Development of the Vasculature

Second Source of Hematopoietic Cells

Although hematopoietic cells colonizing the developing liver initially arise from the yolk sac mesoderm, hematopoietic cells with greater potential and longevity eventually colonize the fetal liver from a separate intraembryonic source. Evidence for this second source of hematopoietic cells came from the use of quail-chick transplantation chimeras (technique covered in Chapter 5 ). In a remarkable experiment, the entire body of a 2-day embryonic chick was removed from the blastoderm, leaving just the yolk sac, and replaced with the entire body of a quail embryo. When examined after a couple of days of incubation, all blood lineages in the chimera were of chicken yolk sac origin, with quail cells providing the stromal (connective tissue) cells. However, within 5 days of grafting, a mixture of quail- and chick-derived blood cells was found circulating within the embryo, and eventually all blood cells were quail derived. Thus, this experiment showed clearly that (1) the yolk sac gives rise to transient embryonic hematopoietic cells; and (2) cells generated in the embryo body eventually replaced these cells, producing the long-lived definitive hematopoietic system. Consequently, for the first time it was shown that by closely following yolk sac hematopoiesis, there had to be a potent second source, an intraembryonic source of hematopoietic cells.

Microscopic analysis of bird, amphibian, mouse, and human embryos at equivalent developmental times identified densely packed clusters of hematopoietic cells adhering to the ventral endothelium of the dorsal aorta in the aorta-gonad-mesonephros (AGM) region and to the endothelium of the vitelline and umbilical arteries. The appearance of the hematopoietic clusters corresponds with the onset of the second wave of hematopoiesis. In humans, hematopoietic clusters appear within the aorta of the AGM region at 27 days of development as small groups of two or three cells, but by day 35, they increase to thousands of cells that extend into vessels adjacent to the umbilical cord region. These cells express transcription factors and cell surface markers associated with early blood progenitors and HSCs (e.g., Gata2, c-Kit, Runx1, Cd34, Cd41, Cd45). Although mammalian intraembryonic hematopoietic cluster cells could be derived from colonizing yolk sac hematopoietic cells, evidence suggests that they are of separate origin (covered later).

Intraembryonic Hematopoietic Cells, a Source of Adult Bone Marrow Hematopoietic Stem Cells

Studies in mice show that the yolk sac mesoderm forms hematopoietic progenitors capable of generating primitive erythrocytes, macrophages, and megakaryocytes. These primitive erythrocytes likely serve as a quickly forming stopgap population of blood cells that fulfill the oxygen needs of the rapidly developing embryo. Progenitors capable of generating definitive myeloid cells (definitive erythrocytes, macrophages, and granulocytes) appear later within the yolk sac. However, studies suggest that these cells must first colonize and interact with the developing liver to acquire long-term myeloid cell–generating capacity.

Definitive HSCs are defined by their ability for long-term, multilineage repopulation of the entire hematopoietic system (erythroid, myeloid, and lymphoid). Studies of mouse embryos show, by in vivo adult transplantation experiments, that the first functional HSCs are generated in the intraembryonic AGM region ( Figs. 13.3, 13.4 ) before their appearance in the liver. One day following HSC generation in the AGM region, HSCs are detected in the liver, suggesting that AGM HSCs migrate to the liver in a second wave of colonization. In fact, live imaging of the mouse midgestation aorta has documented the generation of hematopoietic cells directly from endothelial cells lining the ventral wall of the aorta. Hence, the cellular source of the second wave of hematopoietic cell generation is the hemogenic endothelial cell , generated through a process referred to as endothelial-to-hematopoietic transition. Similar multipotent in vivo adult hematopoietic repopulating cells were found in 4- to 5-week human AGM regions (by xenotransplantations of human cells into immunodeficient mice).

In humans, the liver primordia is colonized first by primitive hematopoietic cells from the yolk sac, possibly as early as day 23 to 24, and then by a second wave containing HSCs from day 30 onward. In vitro assays of human tissue explants of both yolk sac and AGM mesoderm show that both explants can make myeloid cells. However, only the AGM mesodermal tissue gives rise to multipotent hematopoietic cells (that make myeloid and T- and B-lymphocytes). Collectively, these studies suggest the yolk sac extraembryonic mesoderm generates primitive hematopoietic cells necessary for supporting the early cardiovascular needs of the embryo. The definitive HSCs that sustain lifelong adult hematopoiesis are derived from the intraembryonic AGM region (and possibly from hematopoietic clusters in the vitelline and umbilical arteries) and constitute the second wave of cells colonizing the fetal liver before subsequently colonizing the lymphatic organs and bone marrow. The second wave occurs for only a short window of developmental time, ending by day 40 of gestation.

Following the generation of HSCs in the embryo, their survival and proliferation depends on a trophic factor called stem cell factor (Scf, or c-Kit ligand) and its receptor, c-Kit receptor. Mouse mutants completely devoid of the c-Kit receptor or its ligand die in utero of anemia between days 14 and 16 of gestation and contain reduced numbers of erythroid progenitors in the fetal liver. In humans, the C-KIT RECEPTOR protein is expressed in the yolk sac, AGM splanchnic mesoderm, and liver HSCs during all stages of liver hematopoietic development. However, the expression of its ligand, SCF, is temporally regulated. SCF protein is expressed in the AGM region at low levels between days 25 and 34. It is well expressed in human liver by day 34, before levels drop off in late-stage liver development (i.e., by 45 days). Only weak expression of SCF mRNA is detected in the yolk sac (by quantitative real-time polymerase chain reaction; RT-PCR) at 32 days of development. Hence, C-KIT RECEPTOR signaling by SCF in HSCs coincides with the generation of AGM HSCs and with HSC colonization of liver and the appearance of definitive, long-term HSCs. This supports the idea that SCF/c-KIT signaling is involved in survival, differentiation, and proliferation of AGM-derived HSCs in the liver.

What Initiates and Controls Vasculogenesis?

As covered in Chapter 12 , in chick embryos, endodermally derived Bmps in the absence of Wnts induce the cardiogenic cell lineage, whereas Bmps in the presence of Wnts enables blood vessel formation in the splanchnic mesoderm. Moreover, Bmp/Tgfβ signals emanating from extraembryonic endoderm and Wnt signaling in mesoderm prime the extraembryonic mesoderm adjacent to the yolk sac endoderm to form blood islands. In mice, visceral endoderm provides inductive signals (e.g., Bmp, Vegf, Indian hedgehog) that are necessary for inducing the expression of blood island markers in this mesoderm. However, the precise trigger for forming hemangioblasts in vivo, and eventually the hematopoietic and EPC lineages, is still unclear. Both hematopoietic precursors and EPCs share many of the same early expression markers, so they are closely tied to one another with regard to cell lineage specification.

What is known is that Vegf signaling through the vascular endothelial growth factor receptor-2 (Vegfr2) is essential. Knockout mice for Vegfr2 have complete absence of hematopoietic progenitors and EPC lineages and die in utero. Moreover, homozygote knockout mice for vascular endothelial growth factor-A (VegfA) die as the result of lack of blood island formation. Mice lacking vascular endothelial growth factor receptor-1 (Vegfr1) also die, but the defect seems to be a consequence of abnormal EPC proliferation resulting in a disorganized vasculature. Vegf is a powerful promoter of vasculogenesis: injecting Vegf into embryos at the outset of vasculogenesis can vascularize normally avascular areas (e.g., cartilage-forming areas and cornea). Hypervascularization is also observed in transient transgenic gain-of-function quail embryos in which Vegf is overexpressed.

One group of transcription factors that seem to play a key role in endothelial cell specification is the Ets protein family. At least 12 different Ets factors are expressed in endothelial cells, with overlapping functions that work in combination to specify the endothelial cell lineage. One, Etv2, seems especially important, as it regulates expression of several early endothelial genes, and mice null for Etv2 lack endothelial cells, exhibit defects in hematopoiesis, and die early in development.

Angiogenesis Expands and Remodels Initial Vascular Complex

As in vasculogenesis, Vegfs and their receptors play a major role in mediating angiogenesis. In response to VegfA, Vegfr2-expressing endothelial cells, called tip cells , produce long, dynamic filopodia that probe the environment for directional cues (see Fig. 13.7 ). Adjacent endothelial cells called stalk cells proliferate in response to VegfA and express cell adhesion molecules necessary for forming a lumen and maintaining the integrity of the new sprout. So why does the entire endothelium not develop tip cells in response to VegfA? Why do some become stalk cells? The answer involves notch signaling. Notch proteins (notch1–4) are cell-surface receptors for the membrane-bound ligands delta-like 1 (Dll-1), Dll-3, Dll-4, Jag1, and Jag2, which are important in cell fate decision (covered in Chapter 5 ). Mouse embryos deficient in notch1 or notch2 form a normal initial capillary plexus but fail to properly remodel this vasculature. So notch signaling seems to be required for remodeling of the primitive capillary plexus rather than for its initial development.

During angiogenesis, the notch ligand, Dll-4, is predominantly expressed in tip cells, with the strongest notch signaling occurring in the stalk cells. Notch signaling represses sprouting, whereas blocking notch signaling causes excessive tip cell formation. Hence, a tip cell phenotype seems to be the default in the absence of notch signaling, whereas notch signaling promotes the stalk cell phenotype and stabilization of the sprout. In response to VegfA, it is thought that endothelial cells compete via bilateral Dll-4/notch signaling, eventually generating differences between the endothelial cell population of endothelial cells. Those receiving lower levels of notch signaling are specified as tip cells and those receiving higher levels are specified as stalk cells that then downregulate their expression of Vegf receptor (Flt; fms-related tyrosine kinase). As a consequence, these differences in notch signaling alter their subsequent responses to VegfA signaling. This mechanism is consistent with studies showing that mice haploinsufficient for Dll-4 exhibit excessive numbers of tip cells and develop abnormally dense vascular networks.

Xenograph tumor transplant studies in which notch signaling is manipulated also suggest that notch may be an important mediator of tumor vascular patterning by mediating the proper balance in tip and stalk cell numbers. Collectively, these studies show that notch signaling plays a major role in angiogenesis and in remodeling of the primitive capillary plexus by mainly acting as a negative regulator of sprouting.

The stalk-cell phenotype is also actively suppressed, allowing tip formation, by neuropilin-1, a co-receptor for Vegf, and semaphorins, which establishes differential responses to Tgfβ and Bmp signaling in these cells. The Tgfβ family and Tgfβ receptor signaling components—Alk1, Alk5, and endoglin—play critical roles in vasculogenesis and angiogenesis. Knockout mice for the Tgfβ receptors, Alk1 and Alk5, and the Tgfβ-binding protein, endoglin, are defective in angiogenic remodeling as a result of impaired endothelial cell migration and proliferation. They develop abnormal arterial-venous connections, much like those discussed in the “Clinical Taster” for this chapter. Moreover, recruitment of vascular smooth muscle is deficient, leading to poor vascular integrity and vascular instability. Tgfβ exhibits both stimulatory and inhibitory effects on endothelial cells. Recent studies suggest that the decision as to whether to continue angiogenesis or mature into a vessel depends on an interplay between the stimulatory effect of Tgfβ/Alk1 signaling (promoted by endoglin) and the inhibitory effect of Tgfβ/Alk5 signaling.

Another group of receptors and ligands that act in parallel to promote proper angiogenesis are the Tie (tyrosine kinase with immunoglobulin-like and Egf-like domains) receptor/angiopoietin group. Angiopoietin-1 (Ang1) and tyrosine kinase with immunoglobulin-like and Egf-like domains-2 (Tie2) are clearly involved in regulating intussusception of the vasculature, whereas Ang2, in cooperation with the stimulatory effects of Vegf, stimulates sprouting. Mice lacking the Tie2 gene or its ligand Ang1 develop abnormally large and leaky vessels and die in utero. Moreover, these mice exhibit a decrease in endothelial cell number and angiogenic sprouting, as well as failure of vascular intussusception. Tie1 knockout mice develop vessels with holes, and the endothelial cells appear necrotic. These results show collectively that Tie/Ang signaling along with Vegf signaling is essential for expansion and normal remodeling of blood vessels after the initial primitive vasculature is established.

One of the main driving forces for vascularization by angiogenesis is the need to counteract hypoxia. Low oxygen saturation leads to stabilization of the transcription factor, hypoxia-inducible factor-1α (Hif1α). Hif1α upregulates VegfA expression and nitric oxide synthase expression. Nitric oxide production dilates existing blood vessels, thereby increasing the permeability and extravasation of plasma proteins, leading to an increase in expression and activation of the proteases, matrix metalloproteinases, and plasmin. Matrix metalloproteinases and plasmin play major roles in promoting proliferation and migration of endothelial cells by activating growth factors and receptors and increasing extracellular matrix turnover, which is necessary for sprouting. Proper vascular development also involves trimming away vessels that are no longer necessary or that would be detrimental should they remain. For example, in its early development, the retina initially forms excess vascularity that must be trimmed down later. Experimental hyperoxia (i.e., surplus oxygen) in rodents suppresses Vegf levels in the retina. Because Vegf acts as a survival factor for retinal endothelial cells, hyperoxia resulting from the excess blood supply may drive the pruning of excess vessels by decreasing Vegf levels ( Fig. 13.8 ). Hence, the degree of vascularity (i.e., driven by the formation or pruning of vessels) of a tissue may be mediated by oxygen-dependent regulation of Vegf and nitric oxide levels. Finally, mechanical forces associated with luminal and transmural blood flow and endothelial cell metabolism are increasingly being recognized as regulators of sprouting and branching.

Formation of Arteries Versus Veins

The factors that direct and guide vessel remodeling and identity are still unclear, but several ligands and receptors known to play roles in neuronal guidance (covered in Chapter 10, Chapter 9 ) seem to be involved in this process. One such group includes the Eph receptors and their membrane-bound ligands, the ephrins. The binding of ephrins to EphB receptors stimulates transduction signals in the EphB-expressing cells, but this binding can also transduce a reverse signal into the ephrin-expressing cell. Such interactions and signaling events play important roles not only in the development of the nervous system but also in blood vessel remodeling and in the specification of the artery or vein phenotype. EphrinB2 is specifically expressed on the surface of arterial endothelial cells, whereas the EphB4 receptor is specifically expressed on venous endothelial cells ( Fig 13.9 ). When either ephrinB2 or EphB4 is knocked out in mice, the remodeling of primary vascular plexus into arteries and veins fails. How ephrinB2 and EphB4 mediate these changes is unclear, but it has been suggested that during angiogenesis, differential expression of these two molecules may restrain cell migration and create tissue boundaries used to sort out the arterial and venous systems. What is responsible for mediating specific expression of ephrins and Eph receptors is unclear but may involve Tgfβ-mediated signaling, as ephrinB2 expression is absent in Alk1-deficient mice.

Notch signaling may also play a key role in establishing arterial or venous identity upstream of ephrins/Eph receptor signaling even before the formation of the initial vascular complex (see Fig. 13.9 ). In mice, notch1, notch3, and notch4 receptors and their ligands, Dll-4, Jag1, and Jag2, are expressed in arteries but not in veins. Studies in zebrafish show that inhibiting the notch signaling pathway decreases the expression of ephrinB2, resulting in the ectopic expression of venous markers in the dorsal aorta. In contrast, increasing notch signaling decreases the expression of venous markers and induces the expression of arterial markers, including ephrinB2. In mice, expression of the venous transcription factor chicken ovalbumin upstream promoter transcription factor 2 (Coup-tfII) (also known as Nr2f2) actively represses notch signaling and the expression of the arterial marker neuropilin-1 (Nrp1) and promotes the expression of the venous marker Nrp2. Loss of Coup-tfII expression in veins promotes the expression of the arterial marker, Nrp1. Both Nrp1 and Nrp2 are co-receptors for Vegf that bind specific splice variants of Vegf and may mediate differential responses to Vegf signaling in endothelial cells. Therefore, notch signaling may have an important role in mediating not only vascular angiogenesis and remodeling but also arterial/venous specification of EPCs (see Fig. 13.9 ). Finally, notch signal plays an important role in mediating the diameter of developing blood vessels.

Several other ligands and receptors known to play roles in neuronal guidance are involved in guiding arterial and venous blood vessel formation and remodeling, including slit/robo, semaphorins/plexins, and netrins/Unc5b. This may not be too surprising given that vessels and peripheral nerves generally run in parallel with one another.

Angiomas

Blood and lymphatic vessels are stimulated by angiogenic factors to grow into developing organs. If vessel growth is not inhibited at the appropriate time, or if it is stimulated again later in life, blood or lymphatic vessels may proliferate until they form a tangled mass that may have clinical consequences. Excessive growth of small capillary networks is called a capillary hemangioma or nevus vascularis ; a proliferation of larger venous sinuses is called a cavernous hemangioma. Hemangioma of infancy is the most common benign tumor of childhood (incidence of about 2.5% in neonates and up to 10% to 12% in 1-year-olds; Fig. 13.10 ). These tumors grow rapidly and consist mainly of endothelial cells with or without lumens, multilayered basement membranes, and fibrous tissue. Hemangiomas differ from some vascular anomalies like nevus flammeus , which present at birth as birthmarks and grow proportionately with the growth of the child. Most cases of hemangioma of infancy pose no immediate or long-term danger. However, they can be potentially life threatening if they grow in vital organs (e.g., in the skull or vertebral canal, where they can lead to nervous system dysfunction, or in airways, where they can obstruct breathing) or are large enough to create a shunt of physiological significance leading to heart failure. In rare cases, a hemangiosarcoma (metastatic angioma) can develop.

Many hemangiomas seem to have a genetic basis to their origin, as they are associated with developmental syndromes resulting from chromosomal anomalies. In the case of hemangioma of infancy, some cases are linked to chromosome region 5q31-33. This region contains genes coding for Fgf4, Pdgfβ, and fms-related tyrosine kinase—molecules important in blood vessel development. Some hemangiomas are also linked to dysregulation of the Tie/Ang signaling pathway and to FLT2 (VEGFR2) mutations. Multiple hemangioblastomas are associated with a rare, dominantly inherited, familial cancer syndrome called von Hippel-Lindau disease (incidence 1:36,000) characterized by mutations in a tumor suppressor gene located at chromosome 3p25-26. These individuals exhibit life-threatening multiple central nervous system, retinal, and liver hemangioblastomas, renal cell carcinomas, and visceral cysts. Studies show that stromal cells of these tumors produce high levels of Vegf and Hif1α, which could account for the excessive angiogenesis.

Hereditary Hemorrhagic Telangiectasia

Mutations in ALK1 and ENDOGLIN have been linked to hereditary hemorrhagic telangiectasia (HHT). The most common manifestations of HHT (prevalence 1:5000 to 1:8000) are nosebleeds and small vascular anomalies called telangiectases; however, gastrointestinal bleeding and arterial-venous malformations in the lung, brain, and liver progressively develop. Mice heterozygous for mutations in Alk1 or endoglin exhibit very similar progressive pathologies, including the development of large arterial-venous shunts. Mutations in human ALK1 have also been linked to primary pulmonary hypertension involving abnormal remodeling of the pulmonary vasculature. Both HHT and some forms of pulmonary hypertension may stem from an improper balance between Alk1 and Alk5 signaling (covered earlier in the chapter). Endothelial cells from normal healthy patients form extensive, stable, and well-formed vessels in tissue culture, whereas endothelial cells from HHT patients are unable to do so. Endoglin also interacts with the actin cytoskeleton. Endothelial cells from HHT patients exhibit a disorganized actin cytoskeleton, but normal actin organization can be restored if normal Endoglin is overexpressed in these cells. Endothelial cells with a disorganized, abnormal cytoskeleton are more likely to form vessels prone to vascular instability, hemorrhaging, and vascular disarray.