Vertebrates, including humans, have a closed circulatory system for supplying oxygen and nutrients to various tissues in the body, and for removing CO₂ and waste materials from peripheral tissues (Risau, 1997). In addition to the blood circulatory system, vertebrates have a similar tubular system, the lymph vessel system, which is essential for the absorption and delivery of fluids, lipids and immune cells from peripheral tissues to lymph nodes and circulating blood.

The circulatory system plays a crucial role in the etiology of many diseases in humans. The hypothesis that suppressing tumor angiogenesis was a potentially novel anti-cancer strategy was first suggested by J. Folkman in 1970 (for review, Hanahan and Folkman, 1996), although the molecular basis of the regulation of angiogenesis was not clearly characterized prior the late 1980s.

Around 1990^th, the genes encoding vascular endothelial growth factor (VEGF) and its receptor (VEGFR) were isolated and characterized. Based on extensive studies of these molecules, signals mediated by members of the VEGF and VEGFR families were shown to play central roles in angiogenesis and lymphangiogenesis (Leung et al., 1989; Shibuya et al., 1990; Alitalo and Carmeliet, 2002; Ferrara, 2004; Shibuya and Claesson-Welsh, 2006; Shibuya, 2011).

In parallel to the VEGF-VEGFR axis, other regulatory systems, including angiopoietin (Ang)-Tie, Delta-Notch and Ephrin-Eph, have been shown to play a role in angiogenesis (Suri et al., 1996; Wang et al., 1998; Noguera-Troise et al., 2006). Furthermore, studies have demonstrated the existence of a variety of endogenous anti-angiogenic factors, such as thrombospondin-1 (TSP-1), as well as factors involved in negative feedback loops that act to suppress angiogenesis (Watnick et al., 2003).

Based on the knowledge that VEGF signals are key players in tumor angiogenesis, a variety of antibodies and kinase inhibitors which suppress VEGF-VEGFR signaling have been developed that are now widely used in the treatment of cancer. The clinical efficacy of such therapeutics is limited however, and many future studies will focus on improving their performance in the clinic (Hurwitz et al., 2004).

Recently, a soluble form of Flt-1/VEGFR-1, sFlt-1, has been shown to be associated with preeclampsia (PE) (Maynard et al., 2003; Koga et al., 2003; Levine et al., 2004; Young et al., 2010), aged macular degeneration (AMD) (Luo et al., 2013) and nephrotic syndrome (Jin et al., 2012). Furthermore, VEGF signals have been reported to directly regulate neuronal function and survival under certain conditions (Oosthuyse et al., 2001). These results are discussed in this review.

In 1989, two research groups independently isolated a cDNA, one encoding for vascular permeability factor (VPF) and another for VEGF, which proved to be identical, and encoded a single protein, now widely known as VEGF (for review, Dvorak, 2002; Ferrara, 2004). The human genome contains five genes encoding five distinct VEGF family members, namely VEGFA (also called VEGF), placenta growth factor (PlGF), VEGFB, VEGF-C, and VEGF-D (Fig. 1). Structurally, VEGF family proteins are homodimeric, with two subunits of about 120 to 200 amino acids in length. Given their overall structural resemblance to members of the platelet-derived growth factor (PDGF)/macrophage colony-stimulating factor (M-CSF)/stem cell factor (SCF) ligand family, the VEGF and PDGF families are considered to constitute a supergene family, the VEGFPDGF superfamily. With regard to interactions between VEGF ligands and their receptors, VEGF-A, which contains subtypes such as VEGF-121, -165 and -189, binds VEGFR-1/Flt-1 and VEGFR-2 (De Vries et al., 1992), whereas PlGF and VEGF-B bind only VEGFR-1/Flt-1. VEGF-C and VEGF-D bind tightly to VEGFR-3 and more weakly to VEGFR-2. Interestingly, a possibly suppressive ligand, VEGF(xxx)b, has recently been reported (Pritchard-Jones et al., 2007). Although it does not exist in mammalian genome, another VEGF family member, VEGFE encoded in the Orf-viral genome, has been shown to be a VEGFR-2 specific ligand (Shibuya and Claesson-Welsh, 2006).

VEGF-A (+/−) mice, in which a single allele of VEGF-A has been deleted, exhibit embryonic lethality due to immature angiogenesis and cardiovascular insufficiency. Lethality resulting from loss of a single allele of a gene is rare in mammals, and the phenotype of these mice indicates a strict relationship between VEGF dosage and angiogenic homeostasis (Carmellet et al., 1996; Ferrara et al., 1996). Expression of VEGF-A is known to be upregulated under hypoxic conditions, as well as by growth factor signaling, and by hormones such as estrogen (Ferrara, 2004).

In contrast to VEGF-A, PlGF and VEGF-B appear to have a relatively minor role in the regulation of angiogenesis, and have been shown to play a role in cardiac muscle function (Bellomo et al., 2000; Bry et al., 2010). VEGF-C and VEGF-D are initially synthesized as precursor forms that subsequently undergo post-translational processing, and are involved in the regulation of lymphangiogenesis (Alitalo and Carmeliet, 2002). Moreover, they have been demonstrated to be involved in angiogenesis at early stage of embryogenesis, approximately E10.5, as well as in tumor angiogenesis (Dumont et al., 1998; Tammela et al., 2008).

In 1990, our group isolated a cDNA from human placenta that encoded a novel receptor-type tyrosine kinase containing seven extracellular immunoglobulin (Ig)-domains and a kinase insert sequence of approximately 70 amino acids in length. Based on its structural similarity to the Fms receptor, we named it Fms-like tyrosine kinase-1 (Flt-1) (Shibuya et al., 1990). In addition to full length Flt-1, we found that human placenta also expressed a truncated mRNA encoding a protein named sFlt-1, which contained only the ligand-binding region of Flt-1 (Shibuya et al., 1990; Kendall and Thomas, 1993) (Fig. 1). Based on a subsequent study demonstrating its high affinity for VEGF (De Vries et al., 1992), Flt-1 is also referred to as VEGFR-1. Soon after the isolation of Flt-1, VEGFR-2/KDR (Flk-1 in mouse) and VEGFR-3/Flt-4 were isolated, indicating the existence of three distinct VEGFR genes in the mammalian genome.

VEGF-A binds to VEGFR-1/Flt-1 with high affinity (Kd=1–10 pM) and less strongly to VEGFR-2 (Kd=10–100 pM), although the tyrosine kinase (TK) activity of VEGFR-1/Flt-1 is about 10 fold weaker than VEGFR-2 (Keyt et al., 1996; Sawano et al., 1996). Along with the results of gene knockout studies, these data indicate that the major signal transducer in angiogenesis is VEGFR-2 (Shalaby et al., 1995). The VEGF-VEGFR-2 axis regulates angiogenesis in a number of different physiological contexts, including hormone-dependent angiogenesis (Ferrara, 2004; Kim et al., 2013).

Members of the PDGFR/Fms (M-CSF receptor)/Kit family are distantly related to VEGFRs, and are known to signal through PI3K activation domains with tyrosine (Y)-containing motifs such as Y-x-x-methionine (M) and Y-M-x-M, at the TK-insert domain. After autophosphorylation of PDGFR family members, the p85 subunit of PI3K binds the Y-x-x-M and YM-x-M motifs via SH2 domain, resulting in activation of downstream PI3K-Akt and Ras pathways and strong signaling for cell proliferation (Heldin and Westermark, 1999). Very interestingly however, VEGFRs, including VEGFR-2, lack Y-M-x-M and Y-x-x-M motifs in the TK-insert and other regions. In contrast, we found that VEGFR-2 Y1175, a major autophosphorylation site of this receptor, binds the SH2 domain of PLCγ, activating the PLCγ-C kinase-Raf-MEK-MAP kinase pathway to mediate endothelial cell proliferation (Takahashi et al., 1999; Takahashi et al., 2001). Similar to flk-1 (VEGFR-2)−/− mice, VEGFR-2 Y1173F homozygous knock-in mice, in which this tyrosine residue is mutated to phenylalanine (F) (the mouse VEGFR2 protein is two amino acid shorter than the human form), exhibit embryonic lethality due to poor vasculogenesis and angiogenesis (Sakurai et al., 2005). Furthermore, Sase et al. (2009) reported that, using an in vitro ES cell differentiation system into vascular endothelial cells, Y1175F-mutant VEGFR-2 fails to induce endothelial differentiation. Collectively, these reports indicate that VEGF-VEGFR-2 mediated signal for vasculogenesis and angiogenesis is highly dependent on the Phospho (P)Y1175-PLCγ-C kinase pathway.

VEGFR-2 Y1175 has also been shown to be involved in von Willebrand factor release from endothelial cells (Xiong et al., 2009), while VEGFR-2 Y951 is important for vascular permeability and cell migration (Matsumoto et al., 2005). In addition, the VEGFR-2-Hdac6-Hsp90-Bcl2 pathway has been reported to transduce cell survival signals (Dias et al., 2002).

While most TK family kinases engage the Ras activation pathway to stimulate cell proliferation, this pathway appears to be relatively unimportant in VEGFR-2 signaling. The dependence of VEGFR-2 on PLCγ-C kinase in signaling for the MAP kinase activation is therefore unique among TK family members.

The biological characteristics of VEGFR-1/Flt-1 are distinct from those of VEGFR-2. Fong et al. (1995) reported that flt-1 −/− mice are embryonic lethal due to overgrowth of endothelial cells and dysfunction of blood vessels. These results strongly suggest that VEGFR-1/Flt-1 has a negative role in angiogenesis at an early stage of embryogenesis, possibly by maintaining an appropriate level of activation of VEGFR-2 via partial suppression of VEGF. To clarify whether the VEGF-trapping with the binding domain of VEGFR-1/Flt-1 or the TK-dependent negative signaling is crucial for this biological role of VEGFR-1/Flt-1 in embryogenesis, we generated Flt-1 TK-deficient (flt-1 TK−/−) mice. To our surprise, flt-1 TK−/− mice were viable and showed basically normal blood vessel formation (Hiratsuka et al., 1998). These mice, however, exhibited a deficiency in VEGF-dependent migration of macrophages which is in agreement with expression of VEGFR-1/Flt-1 in macrophages and its role in VEGF-dependent macrophage migration (Barleon et al., 1996; Clauss et al., 1996).

Since the flt-1 TK−/− mice lack only signals mediated by VEGFR-1/Flt-1, they are useful for elucidating the importance of VEGFR-1 signals under physiological conditions (Niida et al., 2005), as well as in diseases such as cancer. Studies using these mice by our own group and others have demonstrated that VEGFR-1 TK stimulates angiogenesis in various carcinomas and glioblastomas (Kerber et al., 2008; Muramatsu et al., 2010; Schwartz et al., 2010; Laurent et al., 2011), tumor metastasis (Hiratsuka et al., 2002; Kaplan et al., 2005), inflammatory disease similar to rheumatoid arthritis (Murakami et al., 2006), stroke (Beck et al., 2010), as well as liver repair (Kato et al., 2011) and gastric ulcer healing (Sato et al., 2013). These results indicate that although its TK activity itself is 10-fold weaker than that of VEGFR-2, VEGFR-1/Flt-1 nonetheless represents a potentially important therapeutic target in a variety of diseases, particularly cancer (Shibuya, 2006).

VEGFR-3 is highly expressed in lymphatic endothelial cells, and VEGF-C/D-VEGFR-3 signals have been shown to stimulate lymphangiogenesis and lymph-node metastasis in cancer. In addition, VEGFR-3 has been shown to play a role in tumor angiogenesis (Tammela et al., 2008; Sallinen et al., 2011).

Recently a variety of angiogenesis-regulatory systems have been elucidated, including: the Ang-Tie receptor axis, which mediates stabilization or destabilization of blood vessels; the Ephrin-Eph receptor pathway, which plays a role in arterio-venous differentiation of vessels; the Delta-Notch pathway, which is involved in regulation of vascular morphology and tip cell-stalk cell communication; and the Netrins-UNC5R pathway, which plays a role in guidance during angiogenesis (Suri et al., 1996; Wang et al., 1998; Noguera-Troise et al., 2006; Freitas et al., 2008). Of these, the Delta-Notch receptor system, particularly that involving Dll4, has attracted attention given its crucial role in sprouting and morphogenesis during angiogenesis. Anti-Dll4 neutralizing antibodies were shown to suppress tumor growth in mice, initially suggesting that this might represent an alternative strategy for blocking tumor angiogenesis (Noguera-Troise et al., 2006). However, although Dll4 blockade using these antibodies suppressed tubular formation by endothelial cells, these endothelial cells survived in tissues and gave rise to angiomas (Yan et al., 2010). This finding resulted in the termination or significant suppression of development of anti-tumor drugs based on Dll4 blockade.

In addition to these angiogenesis-regulatory systems, a variety of endogenous angio-suppressive factors have been identified. TSP-1, for example, has been shown to suppress angiogenesis by inducing apoptosis or cell cycle arrest of vascular endothelial cells via upregulation of p21 (Watnick et al., 2003; Yamauchi et al., 2007). Down syndrome patients show lower incidence of cancer. A gene DSCR-1 responsible for anti-cancer was recently identified (Minami et al., 2004). DSCR-1 protein has an anti-angiogenic activity, and an increase in the gene copy number of DSCR-1 gene in Down syndrome patients appear to partly suppress tumor angiogenesis. In addition, VASH1, which is induced in endothelial cells subsequent to activation of VEGF-VEGFR, have been shown to possess anti-angiogenic activity (Sato, 2013). Moreover, angiostatin and endostatin have been characterized as endogenous tumor suppressors in animal models, although it is not clear whether they possess similar roles in humans.

Future studies will be required to clarify which of these factors and signaling pathways are involved in suppressing tumor angiogenesis and which, if any, are suitable for development of therapeutics for clinical use.

Based on the evidence that VEGF-VEGFR signals play central roles in angiogenic processes in a variety of diseases such as cancer, various VEGF signal inhibitors, including anti-VEGF neutralizing antibodies and VEGFR kinase/multi kinase inhibitors, have been successfully developed and now widely used in the clinic (Kim et al., 1993; Hurwitz et al., 2004) (Fig. 2, 3). These drugs are effective in prolongation of PFS (progression-free survival) as well as OS (overall survival) in the treatment of cancer, and have been approved for a variety of solid tumors such as colorectal cancer, lung cancer (nonepithelial, NSCLC), breast cancer, glioblastoma, liver cancer and renal cell carcinoma (Hurwitz et al., 2004; Cohen et al., 2007). The efficacy of anti-VEGF antibody in breast cancer is complex however, in that unlike previous relatively small size phase III studies, recent large scale phase III studies have indicated that a combination of chemotherapy plus anti-VEGF antibody treatment was better than chemotherapy-only treatment with respect to PFS, but not OS. Based on this result, the United States Food and Drug Administration withdrew its approval for the use of anti-VEGF antibody in breast cancer treatment. In contrast however, citing its improvement of PFS in patients, Japan and other countries approved anti-VEGF antibodies for the treatment of breast cancer in 2011. More recently an anti-VEGF antibody has been approved for the treatment of ovarian cancer based on the successful clinical trials.

Given the potential of adverse clinical effects of small molecules kinase inhibitors, novel compounds are currently undergoing clinical trials to identify drugs with fewer side effects.

To date, there have been no successful results in the use of anti-VEGF signaling drugs in the treatment of pancreatic or gastric cancer.

While the clinical efficacy of anti-VEGFs has been demonstrated in solid tumors, it is not clear whether this benefit continues for long periods. OS curves in some phase III trials indicate that over time their efficacy decreases, and the potential for resistance or refractoriness exists in later periods. Refractoriness may be due to several possible mechanisms as follows: (1) other angiogenic factors such as HGF and FGF compensate for the loss of the VEGF proangigenic signal (Pàez-Ribes et al., 2009), (2) vascular endothelial cells in tumor tissue acquire increased resistance to anti VEGF-VEGFR drugs (Hida et al., 2013), or (3) in response to the conditions of hypoxia and low nutrition which result from anti-angiogenic therapy, cancer cells may become more invasive and/or resistant to apoptosis by induction of phospho-Akt and upregulation of histone-demetylases such as JHJM1A (Huveldt et al., 2013; Osawa et al., 2013) (Fig. 4). Extensive studies are required to elucidate the actual mechanism of resistance in this context.

Anti-VEGF signal inhibitors have been shown to be effective in suppressing symptoms of AMD. A recovery of visual acuity was demonstrated in response to intraocular injection of an anti-VEGF neutralizing antibody.

PE, which occurs in 5–7% of pregnancies, causes hypertension and proteinuria in the mother and growth retardation in the fetus, and Cesarean section is often required to save the fetus. sFlt-1 is present at high levels in the serum of PE patients, and the degree of sFlt-1 overexpression has been shown to correlate with the severity of PE (Koga et al., 2003; Maynard et al., 2003) (Fig. 1). Furthermore, Levine et al. (2004) found that elevated sFlt-1 in the serum in asymptomatic women at early stages of pregnancy predisposed to the development of PE in the later stages of pregnancy. In animal models, inoculation of sFlt-1 expression vector into pregnant rats induced hypertension and proteinuria, similar to PE in humans (Maynard et al., 2003).

Since sFlt-1 is derived from the ligand-binding region of VEGFR-1/Flt-1, its major biochemical function is thought to be trapping of VEGF for suppression of VEGF signals (Tanaka et al., 1997; Shibuya, 2006). To our surprise, cancer patients treated with anti-VEGF signal drugs often develop side effects such as hypertension and proteinurea that are similar to the symptoms of PE (Hurwitz et al., 2004). Taken together, these results indicate that the abnormally high levels of sFlt-1 secreted from trophoblasts in placenta contribute to the development of PE, and that strategies targeting this protein have considerable potential in the treatment of this disease (Mezquita et al., 2003; Nagamatsu et al., 2004; Gilbert et al., 2007; Foidart et al., 2009; Kumasawa et al., 2011; Thadhani et al., 2011).

Furthermore, sFlt-1 is expressed in lens epithelial cells as well as pigment epithelial and photoreceptor cells in the eye, where it plays a role in maintaining avascularity in the cornea and in tissues outside of the retina (Ambati et al., 2006; Luo et al., 2013) (Fig. 5). Decreased expression of sFlt-1 in photo-receptor and pigment epithelial cells has been reported in AMD patients, suggesting that an increase in free VEGF may stimulate pathological angiogenesis in the retinas of these individuals.

sFlt-1 is also expressed in podocytes in the kidney, where it binds to the cell membrane lipid to maintain the physiological functions of podocytes and vascular endothelial cells in renal glomeruli (Jin et al., 2012). Interestingly, sFlt-1 does not necessarily sequester VEGF in podocytes, and sFlt-1 localized in lipid rafts in the cell surface of podocytes is sufficient to support physiological secretion of primary urine from glomerular capillaries. Indeed, podocyte-specific knock-out of the flt-1 gene in mice has been shown to induce chronic proteinuria, a condition characteristic of nephrotic syndrome in humans (Jin et al., 2012).

Another exciting advance in the VEGF-VEGFR field in past 12 years is the discovery of an intimate relationship between VEGF signaling and the neuronal system. Although VEGF signaling in neuronal tissues was thought to function directly upon blood vessels rather than neuronal cells, an extensive study clearly showed that this is not the case (Oosthuyse et al., 2001). A hypoxia response element (HRE) sequence in the transcriptional regulatory region of VEGF gene mediates hypoxia-inducible factor (HIF) binding, and is essential for hypoxia-responsive upregulation of VEGF. Deletion of this HRE sequence in mice (VEGFδ/δ mice) resulted in motor neuron degeneration several months after birth, whose pathological alterations in neurons are comparable to amyotrophic lateral sclerosis (ALS) in humans. They also showed that although decreased angiogenesis due to reduced VEGF levels contributes to this phenomenon in part, a direct effect of VEGF on motor neuron also appears to be important. In vitro studies have shown that purified motor neurons express VEGFR-2, and that VEGF signals via VEGFR-2 to stimulate cell survival. Furthermore, treatment of VEGFδ/δ mice with VEGF results in partial suppression of their motor neuron degeneration, strongly suggesting that stimulation of VEGF signaling is an attractive new strategy for the treatment of ALS patients.

Sensory nerve cells in the dorsal root ganglion of mice also express VEGFR-2, and VEGF signals are required to maintain the healthy condition of these cells (Verheyen et al., 2012). Blockade of this signal by drugs such as anti-VEGF neutralizing antibodies results in painful sensory neuropathy, an adverse effect of anti-VEGF signaling therapy. Further studies on this problem are required to improve PFS in cancer patients undergoing anti-VEGF therapy.

On the other hand, olfactory sensory neurons express VEGFR-1, and the VEGF-VEGFR-1 pathway has been shown to be important for physiological function of these neurons (Wittko et al., 2009; Dhondt et al., 2011). Various neuronal diseases should be carefully characterized in the VEGF signal point of view.

VEGF signaling plays a crucial role not only in cancer but also in a variety of other diseases, including neuronal degeneration and nephrotic syndrome. We anticipate that further studies on VEGF signaling and its modulators will herald a new era in which the severe diseases whose etiology is associated with abnormal VEGF signaling can be brought under control.