Genetic and epigenetic mechanisms in the early development of the vascular system

Abstract

The cardiovascular system plays a critical role in vertebrate development and homeostasis. Vascular development is a highly organized sequence of events that requires the correct spatial and temporal expression of specific sets of genes leading to the development of a primary vascular network. There have been intensive efforts to determine the molecular mechanisms regulating vascular growth and development, and much of the rationale for this has stemmed from the increasing clinical importance and therapeutic potential of modulating vascular formation during various disease states.

Introduction

The cardiovascular system is the first functional organ system to develop in the vertebrate embryo. Several genetic (Table 1) and epigenetic (vascular branching, pruning, remodelling) mechanisms are involved in the early development of the vascular system. During embryonal life, blood vessels first appear as the result of vasculogenesis, i.e. the formation of capillaries from endothelial cells (EC) differentiating in situ from groups of mesodermal cells. The primitive heart and primitive vascular plexus are formed in this way (Risau & Flamme, 1995).

Table 1.

Genetic molecular pathways involved in vessel wall maturation

A. Signalling pathways (ligand/receptor)

VEGF/VEGFR-1, VEGFR-2, NRP-1

Ang-1-Ang-2/Tie-1, Tie-2

TGF-β/ALK-1, ALK-5

Eph-B2/Eph-B4

Notch pathway

B. Molecules governing cell–cell interactions

VE-cadherin

N-cadherin

Connexins

Tight junctions

C. Molecules governing cell–matrix interactions

Netrins

Semaphorins

Fibronectin

Integrins

Vasculogenesis leads to the formation of the first major intra-embryonic blood vessels, such as the dorsal aorta and the cardinal veins, and to the formation of the primary vascular plexus in the yolk sac. With the onset of embryonic circulation, these primary vessels have to be remodelled into arteries and veins in order to develop a functional vascular loop. Remodelling of the primary vascular plexus into a more mature vascular system is thought to occur by a process termed angiogenesis. Although, as a general rule, establishment of the vasculature of most organs occurs by angiogenesis, development of the vascular network of certain endodermal organs, including the liver, lungs, pancreas, stomach/intestine and spleen, occurs by vasculogenesis (Pardanaud & Dieterlen-Lievre, 1999).

The term angiogenesis, applied to the formation of capillaries from pre-existing vessels, i.e. capillaries and postcapillary venules (Risau, 1995), is based on endothelial sprouting or intussusceptive (non-sprouting) microvascular growth (IMG) (Ausprunk & Folkman, 1977; Burri & Tarek, 1990). The vascularization of many extra-embryonic and intra-embryonic tissues, including the yolk sac, embryonic kidney, thymus, brain, limb and choroid plexus, occurs by sprouting angiogenesis.

IGM constitutes an additional and/or alternative mechanism for endothelial sprouting and is not dependent on local endothelial cell proliferation or sprouting. Once accumulated, the EC can spread out and attenuate in order to cover the additional surface created by pillar formation (Djonov et al. 2000). With IGM, the capillary network increases its complexity and vascular surface by inserting a multitude of transcapillary pillars, through four consecutive steps: (1) creation of a zone of contact between opposite capillary walls; (2) reorganization of the intercellular junctions of the endothelium, with central perforation of the endothelial bilayer; (3) formation of an interstitial pillar core; and (4) subsequent invasion of the pillar by cytoplasmic extensions of myofibroblasts and pericytes, and by collagen fibrils. It is thought that the pillars then increase in diameter and form a capillary mesh.

Vascular development can be thought of as involving the following processes: formation, stabilization, branching, remodelling, pruning and specialization (Carmeliet, 2003, 2004). Vessels need to be specified into different calibres and types, including division into arteries, veins, capillaries and lymphatics. In addition, they need to recruit supporting cells, smooth muscle cells and pericytes to ensure cell stability of the vessels formed.

Mechanisms of prenatal vasculogenesis

Vasculogenesis begins very early after the initiation of gastrulation in the mammalian embryo, with the formation of blood islets in the yolk sac and angioblast precursors in the head mesenchyme and posterior lateral plate mesoderm. The most typical and earliest sites of prenatal vasculogenesis are the blood islets in the yolk sac, composed of hemangioblasts, the putative common precursors of endothelial and hematopoietic cells (Choi et al. 1998). Angioblasts, the peripheral cells of the blood islets, connect to construct a primitive network composed of capillaries, arteries and veins. Similarly, within the embryo itself, angioblasts start to join together to form a capillary network in the head mesenchyme and posterior lateral plate mesoderm.

Because blood vessel growth is a complex, multigene event (Lee et al. 2004), it is possible that multiple growth factors acting at different times may be required. Thus, while one agent may stimulate the growth of new vascular structure, another may induce their maturation.

Role of fibroblast growth factor-2 (FGF-2)

When quail blastodiscs are explanted and disrupted in culture, the mesoderm cells do not form angioblasts or express proteins associated with EC lineage (Flamme & Risau, 1992), unless they are treated with FGF-2, in which case blood island-like structures are formed (Flamme & Risau, 1992). Flamme et al. (1997) showed that FGF-2 induces pluripotent cells of the quail blastodisc to undergo vasculogenesis, and experiments in the chick have suggested that FGF signalling is important for initiation of angioblast specification (Cox & Poole, 2000).

FGF-2 has been identified in the chick chorioallantoic membrane (CAM) during vasculogenesis and can promote vessel growth when exogenously added to the CAM during embryo development (Ribatti et al. 1995). Moreover, neutralizing antibodies to FGF-2 inhibited vessel growth when applied locally, suggesting that FGF-2 normally functions to promote vessel growth, possibly by inducing angioblast formation from mesoderm (Ribatti et al. 1995).

Role of vascular endothelial growth factor (VEGF)

VEGF and VEGF receptors (VEGFR) are the first EC-specific signal transduction pathways activated during vasculogenesis. VEGF-deficient embryos die between day 8.5 and 9.0 and their primitive vascular structures are severely deficient (Carmeliet et al. 1996; Ferrara et al. 1996), while VEGFR-2-deficient mice die early as a result of blocked migration of angioblasts to the initial sites of vasculogenesis (Shalaby et al. 1995, 1997). Embryos lacking VEGFR-2 die at around 9 days of development and show no development of any blood vessels or hematopoietic cells (Shalaby et al. 1995). The loss of both lineages could suggest that VEGFR-2 is required in the common precursor, the hemangioblast. However, chimera studies suggest that the failure to form the hematopoietic lineage is a result of the incorrect migration and location of the progenitors of the endothelium and the blood system in the mutant yolk sac (Shalaby et al. 1997). VEGFR-2 is highly expressed later in EC (Millauer et al. 1993), but down-regulated in most hematopoietic cells (Kabrun et al. 1997). In the yolk sac, VEGFR-2 expression appears to mark cells prior to the distinction between the outer EC and the enclosed hematopoietic cells (Yamaguchi et al. 1993), which led to the suggestion that it might also mark the hemangioblasts (Eichmann et al. 1997; Kabrun et al. 1997). EC develop in VEGFR-1-deficient embryos, but death before day 9.5 still occurs due to the defective organization of the primitive vascular system (Fong et al. 1995). The phenotype is more suggestive for an overgrowth of EC than a loss of cells, as confirmed by chimera studies (Fong et al. 1999). This suggests that VEGFR-1 could be a negative regulator of VEGF signalling in the early embryo, acting to modulate VEGF by sequestering ligand. Inactivation of the gene for VEGFR-3 also affects lumen formation in large vessels during early development (Dumont et al. 1998). Early microinjection of VEGF into quail embryos caused vasculogenesis in normally avascular areas, while the normal polygonal pattern of vessels, characteristic of early embryos, was replaced by abnormal, enlarged vascular channels (Drake & Little, 1995).

Role of Tie-1, Tie-2 and angiopoietins (Ang)

The Ang family comprises at least four secreted proteins, Ang-1, -2, -3 and -4, all of which bind to the endothelial-specific receptor tyrosine kinase Tie-2, while Tie-1 is an orphan receptor tyrosine kinase. It is well documented that Ang play a critical role in endothelial sprouting, vessel wall remodelling and mural cell recruitment (Thurston, 2003).

Vasculogenesis proceeds normally in embryos lacking both Tie-1 and Tie-2, although they die early due to multiple cardiovascular defects (Suri et al. 1996; Puri et al. 1999). Mutation of Tie-2 does not affect initial formation of blood vessels, but embryos died in midgestation with major defects in vascular remodelling and stability (Dumont et al. 1994; Sato et al. 1995). Blood vessels were enlarged, with fewer branches, and EC tended to round up, dissociate from the underlying support cells and extracellular matrix, and undergo apoptosis. Chimera studies showed that the role of Tie-2 was specific to the late stages of capillary formation (Partanen et al. 1996).

Ang-2 can bind to the Tie-2 receptor but does not activate it; rather, it seems to act as an antagonist, counteracting the effects of Ang-1. Consistent with this, overexpression of Ang-2 in the embryos produces a phenotype similar to loss of function of Ang-1 or Tie-2 (Maisonpierre et al. 1997).

Knockout embryos lacking Ang-1, an activator of Tie-2, display failure of EC adherence and interaction with perivascular cells and extracellular matrix (Davis & Yancopoulos, 1999). In the absence of mural cells, recombinant Ang-1 restored a hierarchial order of the larger vesssels, and rescued oedema and haemorrhage in the growing retinal vasculature of mouse neonates (Uemura et al. 2002). Transgenic overexpression of Ang-1 in skin results in pronounced hypervascularization with the production of many compact stable vessels (Suri et al. 1998; Thurston et al. 1999). This contrasts with the effects of overexpression of VEGF, which also leads to hypervascularization, but with the formation of large, leaky, simple endothelial tubes, unprotected by supporting cells (Drake & Little, 1995). If Ang-1 normally acts to stabilize vessels, then there has to be a mechanism to modulate its action to allow vascular remodelling, with it acting locally to block Ang-1 action and destabilize vessels (Maisonpierre et al. 1997). If VEGF is also present, this localized angiogenesis can occur. In the absence of VEGF, and in the presence of Ang-2, vessels destabilize and EC undergo apoptosis, resulting in vessel regression.

Role of transforming growth factor beta (TGF-β)

Studies of targeted knockout mice have provided evidence of an essential role for TGF-β signalling in vascular development, suggesting that this biphasic response may result from the balance between ALK-1 and ALK-5 signalling in EC. The TGF-β–ALK-1 pathway induces EC and fibroblasts to express Id1, a protein required for proliferation and migration. TGF-β, ALK-1 and endoglin are positive regulators of EC migration and proliferation, whereas the TGF-β–ALK-5 pathway is a positive regulator of vessel maturation (Goumans et al. 2002). Mutations in mutiple components of the TGF-β signalling pathway cause embryonic lethality as a result of vascular defects (Dunker & Krieglstein, 2000).

When mesenchymal cells are co-cultured with EC or treated with TGF-β-1, they express smooth muscle cell (SMC) markers, indicating differentiation toward a SMC lineage (Hirschi et al. 1998). The differentiation can be blocked with neutralizing antibodies against TGF-β (Hirschi et al. 1998). TGF-β-1 has also been shown to direct neural crest cells toward a smooth muscle lineage (Shah et al. 1996). In the absence of endoglin, a protein that binds the TGF-β ligand receptor complex, blood vessels are formed, but not invested with SMC (Li et al. 1999).

Role of neuropilins (NRP)

The NRP are a small family of transmembrane proteins. NRP bind certain members of the VEGF family: NRP-1 and NRP-2 bind VEGF165 and placental growth factor (PlGF); NRP-1 also binds VEGF-B; and NRP-2 also binds VEGF145 and VEGF-C (Karkkainen et al. 2001; Neufeld et al. 2002).

The VEGF binding ability of NRP suggests that NRP may function as regulators of vasculogenesis. Mice lacking a functional NRP-1 gene display vascular abnormalities (Kawasaki et al. 1999), but the vasculature of mice lacking a functional NRP-2 gene develops almost normally except for minor defects in lymphatic vessels (Giger et al. 2000; Yuan et al. 2002). Nevertheless, NRP-2 does play an important role in vasculogenesis, because mice lacking both NRP-1 and -2 display vascular abnormalities that are much more severe than the abnormalities seeen in mice lacking a functional NRP-1 gene (Takashima et al. 2002). These phenotypes suggest a continued requirement for VEGF signalling to promote vascularization of developing organs, such as brain and heart, as well as an expansion and growth of major blood vessels. Overexpression of NRP-1 in transgenic embryos led to excess and dilation of blood vessels and heart malformation (Kitsukawa et al. 1995) consistent with NRP-1 acting as co-activator of the VEGF receptor.

Role of netrins and hedgehog (Hh)

Netrins are a family of secreted laminin-related molecules (Serafini et al. 1994). In addition to their well-demonstrated role in axon-guidance, netrins have recently been implicated in angiogenesis. Park et al. (2004) have reported in vitro experiments that suggest a pro-mitogenic and pro-migratory effect of netrin-1 on EC. By contrast, Lu et al. (2004) have provided evidence for a negative role of netrin in vessel guidance, suggesting that netrins may act as attractants or repellents in both the nervous and the vascular system.

Hedgehog (Hh) is a secreted signalling molecule that serves multiple roles during embryonic development (Ingham, 2001). Recent evidence in both zebrafish and mouse suggests a role for Hh signalling in both vasculogenesis and angiogenesis (Byrd & Grabel, 2004). It was shown that zebrafish embryos lacking Hh activity fail to undergo arterial differentiation, as defined by the expression of artery-specific markers, such as ephrin B2 (Eph-B2) (Lawson et al. 2002). However, injection of mRNA encoding Hh into the zebrafish could induce ectopic vascular expression of Eph-B2, as did the injection of VEGF mRNA (Lawson et al. 2002). Hh signalling can target EC directly or can stimulate blood vessel support cells to secrete angiogenic cytokines.

Role of spoutry (Spry)

Spry was first identified in Drosophila as an inhibitor of FGF signalling, during tracheal development (Mailleux et al. 2001). Direct evidence of a role for spry in angiogenesis comes from a study in which the mouse Spry-4 was overexpressed in the developing endothelium of a mouse embryo using an adenoviral vector (Lee et al. 2001).

Role of semaphorins

Mammalian NRP-1 and NRP-2 encode transmembrane proteins that act as receptors for the axon repellant factors of the class-3 semaphorin subfamiliy (Chen et al. 1997; He & Tessier-Lavigne, 1997; Kolodkin et al. 1997; Giger et al. 1998). It is possible that these semaphorins may also function as regulators of vasculogenesis and artery/vein differentation. Several studies indicate that class-3 semaphorins function as inhibitors of angiogenesis. Semaphorin-3A antagonizes the effects of VEGF in an in vitro angiogenesis assay as well as in vivo (Miao et al. 1999; Bates et al. 2003) and both semaphorin-3A and -3F were shown to inhibit vascular remodelling during embryonic development through an effect on integrin-mediated cell adhesion (Serini et al. 2003). Furthermore, is was observed that semaphorin-3F can inhibit FGF-2 as well as VEGF-induced angiogenesis in vivo (Kessler et al. 2004).

Conrotto et al. (2005) demonstrated that semaphorin-4D is angiogenic in vitro and in vivo and that this effect is mediated by a high-affinity receptor, plexin B1. Moreover, they proved that biological effects elicited by plexin B1 required coupling and activation of the Met tyrosine kinase.

Role of fibronectin

Vasculogenesis takes place in a fibronectin-rich extracellular matrix (Risau & Lemmon, 1988). As soon as the basic vascular network is established, fibronectin decreases in the vicinity of developing blood vessels and EC begin to produce laminin and collagen type IV in increasing amounts (Risau & Lemmon, 1988; Drake et al. 1990; Ausprunk et al. 1991).

Fibronectin-null mice die in utero at E9 to E10 and exhibit severe defects in blood vessel and heart development (George et al. 1993).

Role of integrins

Integrins, which are cell surface heterodimers, play a critical role in angiogenesis (Brooks et al. 1994a). Integrin α_vβ₃ is highly expressed on angioblasts and an α_vβ₃ antagonist, LM 609, prevents the maturation of primordial EC into blood vessels (Drake et al. 1995). Moreover, when CAM were implanted with melanoma tumour fragments and then treated with α_vβ₃-blocking antibody or cyclic peptide antagonist to α_vβ₃, the vessel investment normally observed in these tumours was prevented (Brooks et al. 1994b).

β1 integrin, too, plays a pivotal role in vasculogenesis; when avian embryos are injected with the anti-β1 integrin antibody CSAT, they display only cord-like assemblies of EC at stages and in positions where normal vessels with a lumen should have formed (Drake et al. 1992).

Stabilization of immature vasculature

The nascent vessels are stabilized by recruiting mural cells and by generating extracellular matrix (ECM). The mural cells that form a non-continuous abluminal layer at the level of the capillary and postcapillary venule are referred to as pericytes (Sims, 1986). In large vessels, such as arteries, the mural cells are referred to as SMC and form a multilayered sheath around the elastic artery wall. Interactions with the ECM are likeky to be involved in lumen formation. The ECM protein fibronectin has been shown to be present in the area of migrating EC and immature capillaries (Risau & Lemmon, 1998). The α_vβ₃ integrin remains a good candidate for regulating at least some of the cell–ECM interactions that are involved in vascular development.

Role of platelet-derived growth factor-B (PDGF-B)

PDGF-B is secreted by EC, presumably in response to VEGF, and facilitates recruitment of mural cells. Mutation of PDGF-B caused failure of recruitment of pericytes (Lindahl et al. 1997, 1998). A detailed analysis of vessel development in both PDGF-B and PDGF-B-receptor (PDGF-BR) mutant embryos showed that SMC and pericytes initially form around the vessels but, as vessels sprout and enlarge, PDGF signalling is required for co-migration and proliferation of supporting cells (Hellstrom et al. 1999).

The similarity between phenotypes of PDGF-B/PDGF-BR and endothelial differentiation sphingolipid-G-protein-coupled-receptor-1 (EDG1) knock-out mice (failure of mural cells to migrate to blood vessels) indicates that signalling through EDG1 receptor, which is expressed by mural cells, is another key pathway for mural cell recruitment (Kluk & Hla 2002).

Vascular branching

The gross vascular anatomy of the vascular system is characterized by highly reproducible branching patterns. For example, there are fixed branching sites, branching angles, curvature and size gradation from the aorta for arteries supplying the head, internal organs and legs (Dor et al. 2003). There are also designated sites for secondary branches (i.e. intersomitic vessels and main vessels penetrating different organs), whereas microvessels formed by sprouting angiogenesis are mostly non-sterotyped.

Embryonic vessel formation is also highly dynamic and subject to intense pruning and remodelling throughout development, with vessel tracts appearing and disappearing and links between vessels severing and then reconnecting in entirely new patterns (Isogai et al. 2001).

Local alterations in perfusion produce dramatic changes in vascular patterning throughout the embryo (Le Noble et al. 2004). In adult vessels, vessel segments can adapt to the amount of flow carried (Skalak & Price, 1996; Peirce & Skalak, 2003).

Pruning

Pruning was first described in the embryonic retina and involves the removal of supernumerary blood vessels from redundant channels (Ashton, 1996). Blood flow generally ceases in these excess capillaries, the lumina are obliterated and the EC retract toward adjacent capillaries. Intussusception may also contribute to pruning by formation of asymetrically localized pillars at the branching points of larger vessels. As a result, daughter branches becomes partially obstructed or totally separated from the mother vessels (Djonov et al. 2002).

Remodelling

As the vascular system develops, the initial plexus becomes remodelled into a complex and heterogeneous array of blood vessels, including larger vessels such as arteries and veins, and smaller vessels such as venules, arterioles and capillaries.

Remodelling is not well understood, but it is known to involve the growth of new vessels and the regression of others, as well as changes in the diameter of vessel lumen and vascular wall thickness. The developing vasculature responds dynamically to the growing needs of the embryo by remodelling vessels as required. Some vessels may fuse to form a larger one, as occurs with fusion of the paired dorsal aortae, or they may establish new connections such as the coronary vessels that connect to the aorta (Bogers et al. 1989). It is likely that only a small number of embryonic blood vessels persist into adulthood (Risau, 1995), with most capillaries of the embryonic plexus regressing at some time in development to allow the differentiation of other tissues.

Angiogenesis in the corpus luteum of the cycling ovary, in which there are rapid changes in the vasculature, is illustrative of vascular remodelling (Augustin et al. 1995). Other examples include the regression of capillaries in prechondrogenic regions to allow the differentiation of cartilage (Hallmann et al. 1987), the regression of the hyaloid vasculature to allow the development of the vitreous body in the eye (Latker & Kuwubara, 1981) and the retinal vasculature, which undergoes dramatic vascular remodelling during the formation of the mature vasculature (Benjamin et al. 1998).

Formation of arteries and veins

Differentiation of arteries and veins was thought to be governed by haemodynamic forces, moulding these vessels from the primary vascular plexus. Thoma (1893) observed that vessels carrying a lot of blood widen, whereas those that carry little flow regress. Murray (1926) postulated that vessels adapt to flow in order to optimize the shear stress to which they are subjected. These studies have shown that flow can alter lumen dimensions of arterial segments.

However, recent findings indicate that EC fate is determined before the onset of circulation. Labelling experiments in zebrafish indicate that the arterial and venous fate of endothelial precursors may be determined before the formation of the blood vessels (Zhong et al. 2001). These authors followed individual angioblasts and found that, contrary to expectations, all the progeny of a single angioblast formed either veins or arteries, never both. In other words, each angioblast was already specified as to whether it would form aorta or cardinal vein.

The discovery that members of the Eph family are expressed differentially in arteries and veins from very early stages of development, before the development of functional circulation, was one of the first indications that artery–vein identity is intrinsically programmed. Mutations of Eph-B2 and of Eph-B4 both lead to early embryonic lethality around E9.5 (Wang et al. 1998; Adams et al. 1999; Gerety et al. 1999; Gerety & Anderson, 2002). Remodelling of the primary vascular plexus into arteries and veins was arrested in both mutants. These findings suggest important roles for Eph-B2/Eph-B4 interactions on arterial and venous EC, respectively. Eph-B2 marks arterial EC and SMC, while Eph-B4, a receptor for Eph-B2, only marks veins. Moreover, Eph-B2–Eph-B4 signalling participates in the formation of arteriovenous anastomoses through arresting VEGF- and Ang-1-induced EC proliferation/migration at the arterial–venous interface (Wang et al. 1998; Gerety et al. 1999; Zhang et al. 2001). Disruption of the Eph-B2 gene in mice results in retarded growth and embryonic lethality at E10.5. Vasculogenesis is halted at the primary plexus stage, EC are disorganized and many features of angiogenic remodelling are absent (Adams & Klein, 2000).

Other specific markers for arteries and veins include for the arterial system NRP-1 and members of the Notch family, including Notch-3, DDL4 and GRIDLOCK (Grl), described in zebrafish, chick and mouse (Shutter et al. 2000; Herzog et al. 2001; Lawson et al. 2001; Moyon et al. 2001a,b; Villa et al. 2001). Venous markers include NRP-2 (Herzog et al. 2001), which at later developmental stages becomes restricted to lymphatic vessels in chick and mouse (Herzog et al. 2001; Zhong et al. 2001). The different expression of NRP-1 and NRP-2 in arteries and veins of early chick and mouse embryos (Herzog et al. 2001; Moyon et al. 2001a; Yuan et al. 2002) suggests that NRP may regulate, at least in part, the segregation of the vascular system into arteries and veins. Herzog et al. (2005) have examined the expression patterns of NRP-1 and NRP-2 during the early stages of vasculogenesis and concluded that, before the initiation of flow, the primitive vessels of the extra-embryonic vascular plexus are already segregated into veins and arteries. The vessels in the venous and arterial parts of the plexus are located around the regions in which the vitelline vein and the vitelline artery will later be formed. Eph-B2 expression cannot be seen in the arterial part of the extra-embryonic vascular plexus of 13-somite chick embryos even though the expression of NRP is already segregated. These observations suggested that Eph-B2, in contast to NRP-1, is a relatively late marker of arteries.

Grl mutants exhibit an aberrant development of the aorta (Weinstein et al. 1995) and Notch signalling lies upstream of grl and is required for arterial–venous differentiation (Lawson et al. 2001; Zhang et al. 2001). However, when the Notch signalling pathway activates grl, this directs endothelial precursors to assume an arterial fate.

Notch and Delta

Notch signalling is a highly conserved pathway, initially discovered in Drosophila development (Baron et al. 2002). There are four Notch receptor (Notch 1–4) and five ligands (Jagged-1 and -2, Delta-1, -3, -4) (Iso et al. 2003). All the receptors and ligands have been expressed in at least one vascular compartment, e.g. arteries, veins, capillaries, vascular smooth muscle cells or pericytes.

Notch signalling is required for remodelling the primary plexus into the hierarchy of mature vascular beds and maintaining arterial fate, and is essential for the homeostatic functions of fully differentiated arteries (Alva & Iruela Arispe, 2004). Genetic studies in zebrafish and mice have suggested a key role for Notch signalling, downstream of VEGF-A, in specifying arterial vs. venous fate (Rossant & Hirashima, 2003). Mice with defects in genes encoding Notch, Notch ligands and components of the Notch signalling cascade display vascular defects, such as Alagille's syndrome, a developmental disorder with vascular defects and CADASIL (cerebral autosomal-dominant arteriopathy with subcortical infarcts and leukoencephalopathy) (Shawber & Kitajewski, 2004).

Formation of homotypic and heterotypic junctions

Homotypic and heterotypic junctions (including EC–EC, EC–mural cell and gap junctions) facilitate cell-to-cell communication and regulate vessel permeability.

Role of cadherins

EC express both neural (N)-cadherin (Liaw et al. 1990; Salomon et al. 1992) and vascular endothelial (VE)-cadherin (Lampugnani et al. 1992), and targeted deletion of these genes in mice leads to early embryonic death with associated severe vascular anomalies (Radice et al. 1997; Carmeliet et al. 1999; Gory-Faure et al. 1999). VE-cadherin homozygous null embryonic stem cells show disorganized vessel formation but express normal levels of other EC markers (Vittet et al. 1997). Function-blocking antibodies to VE-cadherin inhibit formation of capillary tubes and disrupt preformed capillary networks in the fibrin/collagen gel model in vitro (Bach et al. 1998). Expression analysis in mouse embryos reveals VE-cadherin gene expression in the endothelial precursor cells of the blood islands, and later in the vasculature of the organs examined (Breier et al. 1996).

Gory-Faure et al. (1999), noting defects in the yolk sac vessels concomitant with apparent normal intra-embryonic vessels (i.e. dorsal aortae), concluded that the extra-embryonic vasculogenesis was dependent on VE-cadherin activity, whereas intra-embryonic vasculogenesis was not. More recently, Crosby et al. (2005) showed that events of de novo blood vessel formation up to the point which a vascular epithelium forms are not dependent on VE-cadherin and that VE-cadherin, expression of which is up-regulated following vascular epithelization, is required to prevent the disassembly of nascent blood vessels.

Role of gap junctions (GJ)

GJ made of connexins, such as Cx37, Cx40 and Cx43, facilitate communication between EC, and between EC and perivascular cells, and play a critical role not only in differentiation of the vasculature, but also in vessel function in response to changes in flow and pressure and in maintenance of vascular tone.

The expression of connexin genes varies throughout the vasculature. Cx37, Cx40 and Cx43 show different expression patterns in aortic coronary vessels among bovine, pig and rat aorta (Van Kempen & Johugmsa, 1999). In vitro evidence has shown that Cx43 expression is rapidly up-regulated in bovine aorta-derived EC in response to non-laminar flow (De Paola et al. 1999) and in SMC by stretch (Cowan et al. 1998). Cx43 has been shown to be expressed in vivo at vessel branch points and regions of non-laminar flow, whereas Cx43 and Cx40 are primarily localized to regions of laminar flow in rat aortic endothelium (Gabriels & Paul, 1998).

Role of tight junctions (TJ)

Tight junctions include TJ peripheral proteins, such as the zonula occuldens (ZO)-1, -2 and -3 and MAGI proteins, as well as TJ-associated transmembrane proteins, such as the junctional adhesion molecules (JAMS), occludin and claudin, which physically occlude the intercellular cleft domain (Gonzales-Mariscal et al. 2003).

Endothelial TJ limit the paracellular flux of hydrophilic molecules across the blood–brain barrier (BBB), which is composed, in addition to EC, of the capillary basement membrane, astrocyte endfeet ensheathing the vessels and pericytes embedded within the basement membrane.

EC differentiation to form organ-specific capillary structures

One of the most intriguing aspects of any developmental process is how differentiating cells migrate to the proper location in the correct spatial and temporal organization to form specific structures, such as organs and tissues. Every organ has a different interaction with the circulatory system. Therefore, it is not surprising that blood vessels display organ-specific features. This is true in particular for microvessels, which have the closest contact with the surrounding organ-specific cells.

EC are a heterogeneous population. There are differences between the endothelia of different species, between large and small vessels, and between EC derived from various microvascular endothelial beds. These differences have been ascribed to genetic predisposition and environmental influences (Page et al. 1991; Aird et al. 1997). The continuous endothelium is either thick or thin. Continuous thick capillaries (EC > 2 µm thick) are found in skeletal tissue, cardiac smooth muscle, testes and ovary tissues, whereas continuous thin capillaries (EC < 1 µm thick) are typical of the central nervous system (CNS) and dermis. Brain EC interact with astrocyte endfeet to produce the BBB by forming continuous endothelium with complex TJ and highly regulated polarized endocytosis and transcytosis (Abbott, 2002). The liver, spleen and bone marrow sinusoids are lined by discontinuous EC that allow cellular trafficking between intercellular gaps. The endocrine glands and kidneys are lined by fenestrated EC that facilitate selective permeability required for efficient absorption, secretion and filtering (Dejana, 1998). The only difference between fenestrated and discontinuous capillaries is the diameter of the pore and the presence or absence of a diaphragm.

EC heterogeneity is also evident in individual organs. For example, the kidney contains fenestrated EC in its peritubular capillaries, discontinuous EC in its glomerular capillaries and continuous EC in its other regions (Risau, 1995).

In addition to morphological heterogeneity, there is functional heterogeneity of EC, including roles in vasoconstriction and vasodilation, blood coagulation and anticoagulation, fibrinolysis, leukocyte homing and diapedesis, acute inflammation and wound healing, atherogenesis, antigen presentation and catabolism of lipoproteins (Gerritsen, 1987).

Role of oxygen

Oxygen can be considered a repelent, while the lack of oxygen is a strong attractant for vessel sprouts. In cells further than the oxygen diffusion limit from a vessel, hypoxia activates hypoxia-inducible transcription factors (HIF-1α, 2α) which turn on the expression of angiogenic genes such as VEGF, inducing vessels to branch towards the hypoxic tissue (Pugh & Ratcliffe, 2003). Thus, when HIF-dependent VEGF expression is genetically dysregulated, organs fail to grow and function normally, because of insufficient vessel branching and growth (Carmeliet et al. 1999a,b; Gerber et al. 1999; Mattot et al. 2002).

Haemodynamic forces

The luminal surface of blood vessels is constantly exposed to haemodynamic forces, primarily to shear stress, which is the tangential force acting upon the EC surface by blood flow. An intraluminal stimulus such as increased shear stress induces capillary expansion without branching, while an abluminal stimulus causes vessel sprouting.

Shear stress is considered to be the driving force behind arteriogenesis (Schaper & Scholz, 2003), which operates to increase the diameter of those vessels forced to handle more flow and hence subjected to an elevated shear stress (Van Royen et al. 2001).

Flow is also critical to mainatin vessel branches (Le Noble et al. 2005). A process termed intussusceptive branching remodelling (IBR) has been described and shown to operate in changing branching angles (Djonov et al. 2002) in a manner optimizing the bifurcation exponent (Bennet et al. 2000). Experimental changes in blood flow dynamics triggered changes in branching angles through IBR towards optimality (Djonov et al. 2002). In addition, haemodynamic forces are critical in reshaping nascent vascular networks and branch angles to optimize flow, probably via shear stress-dependent release of angiogenic signals (Djonov et al. 2002).

A role has been suggested for shear-stress-induced chemoattractants such as MCP-1 in attracting cells that produce endothelial mitogens (Ito et al. 1997). Other studies have demonstrated that shear stress itself can induce signalling by several tyrosine-kinase receptors, including VEGFR-2, without a ligand (Chen et al. 1999).

Regression of certain transient vascular networks may indeed be triggered by cessation of flow, presumably through the down-regulation of shear-regulated vascular survival factor (Meeson et al. 1999).

Neural guidance molecules regulate vascular remodelling and vessel guidance

Like blood vessels, nerves are complex branched systems, and the patterning of nerves and vessels is often congruent in peripheral tissues (Mukouyama et al. 2005). Emerging evidence suggests that axon growth cones and capillary tip cells use common signalling cues (Eichmann et al. 2005a,b).

Several molecules initially discovered for axons, including Eph, netrins, slits and semaphorins, are also involved in vascular remodelling and vessel guidance (Eichmann et al. 2005a,b). In particular, semaphorin 3E and its receptor plexin D1 in addition to the netrin receptor UNC5B have been shown to direct endothelial tip cell navigation (Serini et al. 2003; Lu et al. 2004; Gu et al. 2005).

Variations in EC gene expression

There are also differences in the repertoire of genes expressed by EC of different vessel beds. Distinct peptides could mediate phage homing to different populations of EC throughout the vasculature, including the lung, kidney and brain (Pasqualini & Ruoslahti, 1996; Rajotte et al. 1998). Each peptide binds to a distinct EC surface protein, some of which have been identified (Burg et al. 1999; Koivunen et al. 1999).

By comparing the gene expression patterns of normal and tumour vessels, it has been demonstrated that 46 transcripts were specifically elevated in the tumour endothelium of adult tissues (St Croix et al. 2000; Carson-Walter et al. 2001). Further analysis of a number of these genes, however, indicated that were also expressed during embryonic vascular development.

Interplay between genetic and epigenetic factors

Several studies suggest that EC are not genetically committed to an arterial or venous phenotype, but are plastic and adapt to the expression of the arterial- or venous-specific genes based on local environmental cues (Moyon et al. 2001a; Othman-Hassan et al. 2001).

Le Noble et al. (2004) used a time-lapse video microscopy system and examined arterial–venous differentiation in the developing yolk sac of chick embryo. They observed that prior to the onset of flow, EC expressing arterial- and venous specific markers are localized in a posterior-arterial and anterior-venous pole. Ligation of one artery by means of a metal clip, lifting the artery and arresting arterial flow distal to the ligation site could morphologically transform the artery into a vein. When the arterial flow was restored by removal of the metal clip, arterial makers was re-expressed, suggesting that the genetic fate of arterial EC is plastic and controlled haemodynamics.

Heil et al. (2004) have demonstrated that arteriogenesis, the process of developing an enlarged artery from a pre-existing small arteriole, is in part mediated by activated monocytes under conditions of acute arterial obstruction with concomitant increase in flow or circumferential wall stresses to selected downstream vessels, and is impaired in mice lacking the CC-chemokine receptor-2.

Concluding remarks

Our understanding of how blood vessels are formed, and what regulates this, has grown dramatically in recent years. Most of what we know about the blood vascular system is derived from the study of pathological and reparative events in adult organisms: inflammation, tumour growth and a variety of diseases with angiogenic or angiodegenerative components. These studies have produced a wealth of information about the structure, physiology and pathobiology of the vascular system.

Vascular development is a complex process that involves the formation, migration, proliferation, cytodifferentiation and regression of EC. The influence of ECM on cell differentiation and the mechanical forces of haemodynamics must also be considered.

The classic literature provides a thorough coverage of descriptive aspects of development of the embryonic vascular system. Modern techniques, such as the technology of cell fusion, cell sorting and image analysis, can now provide insights into the mechanisms of these dynamic events during which vessels form and regress, and blood flow changes directions in the same vessels, and can provide powerful tools for examining the processes that govern the development of the vascular system.