Slits are secreted molecules found throughout invertebrate and vertebrate species. There are three described genes in mammals (Slit1, 2, and 3), as well as several splice variants of all three members [48]. They contain leucin-rich repeats (LRRs), endothelial growth factor (EGF) repeats, and a laminin G-like domain (Fig. 11.2a). The known receptors for slits belong to the Robo family. In mammals, there are four Robo receptors, Robo1, 2, 3, and 4 (Fig. 11.2b), together with their described isoforms Dutt1 (for Robo1), Robo2_tv2 (for Robo2, described in zebrafish and rat), as well as Robo3.1 and Robo3.2 (for Robo3) [48]. While Robo1, 2, and 3 are similar and are mainly expressed in the nervous system, Robo4 appears to be special, showing a different molecular structure accompanied with a specific expression pattern within the vascular compartment in mammals [49–51] (Fig. 11.2a). In addition, controversy exists regarding the binding of slits to Robo4 (see below).

Slit/Robo signaling is involved in different neuronal processes, such as axon guidance, axon fasciculation, and dendritic branching [52, 53]. In the developing nervous system of invertebrates and vertebrates, slits were described as exerting a repulsive effect on axons [54, 55]. It was thereby found that Slit/Robo signaling regulates the midline crossing of commissural axon projections by a mechanism that involves differential expression of Robo receptors in pre- and postcrossing axons [53, 56]. Interestingly, interactions between Robo1 and the netrin receptor DCC have been identified in commissural axons [57].

There is a growing body of literature indicating that Slit/Robo functions are not simply restricted to axonal guidance, and do indeed also play a role in the vascular system. Since Robo4 is specifically expressed in the vascular system, the majority of studies have focused on its endothelial-specific function. We summarise the main findings below.

In situ hybridization of whole-mount mouse embryos shows that Robo4 expression starts in large axial vessels and shifts from the dorsal aorta to the intersomitic vessels and capillaries as development proceeds [50]. Robo4 is also expressed in the vasculature of adult tissues [50]. Robo4 knockout mice are fertile and viable, and show normal patterning of blood vessels in different embryonic structures, indicating that Robo4 is either dispensable for developmental angiogenesis or its absence is compensated by yet an unknown mechanism [58]. In contrast, studies in zebrafish show that Robo4 depletion via morpholinos injection, and also overexpression, leads to reduced and defective formation of intersomitic vessels via a mechanism that involves Cdc42 and Rac1 Rho GTPases [59] (Fig. 11.2b).

Slit1, 2 and 3 have been proposed as ligands for Robo4 in ECs [50, 58, 60, 61]; however, other studies challenge whether slits really bind to Robo4, raising the question of potential, still unknown, ligands or a ligand-independent activation of Robo4. Here we describe studies showing activation of Robo4 by slits , as well as studies where slits did not bind to Robo4.

During the growth of a new blood vessel sprout, stalk cells follow behind the leading angiogenic endothelial tip cell and participate in maturation and lumen formation of the new growing vessel [58, 62]. Phalanx cells are the most quiescent ECs in the vessels and are covered by pericytes and form tight junctions with each other (Fig. 11.2). A detailed analysis of Robo4 in the developing retina vasculature revealed that stalk cells express Robo4, while tips cells are Robo4-negative [58] (Fig. 11.2), suggesting that Robo4 might not function as a typical guidance receptor in the endothelium but rather as factor-regulating vascular stability . In support of this concept, even though Robo4 knockout mice do not show obvious differences in vascular patterning, Robo4-deficient animals possess a higher basal level of vascular permeability [58]. Moreover, when subjecting these mice to oxygen-induced retinopathy, they respond with increased vascular leakage compared with wild-type mice [58]. Robo4 stabilizing function is strengthened by results showing that Robo4 (in this case activated by Slit2 or Slit3) inhibited VEGF-induced vascular destabilization and blocked VEGF-induced EC migration, tube formation, and permeability in vitro. These effects were due to a Robo4-dependent blockage of VEGF-induced activation of the Src-family kinases [58, 63] (Fig. 11.2b). Another study in the mouse mammary gland also found a role for Slit/Robo4 signaling in restricting VEGF function by downregulating VEGFR2 activation of Src and focal adhesion kinase (FAK) [62] (Fig. 11.2b).

Interestingly, direct binding of slits to Robo4 is questioned by structural analysis as well as functional investigations of the binding properties of slits [64–66]. In a protein–protein interaction screen to identify Robo4 binding partners, no association of slits to Robo4 was detected. In contrast, an interaction with Unc5B, a Netrin1 receptor (see above), was identified [64] (Fig. 11.2c). Robo4 and Unc5B are co-expressed in ECs, and stimulation of Unc5B-transfected porcine aorta ECs (PAECs) with the extracellular domain of Robo4 (sRobo4) led to Unc5B internalization and cell retraction. Robo4/Unc5B signaling leads to the association of Src kinase to Unc5B and its activation. Moreover, Robo4/Unc5B signaling counteracts VEGF-induced Src activation by sequestering Src away from VEGFR2 [64] (Fig. 11.2c). Consistently, the intraperitoneal injection of an antibody that specifically blocks Robo4/Unc5B interaction led to increased angiogenesis and vascular hyperpermeability in vivo. Since Unc5B is expressed on all cells of the angiogenic sprout (see above), and the expression of Robo4 is excluded from the tip cell, their trans-interaction might represent an additional mechanism regulating tip and stalk cells and, consequently, the angiogenic sensitivity of the vascular sprout (Fig. 11.2c).

Altogether, the studies described above clearly demonstrate the function of Robo4 signaling for vascular stabilization, either by acting as a ligand for Unc5B or by acting as a receptor for slits. These two possible scenarios are not exclusive when considering that other studies have shown the formation of heterodimers between Robo1 and Robo4 in HUVECs [67], thus opening the possibility that slits binding to Robo1–Robo4 complexes also leads to the described effects via a Robo4-dependent, Unc5B-independent mechanism. Moreover, in vitro studies showed that Slit2 via activation of Robo1 and Robo4 inhibits pericyte migration, which might be interpreted as an additional mechanism for stabilizing endothelial integrity [51] (Fig. 11.2d). These results suggest that signal transduction of Slit/Robo4 interaction depends on Robo1 regulating vascular stabilization and thereafter influencing endothelial sprouting.

Robo4, which is strongly expressed in highly active endothelium, is also regulated by hypoxia [49]. However, to date no studies have been conducted proposing a role for Robo4 in tumor angiogenesis. Thus, whether Robo4 also plays a role in tumor angiogenesis, or whether Robo4 is only required in healthy conditions and Robo1 in pathological circumstances, remains to be determined. In support of the second hypothesis, a quantitative analysis of Robo1, Robo4, and Slit2 messenger RNA (mRNA) in colorectal cancer tissue versus normal tissue revealed that while Robo1 was significantly upregulated in tumor tissue, no changes where observed for Robo4 or Slit2 [72].

Taken together, the available evidence suggests that Slit/Robo signaling in the vasculature possesses a stabilizing and maturating role. Further studies would need to clarify the situations where they might have an antiangiogenic or proangiogenic effect.

Eph receptors are named after the Erythropoietin-producing human hepatocellular carcinoma cell line in which they were discovered. Eph receptors are divided into two subgroups—EphAs (EphA1–10 in vertebrates) and EphBs (EphB1–6 in vertebrates)—depending on the sequence homology of their extracellular domain [73, 74] (Fig. 11.3a). Eph receptors interact with the membrane-bound proteins termed ephrins (greek ‘ephros’ for controller), which are also divided into two subclasses—EphrinAs and EphrinBs (Fig. 11.3a). The five homologous members EphrinA1–5 are anchored in the plasma membrane by a GPI linker, and bind almost exclusively to EphA receptors [73]. In contrast, EphrinBs are transmembrane proteins possessing a C-terminal The name PDZ is derived from the first three proteins in which these domains were found: PSD-95 (a 95 kDa protein involved in signaling in the post-synaptic density), Dlg (the Drosophila discs large protein), and ZO1 (the zonula occludens 1 protein involved in maintaining epithelial cell polarity). PDZ motif, and mainly signal via EphBs [73] (Fig. 11.3a). However, reports show that members of the A and B classes interact with each other, as exemplified by EphB2 and EphrinA5 binding [75].

Eph/Ephrin signaling in neurons is extensively investigated and is involved in many contexts throughout the development of the CNS. Based on their structural characteristics, Eph/Ephrin signaling is considered a short-range neuronal wiring signal. Eph/Ephrin signaling controls the formation of topographic maps [76, 77] and, in particular, is essential in pre- and postsynaptic development, as well as synaptic plasticity [78, 79]. The molecular mechanisms through which Eph/Ephrin signaling takes place are very sophisticated and include special features such as receptor clustering and the ability to signal bidirectionally and autonomously [80]. The high degree of possible receptor/ligand combinations, as well as interactions with co-receptors, increases the signal diversity of Eph receptors and ephrins .

In the field of angiogenesis, EphA2 and EphrinA1 are among the best-studied members of the Eph/EphrinA classes, even though almost all EphAs are expressed in ECs (for example, in human brain microvascular ECs) [81]. EphrinA1 is expressed at sites of developmental angiogenesis , and soluble EphA2-Fc inhibits EC migration, sprouting, survival, and corneal angiogenesis induced by VEGF [82, 83]. EphA2 knockout mice initially showed a normal vascular development [84]; however, more detailed studies have described that EphA2-deficient mice possess fewer and enlarged lung capillaries and increased numbers of endothelial sprouts, together with less pericyte coverage [85]. In addition, EphA2-deficient mice hyperreact to bacterial infection of the lung and ovalbumin sensitization by enhanced neoangiogenesis, immune cellular infiltration, and cytokine release [85]. Taken together, these investigations point to a functional role of EphA2 and EphrinA1 signaling in maintaining vascular integrity by stabilizing intracellular connection not only between ECs but also endothelial and mural cells.

EphB4 and EphrinB2 are the best-studied B-class members in angiogenesis, particularly because of their distinctive expression pattern, with higher expression of EphrinB2 in arterial ECs and EphB4 in the venous endothelium [86–88] (Fig. 11.3b). This specific expression pattern, together with the fact that EphB4/EphrinB2 function as a receptor-ligand pair able to signal bidirectionally, outlines the possibility of controlling cellular segregation and boundary formation as functional features of B-class signaling in ECs [89, 90]. An example of this is found in the zebrafish model organism, where the dorsal aorta and caudal vein derive from the same primordial structure that subsequently segregates into arterial and vein ECs. Angioblast cells migrate ventrally from this primary vessel to form the caudal vein in an EphB4/EphrinB2-dependent manner, whereas the remaining cells later form the dorsal aorta [91]. In this regard, a loss-of-function study in mice specifically targeting endothelial EphB4 pointed out that the dorsal aorta of the embryos was enlarged, whereas the formation of cardinal veins was negatively disturbed [90]. More importantly, it was observed that venous ECs were misallocated in arteries [90]. These findings highlight the prominent role of EphB4/EphrinB2 signaling for intercellular communication and tissue boundary formation (Fig. 11.3b).

The generation and evaluation of different EphrinB2 transgenic mouse lines demonstrated the essential role of EphrinB2 for proper vascular development. EphrinB2 knockout mice, or mice with ubiquitous overexpression of EphrinB2, are embryonic or neonatal lethal, respectively, due to severe vascular defects [92, 93]. The lethal phenotype of the ubiquitous EphrinB2 deletion could be rescued by crossing this line with an endothelial EphrinB2 overexpressing mouse line [92, 93], indicating that EphrinB2 expression in the endothelium is essential for survival. Analysis of a mouse line where EphrinB2 overexpression in the endothelium is temporally controlled revealed that overexpression of EphrinB2 during early development also leads to severe vascular defects and embryonic lethality. The severity of the phenotype decreases when EphrinB2 overexpression is induced in the more advanced developmental stages [88]. Consistent with these findings, the endothelial-specific knockout of EphrinB2 under the Cdh5 (vascular endothelial [VE]-cadherin) promoter showed a less-developed vascular system [88]. Therefore, these loss-of-function and gain-of-function studies are evidence that EphrinB2 is essential for vascular development in a specific temporal frame and proangiogenic manner. Interestingly, transgenic mice lacking the cytoplasmic domain of EphrinB2 also die during embryonic development due to defects in vasculogenesis and angiogenesis (similar to the EphrinB2 null mutants) [94], thus showing the EphrinB2 intracellular domain and reverse signaling is important for vascular development . The requirement of the EphrinB2 intracellular domain for developmental angiogenesis was also confirmed with a transgenic mouse line containing deletion of a single valine residue in the cytoplasmic domain (EphrinB2ΔV), which disrupts PDZ-dependent reverse signaling. EphrinB2ΔV showed impaired retina and brain vascularization with reduced vessel branching and vascular sprouts [95]. Similarly, in vitro experiments showed that EphrinB2 is a potent regulator of EC behavior by regulating the actin cytoskeleton through Rho-family small GTPases [96]. In addition, in vitro analyses show that the membrane-bound extracellular EphB4 domain is sufficient to induce EphrinB2-dependent EC invasion, proliferation, and overall survival [97] (Fig. 11.3c). While these studies clearly highlight the importance of EphrinB2 reverse signaling in developmental angiogenesis , the function of EphB4 forward signaling (upon EphrinB2 binding) in this process still remains unknown. Nevertheless, ablation of the EphB4 gene leads to embryonic lethality before E11.5 [98].

EphrinB2 is upregulated in ECs in response to angiogenic factors such as VEGF, fibroblast growth factor (FGF), hepatocyte growth factor (HGF), or activin receptor-like kinase-1 (Alk1) [a member of the transforming growth factor (TGF)-β family], whereas angiopoietin-1 (Ang1) attenuates its expression levels [87, 99]. In vivo and in vitro experiments show that EphrinB2 controls developmental angiogenesis and lymphangiogenesis by controlling EC sprouting activity rather than EC proliferation [88, 100]. EphrinB2 co-localizes and co-immunoprecipitates with VEGFR2 and VEGFR3. EphrinB2 reverse signaling (activated by EphB4) plays a crucial role in VEGFR2 and VEGFR3 receptor internalization, phosphorylation, and downstream signaling (Rac1, Akt, and Erk1/2) after VEGF stimulation [88, 95] (Fig. 11.3c). A recent study showed that a protein complex containing EphrinB2, the clathrin-associated sorting protein Dab2, and the cell polarity regulator Par3 associates with VEGFR2 in the clathrin-coated vesicle and mediates VEGFR2 trafficking towards the early endosome [101] (Fig. 11.3c, d). Atypical PKC (aPKC) acts as a negative regulator of this process by phosphorylating Dab2 and reducing VEGFR2 endocytosis [101] (Fig. 11.3d). Interestingly, sprouting ECs at the leading front of the mouse retina have lower levels of activated aPKC and higher rates of VEGFR endocytosis and turnover compared with nonsprouting ECs (Fig. 11.3d). Consistent with this model, EC-specific knockout mice for Dab2, Par3, and aPKC show vascular phenotypes that resemble those of EphrinB2 mutants [101].

Expression of EphrinB2 does not only localize to ECs but also to pericytes or SMCs [102, 103] (Fig. 11.3). EphrinB2 deficiency in the mural cell compartment led to embryonic vascular malformation and perinatal lethality [89, 103]. Analysis of the vasculature of these mice showed disrupted vessels and defects in the spatial organization of pericytes and SMCs, as revealed by microvessels showing scattered coverage with mural cells [102, 103]. In vitro, EphrinB2 expression in SMCs participated in SMC adhesion, migration, polarization, and formation of focal adhesions via a molecular mechanism involving Crk and p130(CAS) [89, 103].

In summary, EphrinB2 and EphB4 signals are essential for cellular segregation and vascular development, affecting ECs as well as mural cells in a proangiogenic fashion. Moreover, their interaction with VEGFRs, as well as their essential contribution to receptor turnover and signal transduction , makes them an integrative unit for different pathways in the concept of a neurovascular link.

Several studies report a role for EphrinB2 and EphB4 in tumor angiogenesis. In malignant brain tumors, EphrinB2 and EphB4 are expressed in both ECs and tumor cells [104]. EphB4 expressed in tumor cells leads to enhanced tumor angiogenesis via EphrinB2 reverse signaling in the endothelium [97] (Table. 11.1). Consistently, EphrinB2 functional blocking antibodies cause a reduction in angiogenesis and lymphangiogenesis in xenografted mice and a concomitant reduction in tumor growth [105]. Confirmation of the role of EphrinB2 reverse signaling in tumor angiogenesis was demonstrated using an orthotopic glioma tumor model and a skin heterotopic tumor model in the already-mentioned EprhinB2ΔV mice [95]. In these tumor models, angiogenesis and tumor growth was severely reduced compared with tumors in control mice. Moreover, similar to the studies using an EphrinB2 functional blocking antibody, tumor blood vessels in EprhinB2ΔV mice were devoid of sprouts and filopodia [95, 105]. EphrinB2 reverse signaling in tumor vessels was also shown to promote enlargement of blood vessels, the interaction between ECs and mural cells, and reduction of vessel permeability, leading to more stable tumor vessels [97, 104]. A soluble monomeric extracellular domain of EphB4 was shown to act as an antagonist of EphB4/EphrinB2 signaling and inhibited angiogenesis in vitro and in vivo as well as tumor growth in xenograft tumor models [100, 106].

With respect to other Ephrins and Eph members, few studies have involved EphA2 and its ligand EphrinA1 in tumor angiogenesis. EphrinA1 and EphA2 are expressed in tumor cells and tumor ECs [107]. Tumors grown in EphA2-deficient mice are smaller and less vascularized compared with tumors in wild-type mice [108, 109]. In addition, injection of soluble EphA2 and EphA3-Fc into tumor-bearing mice results in reduced tumor growth and reduced tumor angiogenesis [110, 111].

Semaphorins (Sema) are membrane-bound, soluble signaling molecules that are able to signal long range via diffusion gradients, as well as short range via cell-to-cell contact. Dependent on their molecular structure, semaphorins are grouped into eight subclasses, with members of classes 3–7 expressed in vertebrates [112] (Fig. 11.4a). Semaphorins were first described in the nervous system where they function mainly as chemorepellents during axon guidance and neuronal wiring. However, similar to netrins and slits , there are reports highlighting a bifunctional role, whereby semaphorins can either repel or attract axonal growth cones, and also harbor both capacities [113]. Semaphorin signaling often induces cytoskeletal remodeling, affecting cellular properties such as shape, migration, and cellular connectivity [114]. The main receptors of semaphorins are plexins and neuropilins (Nrps). The nine members of the plexin family are divided into four subclasses (A, B, C, and D), whereas the group of neuropilins has two representative receptors, Nrp1 and Nrp2 [115] (Fig. 11.4a). Plexins are able to directly interact with membrane-bound semaphorins , whereas soluble semaphorin members (class 3 semaphorins only) often bind to neuropilins and signal through a heterodimer complex of neuropilin and plexin (with the exception of Sema3E, which binds directly to PlexinD1 and can induce signaling independent of Nrp1) [115]. The fact that neuropilins are known to be a co-receptor for other tyrosine kinase receptors (such as VEGFRs) makes semaphorin signaling an ideal pathway for studying their neurovascular properties [116].

Class 3 semaphorins are the best-studied semaphorins in the vascular system. Therefore, in this section we will focus on describing the fundamental aspects and functions of this group of molecules and their receptors in blood vessels. The emerging role of class 4 and 6 semaphorins in the vascular system will be briefly mentioned.

Among the different plexin receptors, PlexinD1 is highly expressed in ECs of developing blood vessels [117, 118]. Gene targeting studies in zebrafish show the importance of this receptor in vascular development . In zebrafish, a PlexinD1 loss-of-function mutation results in increased intersomitic vessel branching [119, 120]. Sema3A orthologs (Sema3a1 and Sema3a2) are expressed in the adjacent developing somites and restrict the growth of angiogenic sprouts to the intersomitic boundaries via PlexinD1 binding [118, 119]. In vitro studies show that Sema3A stimulation of HUVECs results in loss of actin stress fibers and inhibition of VEGF-induced cell migration, thus suggesting that Sema3A, expressed in the somites, repels PlexinD1-expressing intersomitic vessels and therefore guides vessels along the somite boundaries [118]. Further studies reporting on the molecular mechanisms of Sema/PlexinD1 signaling in intersomitic vessels show that Sema/PlexinD1 signaling antagonizes VEGF proangiogenic activity by controlling the levels of soluble VEGFR1 (sFlt1) expressed by ECs [119].

In a similar manner, PlexinD1 knockout mice present blood vessel patterning defects and die soon after birth [121]. Endothelial-specific PlexinD1 knockout mice (using the Tie2-Cre mouse line) consistently present defective development of the vasculature, heart, and skeleton [122]; however, Sema3A knockout mice show none or mild vascular defects, highly depending on genetic background [123, 124]. Sema3E in mice, in contrast to zebrafish , seems to be the required ligand for PlexinD1 signaling in the vascular system (independently of Nrp1). Sema3E is expressed in a caudal-rostral gradient in the developing somites, and Sema3E null mouse embryos present a similar phenotype as PlexinD1 knockout mice; however, they are viable throughout adulthood [121]. Interestingly, it was recently shown that despite severe initial defects in the dorsal aorta of Sema3E null mice (which would lead to lethality), these defects are resolved during development via morphogenetic cellular rearrangements whereby abnormal vessels remodel into normal vessels instructed by guidance cues from the lateral plate mesoderm [125]. Sema3E/PlexinD1 signaling in the vasculature has also been investigated in the developing mouse retina, where astrocyte-derived VEGF induces PlexinD1 expression in ECs of the angiogenic front (Fig. 11.4b). Even though Sema3E expression is evenly distributed in the retinal ganglion cell layer, local PlexinD1 expression at the angiogenic front restricts Sema3E signaling to the sprouting front of the retina vasculature [126, 127] (Fig. 11.4bb). Further investigation into the molecular mechanisms revealed that Sema3E/PlexinD1 signaling negatively regulates Dll4 expression in the retinal vasculature and that, in turn, Sema3E/PlexinD1 signaling is required for a balanced ratio of tip and stalk cells [127] (Fig. 11.4c).