PlGF and VEGF-B bind and activate the VEGFR-1 receptor but do not bind to VEGFR-2 or to VEGFR-3.74 PlGF was reported to potentiate VEGF signaling through activation of VEGFR-1/VEGFR-2 heterodimers.75 Although PlGF does not seem to affect vasculogenesis and developmental angiogenesis, it does play an important role in pathological angiogenesis, possibly through the recruitment of monocytes.61,76 VEGF-B was found to bind to np1.5 while the heparin binding form of PlGF, PlGF-2, binds to both neuropilins.3,4 The biological function of this binding ability is still unclear. It was recently claimed, in addition, that np1 contains a domain that mimics the structure of the glycosaminoglycan heparin, thus allowing np1 to bind to a wide range of heparin binding growth factors. It was even claimed that neuropilins participate in signal transduction of growth factors such as basic fibroblast growth factor.77 However, this observation seems to contradict previous observations which indicate that heparin enhances the binding of VEGF165 to neuropilins.1

6. Signal Transduction by Neuropilins

The intracellular domain of the neuropilins is short, and it was therefore assumed that it does not suffice to for the transduction of biological signals. This view was supported by experiments that have shown that although np1 is required for s3a-induced collapse of axonal growth cones, deletion of the cytoplasmic domain of np1 does not inhibit s3a activity, suggesting the existence of independent signal transducing moieties.52 These were later found to be the products of genes encoding type-A plexins.12,13 (Fig. 1). Type-A plexins were found to form complexes with neuropilins, and to serve as the signal transducing components in plexin/neuropilin holo-receptors for various class-3 semaphorins.12,13,33,78 Plexins belonging to the other three plexin subfamilies may also be able to form functional complexes with neuropilins, as demonstrated in the case of plexin-B1 and np1,12,79 and recently also in the case of plexin-D1 and np1.80,81 However, the assumption which predicts that neuropilins cannot transduce biological signals on their own due to their short intracellular domains was recently challenged in experiments in which the extracellular domain of the epidermal growth

12 G. Neufeld et al.

factor (EGF) receptor was fused to the intracellular and transmembrane domains of np1. These experiments indicate that the chimeric receptor can promote cell migration in response to EGF, suggesting that the intracellular domain of np1 does transduce biological signals independently,82 perhaps via binding to the PSD-95/Dlg/ZO-1 domain of the NIP protein.51 These experiments therefore indicate that the short intracellular domain of the neuropilins is not devoid of function.

Nevertheless, it is now widely accepted that to transduce signals of class-3 semaphorins such as s3a, s3f and s3c, neuropilins form complexes with either A-type plexins12,13 or with plexin-D1.81 Activation of plexins by semaphorins, either directly or indirectly via neuropilins, leads to diverse biological responses. One of the best studied responses is the s3aand s3f-induced repulsion of axonal growth cones. The repulsion is apparently triggered by local changes in cell adhesion and actin cytoskeleton organization. Activation of A-type plexins occurs in response to the binding of a class-3 semaphorin to neuropilin, leading to the phosphorylation of tyrosine residues in the cytoplasmic domain, activation of small G proteins and subsequent effects on cell adhesion, cell shape cell migration and cell proliferation. The phosphorylation of plexins is the result of semaphorin-induced recruitment of cytosolic tyrosine kinases. The binding of s3a to np1 induces the association of the tyrosine-kinase Fes/Fps with plexin-A1 leading to plexin-A1 phosphorylation. In growth cones of s3a responsive nerve cells, Fes/Fps forms complexes with brain-specific collapsin response mediator protein-2 (CRMP-2) and with CRMP associated molecule (CRAM). These two proteins are required for s3a signaling to the actin cytoskeleton in responsive nerve cells and are also phosphorylated by Fes/Fps in response to s3a, although their exact role is still unclear.35 Another cytosolic tyrosine-kinase that was found to associate with the intracellular part of plexin-A1 as well as plexin-A2 is fyn. Fyn phosphorylates plexin-A2 in response to s3a and binds cdk5 kinase, which is also phosphorylated by Fyn. Activation of cdk5 by fyn was found to be essential for s3a-mediated growth cone repulsion.36

Another signal transduction mechanism that is activated by class- A plexins involves activation of MICALs (Molecule interacting with CasL). It was found that plexin-A, a Drosophila class-A plexin receptor,

associates with MICAL in response to the Drosophila semaphorin s1a. The mammalian homologue, MICAL-1, interacts with intermediate filaments, as well as with actin and small GTPases like Rab1. It also functions as a putative flavoprotein monooxygenase, and its oxidative activity may be important for semaphorin signaling since monooxygenase inhibitors inhibit MICAL-mediated s3a signal transduction.39,83,84 Another protein, Nervy, a member of the myeloid translocation gene family of A-kinase anchoring proteins (AKAPs), regulates repulsive axon guidance in Drosophila by linking the cyclic adenosine monophosphate (cAMP)-dependent protein kinase (PKA) to the Semaphorin-1a (Sema-1a) receptor Plexin-A (PlexA), and nervey homologues may fulfill similar roles in vertebrates.85

Plexins can also affect the organization of the actin cytoskeleton by modulating the activity of small GTPases such as RhoA, Rnd1, Rac1 and CDC42, all of which have been shown to control the organization of the actin cytoskeleton.86,87 Activation of Rac and CDC42 usually triggers the formation of lamellipodia and filopodia, respectively. In contrast, activation of the Rho family members Rho1 and Rnd1 leads to the formation of stress fibers. S3a induces activation of Rac1 but the mechanism by which it does so is still unclear. This is probably triggered by the binding of Rac1 to plexin-A1 since the interaction is required for the collapsing activity of s3a,88 although it is unclear why activation of Rac1, usually associated with formation of filopodia and lamellipodia would be associated with the collapse observed in response to s3a. Out of the Rho family GTPases tested only RhoD and Rnd1 were found to bind directly to plexin-A1. Interestingly, Rnd1 binding to plexin-A1 leads to a collapse of the actin cytoskeleton even in the absence of s3a. RhoD on the other hand antagonizes the effects of Rnd1 even though both GTPases belong to the Rho family of GTPases.89 Various guanine nucleotide exchange factors (GEFs) and GTPase activating proteins (GAPs), which function as regulators of GTPases, also modulate the activity of GTPases that bind to plexins.43,44 The GTPases in turn regulate the activity of downstream effectors such as the LIM kinase which in turn regulates the phosphorylation state of cofilin, an actin binding/cleaving protein that is required for s3a-induced growth cone collapse.90 Similar mechanisms are presumably activated in endothelial

14 G. Neufeld et al.

cells in response to semaphorins such as s3a and s3f. Indeed, s3f and s3a repulse endothelial cells and inhibit angiogenesis, presumably as a result of their effects on the actin cytoskeleton.91−94

7.The Role of the Neuropilins in the Regulation of Vasculogenesis and Angiogenesis

Neuropilins as modulators of VEGF function: Since both neuropilins function as splice form Specific VEGF receptors, it was not surprising that they were found to affect VEGF signaling and function in various experimental systems. Initially, the binding of VEGF165 to np1 was found to enhance VEGF165-induced migration of endothelial cells in cells that express in addition to np1 the VEGF receptor VEGFR-2.2 It was subsequently observed that soluble dimers of the np1 extracellular domain enhance VEGF-induced vascular development while monomers of the soluble extracellular domain function as VEGF165 traps and inhibit VEGF-induced vascular development.95 The role of np1 in embryonic vascular development was also studied in gene targeting experiments. Mice lacking functional np1 receptors suffer from impaired neural vascularization and from defects in the development of large arteries such as branchial arch arteries. In addition, the development of the heart was strongly impaired in these mice, and failure of heart function was responsible for their premature death.96 Binding/competition experiments have demonstrated that the VEGF165 and s3a binding domains of np1 overlap.49 Knock-in mice expressing an np1 variant lacking s3a binding ability but retaining VEGF binding displayed normal vascular development but abnormal neural development indicating that the VEGF binding ability of np1 is critical for proper vascular development. In contrast, the s3a binding ability is required in addition to the VEGF binding ability for proper heart development. These results are strengthened by experiments showing that mice in which np1 was targeted in endothelial cells but not in other cell types also suffer from severe vascular abnormalities,97 and by experiments which show that proper development of the vasculature in zebrafish requires np1.98

The mechanism by which np1 enhances VEGF165-induced signal transduction via the VEGF receptor VEGFR-2 is unclear. It was

suggested that np1 binds VEGF165 and presents it to the VEGFR-2 receptor, thereby increasing responsiveness to VEGF165. Such a mechanism should function in trans too, and it was indeed found that angiogenesis is enhanced in tumors containing tumor cells expressing high levels of np1.99 It was recently suggested that np1 contains a heparin-like binding domain that enables np1 to bind a wide variety of heparin binding growth factors such as basic fibroblast growth factor and not just VEGF165, and that as a consequence np1 is able to potentiate the activity of a wide variety of heparin binding growth factors.77 On the other hand, it was also reported that np1 forms complexes with VEGFR-2 directly.49,100,101 The formation of such complexes may account, at least partially, for the np1-dependent potentiation of VEGF165 activity.

The role of np2 in VEGF-induced vasculogenesis and angiogenesis is less clear. Np2 binds VEGF165 with a somewhat lower affinity than that of np1. However, the vasculature of mice lacking a functional np2 receptor develops normally except for defects observed at birth in small lymphatic vessels.73,102 This does not mean that such mice are normal with respect to their responses to VEGF. Indeed, it was recently reported that mice lacking a functional np2 gene do not respond to VEGF165 by retinal angiogenesis.103 The importance of np2 to vascular development is highlighted in experiments in which mice lacking both functional neuropilins were generated. These mice display a total lack of endothelial cells,104,105 and their phenotype therefore resembles the phenotype of mice lacking functional VEGFR-2 receptors. Furthermore, mice lacking a functional np2 gene and containing only one functional np1 gene also displayed vascular abnormalities that were more severe than those observed in mice that lack both np1 alleles.104 These experiments therefore indicate that np2 does have an important role in vasculogenesis and developmental angiogenesis as well as in the development of the lymphatic system.

Neuropilins as transducers of semaphorin signals in angiogenesis: Neuropilins bind, in addition to VEGF, several class-3 semaphorins. S3a binds to np1 but not to np2.9,10 It was therefore natural to assume that s3a may function as an inhibitor of angiogenesis by interfering with np1mediated VEGF signaling. Indeed, it was found that s3a inhibits the pro-angiogenic effects of VEGF165 in in vitro angiogenesis experiments, and that it inhibits VEGF165 binding to np1.106 Subsequent experiments

16 G. Neufeld et al.

demonstrated that s3a can inhibit developmental angiogenesis in chick embryo forelimbs94 and vascular branching in the developing chick brain.93 However, no effects of s3a on tumor progression or tumor angiogenesis have been reported to date.

Even though VEGF binds efficiently to np2,3 it does not inhibit the binding of the np2-specific s3f to np2.50 Nevertheless, in vitro experiments have shown that s3f can inhibit both VEGF-and bFGF-induced proliferation of endothelial cells, and that s3f is also able to inhibit bFGFand VEGF165-induced angiogenesis.91 S3f also repels endothelial cells in in vitro experiments indicating that it could also affect angiogenesis through repulsion of migrating endothelial cells.92 Both S3f and s3b were identified as tumor suppressor genes whose loss is associated with tumor progression of small cell lung carcinoma.25,26,107 It is therefore not surprising that both s3f and s3b inhibit tumor formation from small cell lung carcinoma-derived cells.107,108 Interestingly, it was recently shown that s3f also affects the behavior of tumor cells directly, via np1 and np2 receptors expressed on the tumor cells. S3f inhibits the VEGF-induced spreading, of MCF-7 breast cancer cells by inhibiting np1 mediated signaling,109 and repulses C100 breast cancer cells as a result of its binding to np2 receptors expressed by this cell type.110 Similarly, s3b was found to antagonize the anti-apoptotic effects of VEGF in NCI-H1299 lung cancer-derived cells, probably by interfering with neuropilin-mediated VEGF signaling in these cells.111 These findings indicate that s3f can suppress tumor angiogenesis and that s3b and s3f can directly affect the behavior of tumor cells expressing neuropilins.

S3c is a semaphorin that seems to signal through np2 or through np1/np2 complexes. Like s3f, it can bind to np1 and as such can act as an antagonist of s3a.112 Recently, it was found that in the presence of plexin-D1, s3c can induce signal transduction via np1 as well as via np2.81 The heart of mice lacking a functional s3c gene does not develop normally, and mice lacking a functional plexin-D1 gene suffer from heart defects and vascular patterning defects.24,80,81 It follows that plexin-D1-mediated s3c signaling plays an important role in vascular development, although it is unclear whether these effects are the result of inhibition of endothelial cell migration and proliferation.