Blood vessel formation is a highly dynamic tissue-remodeling event that can be observed from early development in vertebrate embryos. Dorsal aortae, the first functional intra-embryonic blood vessels, arise as two separate bilateral vessels in the trunk and undergo lateral-to-medial translocation, eventually fusing into a single large vessel at the midline. After this dramatic remodeling, the dorsal aorta generates hematopoietic stem cells. The dorsal aorta is a good model to use to increase our understanding of the mechanisms controlling the establishment and remodeling of larger blood vessels in vivo. Because of the easy accessibility to the developing circulatory system, quail and chick embryos have been widely used for studies on blood vessel formation. In particular, the mapping of endothelial cell origins has been performed using quail-chick chimera analysis, revealing endothelial, vascular smooth muscle, and hematopoietic cell progenitors of the dorsal aorta. The avian embryo model also allows conditional gene activation/inactivation and direct observation of cell behaviors during dorsal aorta formation. This allows a better understanding of the molecular mechanisms underlying specific morphogenetic events during dynamic dorsal aorta formation from a cell behavior perspective.
The developing chick embryo is easily observed by creating a small hole in the eggshell; extended observation is even possible if the hole is resealed by adhesive tape. Researchers have taken advantage of this built-in culture system for more than a century, using chick embryos as a model for developmental biology. The blood vessel network, which can be observed in the chick embryo from day 2 of incubation, is one of the prominent features of vigorous embryogenesis. Moreover, a beating heart and flowing red blood cells are the first visible vital functions of the circulatory system; therefore, many biologists have been drawn to study the blood vessels in the chick embryo (for a review, see Noden 1990). Dorsal aortae are the first intra-embryonic blood vessels to arise in the trunk. Primary dorsal aortae comprise a pair of longitudinal vessels in which the anterior ends are connected to the nascent heart via outflow tracts and the posterior parts are linked to vitelline arteries at the umbilicus level. In addition to its critical function as the largest vessel to circulate blood through the embryo's body, the dorsal aorta is known to be a place of secondary embryonic hematopoiesis (for reviews, see Dieterlen-Lievre & Le Douarin 2004; Jaffredo et al. 2005; Adamo & Garcia-Cardena 2012). The dorsal aorta also acts as a signaling center, providing instructive signals to induce pancreas differentiation, migration of neural crest cells, and subsequent specification of cell-types (Reissmann et al. 1996; Schneider et al. 1999; Lammert et al. 2001; Yoshitomi & Zaret 2004; Saito et al. 2012). Moreover, the dorsal aorta does not resemble peripheral blood vessels, which are typically seen in small organs and cancer, in terms of its diameter, the number of endothelial cells incorporated within the vessel, and its blood flow volume. Studies of dorsal aorta formation during embryogenesis will provide clues to understanding the mechanisms through which large blood vessel formation is achieved in vivo. This review presents the dynamic process of dorsal aorta formation and describes recent discoveries related to regulatory mechanisms that have been revealed by studies using avian (mostly quail and chick) embryo models.
Overview of the avian model in vascular biology
In 1969, Le Douarin discovered that quail and chick cells have different nuclear morphologies can be distinguished from each other by the Feulgen reaction (Le Douarin 1969, 1973; for a review, see Le Douarin et al. 2008). This finding greatly motivated the fate mapping of various tissues by quail-chick chimera analysis, achieved by micro-scale tissue transplantation between these two embryos under a binocular microscope. Quail-chick chimera analysis has allowed the acquisition of many fundamental ideas and concepts in immunology, hematology, neurobiology, and developmental biology (for reviews, see Le Douarin 2005; Le Douarin et al. 2008). The development of a quail-specific monoclonal antibody, QCPN, generated by Bruce and Jean Carlson (available from the Developmental Studies of Hybridoma Bank (DSHB), University of Iowa, USA), facilitated the identification of the chimera by a quail-nuclear-specific signal (Selleck & Bronner-Fraser 1995). Species- and cell type-specific antibodies have been generated by several groups (Peault et al. 1983; Tanaka et al. 1990; Aoyama et al. 1992; for reviews, see Noden 1990; Le Douarin et al. 2008). QH1 is a cell type-specific monoclonal antibody that recognizes a glycosylated epitope in endothelial and hemangiogenic cells of the quail (available from the DSHB) (Pardanaud et al. 1987). The vascular cell-specific reactivity of the QH1 antibody facilitated the identification of quail donor-derived endothelial cells in chick host embryos; through this, the QH1 antibody has aided in the determination of the blood vessel origins of higher vertebrates (for reviews, see Dieterlen-Lievre & Le Douarin 2004; Jaffredo et al. 2010). Whole-mount ex ovo culture system has permitted the observation and manipulation of avian embryos at early stages (New 1955; Chapman et al. 2001). This culture system enabled the manipulation of the endoderm, which is located at an inaccessible position in ovo (Vokes & Krieg 2002; Kimura et al. 2006). Moreover, whole-mount ex ovo culture technique has allowed the direct observation of fluorescent-labeled cell behaviors by time-lapse imaging microscopy (Kulesa & Fraser 1998).
After the 1990s, techniques permitting gene delivery into developing avian embryos, for example, infection of retroviral vectors, electroporation, and sonoporation of plasmid vectors, became available (Yamagata et al. 1994; Itasaki & Nakamura 1996; Funahashi et al. 1999; Momose et al. 1999; Ohta et al. 2003). These methods have enabled the analysis of gene function in developing avian embryos. In addition, fate maps determined by the quail-chick chimera and fluorescent dye labeling have greatly contributed to the delivery of DNA constructs into progenitors of target tissues at appropriate stages. Tet-on/off inducible vector systems have allowed conditional gene manipulation in avian embryos, thereby giving researchers the ability to analyze common genes in lineage-overlapped tissues (Hilgers et al. 2005; Watanabe et al. 2007). Currently, genomic integration of exogenous genes by the transposon system and tissue-specific conditional gene manipulations are feasible in avian embryos (Sato et al. 2007; Hou et al. 2011; Yokota et al. 2011). These are very important tools to study the direct effects of gene manipulation in a specific morphological event by circumventing multiple abnormalities caused by early gene manipulation in a common progenitor. Inactivation of genes by shRNA and morpholino-oligos can also be achieved in the avian embryo, as well as other model animals (Katahira & Nakamura 2003; Sheng et al. 2003). Moreover, gene editing by zinc-finger nuclease (ZFN) or TAL effector nucleases (TALENs) is going to be feasible (Miller et al. 2007; Christian et al. 2010; Song et al. 2012). In summary, the avian embryo is a historically important and technically feasible model to study the mechanisms underlying blood vessel formation in higher vertebrates. These advantages have motivated scientists to generate a transgenic quail line that carries an endothelial cell-specific fluorescent reporter gene (Sato et al. 2010). With this model transgenic quail line, time-lapse imaging has revealed the dynamic nature of endothelial cell behaviors during development.
Dorsal aorta formation in the avian embryo
The descending aorta is the largest blood vessel in the trunk. Mature descending aorta consists of endothelial cells (ECs) surrounded by a thick layer of vascular smooth muscle cells (vSMCs) and fibroblasts embedded in the extracellular matrix (ECM) at the outer layer (for reviews, see Jain 2003; Jaffredo et al. 2010). This structure, unlike small capillaries, gives elasticity and robustness to the descending aorta and allows specialized functions, such as acting as the central vessel to receive blood flow pressure from the heart. The dorsal aorta is the embryonic precursor of the descending aorta. Nascent dorsal aortae arise from two separated cords at Hamburger-Hamilton (HH) stage 8 (Hamburger & Hamilton 1992) in avian embryos (Fig. 1) (Pardanaud et al. 1987; Coffin & Poole 1988). As the distribution pattern of QH1-positive cells shows, the primary dorsal aorta is not a continuous structure at its beginning. These QH1-positive cells are mostly ECs, according to analysis of VE-cadherin expression, a definitive EC marker, at the same stage (GEISHA ID: Cadherin-5) (Bell et al. 2004). The preliminary dorsal aorta develops a vascular appearance within a few hours, by HH stage 10 (E1.5). The anterior ends of the dorsal aortae are connected with outflow tracts, and the posterior ends gradually elongate toward the tail by connecting with the vascular plexus in the splanchnic mesoderm. Shortly after the heart beat becomes detectable at around HH stage 10–11, circulation of blood cells can be observed in the dorsal aortae by HH stage 12 (early E2). The vascular plexus, which is linked to the dorsal aortae, is remodeled into the vitelline artery after undergoing dynamic fusion and/or reduction of the vessels at the prospective umbilicus level (le Noble et al. 2004). In this way, the primary vascular network in the trunk is established.
The primary dorsal aorta is located underneath the lateral plate mesoderm and gradually changes positions from lateral to medial, eventually fusing with another dorsal aorta that migrates from the opposite side; this allows the formation of a single large blood vessel at the midline (Fig. 1) (Wiegreffe et al. 2007; Garriock et al. 2010). These lateral-to-medial and double-to-single transitions of the dorsal aorta distribution and morphology are conserved between amniote embryos. In the mouse embryo, a pair of nascent dorsal aortae is observed at E8.0, and fusion at the midline begins around E9.5 from anterior (Strilic et al. 2009). In turtle (Pelodiscus sinensis) embryos, the dorsal aorta is highly comparable to chick embryos at each equivalent stage (Nagashima et al. 2005). In amniote embryos, cardinal veins also arise as two separated vessels along the lateral side of somites later than the dorsal aorta and are maintained as symmetric vessels for several days (Bellairs & Osmond 2005). In zebrafish embryos, the dorsal aorta and posterior cardinal veins are formed by a single vessel from the beginning (Jin et al. 2005; for a review, see Swift & Weinstein 2009). Therefore, the processes of large blood vessel formation in the trunk are different between amniote and non-amniote embryos.
The two different origins of dorsal aorta endothelial cells
The splanchnic mesoderm has been implicated as the primary origin of dorsal aorta ECs (Poole & Coffin 1989). The angiogenic potential of the splanchnic mesoderm is supported by the expression of the endothelial/hematopoietic progenitor cell marker, Scl/Tal1 (Drake et al. 1997; Minko et al. 2003). Time-lapse imaging analysis of Cy3-QH1 pulse-labeled quail embryos directly demonstrated that ECs, which migrate from the lateral plate area, participate in primary dorsal aorta formation (Fig. 2A) (Rupp et al. 2004). Quail-chick chimera analysis performed by Pardanaud et al. demonstrated that the ECs of the dorsal aorta are derived from two separate lineages (Pardanaud et al. 1996). Somite-derived ECs mainly distribute into the roof, while splanchnic mesoderm-derived ECs distributed into the floor of the dorsal aorta (Fig. 2B,C). This fate-mapping analysis indicated that the dorsal aorta endothelium has two-distinct origins and is clearly compartmentalized into somite-derived roof and splanchnic mesoderm-derived floor regions. Participation of somite-derived cells in the dorsal aorta endothelium has been consistently observed in other fate-mapping studies of the somite and presomitic mesoderm (PSM) by fluorescent-dye labeling (Stern et al. 1988) and quail-chick chimeras (Wilting et al. 1995; Klessinger & Christ 1996). Similar to avian embryos, murine somite has the potential to provide ECs to the dorsal aorta (Ambler et al. 2001; Esner et al. 2006). The developing dorsal aorta significantly increases its diameter during lateral-to-medial migration, whereas expansive cell proliferation of pre-existing ECs in the dorsal aorta is not observed (Benedito et al. 2008). Considering such non-proliferative growth of the early dorsal aorta, recruitment of new ECs from nearby angiogenic mesoderm is a possibility. Indeed, detailed fate mapping of each dorsolateral, mediolateral, and ventral part of the somite demonstrated that individual compartments provide ECs to the nearest blood vessels, according to their position in the somite (Wilting et al. 1995). This suggests that the ventral part of the somite, rather than the dorsal part, is the preferential origin of the dorsal aorta endothelium.
Pardanad et al. noticed that splanchnic mesoderm-derived ECs form hematopoietic clusters at the floor of the dorsal aorta (Fig. 2D) (Pardanaud et al. 1996). When Ac-LDL-DiI is perfused into the circulation of chick embryos, ECs are specifically labeled with Ac-LDL-DiI through endocytosis. Using this technique, the fate of ECs lining the dorsal aorta can be analyzed in avian embryos (Jaffredo et al. 1998). At the aortic floor, labeled ECs form clusters that are positive for a hematopoietic cell marker, CD45, and negative for an EC marker, vascular endothelial growth factor 2 (VEGFR2) (Jaffredo et al. 1998). This was direct evidence supporting the transition of dorsal aorta ECs into hematopoietic cells. Moreover, extended lineage tracing determined that splanchnic mesoderm-derived ECs found in the floor of the dorsal aorta are transient and are replaced with somite-derived ECs (Fig. 2E) (Pouget et al. 2006). Thus, it appears likely that splanchnic mesoderm-derived ECs undergo an endothelial-to-hematopoietic transition (Pardanaud et al. 1996; Jaffredo et al. 1998; Pouget et al. 2006; for a review, see Jaffredo et al. 2010). Therefore, the somite is the definitive origin of dorsal aorta ECs. Moreover, in addition to the dorsal aorta, somitic ECs are broadly incorporated in cardinal veins, intersegmental vessels (ISVs), spinal vasculature, body wall, and limb bud vessels (Wilting et al. 1995; Pardanaud et al. 1996; Ambler et al. 2001; Pouget et al. 2006).
Somitic cell contribution to the dorsal aorta is regulated by Notch
The somite is known to give rise to musculoskeletal tissues of the adult body via the dermatome, myotome, and sclerotome (for a review, see Aoyama et al. 2005). The regulatory mechanisms underlying somite differentiation into these tissues are well studied at the molecular level (for a review, see Buckingham & Vincent 2009); however, our understanding of the somitic cell contribution to the dorsal aorta had not yet reached the level of molecular mechanisms. Notch signaling mediates signal transduction between neighboring cells by interactions between the transmembrane receptor, Notch (Notch1–Notch4) and the transmembrane Notch ligands Delta-like (Dll1–Dll4) and Jagged (Jag1 and Jag2, also known as Serrate1 and Serrate2 in chicks). Loss of Notch signaling components causes two major severe defects in blood vessel formation and somite segmentation in vertebrate embryos at early stages (for reviews, see Gridley 2007; Lewis et al. 2009). Mouse mutant embryos carrying any one of several Notch signal deficiencies die at early stages of development due to vascular defects; therefore, to understand Notch signaling functions at later stages, conditional gene activation/inactivation studies are required. Notch-activation in cells induces expression of downstream Hes/hairy family genes. In situ hybridization analysis of Hes/hairy genes occasionally allows the identification of Notch-activated tissues; however, in the case of PSM, Hes/hairy transcripts are quickly destabilized due to the oscillatory mechanisms of Notch signaling (for a review, see Pourquie 2011). To map Notch-activated cells, a stable Notch-responsive fluorescent reporter plasmid was introduced into developing chick embryos by electroporation (Sato et al. 2008). Fluorescent reporter signals were found in the posterior half of each somite and eventually localized to dorsal aorta endothelium. These Notch-activated cells in the dorsal aorta were also confirmed to have been derived from the somite by homotopic transplantation of a Notch reporter-expressing somite into a non-electroporated host embryo (Sato et al. 2008). Furthermore, overexpression of a constitutively active form of Notch1 in the somite induced selective migration of Notch-activated cells into the dorsal aorta. In this study, to circumvent Notch hyper-activation in lineage-overlapped PSM, expression of the constitutively-active form of Notch is induced after somite formation by a Tet-on conditional gene manipulation technique (Sato et al. 2007; Watanabe et al. 2007). Consistent with the known process of EC replacement by somite-derived cells (Pouget et al. 2006), Notch-activated cells are initially located in the roof of the dorsal aorta and are subsequently found in both the roof and floor at later stages (Sato et al. 2008). The selective contribution of Notch-activated somitic cells to the dorsal aorta indicates that Notch signaling induces differentiation of somitic cells into endothelial precursor cells (Fig. 3C). The Dll4- and/or Serrate1-expressing dorsal aorta is proximal to the ventral surface of the somite, while Dll1 is expressed in the posterior half of each somite (Myat et al. 1996; Palmeirim et al. 1998; Nimmagadda et al. 2007a). However, which of those ligands are responsible for Notch activation in the somite remains unknown. Defective formation of the dorsal aorta due to aberrant hyperproliferation and -migration of ECs has been reported in Dll4-null mutant mouse embryos (Benedito et al. 2008). In contrast, EC-specific deletion of the Serrate1 gene homologue, Jagged1, in mouse embryos does not result in defective dorsal aorta formation, but does cause failure of vascular smooth muscle cell contributions (Fig. 3D) (High et al. 2008). Thus, Jagged1/Serrate1 in the dorsal aorta is unlikely to be the ligand responsible for the induction of EC differentiation from the somite.
The EphirnB2 expression domain in the posterior half of the somite has been shown to overlap with the Notch-active domain (Sato et al. 2008). Loss-of-function phenotypes of Notch signal components in mouse and zebrafish embryos are associated with downregulation of EphrinB2 in aortic vessels; thus, EphrinB2 is thought to be a downstream genetic component of Notch signaling (Fig. 3B) (Lawson et al. 2001; Krebs et al. 2004; Grego-Bessa et al. 2007). Importantly, somitic cell contribution to the dorsal aorta endothelium by Notch activation is inhibited when EphrinB2 shRNA is co-electroporated. Therefore, EphrinB2 is thought to be a downstream player in the Notch signaling pathway that regulates the migration of somitic cells (Fig. 3C) (Sato et al. 2008). Mouse embryos carrying EC-specific gain-of-function Notch alleles show enlarged dorsal aortae and small cardinal veins, while embryos carrying Notch loss-of-function mutations exhibit small dorsal aortae and large cardinal veins (Kim et al. 2008). Thus, endothelial Notch signaling controls the vessel size balance between the dorsal aorta and cardinal veins. Notch signal activation occurs consecutively in somite and ECs and consistently acts to promote dorsal aorta formation.
The functions of Notch signaling in vascular development have recently been elucidated at the cellular level, particularly in the capillary tip cell model (Fig. 3A) (for reviews, see Roca & Adams 2007; Siekmann et al. 2008; Herbert & Stainier 2011). Tip cells are typically observed around the leading edges of sprouting blood vessels at the initial phase of vascular remodeling and lead vessel sprouting toward the direction of higher VEGF concentrations. The increased number of tip cells resulting from Dll4 disruption causes aberrant blood vessel formation both in vitro and in vivo, suggesting that Dll4 regulates the precise number of tip cells. Dll4 expression is induced by VEGF in tip cells, and this subsequently activates Notch receptor in adjacent stalk cells. Notch activation in the stalk cells leads to downregulation of VEGFR2 (also known as Flk1), suppressing the response of stalk cells in the vicinity to VEGF. Thus, Notch signaling plays a role in the fine-tuning of diffusible VEGF signaling in order to regulate the distribution of tip cells in the micro-capillary environment (for reviews, see Roca & Adams 2007; Siekmann et al. 2008; Herbert & Stainier 2011). Our previous study demonstrated that the dorsal aorta sends attraction signal to the endothelial precursor cells in the somites (Sato et al. 2008). This suggests one explanation for why heterotopically transplanted angiogenic tissues contribute normally to pre-existing dorsal aorta endothelium, not causing secondary dorsal aorta formation at the site of transplantation (Pardanaud et al. 1996) or allowing Notch-activated cells to form an ectopic dorsal aorta inside of the somite (Sato et al. 2008). Similar to the tip cell specification mechanism induced by VEGF-Notch crosstalk, Notch signaling may interact with some diffusible attraction molecules secreted from the dorsal aorta, thereby regulating appropriate population of responsive somitic cells that are destined to be ECs. In fact, the molecular nature of the attraction signal from the dorsal aorta is another important issue that still remains unknown.
A notochord-derived BMP antagonist regulates the laterality of the primitive dorsal aorta
The midline is initially avascular until the dorsal aorta migrates from lateral regions. When the notochord is ablated from quail embryos at earlier stages, blood vessels are inappropriately formed around the midline (Reese et al. 2004). In contrast, ectopic transplantation of the notochord into the paraxial mesoderm area results in prevention of vascular plexus formation around the secondary grafted notochord. Therefore, the notochord is thought to act in opposition to blood vessel formation (Fig. 4A) (Reese et al. 2004). The notochord expresses the bone morphogenetic protein (BMP) antagonists Noggin and Chordin during dorsal aorta formation (Reese et al. 2004; Garriock et al. 2010). Interestingly, implantation of Noggin- or Chordin-secreting cells mimics the effects of ectopic notochord transplantation on the absence of ECs (Bressan et al. 2009). Conversely, implantation of BMP4-secreting cells into the midline induces ectopic vascular plexus formation around the cells, similar to the effect observed in notochord-ablated embryos (Reese et al. 2004). Expression of BMP1, -2, and -4 mRNA can be detected in the lateral plate mesoderm, endoderm, and dorsal edge of the neural plate in early quail embryos (Reese et al. 2004). BMPs secreted from those tissues promote the formation of lateral blood vessels, while, secretion of Chordin from the midline negatively influences blood vessel formation, inhibiting angioblast differentiation, EC assembly, and fusion of dorsal aortae (Bressan et al. 2009).
Garriock et al. (2010) found that Chordin mRNA expression in the notochord was diminished directionally, from anterior to posterior, prior to dorsal aorta fusion. Consistent with this anteroposterior wave of Chordin mRNA attenuation, the negative influence of the notochord on dorsal aortae disappears, and, conversely, the notochord begins to facilitate dorsal aorta fusion from the anterior side (Garriock et al. 2010). In general, BMP antagonists spatially restrict BMP signal activation by physical binding with BMP ligands in the extracellular space, thus preventing the binding of BMP ligands to receptors. Genetic studies have shown that VEGF and Sonic hedgehog (Shh) are morphogens known to be involved in blood vessel formation; these genes are also expressed in the notochord. Inhibition of VEGF and Shh signaling functions in quail embryos leads to failure in dorsal aorta migration toward the midline in a similar manner to Chordin overexpression, indicating that VEGF and Shh are essential for dorsal aorta translocation (Garriock et al. 2010). Moreover, one of the VEGF receptors, VEGFR2 (also known as Quek1 in quails) likely functions downstream of the BMP4 signaling (Nimmagadda et al. 2005; Ben-Yair & Kalcheim 2008; Suzuki et al. 2008). The role of BMP signaling in dorsal aorta formation is hypothesized to be permissive. For example, after Chordin reduction, BMP signal activation allows laterally located ECs to receive instructive signals from the midline (Garriock et al. 2010).
Involvement of Hh signaling in dorsal aorta formation
Hedgehog (Hh) is a diffusible morphogenic ligand known to signal between tissues. Vertebrates have three homologous Hh genes, Sonic, Indian, and Desert hedgehog (Shh, Ihh, and Dhh, respectively). Involvement of hedgehog signaling in blood vessel formation has been reported in various model animals (for reviews, see Byrd & Grabel 2004; Nagase et al. 2008). Ihh, expressed in the visceral endoderm in the yolk sac, is required for proper angiogenesis of the yolk sac vasculature in mouse embryos (Dyer et al. 2001; Byrd et al. 2002). Targeted disruption of an Hh receptor, Smoothened (Smo), also results in severe deformation of the yolk sac vasculature and dorsal aorta (Byrd et al. 2002; Vokes et al. 2004). Since Hh signaling acts frequently in various tissues throughout development, conditional perturbation of Hh signaling is necessary to define its role at specific time points and in specific tissues. Conditional loss-of-function studies of Hh signaling were performed using a broad-spectrum Hh inhibitor, cyclopamine, in avian and mouse embryos. Cyclopamine treatment caused failure of dorsal aorta fusion (Vokes et al. 2004; Nagase et al. 2006b; Moran et al. 2011), and administration of a specific antibody against Shh (5E1) into quail embryos after primary dorsal aorta formation also resulted in failed fusion of dorsal aortae (Kolesova et al. 2008). Contrary to those antagonistic treatments, enhancement of Hh signaling by the Smo agonist, SAG, leads to enlargement of the dorsal aortae in quail embryos (Moran et al. 2011).
Shh expression in the notochord is conserved between vertebrate embryos. In a proposed model of dorsal aorta formation in zebrafish embryos, Shh secreted from the midline tissues induces VEGF expression in the somite (Lawson et al. 2002). This promotes Notch signal activation in prospective dorsal aorta ECs, which are located in the dorsal side of the intermediate cell mass (ICM). Eventually, ephrinB2 expression is activated in the dorsal ICM to display arterial EC identity. This cascade specifies that ICM cells in close proximity to the notochord will become arterial ECs (Fig. 3B) (Lawson et al. 2002). In a zebrafish model of dorsal aorta specification, Shh secreted from the notochord acted indirectly via VEGF; however, during earlier stages, Shh acted directly on the lateral-to-medial migration of angioblasts (Gering & Patient 2005; for reviews, see Hogan & Bautch 2004; Swift & Weinstein 2009). In an avian embryo model, Shh was expressed not only in the notochord but also in the endoderm, and its receptors, Patched1, Patched2 (Ptc1, Ptc2), and Smo, were expressed in developing dorsal aorta ECs, implying direct Shh signal transduction between the endoderm and dorsal aortae (Vokes et al. 2004). The endoderm always comes in contact with migrating dorsal aorta ECs until the cooperation of vascular smooth muscle cells has been achieved, and this tissue plays an important role in the formation of the tubular endothelium in frog and quail embryos (Vokes & Krieg 2002). Failure of vascular plexus formation caused by surgical elimination of the adjacent endoderm can be rescued by implantation of Shh-diffusible beads (Vokes et al. 2004). Moreover, in the same series of endoderm-ablation/bead-implantation experiments, the vascular plexus pattern observed after rescue treatment with VEGF was completely different from the Shh-induced pattern (Vokes et al. 2004). Since Shh induces capillary network formation in vitro independently of VEGF, Shh can act directly on ECs, and the endoderm can be considered as a substantial resource of Shh that influences dorsal aorta formation in the avian embryo. Although Hh is known to act as either a mitogen or survival factor in other tissues, the number of ECs in the dorsal aorta is not influenced by inhibition or activation of the Hh pathway (Vokes et al. 2004; Moran et al. 2011). Moreover, cyclopamine treatment in chick and mouse embryos induced downregulation of BMP4, VEGF, and VEGFR-2 expression (Nagase et al. 2006a; Moran et al. 2011), indicating that Hh signaling also functions as an upstream regulator of these blood vessel-related genes, similar to notochordal Shh in zebrafish embryos. Stage-dependent distinct vascular phenotypes of Hh-inhibited embryos suggest that Hh signaling is a commonly used mechanism influencing various aspects of blood vessel formation, including fusion of the dorsal aortae.
When, where, and how does VEGF affect dorsal aorta formation?
Vascular endothelial growth factor has long been studied as a key regulator of blood vessel formation. Mice lacking VEGF signal components show various developmental defects, including reduced numbers of ECs and failure to form a functional vasculature (for reviews, see Herbert & Stainier 2011; Eichmann & Simons 2012). Due to early death of VEGF signal-deficient mouse embryos, defining the function of VEGF signaling in the formation of specific blood vessels has been difficult. In the trunk regions of quail and mouse embryos, VEGF-A is widely expressed in various tissues, including the notochord, endoderm, neural tube, and somite, and VEGFR2/Flk1/Quek1 is expressed in the blood vessels and somite (Eichmann et al. 1993; Jaffredo et al. 1998; Hogan et al. 2004; Nimmagadda et al. 2004; Vokes et al. 2004; Ema et al. 2006; Garriock et al. 2010; Moran et al. 2011). Administration of VEGF-A into quail embryos induced inappropriate neovascularization in normally avascular areas and excessive fusion of the vessels, resulting in enlarged dorsal aortae, indicating that VEGF signaling strongly accelerates blood vessel formation in vivo (Drake & Little 1995; Vokes et al. 2004). In contrast, a VEGF antagonist, sFlt1 (the soluble form of VEGFR1/Flt1), inhibited angioblast assembly in the primary dorsal aorta (Argraves et al. 2002). The functions of VEGF signaling in vascular network formation are currently well characterized, and the mechanisms through which Notch-, Eph-, and Neuropilin-mediated pathways modulate the VEGF signal output on EC behaviors have also been elucidated (for reviews, see Herbert & Stainier 2011; Eichmann & Simons 2012).
In zebrafish and Xenopus embryos, VEGF-A found near axial or paraxial tissues has been implicated in angioblast migration and the subsequent formation of the dorsal aorta (Fouquet et al. 1997; Cleaver & Krieg 1998; Lawson et al. 2002; for reviews, see Swift & Weinstein 2009). In zebrafish embryos, VEGF-A is expressed by the ventromedial region of the somite (Liang et al. 1998). This site corresponds to the sclerotome of avian and mammalian somites. Moreover, in the mouse embryo, conditional deletion of a single VEGF-A allele by the Collagen2a1-Cre driver results in aberrant dorsal aorta formation, indicating the importance of somitic VEGF-A expression for dorsal aorta formation (Haigh et al. 2000). In its well-characterized role in the neural tube, VEGF-A has been shown to induce EC differentiation via VEGFR2 activation in the presomitic mesoderm (Hogan et al. 2004). These presomitic mesoderm-derived ECs are attracted to VEGF-A-expressing neural tube and eventually generate the perineural vascular plexus (Fig. 4A) (Hogan et al. 2004; James et al. 2009). In normal quail embryos at the dorsal aorta-forming stage, VEGFR2/Quek1 is expressed in the dorsolateral region of the somite (Eichmann et al. 1993; Nimmagadda et al. 2005), corresponding to dorsolaterally located EC origins, unlike that observed in the dorsal aorta (Wilting et al. 1995). Experimental disruption of VEGFR2/Quek1 gene expression in the somite results in reduction of cardinal veins and dorsolaterally located peripheral vessels, suggesting that somitic VEGR2/Quek1 is involved in the formation of laterally located blood vessels (Nimmagadda et al. 2005; Ben-Yair & Kalcheim 2008). To date, contributions of VEGR2-expressing somitic cells on the formation of dorsal aorta endothelium have not been confirmed. However, it has been demonstrated that a single VEGF-A allele deletion in mouse embryos leads to failure in lumenized dorsal aorta formation (Strilic et al. 2009), and VEGF has been shown to regulate Rho-associated protein kinase (ROCK) activity, mediating the recruitment of cytoskeletal molecules involved in formation of the vascular lumen (Strilic et al. 2009). Taken together, these studies suggest that VEGF signaling may have distinctive functions in the formation of the dorsal aorta and peripheral blood vessels.
Molecular mechanisms of dorsal aorta lumen formation
Recently, the molecular basis of dorsal aorta lumen formation has been investigated in mouse embryos. Observation of the ultrastructure of nascent dorsal aorta ECs by electron microscopy demonstrated that the primary dorsal aorta has a cord-like shape without a lumen, and adjacent ECs are in close contact with each other at VE-cadherin-mediated multiple adherence junctions (Strilic et al. 2009). These adherence junctions are remodeled through the anti-adhesive functions of CD34-sialomucins, and subsequent protein kinase C (PKC)- and ROCK-mediated cytoskeletal reorganization mechanisms promote vascular lumen formation (Strilic et al. 2009). Ras-interacting protein1 (Rasip1), an EC-specific regulator of GTPase, suppresses RhoA and downstream ROCK activity (Xu et al. 2011). In Rasip1-deficient mouse embryos, the dorsal aorta displays a cord-like immature structure due to failure in reorganization of junctional proteins, which are essential for proper EC association with surrounding ECM (Xu et al. 2011). Thus, these studies revealed that a GTPase-mediated mechanism regulates dynamic changes in EC morphologies during the formation of the vascular lumen.
A novel in vitro model for vascular fusion, the uniluminal vascular spheroid, has been established from mouse embryonic allantoic tissue. This vascular model contains both ECs and vascular smooth muscle cells in inner and outer layers, respectively (Fleming et al. 2010). The uniluminal vascular spheroid has a larger lumen (approximately 400 μm in diameter) than conventional in vitro microcapillary models. When a pair of the uniluminal vascular spheroids is attached in a hanging drop culture, these two vascular spheroids fuse into a single spheroid within 12 h (Fleming et al. 2010). This simplified model of large blood vessel fusion is expected to facilitate our understanding of the molecular and cellular mechanisms of dorsal aorta fusion.
The contribution of smooth muscle cells to the dorsal aorta
Vascular smooth muscle cells (vSMCs) in the dorsal aorta initially arise from lateral plate mesoderm and cover the ventral aspect of fusing dorsal aortae in amniote embryos (Wiegreffe et al. 2007, 2009; Wasteson et al. 2008). Quail-chick chimera analysis revealed that sclerotomal cells give rise to vSMCs on the roof of the dorsal aorta, whereas myotome-derived smooth muscle cells do not contribute to vSMC (Wiegreffe et al. 2007; Pouget et al. 2008). Interestingly, similar to the ECs, lateral plate-derived smooth muscle cells are eventually replaced by sclerotome-derived cells (Pouget et al. 2006, 2008; Wiegreffe et al. 2007). Sequential contributions of lateral plate- and somite-derived vSMCs to the dorsal aorta have also been observed in mouse embryos by genetic labeling approaches (Wasteson et al. 2008).
Jag1 on the EC surface promotes vSMC differentiation by interaction with Notch3 on prospective vSMCs (High et al. 2008; Liu et al. 2009). The Notch3 gene is highly expressed in vSMCs and Notch3 deletion mutants display defects in vSMC formation during postnatal maturation of arterial vessels (Domenga et al. 2004; Liu et al. 2009). Failure in vSMC formation during embryogenesis can be caused by Notch2/Notch3-double inactivation. In this double mutant mouse embryo, primary blood vessel formation occurs normally, but severe vascular collapse takes place later due to loss of vSMCs (Wang et al. 2012). This indicates that vSMCs are essential for vessel stabilization and EC survival (Fig. 3D). Together, these studies in mouse embryo models allow us to describe mechanisms of cardiovascular anomalies in human congenital diseases, which are closely associated to these Jag1 and Notch3 mutations (Shawber & Kitajewski 2004). Moreover, electroporation of somitic cells with a constitutively active form of Notch1 affects the smooth muscle layer around the dorsal aorta as well as the endothelial layer (Sato et al. 2008), mimicking the vSMC-inducible activity of Notch2 or Notch3 (Ben-Yair & Kalcheim 2008). In extra-embryonic blood islands, Notch1-activated cells preferentially give rise to vSMCs in response to Wnt signaling, preventing common progenitor cells from adopting an EC fate (Shin et al. 2009). The outcome of Notch signaling activation varies according to morphogen circumstances, for example, by crosstalk with the Wnt signaling pathway (Corada et al. 2010; for a review, see Andersen et al. 2012). Human Notch1 was first identified in T-cell leukemia, and Notch signaling is also involved in hematopoiesis (Ellisen et al. 1991; for a review, see Bigas et al. 2010). During dorsal aorta formation and subsequent hematopoiesis, Notch signaling functions in reciprocal cell-fate determination events by either lateral inhibition or lateral induction mechanisms between common progenitor cells.
Other molecules implicated in dorsal aorta formation
Foxc2/Mfh1, a member of the winged helix/forkhead family of transcription factors, is expressed in both the sclerotome and dorsal aortae in avian and murine embryos (Furumoto et al. 1999; Pouget et al. 2008). Double-null mutation of Foxc1 and Foxc2 in mouse embryos causes failure of normal somitogenesis and blood vessel formation (Kume et al. 2001). Foxc2 acts as a key regulator of blood vessel formation by controlling the transcription of Dll4, CXCR4, Integrinß3, and Angiopoetin2 genes. In addition, Foxc2 acts upstream of Notch signaling in the somite (Kume et al. 2001). A recent study using mouse mutant embryos demonstrated that segregation of Pax3 and Fox2 expression in the myotome and sclerotome, respectively, is regulated by a reciprocal downregulation mechanism between Pax3 and Foxc2 (Lagha et al. 2009). Foxc2 directs multipotent progenitors in the somite to vascular cell fates, suppressing muscular cell fates. This suggests higher levels of Foxc2 permit the participation of somitic cells in the dorsal aorta ECs and vSMCs (Lagha et al. 2009).
SDF1, a chemokine ligand, is expressed in the dorsal aorta and approximate somite, and its receptor, CXCR4, is expressed in the dorsal side of the somite (Ohata et al. 2009). Although overexpression of CXCR4 promotes migration of the CXCR4-expressing somitic cells toward SDF1-expressing tissues, these cells do not participate in the dorsal aorta formation (Ohata et al. 2009). In later stages, SDF1 secreted from the dorsal aorta directs migrating neural crest cells, which give rise to sympathetic neurons and the adrenal medulla beside the dorsal aorta (Saito et al. 2012). In zebrafish embryos, Cxcl12b, a fish homologue of murine SDF1, is expressed in the endoderm. Cxcl12b is required for proper guidance of Cxcr4a-expressing angioblasts that give rise to lateral dorsal aortae (Siekmann et al. 2009). Similar to the zebrafish embryo, SDF1 in the endoderm guides angioblasts in chick embryos (Fig. 4) (Katsumoto & Kume 2011). Inhibition of CXCR4 activation by AMD3100, a specific CXCR4 inhibitor, at an earlier stage (i.e., HH stage 6, before dorsal aorta formation) resulted in absence of the dorsal aortae, indicating that the SDF1-CXCR4 chemokine signal acts to guide EC precursors that give rise to the dorsal aorta. Failure to recruit the dorsal aorta precursor cells to the vicinity of the endoderm influences the expression of Pdx1, a pancreas marker, in the endoderm. Therefore, SDF1-CXCR4 signaling mediates reciprocal interactions between primary dorsal aorta precursor cells and the endoderm to instruct pancreas development (Katsumoto & Kume 2011).
Endothelial-to-hematopoietic transition of the dorsal aorta floor
Hematopoietic cell-like cluster localization at the floor of the dorsal aorta has been observed in avian embryos and various mammalian embryos for more than 100 years (Fig. 2D) (for a review, see Adamo & Garcia-Cardena 2012). Initial hematopoietic cells are derived from extra-embryonic blood islands, and these are replaced by intra-embryonic blood cells arising from the floor of the dorsal aorta (for reviews, see Dieterlen-Lievre & Le Douarin 2004; Dieterlen-Lievre et al. 2006). Upregulation of hematopoietic cell-specific markers strongly supports the emergence of hematopoietic cells at the dorsal aorta floor (Jaffredo et al. 1998, 2005). Since hematopoietic stem cells, which colonize the thymus, spleen, bursa, and bone marrow, have intra-embryonic origins, the dorsal aorta has been suggested to be a progenitor of hematopoietic stem cells in avian embryos (for reviews, see Dieterlen-Lievre & Le Douarin 2004; Dieterlen-Lievre et al. 2006; Jaffredo et al. 2010). Pulse labeling of ECs in a tamoxifen-inducible system in mouse embryos confirmed that hematopoietic stem cells are derived from the ECs located at the dorsal aorta floor, similar to that observed in avian embryos (Zovein et al. 2008).
One possible mechanism that may restrict hematopoietic potential in the splanchnic mesoderm was addressed by Pardanaud & Dieterlen-Lievre (1999). They found that transplantation of a quail somite, transiently exposed to the endoderm, into the coelom of a chick host embryo gave rise to hematopoietic clusters at the dorsal aorta floor. This result suggests that the endoderm initiates hematopoietic potential to endothelial precursor cells in the closely associated splanchnic mesoderm (Pardanaud & Dieterlen-Lievre 1999). However, it is not clear whether only splanchnic mesoderm-derived ECs undergo the transition to hematopoietic cells if they are replaced by somite-derived ECs. Quantitative evaluation of somite- and splanchnic mesoderm-derived EC fates at later stages is required to adequately address this question. A better understanding of the molecular mechanisms driving splanchnic mesoderm lineage-specific redifferentiation of ECs into hematopoietic stem cells is necessary to determine the potential therapeutic applications of this knowledge.
Dorsal aorta formation in nonamniotes
Nonamniote embryos do not form the dorsal aorta from fusion of bilateral dorsal aortae. In zebrafish embryos, laterally located angioblasts migrate toward the midline and give rise to the ICM (Fouquet et al. 1997; Gering et al. 1998). These lateral plate-derived cells populate the dorsal aorta endothelium, and the somite is unlikely to contribute to the dorsal aorta. In addition to the formation of the dorsal aorta, the process of cardinal vein formation in zebrafish embryos is completely different from that in avian embryos. Time-lapse imaging analysis of a kdrl-GFP zebrafish embryo revealed that the cardinal vein arises from sprouting angiogenesis of the ventral region of the dorsal aorta (Herbert et al. 2009); in contrast, in avian embryos, cardinal veins are generated de novo by vasculogenesis from the somite (Pardanaud et al. 1996; Pouget et al. 2006). Blood vessel patterns vary in low-oxygen conditions due to upregulation of hypoxia inducible factor (HIF). HIF proteins are bHLH transcription factors that allow cells to survive in low-oxygen environments by controlling the expression of genes involved in energy metabolism, vascular remodeling, hematopoiesis, cellular proliferation, and apoptosis (for a review, see Dunwoodie 2009). HIF2α, a member of the HIF family of proteins, is involved in the transcriptional activation of VEGF-A, VEGFR2/Flk1, and Tie2 genes. Low-oxygen conditions induce HIF2α expression in chick embryos (Ota et al. 2007), implying that vascular cell behaviors are flexibly modified according to the hypoxic state of the body. Accordingly, differences in blood vessel formation processes between amniote and nonamniote embryos may be related to embryo body size and/or growth environment, either on land or underwater.
The endothelial-to-hematopoietic transition at the floor of the aorta is conserved between fish (Kissa & Herbomel 2010; Lam et al. 2010), amphibians (Mills et al. 1999), birds (Pardanaud et al. 1996; Jaffredo et al. 1998), and mammals (Zovein et al. 2008; Boisset et al. 2010), with the exception of the direction of hematopoietic cell release. In avian and mammalian embryos, hematopoietic cells bud dorsally into the lumen; conversely, zebrafish hematopoietic cells bud ventrally and are stored in a subaortic space between the dorsal aorta and cardinal vein (Kissa & Herbomel 2010; Lam et al. 2010). These hematopoetic cells transmigrate into the dorsal aorta lumen during the onset of circulation (Iida et al. 2010). Amniote embryos also have EC-derived subaortic cells underneath the dorsal aorta; however, these cells are unlikely to participate in hematopoiesis (Zovein et al. 2008). Mechanical stress generated by blood flow promotes hematopoiesis from the dorsal aorta endothelium in zebrafish embryos and mouse ES cells (Adamo et al. 2009; North et al. 2009). Runx1 upregulation at the dorsal aorta floor is essential for the endothelial-to-hematopoietic transition, but not thereafter (Chen et al. 2009). Additionally, blood flow is required to induce Runx1 upregulation (North et al. 2009); however, whether the fluid shear stress is biased toward floor endothelium still remains unclear. Further studies are necessary in order to understand how Runx1-expressing hematopoietic stem cells are spatially restricted only in the dorsal aorta floor. In chick embryos, measurement of hemodynamics in vivo has been achieved by noninvasive laser Doppler imaging techniques. This revealed that pulsatile shear is an important exogenous factor directing arterial cell differentiation and vitelline vasculature remodeling (Buschmann et al. 2010).
Dynamic endothelial cell behaviors during dorsal aorta migration
Endothelial cell behaviors during the lateral-to-medial dorsal aorta translocation have been addressed using the transgenic quail embryo, tie1:H2B::eYFP (Sato et al. 2010; see also the Lansford section in this issue). In tie1:H2B::eYFP transgenic quails, whole ECs are specifically labeled with histone 2B-tagged eYFP under the control of the tie1 promoter from embryogenesis until adulthood. Imaging of the tie1:H2B::eYFP embryo by confocal laser microscopy has facilitated our understanding of the three-dimensional structure of the blood vessels and the distribution pattern of the ECs lining. Nuclear-localization of the eYFP signals can be clearly identified and allows us to trace their behaviors automatically using image processing software. Thus, using time-lapse observation of tie1:H2B::eYFP transgenic quail embryos by two-photon laser microscopy, we were able to describe individual EC behavior during dorsal aorta translocation with higher spatial resolution. Dorsal aorta ECs migrate overall in the anterior direction, while the floor and roof regions migrate in a different manner (Sato et al. 2010). Floor ECs move collectively from the lateral to medial region, whereas roof ECs do not move in a collective manner. Such distinct modes of cell migration between floor and roof walls may be related to the distinct origins of these populations (Pardanaud et al. 1996). In the roof endothelium in particular, recruitment of angioblasts from the somite occurs, potentially influencing the migration of pre-existing ECs. In contrast, floor ECs are associated with the endoderm during migration. Mechanical interactions between the endoderm and cardiogenic mesoderm during the collective migration of these tissues in the course of heart morphogenesis were recently described in chick embryos (Varner & Taber 2012). Contraction of the endoderm around the midline pulls the associated cardiogenic mesoderm in the same direction. Dorsal aorta floor cells may use a similar mechanism to move collectively from the lateral to medial region. However, no studies have yet addressed the involvement of specific molecules in these distinct cell behaviors. As described in previous sections, Hh, VEGF, and BMP antagonists are involved in the dynamic change of the dorsal aorta pattern (Fig. 4B). Further quantitative imaging experiments are awaited to understand how these molecules influence the individual EC behavior in the dorsal aorta.
Vascular endothelial, smooth muscle, and hematopoietic cells arise from closely associated progenitors in early development. In addition, many regulatory molecules are involved in various stages of blood vessel formation. Determination of the specific cell progenitors in avian embryos has provided an essential framework for investigating site- and stage-specific molecular mechanisms, enabling us to develop a detailed understanding of gene functions at each elementary step of blood vessel formation and facilitating the investigation of dorsal aorta-specific mechanisms. Studies of dorsal aorta formation have given us insights into the distinct molecular mechanisms of large blood vessel formation in vivo, differentiating this process from that of small capillaries. Further understanding of the mechanisms controlling dorsal aorta endothelial cell behaviors and the floor-specific endothelial-to-hematopoietic transition phenomenon will allow reconstruction of large blood vessels and induction of hematopoiesis in vitro for potential therapeutic applications in the future.
This work was supported by JSPS KAKENHI grant number 22770219 and 23111523. The author thanks past members of Professor Yoshiko Takahashi's and Dr Rusty Lansford's research groups for the works cited here.