Embryonic Hematopoiesis

Hemangiogenic precursor cells first arise in the posterolateral mesoderm during gastrulation and from there migrate to the earliest blood-forming organs (Fig. 17.1). Under the influence of Runx-1, some of their progeny follow the hematopoietic lineage, whereas others, responding to Hoxa3, enter the endothelial lineage. Still other progeny will enter a third lineage and eventually form vascular smooth muscle cells. Although blood cell formation (hematopoiesis) begins in the yolk sac, the yolk sac–derived cells are soon replaced by blood cells that are independently derived from other sites of hematopoiesis (Fig. 17.2).

The blood islands contain pluripotential hematopoietic stem cells, which can give rise to most types of cells found in the embryonic blood. The erythrocytes produced in the yolk sac are large nucleated cells that enter the bloodstream just before the heart tube begins to beat at about 22 days’ gestation. For the first 6 weeks, the circulating erythrocytes are largely yolk sac derived, but during that time, preparations for the next stages of hematopoiesis are taking place.

Analysis of human embryos has shown that, starting at 28 days, definitive intraembryonic hematopoiesis begins in small clusters of cells (para-aortic clusters) in the splanchnopleuric mesoderm associated with the ventral wall of the dorsal aorta and shortly thereafter in the aorta/genital ridge/mesonephros (AGM) region. Precursor cells from the AGM region make their way via the blood to blood-forming sites in the liver, the yolk sac, and the placenta. Hematopoietic stem cells formed in the AGM, the yolk sac, and the placenta become transported to the liver via the circulation to the liver (see Fig. 17.1). By 5 to 6 weeks of gestation, sites of hematopoiesis become prominent in the liver. In both the yolk sac and the early sites of embryonic hematopoiesis, the endothelial cells themselves briefly retain the capacity for producing blood-forming cells. There is now evidence that in the AGM region, nitric oxide gas signaling, resulting from shear stress caused by blood flow on the endothelial cells, can induce their transformation into hematopoietic stem cells.

The erythrocytes produced by the liver are quite different from the erythrocytes derived from the yolk sac. Although still considerably larger than normal adult red blood cells, liver-derived erythrocytes are non-nucleated and contain different types of hemoglobin. By 6 to 8 weeks of gestation in humans, the liver replaces the yolk sac as the main source of blood cells. Although the liver continues to produce red blood cells until the early neonatal period, its contribution begins to decline in the sixth month of pregnancy. At this time, the formation of blood cells shifts to the bone marrow, the definitive site of adult hematopoiesis. This shift is controlled by cortisol secreted by the fetal adrenal cortex. In the absence of cortisol, hematopoiesis remains confined to the liver. Before hematopoiesis becomes well established in the bone marrow, small amounts of blood formation may also occur in the omentum and possibly the spleen.

Cellular Aspects of Hematopoiesis

The first hematopoietic stem cells that arise in the embryo are truly pluripotential in that they can give rise to all the cell types found in the blood (Fig. 17.3). These pluripotent stem cells, sometimes called hemocytoblasts, have great proliferative ability. They produce vast numbers of progeny, most of which are cells at the next stage of differentiation, but they also produce small numbers of their original stem cell type, which act as a reserve capable of replenishing individual lines of cells should the need arise. Very early in development, the line of active blood-forming cells subdivides into two separate lineages. Lymphoid stem cells ultimately form the two lines of lymphocytes: B lymphocytes (which are responsible for antibody production) and T lymphocytes (which are responsible for cellular immune reactions). Myeloid stem cells are precursors to the other lines of blood cells: erythrocytes, the granulocytes (neutrophils, eosinophils, and basophils), monocytes, and platelets. The second-generation stem cells (lymphoid and myeloid) are still pluripotent, although their developmental potency is restricted because neither lymphoid cells nor myeloid cells can form the progeny of the other type.

Stemming from their behavior in certain experimental situations, the hematopoietic stem cells are often called colony-forming units (CFUs). The first-generation stem cell is called the CFU-ML because it can give rise to myeloid and lymphoid lines of cells. Stem cells of the second generation are called CFU-L (lymphocytes) and CFU-S (spleen) (determined from experiments in which stem cell differentiation was studied in irradiated spleens). In some cases, the progeny of CFU-ML and CFU-S are committed stem cells, which are capable of forming only one type of mature blood cell. For each lineage, the forming cell types must pass through several stages of differentiation before they attain their mature phenotype.

What controls the diversification of stem cells into specific cell lines? Experiments begun in the 1970s provided evidence for the existence of specific colony-stimulating factors (CSFs) for each line of blood cell. CSFs are diffusible proteins that stimulate the proliferation of hematopoietic stem cells. Some CSFs act on several types of stem cells; others stimulate only one type. Although much remains to be learned about the sites of origin and modes of action of CSFs, many CSFs seem to be produced locally in stromal cells of the bone marrow, and some may be stored on the local extracellular matrix. CSFs are bound by small numbers of surface receptors on their target stem cells. Functionally, CSFs represent mechanisms for stimulating the expansion of specific types of blood cells when the need arises. Recognition of the existence of CSFs has prompted considerable interest in their clinical application to conditions characterized by a deficiency of white blood cells (leukopenia).

Certain Hox genes, especially those of the Hoxa and Hoxb families, play an important role in some aspects of hematopoiesis. Exposure of bone marrow to antisense oligonucleotides against specific Hox genes results in the suppression of specific lines of differentiation of blood cells. Conversely, engineered overexpression of genes, such as Hoxb8, Hoxa9, and Hoxa10, causes leukemia in mice. Evidence is increasing for the involvement of Hox genes in the pathogenesis of human leukemias. One important function of the Hox genes in hematopoiesis is the regulation of proliferation. Several growth factors, especially bone morphogenetic protein-4 (BMP-4), Indian hedgehog, and Wnt proteins, are important in stimulating and maintaining hematopoietic stem cell activity.

Erythropoiesis

Red blood cell formation (erythropoiesis) occurs in three waves during the embryonic period. The first wave begins with precursors within the yolk sac, which produce primitive nucleated erythrocytes that mature within the bloodstream. The second wave also begins in the yolk sac, but the precursor cells then colonize the embryonic liver and produce the first of a generation of definitive fetal erythrocytes that are dominant during the prenatal period. The third wave consists of precursor cells that enter the liver from the AGM mesoderm and the placenta. Some of these definitive erythroid progenitor cells send progeny directly from the liver into the bloodstream as definitive fetal erythrocytes. Others seed the bone marrow and produce adult-type erythrocytes later in the fetal period.

The erythrocyte lineage represents one line of descent from the CFU-S cells. Although the erythroid progenitor cells are restricted to forming only red blood cells, there are many generations of precursor cells (Fig. 17.4). The earliest stages of erythropoiesis are recognized by the behavior of the precursor cells in culture, rather than by morphological or biochemical differences. These are called erythroid burst-forming units (BFU-E) and erythroid CFUs (CFU-E). Each responds to different stimulatory factors. The pluripotent CFU-S precursors (see Fig. 17.3) respond to interleukin-3, a product of macrophages in adult bone marrow. A hormone designated as burst-promoting activity stimulates mitosis of the BFU-E precursors (see Fig. 17.4). A CFU-E cell, which has a lesser proliferative capacity than a BFU-E cell, requires the presence of erythropoietin as a stimulatory factor.

Erythropoietin is a glycoprotein that stimulates the synthesis of the mRNA for globin and is first produced in the fetal liver. Later in development, synthesis shifts to the kidney, which remains the site of erythropoietin production in adults. Under conditions of hypoxia (e.g., from blood loss or high altitudes), the production of erythropoietin by the kidneys increases, thereby stimulating the production of more red blood cells to compensate for the increased need. In adult erythropoiesis, the CFU-E stage seems to be the one most responsive to environmental influences. The placenta is apparently impervious to erythropoietin, and this property insulates the embryo from changes in erythropoietin levels of the mother and eliminates the influence of fetal erythropoietin on the blood-forming apparatus of the mother.

One or two generations after the CFU-E stage, successive generations of erythrocyte precursor cells can be recognized by their morphology. The first recognizable stage is the proerythroblast (Fig. 17.5), a large, highly basophilic cell that has not yet produced sufficient hemoglobin to be detected by cytochemical analysis. Such a cell has a large nucleolus, much uncondensed nuclear chromatin, numerous ribosomes, and a high concentration of globin mRNAs. These are classic cytological characteristics of an undifferentiated cell.

Succeeding stages of erythroid differentiation (basophilic, polychromatophilic, and orthochromatic erythroblasts) are characterized by a progressive change in the balance between the accumulation of newly synthesized hemoglobin and the decline of first the RNA-producing machinery and later the protein-synthesizing apparatus. The overall size of the cell decreases, and the nucleus becomes increasingly pyknotic (smaller with more condensed chromatin) until it is finally extruded at the stage of the orthochromatic erythrocyte. After the loss of the nucleus and most cytoplasmic organelles, the immature red blood cell, which still contains a small number of polysomes, is a reticulocyte. Reticulocytes are released into the bloodstream, where they continue to produce small amounts of hemoglobin for 1 or 2 days.

The final stage of hematopoiesis is the mature erythrocyte, which is a terminally differentiated cell because of the loss of its nucleus and most of its cytoplasmic organelles. Erythrocytes in embryos are larger than their adult counterparts and have a shorter life span (50 to 70 days in the fetus versus 120 days in adults).

Hemoglobin Synthesis and its Control

Both the red blood cells and the hemoglobin within them undergo isoform transitions during embryonic development. The adult hemoglobin molecule is a complex composed of heme and four globin chains: two α and two β chains. The α and β subunits are products of genes located on chromosomes 16 and 11 (Fig. 17.6). Different isoforms of the subunits are encoded linearly on these chromosomes.

During the period of yolk sac hematopoiesis, embryonic globin isoforms are produced. The earliest embryonic hemoglobin, sometimes called Gower 1, is composed of two ζ (α-type) and two ε (β-type) chains. After passing through a couple of transitional forms (Table 17.1), hemoglobin synthesis enters a fetal stage by 12 weeks, which corresponds to the shift in the site of erythropoiesis from the yolk sac to the liver. Fetal hemoglobin consists of two adult-type α chains, which form very early in embryogenesis, and two γ chains, the major fetal isoform of the β chain. Fetal hemoglobin is the predominant form during the remainder of pregnancy. The main adaptive value of the fetal isoform of hemoglobin is that it has a higher affinity for oxygen than the adult form. This is advantageous to the fetus, which depends on the oxygen concentration of the maternal blood. Starting at about 30 weeks’ gestation, there is a gradual switch from the fetal to the adult type of hemoglobin, with α₂β₂ being the predominant type. A minor but functionally similar variant is α₂δ₂.

Formation of Embryonic Blood Vessels

The early embryo is devoid of blood vessels. Although blood islands appear in the wall of the yolk sac, and extraembryonic vascular channels form in association with them (see Fig. 6.19), much of the vasculature of the embryonic body is derived from intraembryonic sources. During the early period of somite formation, networks of small vessels rapidly appear in many regions of the embryonic body.

The formation of blood vessels in the embryo consists of several phases (Fig. 17.7). The first is the specification of a population of vascular precursors, called angioblasts. These cells become organized into a primary capillary plexus through a process known as vasculogenesis. To keep pace with the rapidly growing embryo, the primary capillary plexus must rapidly undergo reorganization through the resorption of existing vessels and the sprouting of new branches to support the expanding vascular network. This latter process is called angiogenesis. Angiogenesis continues not only in the prenatal period, but also throughout adult life, as tissues and organs continually adapt to changing conditions of life, whether normal or pathological.

Detailed descriptive studies and transplantation experiments involving intrinsic cellular labels or graft-specific monoclonal antibody labels have shown that angioblasts arise from most mesodermal tissues of the body, except notochord and prechordal mesoderm (Table 17.2). Embryonic blood vessels form from angioblasts by three main mechanisms. Many of the larger blood vessels, such as the dorsal aortae, are formed by the coalescence of angioblasts in situ. Other equally large channels, such as the endocardium, are formed by angioblasts migrating into the region from other sites. Other vessels, especially the intersegmental vessels of the main body axis and vessels of the central nervous system, arise as vascular sprouts from existing larger vessels. Many of the angioblasts of the trunk are originally associated with the splanchnic mesoderm.

The developmental processes leading to the initial formation of the aorta are beginning to be understood. The endothelium of the early paired aortae is derived from splanchnopleure and requires an interaction with the underlying endoderm for its development. While the aortae are still in the paired stage, somite-derived cells contribute to their dorsal walls. Concomitantly, the ventral splanchnopleure-derived endothelium begins to give rise to clusters of hematopoietic stem cells. Then the dorsal somite-derived endothelial cells overgrow the ventral splanchnopleure-derived endothelial cells. When this is completed, hematopoiesis in the aorta ceases.

All stages in the formation of the vascular system occur in response to the influence of powerful growth factors and their receptors. The initial phase of recruitment of a population of angioblasts from the mesoderm is characterized by the appearance of a transmembrane vascular endothelial growth factor receptor (VEGFR-2) on their surfaces (see Fig. 17.7). Soon, in response to the production of vascular endothelial growth factor (VEGF-A) by the surrounding mesenchyme, the phase of vasculogenesis occurs, and the angioblasts form the cellular tubes that become the basis for the primary capillary plexus.

The formation of vascular endothelial sprouts, the cellular basis for angiogenesis, occurs against a background of VEGF/VEGFR-1 and VEGF/VEGFR-2 interactions, but with a new set of players added. A sprouting factor, angiopoietin-1, interacts with its receptor, Tie-2, on the endothelial cells at sites where endothelial sprouts will occur. The Notch signaling pathway is also strongly tied to the formation of vascular sprouts (a common denominator with other organ systems that display branching morphogenesis), but its connection to the angiopoietin-1/Tie-2 mechanism remains unclear.

The next step in building a blood vessel is formation of the vascular wall, which in the trunk and extremities is derived from local mesoderm that becomes associated with the endothelial lining of the vessel. In the head and many areas of the aortic arch system, mesenchyme derived from neural crest ectoderm is a major contributor to the connective tissue and smooth muscle of the vascular wall. The neural crest, however, does not give rise to endothelial cells.

Two-way molecular signaling is involved in building up the walls of blood vessels. In response to the angiopoietin-1/Tie-2 interaction that occurs during angiogenesis, the endothelial cells release their own signaling molecule, platelet-derived growth factor, which stimulates the migration of mesenchymal cells toward the vascular endothelium. The release of other growth factors (transforming growth factor-β [TGF-β] and myocardin, a master regulator of smooth muscle formation) by the endothelial cells stimulates the differentiation of the mesenchymal cells into vascular smooth muscle or pericytes.

Research has shed considerable light on the differentiation of the arterial versus venous system. The arterial or venous identity of endothelial cells is established very early in their development, before angiogenesis and before the onset of circulation. The endothelial cells of developing arteries express the membrane-bound ligand Ephrin-B2, whereas the endothelial cells of developing veins express the receptor Eph-B4 on their surface membranes. These characteristic phenotypes are the results of different signaling cascades. Arterial vessels are the first to differentiate from generic endothelial background, and a set of signals, starting with sonic hedgehog (shh), finally leads to the acquisition of an arterial phenotype (Fig. 17.8). Notch, one of the links in that cascade, not only causes progression of the sequence of arterial differentiation leading to Ephrin-B2 expression, but it inhibits the expression of Eph-4 and the pathway leading to the venous phenotype. In what was earlier assumed to be a purely default mode, venous differentiation occurs under the influence of COUP-TFII (Chicken ovalbumin upstream promoter-transcription factor II), which suppresses the arterial pathway by inhibiting Notch signaling, but is a determinant of venous differentiation by acting upstream of Eph-4. Under the influence of Sox-18 and Prox-1, lymphatic vessels form and branch off from the veins. Prox-1 is a master regulator of venous identity. Later, physiological and local factors play a role in the differentiation of blood vessels. When the flow of blood to the yolk sac is greatly reduced, vessels slated to become arteries develop venous characteristics; correspondingly, developing veins exposed to high blood pressure transform into arteries.

As with myoblasts, angioblasts seem to react to local environmental cues that determine the specific morphological pattern of a blood vessel. An unexpected finding is that the pattern of the peripheral innervation often determines the pattern of the smaller arteries. The growing tips of endothelial cell buds and axonal growth cones of outgrowing nerve fibers contain receptors that respond remarkably similarly to the major families of environmental ligands (Table 17.3). VEGF, secreted by the nerve fibers, acts as an effective patterning agent for the blood vessels. The smooth muscle cells of the developing arteries secrete a factor, artemin, that guides the extension of sympathetic nerve fibers along the vessel wall.

Tracing studies of transplanted angioblasts have shown that some can of these cells migrate long distances. Angioblasts that have migrated far from the place into which they were grafted become integrated into morphologically normal blood vessels in the areas where they settle.

Local factors also influence the initiation of vasculogenesis. In some organs (e.g., the liver) or parts of organs (e.g., the bronchi of the respiratory system), the blood vessels supplying the regions arise from local mesoderm, whereas other organs (e.g., the metanephric kidneys) or parts of organs (e.g., the alveoli of the lungs) are supplied by blood vessels that grow into the mesenchyme from other tissues. In the latter type of vascularization mechanism, evidence is increasing that these organ primordia produce their own angiogenesis factors that stimulate the growth of vascular sprouts (by promoting mitosis of endothelial cells) into the glandular mesenchyme. Nearby blood vessels, in turn, influence the morphogenesis and differentiation of many structures (e.g., pancreas, glomerulus, liver) with which they are associated.

The dorsal aorta forms from the direct aggregation of endothelial precursor cells derived from the lateral plate mesoderm. These cells form a vessel directly by a vasculogenesis mechanism. Vasculogenesis is stimulated by VEGF and other factors produced by the endoderm and BMP in the lateral mesoderm. When first formed, the cranial part of the dorsal aorta is a paired, with each member located lateral to the midline. The reason is that in the midline the notochord secretes the BMP antagonists noggin and chordin, which inhibit the activity of BMP and also inactivate the vasculogenic influences from the endoderm. Late in the fourth week, hematopoietic stem cells form in the lining of the ventral part of the aorta (see p. 409).

The system of aortic arches in early human embryos is organized along the same principles as the system of arteries supplying blood to the gills of many aquatic lower vertebrates. Blood exits from a common ventricle in the heart into a ventral aortic root, from which it is distributed through the branchial arches by pairs of aortic arches (Fig. 17.9A). In gilled vertebrates, the aortic arch arteries branch into capillary beds, where the blood becomes reoxygenated as it passes through the gills. In mammalian embryos, the aortic arches remain continuous vessels because gas exchange occurs in the placenta and not in the pharyngeal arches. The aortic arches empty into paired dorsal aortae where the blood enters the regular systemic circulation. In human embryos, all aortic arches are never present at the same time. Their formation and remodeling show a pronounced craniocaudal gradient. Blood from the outflow tract of the heart (the truncoconal region) flows into an aortic sac, which differs from the truncoconal region in the construction of its wall. The aortic arches branch off from the aortic sac.

The developmental anatomy of the aortic arch system illustrates well the principle of morphological adaptation of the vascular bed during different stages of embryogenesis (Table 17.4). Continued development of the cranial and cervical regions causes components of the first three arches and associated aortic roots to be remodeled into the carotid artery system (see Fig. 17.9). With the remodeling of the heart tube and the internal division of the outflow tract into aortic and pulmonary components, the fourth arches undergo an asymmetrical adaptation to the early asymmetry of the heart. The left fourth aortic arch is retained as a major channel (arch of the aorta), which carries the entire output from the left ventricle of the heart. The right fourth arch is incorporated into the right subclavian artery.

Embryology textbooks traditionally depict the aortic arch system as consisting of six pairs of vascular arches, but the fifth and sixth arches never appear as discrete vascular channels similar to the first through fourth arches. The fifth aortic arch, if it exists at all, is represented by no more than a few capillary loops. The sixth (pulmonary arch) arises as a capillary plexus associated with the early trachea and lung buds. The capillary plexus is supplied by ventral segmental arteries arising from the paired dorsal aortae in that region (Fig. 17.10). The equivalent of the sixth arch is represented by a discrete distal segment (ventral segmental artery) connected to the dorsal aorta and a plexuslike proximal segment that establishes a connection between the aortic sac at the base of the fourth arch and the distal segmental component. As the respiratory diverticulum and early lung buds elongate, parts of the pulmonary capillary network consolidate to form a pair of discrete pulmonary arteries that connect to the putative sixth arch. Although the term sixth aortic arch is frequently used in anatomical and clinical literature, pulmonary arch is a more appropriate term because it does not imply equivalence to the other aortic arches.

Similar to the fourth aortic arch, the pulmonary arch develops asymmetrically. On the left side, it becomes a large channel. Its distal segment, which was derived from a ventral segmental artery, persists as a major channel (ductus arteriosus) that shunts blood from the left pulmonary artery to the aorta (see Fig. 17.9C). The lungs are protected by this shunt from a flow of blood that is greater than what their vasculature can handle during most of the intrauterine period. On the right side, the distal segment of the pulmonary arch regresses, and the proximal segment (the base of the right pulmonary artery) branches off from the pulmonary trunk.

The asymmetry of the derivatives of the pulmonary arch accounts for the difference between the course of the right and left recurrent laryngeal nerves, which are branches of the vagus nerve (cranial nerve X). These nerves, which supply the larynx, hook around the pulmonary arches. As the heart descends into the thoracic cavity from the cervical region, the branch point from the vagus of each recurrent laryngeal nerve is correspondingly moved. On the left side, the nerve is associated with the ductus arteriosus (see Fig. 17.9C), which persists throughout the fetal period, so it is pulled deep into the thoracic cavity. On the right side, with the regression of much of the right pulmonary arch, the nerve moves to the level of the fourth arch, which constitutes an anatomical barrier. The positions of the right and left recurrent laryngeal nerves in an adult reflect this asymmetry, with the right nerve curving under the right subclavian artery (fourth arch) and the left nerve hooking around the ligamentum arteriosum (the adult derivative of the ductus arteriosus, the distal segment of the left pulmonary arch).

As the aortic arches become modulated into their adult configuration, specialized baroreceptors begin to form. One of these is the carotid sinus, which is located in the proximal portion of each internal carotid artery and is innervated by fibers of the glossopharyngeal nerve. Similarly, baroreceptors located in the proximal part of the right subclavian artery and in the aortic arch between the left common carotid and subclavian arteries are innervated by branches of the vagus and recurrent laryngeal nerves. Chemosensory structures associated with the aortic arches are the paired carotid bodies, which are located at the bifurcation of the internal and external carotid arteries and are innervated by a sensory branch of the glossopharyngeal nerve and sympathetic nerve fibers from the superior cervical ganglion.

Major Branches of the Aorta

In the early embryo, when the dorsal aortae are still paired vessels, three sets of arterial branches arise from them—dorsal intersegmental, lateral segmental, and ventral segmental (Fig. 17.11). These branches undergo a variety of modifications in form before assuming their adult configurations (Table 17.5). The ventral segmental arteries arise as paired vessels that course over the dorsal and lateral walls of the gut and yolk sac. With the closure of the gut and the narrowing of the dorsal mesentery, certain branches fuse in the midline to form the celiac, superior, and inferior mesenteric arteries.

Table 17.5

Major Arterial Branches of the Aorta

Embryonic Vessels	Adult Derivatives
Dorsal Intersegmental Branches (Paired)
Cervical intersegmental (1-16)	Lateral branches joining to become vertebral arteries
Seventh intersegmentals	Subclavian arteries
Thoracic intersegmentals	Intercostal arteries
Lumbar intersegmentals	Iliac arteries
Lateral Segmental Branches
Up to 20 pairs of vessels supplying the mesonephros	Adrenal arteries, renal arteries, gonadal (ovarian or spermatic) arteries
Ventral Segmental Branches*
Vitelline vessels	Celiac artery, superior and inferior mesenteric arteries
Allantoic vessels	Umbilical arteries

^*Originally paired in areas where the embryonic aorta itself consists of paired components.

The umbilical arteries begin as pure ventral segmental branches supplying allantoic mesoderm, but their bases later connect with lumbar intersegmental vessels. The most proximal umbilical channels then regress, and the intersegmental branches become their main branches off the aorta. Like their subclavian counterparts in the arms, the initially small arterial branches (iliac arteries) supplying the leg buds appear as components of the dorsal intersegmental (lumbar) branches of the aorta. After the umbilical arteries incorporate the proximal segments of the intersegmental vessels, the iliac arteries arise as branches off the umbilical arteries.

Arteries of the Head

The arteries supplying the head arise from two sources. Ventrally, the aortic arch system (first to third arches and corresponding roots) gives rise to the arteries supplying the face (external carotid arteries) and the frontal part of the base of the brain (internal carotid arteries) (see Fig. 17.9).

At the level of the spinal cord, the vertebral arteries, which form through connections of lateral branches of dorsal intersegmental arteries, grow toward the brain. Soon, they veer toward the midline and fuse with the basilar artery, which has already formed earlier from the fusion of the bilateral longitudinal neural arteries (Fig. 17.12). The longitudinal neural arteries, which run along the ventral sides of the brain from the midbrain through the caudal end of the hindbrain, represent the major arterial supply to the brain during the fourth week of development. The basilar artery runs along the ventral surface of the brainstem and supplies the brainstem with a series of paired arteries. As the basilar artery approaches the level of the diencephalon and the internal carotid arteries, sets of branches from each of these major vessels grow out and fuse, to form posterior communicating arteries, which join the circulations of the basilar and internal carotid arteries. Two other small branches off the internal carotid system fuse in the midline to complete a vascular ring (circle of Willis), which underlies the base of the diencephalon and encircles the optic chiasm and pituitary stalk. The circle of Willis is a structural adaptation that ensures a continuous blood supply in the event of occlusion of some major arteries supplying the brain. In addition, it is a structural landmark that marks the transition between mesodermally derived blood vessels of the trunk and those of the head, which are primarily of neural crest origin.

Development of Veins

Veins follow a morphologically complex pattern of development characterized by the formation of highly irregular networks of capillaries and the ultimate expansion of certain channels into definitive veins. Because of the multichanneled beginnings and the number of options (Fig. 17.13), the adult venous system is characterized by a higher incidence of anatomical variations than the arterial system. A detailed description of the development of venous channels is beyond the scope of this text.

Cardinal Veins

The cardinal veins form the basis for the intraembryonic venous circulation. Several sets of cardinal veins appear at different times and in different locations. Within any set of cardinal veins, some segments regress, and others persist, either as independent channels or as components of composite veins that also include portions of other cardinal veins.

The earliest pattern of cardinal veins consists of paired anterior and posterior cardinal veins, which drain blood from the head and body into a pair of short common cardinal veins (see Fig. 17.13). The common cardinal veins, in turn, empty their blood into the sinus venosus of the primitive heart (see Fig. 17.15).

In the cranial region, the originally symmetrical anterior cardinal veins are transformed into the internal jugular veins (Fig. 17.14). As the heart rotates to the right, the base of the left internal jugular vein is attenuated. At the same time, a new anastomotic channel, which ultimately forms the left brachiocephalic vein, connects the left internal jugular vein to the right one. Through this anastomosis, the blood from the left side of the head is drained into the original right anterior cardinal vein, which ultimately becomes the superior vena cava, emptying into the right atrium of the heart. Meanwhile, the proximal part of the left common cardinal vein persists as a small channel, the coronary sinus, which is the final drainage pathway of many of the coronary veins, also into the right atrium of the heart.

In the trunk, a pair of subcardinal veins arises in association with the developing mesonephros. The subcardinal veins are connected with the posterior cardinal veins and to each other through numerous anastomoses. Both the postcardinal and the subcardinal veins drain the mesonephric kidneys through numerous small side branches. As the mesonephric kidneys start to regress, the veins draining them also begin to break up. At this point, a pair of supracardinal veins appears in the body wall dorsal to the subcardinal veins. Over time, all three sets of cardinal veins in the body break up to varying degrees, with surviving remnants incorporated into the inferior vena cava. The inferior vena cava forms as a single asymmetrical vessel that runs parallel to the aorta on the right (see Fig. 17.13). Most of the named veins of the thoracic and abdominal cavities are derived from persisting segments of the cardinal vein system.

Vitelline and Umbilical Veins

The extraembryonic vitelline and umbilical veins begin as pairs of symmetrical vessels that drain separately into the sinus venosus of the heart (Fig. 17.15). Over time, these vessels become intimately associated with the rapidly growing liver. The vitelline veins, which drain the yolk sac, develop sets of anastomosing channels within and outside the liver. Outside the liver, the two vitelline veins and their side-to-side anastomotic channels become closely associated with the duodenum. Through the persistence of some channels and the disappearance of others, the hepatic portal vein, which drains the intestines, takes shape. Within the liver, the vitelline plexus becomes transformed into a capillary bed that allows the broad distribution of food materials absorbed from the gut through the functional parts of the liver. From the hepatic capillary bed, the blood that arrives from the hepatic portal vein passes into a set of hepatic veins, which empty the blood into the sinus venosus.

The originally symmetrical umbilical veins soon lose their own hepatic segments and drain directly into the liver by combining with the intrahepatic vascular plexus of the vitelline veins. Soon, a major channel, the ductus venosus, forms and shunts much of the blood entering from the left umbilical vein directly through the liver and into the inferior vena cava. The ductus venosus is an important adaptation for maintaining a functional embryonic pattern of blood circulation. Shortly thereafter, the right umbilical vein degenerates, leaving the left umbilical vein the sole channel for bringing blood that has been reoxygenated and purified in the placenta back to the embryonic body. The ductus venosus permits the incoming oxygenated placental blood to bypass the capillary networks of the liver and to distribute it to the organs (e.g., brain, heart) that need it most.

Pulmonary Veins

The pulmonary veins are phylogenetically recent structures that form independently, rather than taking over portions of the older cardinal vein systems. From each lung, venous drainage channels converge until they ultimately form a single large common pulmonary vein, which empties into the left atrium of the heart. As the atrium expands, the common pulmonary vein becomes incorporated into its wall (Fig. 17.16). Ultimately, the absorption passes the first and second branch points of the original pulmonary veins, with resulting entrance of four independent pulmonary veins into the left atrium.

Development of Lymphatic Channels

The lymphatic system arises by cells budding off from the lateral surfaces of the anterior and posterior cardinal veins (see Fig. 17.8). As they are preparing to bud off from the veins, the prelymphatic endothelial cells express the homeobox genes Sox18 and Prox-1. This is followed by the expression of VEGFR-3 and their specification to become lymphatic endothelial cells. These cells respond to two specific isoforms of VEGF: VEGF-C and VEGF-D. The emigrating lymphatic endothelial cells aggregate in areas where the lateral plate mesoderm produces VEGF-C to form primary lymph sacs. The first of these are the two jugular lymph sacs arising from the anterior cardinal veins and abdominal lymph sacs, which arise from the endothelium of the posterior cardinal veins (Fig. 17.17). In the abdomen, a retroperitoneal lymph sac forms on the posterior body wall at the root of the mesentery during the eighth week. Later, a cisterna chyli forms at the same level, but dorsal to the aorta. At about the same time, a pair of posterior lymph sacs arises at the bifurcation of the femoral and sciatic veins. By the end of the ninth week, lymphatic vessels connect these lymph sacs.

Two major lymphatic vessels connect the cisterna chyli with the jugular lymph sacs. An anastomosis forms between these two channels. A single lymphatic vessel consisting of the caudal part of the right channel, the anastomotic segment, and the cranial part of the left channel ultimately becomes the definitive thoracic duct of an adult. The thoracic duct drains lymph from most of the body and the left side of the head into the venous system at the junction of the left internal jugular and subclavian veins. The right lymphatic duct, which drains the right side of the head and upper part of the thorax and right arm, also empties into the venous system at the original location of the right jugular sac. In contrast to the lymph sacs, many of the peripheral lymphatic vessels originate from local lymphangioblasts.

The heart arises from splanchnic mesoderm. The first part of the heart to develop is the primary heart field (cardiac crescent; see Fig. 6.14B), which forms the left ventricle and the atria, the phylogenetically most primitive components of the mammalian heart. Exposure of the most posterior cells of the cardiac crescent to a retinoic acid gradient arising in the posterior mesoderm conditions these cells to adopt an atrial identity, whereas the more anterior cells not exposed to retinoic acid by default assume a ventricular identity (Fig. 17.18). The appearance of the secondary heart field provides the cellular material for formation of the evolutionarily more recent right ventricle and outflow tract. Cells of the secondary heart field also form the proepicardium (see p. 105) and contribute myocardial cells to the inflow (atrial) areas of the heart.

Cells of the secondary heart field arise from multipotent precursors within the pharyngeal mesoderm. These precursor cells can form either skeletal or cardiac muscle. Those from the first pharyngeal arch can either migrate to the head and form the masticatory muscles or migrate caudally into the secondary heart field, where they later become incorporated into the right ventricle. Similarly, precursor cells from the second pharyngeal arch can give rise to the muscles of facial expression or contribute to the outflow tract of the heart.

Several sets of molecules (MEF2, NKX2, GATA, Tbx, and Hand) form what has been called the core regulatory network that guides the differentiation of cardiac tissue. These sets of molecules are differentially regulated, however, by upstream activators that are specific to either the primary or the secondary heart fields (see Fig. 17.18). Even within the core regulatory network, molecular variants are unique to derivatives of the primary or secondary heart fields. Hand-1 is expressed in cells derived from the primary heart field, and in its absence left ventricular anomalies are seen. Hand-2 is expressed in cells derived from the secondary heart field, and if it is not expressed, the right ventricle does not form. Research has shown that the actions of microRNAs are tightly integrated into the overall regulatory programs that guide development of the heart.

As the outflow tract develops, it receives additional cellular contributions from various sources. Endothelial components arise from cephalic paraxial and lateral mesoderm in the region of the otic placode. A major cellular component of the wall of the outflow tract is derived from cardiac neural crest, specifically from the level between the midotic placode to the caudal end of the third somite (see Fig. 12.11). These cranial components integrate with the bilateral cardiac primordia while in the cervical region. As the heart descends into the thoracic cavity, the cranially derived cells of the outflow tract accompany it.

Another source of cells for the developing heart is the proepicardium (see Fig. 6.18). The proepicardium, which consists of an aggregate of mesothelial cells in the pericardium near the inflow area of the early heart, gives rise to the epicardium, most of the interstitial cells of the heart, and the coronary vasculature. Through an epitheliomesenchymal transformation mechanism, some of the epicardial cells transform into the fibroblasts, which invade the heart musculature and constitute most of the interstitial cells of the heart, as well as the smooth muscle cells of the coronary vessels.

The endocardium arises by a process of vasculogenesis within the forming cardiac tube. The endocardial cells arise from mesodermal cells of the cardiac crescent, but it is uncertain whether, within the crescent, endothelial precursor cells are already fully committed to the endothelial lineage or whether the precursor cells are bipotential and are also capable of differentiating into cardiac muscle cells.

Looping of the Heart

The cardiac tube, which takes shape late in the third week, is bilaterally symmetrical (see Fig. 6.17) and is composed principally of cells derived from the primary heart field. As it develops, the heart tube incorporates cells from the secondary heart field. Soon, it undergoes a characteristic dextral looping, making it the first asymmetrical structure to appear in the embryonic body. (The molecular basis for asymmetry of the embryonic body is discussed on p. 87.)

Certain heart-specific genes also seem to be involved in the early looping of the heart tube. Knockouts of four types of cardiac transcription factors (Nkx2.5, MEF-2, Hand-1, and Hand-2) all are characterized by the blocking of heart development at the stage of looping. The first molecular indication of asymmetrical development of the heart tube is the shift in expression of the transcription factor, Hand-1 (e-Hand), from both sides to the left side of the caudal heart tube. Hand-2 (d-Hand) is expressed predominantly in the primordium of the right ventricle. The Hand molecules may play a role in interpreting earlier asymmetrical molecular signals and translating this information into cellular behavior that results in looping.

As the straight heart tube begins to undergo looping, the originally ventral surface of the heart tube becomes the outer border of the loop, and the originally dorsal surface becomes the inner border of the loop. When the individual heart chambers begin to form, they arise as outpocketings from the outer border of the heart loop. Characterization of the cellular basis for cardiac looping has proven very difficult. Although it has been clearly shown that looping is an intrinsic property of the developing heart tube, the mechanisms by which molecular instructions are translated into structural changes remain obscure.

Research has revealed that the outpocketing (ballooning) of the future ventricular and atrial chambers is accompanied by the development of important structural and functional properties. The early symmetrical heart tube (called the primary myocardium) is characterized by slow growth, slow conduction of impulses, slow contraction, and the ability to undergo spontaneous depolarization. As the heart begins to loop, bulges that represent the incipient ventricular and atrial chambers appear on the outer surfaces of the loop and at the inflow end of the heart tube (Fig. 17.19).

In contrast to the primary myocardium, the cells of the chamber myocardium are characterized by a high proliferative capacity, strong contractility, high conduction velocity, and a low capacity to generate spontaneous impulses. Development of the primary heart tube is guided by the transcription factor Tbx-2, whereas development of the ventricular chambers is controlled by Tbx-5. As the heart undergoes developmental remodeling, the chamber myocardium constitutes the bulk of the atrial and ventricular chambers, whereas the primary myocardium is retained in the outflow tract (where it is later joined by cells from the neural crest), the atrioventricular canal and atrioventricular valves (where it is joined by cells originating from the proepicardium), and parts of the conducting system (the sinoatrial and atrioventricular nodes and the atrioventricular bundle).

The result of cardiac looping is an S-shaped heart in which the originally caudal inflow part of the heart (atrium) becomes positioned dorsal to the outflow tract. In the early heart, the outflow tract is commonly called the bulbus cordis, and it leads directly into the aortic sac and the incipient aortic arch system (Fig. 17.20). In early looping, the cranial limb of the S represents the bulbus cordis, and the middle limb represents the ventricular part of the heart, with the primordium of the right ventricle closest to the outflow tract, and that of the left ventricle next to the caudal limb. The caudal limb represents a common atrium. Later, the common atrium bulges on either side (Fig. 17.21), and an internal septum begins to divide the common ventricle into right and left chambers. The outflow tract (bulbus cordis of the early heart) retains its gross tubular appearance. Its distal part, which leads directly into the aortic arch system, is called the truncus arteriosus. The shorter transitional segment between the truncus and the ventricle is called the conus arteriosus. The conus is separated from the ventricles by faint grooves.

Early Atrioventricular Partitioning of the Heart

Early in heart development, the atrium becomes partially separated from the ventricle by the formation of thickened atrioventricular cushions. A similar but less pronounced thickening forms at the junction between the ventricle and the outflow tract (Fig. 17.22). In these areas, the cardiac jelly, which is organized like a thick basement membrane, protrudes into the atrioventricular canal. The endocardial cushions function as primitive valves that assist in the forward propulsion of blood.

In response to an inductive action by the underlying primary myocardium, certain antigenically distinct cells from the endocardium of the endocardial cushions lose their epithelial character and become transformed into mesenchymal cells, which migrate into the cardiac jelly. The inductive stimulus, which emanates from primary, but not chamber, myocardial cells, was initially described as being packaged as 20- to 50-nm particles, called adherons (see Fig. 17.22). Current data suggest that TGF-β3 and BMP-2, which act upstream of Notch and Tbx-2, are important components of the overall inductive signal, as a result of which certain endothelial cells express Snail-1/2, Msx-1, and Twist-1. Endocardial cells express the neural cell adhesion molecule (N-CAM) on their surfaces. Those cells that transform into mesenchymal cells downregulate the production of N-CAM, which facilitates their transformation into motile cells. The transformed mesenchymal cells secrete proteases, which destroy inductively active adherons and restore morphogenetic stability to the endocardial cushion regions.

These cellular and molecular events form the basis for the early formation of the major heart valves. Disturbances in these processes can account for many malformations of the heart.

The endocardial cushions (see Fig. 17.22), which ultimately become transformed into dense connective tissue, form on the dorsal and ventral walls of the atrioventricular canal. As they grow into the canal, the two cushions meet and separate the atrioventricular canal into right and left channels (Figs. 17.23 and 17.24). The early endocardial cushions serve as primitive valves that assist in the forward propulsion of blood through the heart. Later in development, thin leaflets of anatomical valves take shape in the atrioventricular canal. The definitive valve leaflets do not seem to come from endocardial cushion tissue as much as from invagination of superficial epicardially derived tissues of the atrioventricular groove. The valve that protects the right atrioventricular canal develops three leaflets (tricuspid valve), but the valve in the left canal (mitral, or bicuspid, valve) develops only two. At the molecular level, inhibition of expression of Tbx-2 on the atrial and the ventricular sides of the atrioventricular canal effects the separation between the atria and ventricles.

Partitioning of the Ventricles

When the interatrial septa are first forming, a muscular interventricular septum begins to grow from the apex of the ventricular loop between the ballooning right and left ventricular chambers toward the atrioventricular endocardial cushions (see Fig. 17.24C). The early division of the common ventricle is also reflected by the presence of a groove on the outer surface of the heart (Fig. 17.25). Although an interventricular foramen is initially present, it is ultimately obliterated by (1) further growth of the muscular interventricular septum, (2) a contribution by truncoconal ridge tissue that divides the outflow tract of the heart, and (3) a membranous component derived from endocardial cushion connective tissue.

Partitioning of the Outflow Tract of the Heart

In the very early tubular heart, the outflow tract is a single tube, the bulbus cordis. By the time the interventricular septum begins to form, the bulbus has elongated and can be divided into a proximal conus arteriosus and a distal truncus arteriosus (see Fig. 17.20). Closest to the heart, the wall of the outflow tract is composed largely of cells derived from the secondary heart field; more distally, cells derived from the neural crest predominate. Although initially a single channel, the outflow tract is partitioned into separate aortic and pulmonary channels through the appearance of two spiral truncoconal ridges, which are derived largely from neural crest mesenchyme. These ridges bulge into the lumen and finally meet, thus separating the lumen into two channels. The aortic sac, which is located distal to the truncoconal region, does not contain ridges. Partitioning of the outflow tract begins near the ventral aortic root between the fourth and sixth arches and extends toward the ventricles, spiraling as it goes (Fig. 17.26). This accounts for the partial spiraling of the aorta and the pulmonary artery in the adult heart.

Before and during the partitioning process, the neural crest–derived cells of the wall of the outflow tract begin to produce elastic fibers, which provide the resiliency required of the aorta and other great vessels. Elastogenesis follows a gradient, first through the outflow tract, then into the aorta itself, and ultimately into the smaller arterial branches off the aorta.

At the base of the conus, where endocardial cushion tissue is formed in the same manner as in the atrioventricular canal, two new sets of semilunar valves form (Fig. 17.27). These valves, each of which has three leaflets, prevent ejected blood from washing back into the ventricles. Cranial neural crest cells and cardiac mesoderm contribute to the formation of the semilunar valves. As previously stated, the most proximal extensions of the truncoconal ridges contribute to the formation of the interventricular septum. Just past the aortic side of the aortic semilunar valve, the two coronary arteries join the aorta to supply the heart with blood.

Innervation of the Heart

Although initial heart development occurs independently of nerves, three sets of nerve fibers ultimately innervate the heart (Fig. 17.28). Sympathetic (adrenergic) nerve fibers, which act to speed up the heart beat, arrive as outgrowths from sympathetic ganglia of the trunk. These nerve fibers are derived from trunk neural crest. Parasympathetic (cholinergic) innervation is derived from the cardiac neural crest. Neurons of the cardiac ganglia, which are the second-order parasympathetic neurons, migrate directly to the heart from the cardiac neural crest. These synapse with axons of first-order parasympathetic neurons that gain access to the heart via the vagus nerve. Sensory innervation of the heart is also supplied via the vagus nerve, but the sensory neurons originate from placodal ectoderm (nodose placode) (see Fig. 13.1). The direct innervation of the heart thus has three separate origins.

If the cardiac neural crest is removed in the early chick embryo, cholinergic cardiac ganglia still form. Experiments have determined that the nodose placodes compensate for the loss of neural crest by supplying neurons that replace the normal parasympathetic ones.

Initiation of Cardiac Function and the Conducting System of the Heart

The human heart begins to beat 22 to 23 days after fertilization, when it is still in the stage represented by the primary heart myocardium (see p. 107). The beat is slow (<40 beats/minute), and pacemaking activity begins near the inflow region of the heart and spreads toward the outflow tract through spontaneous depolarization of the cells. At this stage, the heart functions like a simple peristaltic pump.

As the atrial and ventricular chambers take shape, the differentiating cardiomyocytes are unable to generate or propagate beats in the same way as the cells of the primary myocardium. To coordinate the beat of the expanding chambers, it is necessary for the mammalian heart to develop a specialized conducting system, which takes advantage of some elements from the primary myocardium in the atrial region and adds to them a phylogenetically newer conducting system within the ventricular myocardium.

The sinoatrial node is the pacemaker of the mature heart (Fig. 17.29), and it is a direct descendant of the cells in the primary myocardium that initiate the first coordinated heart beats. Terminals of sympathetic and parasympathetic nerve fibers grow into the area to modulate the heart beat. The contractile stimulus then passes to the atrioventricular node through mechanisms still not well understood. From early development, the atrioventricular node, which is also a direct derivative of the primary myocardium, functions to slow down the conductive impulse to separate the contractions of the atrial and ventricular chambers. Activity of the transcriptional repressor Tbx-3 prevents the primary myocardial cells destined to form the sinoatrial and atrioventricular nodes from differentiating into the more highly contractile and more poorly conducting cells that characterize the ventricular chambers. From the atrioventricular node, the pacemaking impulse then passes with increasing velocity down the atrioventricular bundle and into left and right bundle branches before spreading out over the ventricular myocardium as the Purkinje fibers.

The atrioventricular node and bundle arise from a segment of a ring of myocardial cells that initially surround the interventricular foramen and later in some species become translocated to form a figure eight–shaped ring at the atrioventricular junction (see Fig. 17.29). Branches from this ring ultimately pass along either side of the interventricular septum and then arborize along the ventricular walls as Purkinje fibers.

The part of the conducting system consisting of bundle and Purkinje fibers represents a network of highly modified cardiac muscle fibers, whose structural and functional characteristics have been highly modified during development by paracrine influences. Stimulated by hemodynamic forces from the beating ventricles, endothelial cells produce an enzyme that activates the peptide endothelin-1, which along with other factors, such as neuregulin from the endocardium, stimulates the transformation of early cardiomyocytes into conducting cells of the Purkinje system. Purkinje cells elaborate connexins, which facilitate rapid conduction from one cell to another. Very rapid conduction is necessary to ensure a nearly simultaneous beat throughout the ventricle.