atriums balloon in parallel, the pouches that will eventually form the apexes of the left and right ventricles balloon in sequence from the ventricular loop ( Fig. 3-9 ). The apical part of the developing left ventricle balloons from the inlet part of the loop, whilst the apical part of the developing right ventricle takes its origin from the outlet part of the loop. Concomitant with the ballooning of these two parts, so the muscular ventricular septum is formed between the apical components (see Fig. 3-9 ), albeit that the cells of the septum largely belong to the left ventricle. The cells making up the walls of the ballooned segments are composed of secondary, or chamber, myocardium (see Fig. 3-6 ). Ballooning of the chamber myocardium sets the scene for septation of the atrial and ventricular chambers. For this to be achieved, however, appreciable remodelling is required in the initial cavity of the primary heart tube. This is because, subsequent to looping and the initial phases of ballooning, the circumference of the atrioventricular canal is attached in its larger part to the inlet of the heart tube, albeit that a direct connection already exists from its right side to the developing outflow tract ( Fig. 3-10 ).
DEVELOPMENT OF THE VENOUS COMPONENTS
Subsequent to the process of looping, the venous pole of the heart tube shows a good degree of symmetry. Venous tributaries from both sides of the embryo itself, along with bilateral channels from the yolk sac, and from the placenta, drain into the atrial component of the tube through confluent orifices ( Fig. 3-11 ). These structures are often described as the sinus venosus, with the channels draining to the atrial component identified as the horns of this venous sinus. Such a discrete component of the heart is to be found in lower animals, such as fish. No anatomically discrete structure is seen, however, in the early stages of development of the mammalian heart. 14 The venous tributaries on both sides of the embryo simply empty into the atrial component through the confluent right- and left-sided channels (see Fig. 3-11 ). At the initial stages, there are no landmarks indicating the junction of these venous channels with the atrial component. It is not until the systemic venous tributaries have remoulded so as to drain asymmetrically to the right side of the atrial component that structures are seen demarcating their borders, these structures then being recognised as the valves of the systemic venous sinus ( Fig. 3-12 ). A key part of normal development, therefore, is remoulding of the systemic venous tributaries so that they open exclusively to the right side of the developing atrial component. This process involves the formation of anastomoses between right- and left-sided components of the various venous systems so that the left-sided venous return is shunted to the right side of the embryo. The major anastomosis formed in the caudal part of the embryo results in all the umbilical venous return from the placenta being diverted to the caudal part of the cardinal venous system. This anastomotic channel persists as the venous duct. The vitelline veins largely disappear, with some of these structures being incorporated into the venous system of the liver as this structure develops in the transverse septum. A second important anastomosis develops in the cranial part of the embryo, the left brachio-cephalic vein, this channel serving to divert the venous return from the left-sided to the right-sided cardinal vein. With this shift of the cranial venous return to the right-sided cardinal channel, and with the disappearance of the left-sided vitelline and umbilical veins, there is gradual diminution in size of the atrial orifice of the left-sided venous confluence. As this structure diminishes in size, so its walls become incorporated into the left half of the developing atrioventricular junction ( Fig. 3-13 ). It has been suggested that a left sinuatrial wall is required to separate this channel from the left side of the developing atrium. 25 This is not the case. The left venous confluence simply retains its own walls as it is incorporated into the developing left atrioventricular groove. Subsequent to the rightward shift of the systemic venous orifices, the valvar structures appear, which then permit anatomical distinction between the systemic venous sinus and the remainder of the developing right atrium. The cranial and caudal right-sided cardinal veins, along with the orifice of the left systemic venous confluence, now open within their confines.
Remodelling of the systemic venous sinus also sets the scene for development of the pulmonary venous system. The pulmonary veins, of course, cannot appear until the lungs themselves have formed. These develop as buds on the ends of the bifurcating tracheo-bronchial tube, this structure extending from the ventral aspect of the gut, the lungs developing in the ventral part of the mediastinal mesenchyme. A further venous channel then develops from a mid-line strand formed within the mediastinal tissues. 26 This channel, when canalised, drains the developing intrapulmonary venous plexues from both lungs, and joins the heart at the site of the persisting part of the dorsal mesocardium ( Fig. 3-14 ; see also Fig. 3-13 ). This part of the mesocardium persists as most of the initial structure breaks down during looping, and continues to attach the most dorsal part of the atrial component of the heart tube to the mediastinum. When the channel is viewed internally, its edges are seen as two ridges that bulge into the lumen of the atrial cavity (see Figs. 3-11 and 3-12 ). These are the pulmonary ridges. After the pulmonary venous channel has canalised within the mediastinum, it opens to the atrial cavity between these ridges, appearing initially as a midline structure, with its opening directly adjacent to the developing atrioventricular junction ( Figs. 3-15 and 3-16 ).
For over a century there has been controversy as to the relationship between this newly formed pulmonary venous confluence and the tributaries of the systemic venous sinus. From the stance of the morphology, 15,27,28 and the lineage of the cells forming the pulmonary vein, 13,24,29 the evidence is now overwhelming that, during normal development, the pulmonary venous structure has never had any connection with the systemic venous tributaries. It forms as a new structure within the mediastinum, and opens within the mediastinal myocardium to the left atrium. As such, it is positioned to the left of the site of appearance of the primary atrial septum, which is also derived from mediastinal myocardium (see Fig. 3-8 ). The pulmonary venous structures are also recognisable from the outset as being derived from mediastinal myocardium, whereas the systemic venous tributaries initially possess a primary myocardial lineage. We have now also shown that the systemic venous tributaries can be identified in molecular terms by their expression of the transcription factor Tbx18. The pulmonary veins, in contrast, do not contain this protein. 30 When the pulmonary vein first appears, it is a mid-line structure which drains to the heart directly adjacent to the atrioventricular junction. It is only appreciably later in the development in the human heart that the venous component of the left atrium is remodelled so that, at first, separate orifices appear to drain the blood from the right and left lungs ( Fig. 3-17 ). And it is then later still, indeed not until the completion of atrial septation, that the pulmonary venous component achieves its definitive position at the roof of the left atrium, with separate orifices on both sides for the superior and inferior veins from each of the two lungs ( Fig. 3-18 ).
Only at this stage, when there are four pulmonary venous orifices, is it possible to see formation of the so-called secondary atrial septum. This so-called septum, in the postnatal heart, is no more than the fold between the right-sided pulmonary veins and the systemic venous tributaries ( Fig. 3-19 ). It is not produced during development until the pulmonary channels are incorporated into the atrial roof (see Fig. 3-18 ). Such a superior interatrial fold is lacking in abnormal human hearts having totally anomalous pulmonary venous connection.
SEPTATION OF THE ATRIAL CHAMBERS
With the initial rightward shift of the tributaries of the systemic venous sinus, the stage is set for septation of the atrial component of the heart. As the systemic venous sinus reorientates relative to the developing atrium, the addition of the new mediastinal myocardium forms the larger part of the body of the developing atrial component (see Fig. 3-8 ). The atrioventricular canal, of course, was present from the outset, and is composed of primary myocardium. The myocardium of the atrial component itself was also initially composed of primary myocardium, but as we have shown, the two appendages bud dorsocranially in symmetrical lateral fashion from this lumen, passing to either side of the developing outflow tract. With the rightward shift of the systemic venous tributaries, and the appearance of the mediastinal myocardium, so there is also a rightward shift of the dorsal corridor of primary myocardium that continues to form the floor of the systemic venous sinus. It is at this stage that we first see the appearance of the primary atrial septum, or septum primum, which grows as an interatrial shelf from the atrial roof (see Fig. 3-16 ). By the time that the primary atrial septum appears, endocardial cushions have also developed within the atrioventricular canal, these structure growing towards each other so as eventually to divide the canal itself into right-sided and left-sided channels ( Fig. 3-20 ). As the cushions grow towards each other to divide the canal, so the primary septum grows towards the cushions, carrying on its leading edge a further collection of endocardial tissue, the so-called mesenchymal cap. By the time the primary septum and mesenchymal cap approach the cushions, the cranial part of the septum, at its origin from the atrial roof, has broken down, creating the secondary interatrial foramen. The primary foramen is the space between the mesenchymal cap and the fusing atrioventricular endocardial cushions ( Fig. 3-21 ). It is then fusion of the mesenchymal cap with the endocardial cushions that obliterates the primary atrial foramen. This process occurs to the right side of the pulmonary ridges, so that the solitary opening of the newly canalised pulmonary vein is committed to the left side of the dividing atrial component, the orifices of the systemic venous tributaries, enclosed within the systemic venous valves, obviously being committed to the right side of the atrium by this selfsame process. The base of the newly formed atrial septum, formed by the mesenchymal cap, is then further reinforced by growth into the heart of mesenchymal tissue through the right pulmonary ridge (see Fig. 3-14 ). This process was initially illustrated by Wilhelm His the Elder, who called the protrusion seen in the caudal wall of the atrium the vestibular spine, or spina vestibuli. 31
The mesenchymal tissue of the spine, together with the mesenchymal cap on the primary septum, then muscularises to form the buttress at the base of the atrial septum, anchoring the septum firmly against the central fibrous body, itself formed from the fused atrioventricular cushions (see Figs. 3-17 and 3-18 ). The so-called sinus septum is no more than the bifurcation between the orifice of the inferior caval vein and the coronary sinus within the confines of the valves of the systemic venous sinus. The right valve itself persists to varying degrees in the definitive heart, remaining as the Eustachian valve adjacent to the opening of the inferior caval vein, and the Thebesian valve in relation to the opening of the coronary sinus. These two valves join together in the musculature of the so-called sinus septum and, in the definitive heart, a fibrous structure extends through the newly muscularised buttress at the base of the atrial septum (see Fig. 3-17 ), extending forward to insert into the central fibrous body. This is the tendon of Todaro, an important landmark to the site of the atrioventricular conduction axis.
Subsequent to these changes, the base of the atrial septum separates the newly formed right and left atriums, but a foramen is still present at the atrial roof. This hole is an essential part of the fetal circulation, permitting the richly oxygenated placental return to reach the left side of the developing heart so as to pass to the developing brain. As explained, not until the pulmonary veins are incorporated into the atrial roof does the upper margin of this foramen become converted into the interatrial fold, often described inaccurately as the secondary atrial septum. It is the formation of the fold that provides the buttress for the flap valve of the definitive oval foramen, the flap itself being formed by the primary atrial septum (see Fig. 3-19 ). This process is not completed until well after the finish of definitive cardiac septation. 28
After the formation of the primary atrial septum, and its reinforcement by the vestibular spine, the atrial chambers have effectively been separated one from the other. As also explained, the septum initially grows between the site of the left venous valve and the orifice of the newly formed pulmonary vein. All the tissue to the left of the venous valve is mediastinal myocardium (see Fig. 3-8 ), and it is this tissue which represents the body of the developing atrium. The larger part of this body is, therefore, committed to the definitive left atrium subsequent to atrial septation. If we summarise the development of the atrial components, each atrium possesses a part of the body derived from mediastinal myocardium, with the larger part committed to the morphologically left atrium. Each atrium also possesses an appendage, formed by budding from primary myocardium of the heart tube and differentiation into secondary, or chamber, myocardium, and a vestibule, derived from the initial primary myocardium of the atrioventricular canal. The venous components, in contrast, have disparate origins. The systemic venous myocardial component is formed by differentiation of Tbx18-positive mesenchyme into myocardium. The pulmonary venous myocardial component, in contrast, is derived, along with the atrial septum, from Islet1-positive mediastinal mesenchyme, itself developing from what some call the secondary heart field.
THE ATRIOVENTRICULAR CANAL
In early stages, the junction between the developing atrial component and the inlet of the ventricular loop is indeed a canal with a finite length (see Fig. 3-15 ). The canal itself is septated by fusion of the superior and inferior atrioventricular endocardial cushions (see Fig. 3-20 ). His 31 described the fused cushions as producing the intermediate septum. It is this part of the septum that is buttressed by the ingrowth of the vestibular spine, with muscularisation of the spine and mesenchymal cap forming the prominent infero-anterior rim of the oval fossa seen in the definitive heart (see Figs. 3-17 and 3-18 ). The cushions themselves, as we will describe subsequently, provide the foundations for formation of the aortic leaflet of the mitral valve, and the septal leaflet of the tricuspid valve. They also contribute to closure of the interventricular foramen, forming in the process the membranous part of the septum. Only subsequent to delamination of the septal leaflet of the tricuspid valve, a relatively late event, does this part become separated into atrioventricular and interventricular portions. 32
Much has been learned in recent years concerning the endothelial to mesenchymal transformations that take place during formation of the atrioventricular cushions. 33 These events are not directly relevant to an understanding of cardiac development. It was thought, in the past, that failure of fusion of the cushions underscored the development of hearts with atrioventricular septal defect and common atrioventricular junction, for quite some time these lesions being labelled as endocardial cushion defects. 34 We now know that abnormal hearts can develop with all the features of atrioventricular septal defect subsequent to fusion of the cushions, albeit that the cushions themselves are abnormal. The problem underscoring the abnormality is one that permits the retention of the common junction, probably involving abnormal formation of the vestibular spine. 35,36 Irrespective of such niceties, there is no question but that, during normal development, the atrioventricular cushions fuse with each other to divide the atrioventricular canal into right-sided and left-sided channels (see Fig. 3-20 ). With ongoing development, part of the musculature of the atrioventricular canal becomes sequestrated as the atrial vestibules. We know this because of studies made using an antibody prepared against the nodose ganglion of the chicken. 37 These studies proved that this antibody, subsequent to formation of the ventricular loop, marked serendipitously a ring of cells in the developing human heart that surrounded the primary foramen ( Fig. 3-22 ). The marked area extended from the crest of the muscular ventricular septum and included the right side of the atrioventricular canal. When human embryos were then studied subsequent to formation of the right atrioventricular junction, and commitment of the right atrium to the right ventricle, the area of the atrioventricular canal initially seen to have been marked by the antibody was located in the vestibule of the right atrium, and was separated from the ventricular myocardium by the forming insulating tissues of the right atrioventricular junction ( Fig. 3-23 ). The insights provided by the studies using this antibody showed, therefore, that part of the atrioventricular canal musculature became sequestrated as the atrial vestibules with ongoing development. The studies also provided important insights into remodelling of the primary interventricular foramen, which we will discuss further in our next section. We now know that the parietal wall of the inlet of the left ventricle is also derived from the initial primary myocardium of the atrioventricular canal, at least in the mouse heart. This knowledge comes from the development of a mutant mouse in which the gene Tbx2 was used to mark the lineage of the initial myocardium in the atrioventricular canal ( Fig. 3-24 ). The studies, as yet unpublished, showed that, whilst the parietal wall of the left ventricular inlet is marked by the gene-encoded product the greater part of the septum is free from marked cells. The septum, therefore, along with the apical component develops from the chamber myocardium of the left embryonic ventricle.
FURTHER DEVELOPMENT OF THE VENTRICULAR LOOP
The confusion that existed concerning the way in which the ventricular loop became converted into the definitive ventricles, 38 and which received considerable attention in the previous edition of this book, 2 has been resolved in part by the use of descriptive terms for the inlet and outlet parts of the loop, and in part by recognition that the apical parts of the ventricles balloon out in series from the lumen of the primary tube, the apical part of the left ventricle ballooning from the inlet, and the right ventricular apical part from the outlet (see Fig. 3-9 ). It had long been recognised that functional separation of the left-sided and right-sided bloodstreams had taken place long before the completion of ventricular septation, so that two parallel bloodstreams, instead of a single one, traverse the serial segments (see Fig. 3-10 ). Development of the ventricles simply proceeds by partitioning these bloodstreams so that the one originating from the right side of the atrioventricular canal becomes channeled to the pulmonary trunk, whilst the one commencing at the left side of the atrioventricular canal is committed to the aorta. In addition to requiring marked remodelling of the inner heart curvature, this process also requires appropriate septation of the outlet component of the primary heart tube.
Expansion of the right side of the atrioventricular junction is sufficient to place the cavity of the right atrium in more direct communication with the apical part ballooned from the outlet part of the ventricular loop. The mechanics of this expansion are well illustrated by the fate of the ring of cells marked by the antibody to the nodose ganglion of the chick, and illustrated in Figs. 3-22 and 3-23 . 37 The fate of this ring of marked cells also shows that, during ongoing development, the outlet part of the heart tube is reorientated so that its dorsal half becomes the outlet from the left ventricle. Concomitant with this reorientation, the proximal parts of the cushions that have developed and fused to divide the outlet segment of the primary tube into pulmonary and aortic channels are brought into alignment with the crest of the muscular ventricular septum ( Fig. 3-25 ), the latter structure, as we have already shown, being formed by the apical ballooning of the chamber myocardium of the right and left ventricles (see Fig. 3-9 ). This remodelling of the cavity of the initial heart tube, providing an effective inlet to the apical part of the right ventricle, and an outlet for the apical part of the left ventricle, then permits the middle part of the initial foramen to be closed by apposition of tissue derived from both the atrioventricular and outlet endocardial cushions 39,40 ( Fig. 3-26 ).
Before moving on to consider the formation of the atrioventricular valves, obviously crucial features of the definitive ventricles, it is convenient first to discuss the changes that take place with the outlet segment of the heart tube, and the arteries it feeds within the developing pharyngeal mesenchyme.
THE OUTLET SEGMENT
It is, perhaps, the outlet segment of the heart about which we have learnt most since the appearance of the second edition of our book. 2 In the first place, as we have now emphasised several times, we now know that this part of the developing heart is derived from a secondary source, different at least in temporal terms from that producing the initial linear heart tube. In the second place, although the larger parts of the intrapericardial outflow tracts in the definitive heart have arterial walls, the entirety of the outlet segment, when initially formed, has walls made of myocardium. In the third place, we now know that the so-called aortic sac is little more than a manifold giving rise to the arteries that extend through the arches of the pharyngeal mesenchyme. In previous accounts, notions of division of this sac, and the outlet segment itself, by growth of an aortopulmonary septum have been grossly exaggerated. We also now know that migration of cells from the neural crest is crucial for normal development of this part of the heart, and for its separation into the pulmonary and aortic channels. In this section, therefore, we will seek to correlate our own findings and observations with currently existing concepts of development.
The distal extent of the developing outlet segment of the heart tube, subsequent to the completion of ventricular looping, is marked by the margins of the pericardial cavity. At this stage of development, the outlet is a tube with a solitary lumen, taking its origin from the outlet of the ventricular loop. At its distal extent, marked by the pericardial reflections, the lumen becomes confluent with the area known as the aortic sac ( Fig. 3-27 ). At this early stage, the walls of the tube are exclusively myocardial. Endocardium is then formed in the luminal lining, again by a process of mesenchymal to endothelial transformation as occurs with the atrioventricular cushions, but the endocardial jelly initially lines the entirety of the tube in circumferential fashion. When viewed externally at this stage, the tube has an obvious bend, permitting the distinction of proximal and distal parts ( Fig. 3-28 ). At the margins of the pericardial cavity, the lumen is continuous with the space in the ventral pharyngeal mesenchyme from which originate the arteries that initially run symmetrically through the developing pharyngeal arches. These arteries encircle the gut and the developing tracheobronchial groove, uniting dorsally to form the descending aorta. Although it is frequent for cartoons representing this stage to show five pairs of arteries, in reality there are never more than two or three pairs of arches, along with their arteries, to be seen at any one time. At the earliest stage, at least in the mouse, the mediastinal space has right and left horns, with each horn giving rise to the arteries of the first to third arches ( Fig. 3-29 ). These arteries rapidly become assimilated into the arterial system of the head and face. By the time it becomes possible to recognise the arteries of the fourth and sixth arches, the arteries within the fourth arch are feeding the arteries of the third arches, and it is no longer possible to recognise the arteries of the initial two arches as encircling structures ( Fig. 3-30 ). The cavity of the so-called aortic sac by this stage is little more than the continuation of the lumen of the outlet segment beyond the pericardial boundaries. It is still possible, nonetheless, to recognise obvious proximal and distal parts of the outlet segment, with a marked dog-leg bend between them. Within the lumen, throughout the tract, the endocardial tissue has now thickened to form opposing cushions, or ridges. When traced proximally to distally, the ridges spiral. The ridge that is parietal at the proximal end of the outlet turns beneath the other ridge at the bend, and achieves a caudal location within the distal outflow tract. The ridge that is septal proximally spirals to become positioned cranially at the distal extent of the outlet. This means that, as the ridges approximate one another, fusing along their facing surfaces, the proximal outflow tract will eventually be separated into ventral and dorsal channels, whilst fusion of the ridges distally will produce right-sided and left-sided channels. Fusion of the ridges, or cushions, however, does not occur at the same time, but rather commences distally, with the act of closure moving in proximal direction. Prior to the commencement of fusion, important changes have also taken place in the aortic sac. There is marked diminution in size of the right-sided arteries running from the sac to join the descending aorta ( Fig. 3-31 ).
As the dorsal parts of the arteries running with the right pharyngeal arches begin to involute, so it becomes possible to recognise the developing pulmonary arteries, which course caudally within the ventral mesenchyme of the mediastinum to feed the rapidly growing lung buds. The effect of these changes is that the aortic sac now gives rise to only two sets of arteries, located cranially and caudally, with the orifices of the left-sided arteries being appreciably larger than those seen to the right, particularly for the arteries running within the sixth arch. Throughout this process of remoulding, it has been the dorsal wall of the pharyngeal mesenchyme, separating the origins of the pairs of arteries running through the fourth and sixth arches, which represents the so-called aortopulmonary septum. It is this tissue that separates the flow from the cranial part of the outflow tract and aortic sac to the fourth arch and the developing brachiocephalic arteries from that flowing to the caudal part, which now feeds the two pulmonary arteries and the artery of the left sixth arch, representing the arterial duct, the right sixth arch having involuted (see Fig. 3-31 ). Concomitant with the remodelling of the arteries within the pharyngeal arches, cells have begun to grow from the pharyngeal mesenchyme into the distal ends of the outlet segment, growing parietally between the ends of the distal ridges or cushions to replace the initially myocardial walls. The tongues of tissue thus formed, which rapidly arterialise, produce at the same time a fishmouth appearance for the remaining myocardial margins of the outflow tract, as initially emphasised by Bartelings and Gittenberger-de Groot 41 ( Fig. 3-32 ). The tongues run between the distal extents of the cranial and caudal ridge, which reach almost to the margins of the pericardial cavity.
