Snake Venom Vascular Endothelial Growth Factors (VEGFs) Exhibit Potent Activity through Their Specific Recognition of KDR (VEGF Receptor 2)*

Vascular endothelial growth factor (VEGF165) exhibits multiple effects via the activation of two distinct endothelial receptor tyrosine kinases: Flt-1 (fms-like tyrosine kinase-1) and KDR (kinase insert domain-containing receptor). KDR shows strong ligand-dependent tyrosine phosphorylation in comparison with Flt-1 and mainly mediates the mitogenic, angiogenic, and permeability-enhancing effects of VEGF165. Here we show the isolation of two VEGFs from viper venoms and the characterization of their unique biological properties. Snake venom VEGFs strongly stimulated proliferation of vascular endothelial cells in vitro. Interestingly, the maximum activities were almost twice that of VEGF165. They also induced strong hypotension on rat arterial blood pressure compared with VEGF165 in vivo. A receptor binding assay revealed that snake venom VEGFs bound to KDR-IgG with high affinity (Kd = ∼0.1 nm) as well as to VEGF165 but did not interact with Flt-1, Flt-4, or neuropilin-1 at all. Our data clearly indicate that snake venom VEGFs act through the specific activation of KDR and show potent effects. Snake venom VEGFs are a highly specific ligand to KDR and form a new group of the VEGF family.

physiologic and pathologic angiogenesis (1,2). In vitro studies show VEGF 165 is a potent and specific mitogen for vascular endothelial cells (3)(4)(5). VEGF 165 is also known as vascular permeability factor based on its activity to provoke vascular leakage in animal models (6,7). These distinct biological effects are closely involved with each other and serve for the vasculogenic and angiogenic activities of VEGF 165 (8,9). In vascular endothelial cells, the biological function of VEGF 165 is mediated through binding to two receptor tyrosine kinases, fms-like tyrosine kinase-1 (Flt-1) (VEGF receptor 1) (10,11) and kinase insert domain-containing receptor (KDR) (VEGF receptor 2) (12,13). Studies using transgenic animals and cells for either receptor revealed that Flt-1 acts in the later phase of angiogenesis, such as in tube formation (14,15), while KDR acts in early phase of vasculogenesis and angiogenesis (16,17). The binding affinities of the two receptors differ by between ϳ10 and 90 times: the K d for VEGF 165 binding to Flt-1 is 10 pM (11,18) and ϳ75-770 pM for binding to KDR (12,13,18,19). KDR shows strong ligand-dependent tyrosine phosphorylation in comparison with Flt-1 (18) and mainly mediates the mitogenic, angiogenic, and permeability-enhancing effects of VEGF 165 (2,8).
Snake venoms are an attractive source of molecules that have unique biological activities. VEGF homologous potent hypotensive factor (HF) was isolated from Vipera aspis aspis (Aspic viper) venom (20). It has been reported that HF induces dramatic hypotension compared with VEGF 165 and other growth factors (20), but the detailed characterization of its potent activity remains to be resolved.

EXPERIMENTAL PROCEDURES
Toxin Purification-Vammin was purified by five successive chromatographic steps as follows: 200 mg of Vipera ammodytes ammodytes venom (Sigma) was dissolved in 2 ml of 50 mM Tris-HCl, pH 8.0. The venom sample was applied onto a Superdex 200pg gel-filtration chromatography column in 50 mM Tris-HCl, pH 8.0, at a flow rate of 2 ml/min. Fractions containing vammin were detected by enzyme-linked immunosorbent assay using anti-HF antiserum and loaded onto a Hi-Trap TM heparin HP column using the same buffer. The column was eluted with a linear gradient of NaCl at a flow rate of 1 ml/min. The vammin fractions were dialyzed against 50 mM Tris-HCl, pH 8.0, and applied to a Q-Sepharose HP column. The flow-through fractions were pooled and dialyzed against 20 mM imidazole HCl at pH 6.0 before loading onto a SP-Sepharose HP column. The column was separated by a linear NaCl gradient at a flow rate of 1 ml/min. Finally, the vammincontaining fraction was purified with a Mono S column. For the purification of VR-1 and HF, a similar isolation method to that for vammin was employed.
Measurement of Hypotensive Effect of Toxins-These are provided in Supplementary Fig. 1.

RESULTS AND DISCUSSION
Here we isolated two anti-HF antiserum reactive proteins, vammin and VR-1, from the venoms of Vipera ammodytes ammodytes (Western sand viper) and Daboia russelli russelli (also known as Russell's viper), respectively. The primary structures of vammin and VR-1 were determined by peptide sequencing. The pyrrolidone ring was opened by methanolysis because N-terminal residues of both intact proteins were blocked by a pyroglutamyl residue. Each chain of vammin comprised a total of 110 amino acid residues, while the number of residues in each chain of VR-1 was 109. As expected, the primary structure of vammin and VR-1 shared identity with members of the VEGF family, and the cysteine knot motif, a * This work was supported in part by Scientific Research grants-inaid from the Ministry of Education, Science, and Culture of Japan (to T. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § To whom correspondence should be addressed: Dept. of Biochemistry, Meiji Pharmaceutical University, 2-522-1, Noshio, Kiyose, Tokyo 204-8588, Japan. Tel./Fax: 81-424-95-8479; E-mail: tmorita@mypharm.ac.jp. 1 The abbreviations used are: VEGF, vascular endothelial growth factor; eNOS, endothelial nitric-oxide synthase; Flt-1, fms-like tyrosine kinase-1; HF, hypotensive factor; ICPP, increasing capillary permeability protein; KDR, kinase insert domain-containing receptor; NO, nitric oxide; L-NNA, N -nitro-L-arginine; svVEGF, snake venom vascular endothelial growth factor; DBP, diastolic blood pressure. characteristic of the VEGF family proteins, was completely conserved (Fig. 1). The sequence identities with human VEGF 165 were 47.6 and 48.1%, respectively ( Fig. 1). Calculated molecular weights were 25,129 and 25,093 for vammin and VR-1, respectively, as homodimers ( Fig. 1).
In vitro studies reveal VEGF 165 to be a potent and specific mitogen for vascular endothelial cells (3)(4)(5). We analyzed the growth factor activity of snake venom VEGFs on cultured vascular endothelial cells and found that they could stimulate endothelial proliferation with an EC 50 of ϳ100 pM, as well as that of VEGF 165 (EC 50 of ϳ70 pM) (Fig. 2a). It is interesting to note that there is a marked difference on the maximal growth level between snake venom VEGFs and VEGF 165 . VEGF 165 induced endothelial proliferation at a level of 190.1 Ϯ 16.1% compared with the control, while vammin induced endothelial proliferation at a higher level of 265.9 Ϯ 17.8% compared with the control (Fig. 2a). These data strongly suggest that snake venom VEGFs stimulate endothelial proliferation through a distinct mechanism from VEGF 165 .
Next we examined the potent effects of snake venom VEGFs in vivo. Intravascular administration of vammin (0.3 g/g) resulted in very rapid and drastic hypotension (Fig. 2b). Hypotension reached a maximal level within 3-5 min (Fig. 2b) and was clearly dose-dependent (Fig. 2c). Injection of 30 ng/g vammin induced significant hypotension, and the maximal effect was observed at 0.3 g/g administration (Fig. 2c). The doseresponse curve shows a more distinct hypotensive effect for diastolic blood pressure (DBP) than systolic blood pressure (SBP) (Fig. 2c). EC 50 values for DBP and SBP were 24 and 29 ng/g, respectively. VR-1 also induced dramatic hypotension, as with HF (see Supplemental Data). EC 50 values for VR-1 and HF were 10 and 15 ng/g on DBP and 15 and 18 ng/g on SBP, respectively. The maximal decreases in blood pressures were 56.0 Ϯ 6.8% in DBP and 24.7 Ϯ 5.2% in SBP for vammin (n ϭ 3), 48.4 Ϯ 1.7% and 16.7 Ϯ 1.1% for VR-1 (n ϭ 3), and 52.0 Ϯ 1.9% and 24 Ϯ 5.9% for HF (n ϭ 3), showing that the three toxin proteins essentially possessed an equally efficacious effect on blood pressures, although VR-1 was slightly less than the others. In contrast, VEGF 165 induced only a slight decrease in blood pressure (83.5 Ϯ 3.9% decrease in mean arterial pressure compared with 60.6 Ϯ 4.1% in vammin, Fig. 2d), indicating snake venom VEGFs exhibit stronger activity than VEGF 165 even in vivo, which is similar to growth factor activity in vitro.
Next we attempted to determine the pathway responsible for inducing hypotension. Snake venom VEGFs appear to affect peripheral blood vessels, for example, via a nitric oxide (NO)mediated effect because of the rapid and preferential effect on DBP. Pretreatment with N -nitro-L-arginine (L-NNA), a NO synthase inhibitor, completely blocked toxin protein-induced hypotension but pretreatment with phenylephrine, an ␣-adrenoreceptor agonist, did not (Fig. 2e). These data strongly indicate that snake venom VEGF-induced hypotension is probably mediated by NO. Recent studies revealed that human VEGF 165 -mediated hypotension is endothelium-dependent and that VEGF 165 stimulates NO production via at least three different pathways in vascular endothelial cells: 1) up-regulation of endothelial NO synthase (eNOS) protein level over ϳ12 h to 4 days via protein kinase C (PKC) activation (21,22), 2) transient activation of eNOS by Ca 2ϩ /calmodulin upon intracellular Ca 2ϩ release (23), and 3) enhancement of eNOS activity by phosphorylation of serine 1177 over ϳ30 min (24). Hypotension induced by snake venom proteins was very rapid and sustained for at least ϳ30 min (Fig. 2b). Therefore, we next sought to determine whether the pathway of eNOS activation is involved in the phosphorylation of eNOS on serine 1177 by immunoblot analysis using phosphoserine 1177-specific antibody on cultured vascular endothelial cells. Vammin exposure (1 nM) stimulated eNOS phosphorylation of serine 1177, and the time course exhibited a maximal effect at 5 min and returned to basal levels within 60 min (Fig. 2f). These data indicated that snake venom VEGFs produce nitric oxide via activation of eNOS by phosphorylation of serine 1177, similar to VEGF 165 , despite their strong hypotensive activity.
To obtain a detailed biochemical characterization of potent activity of snake venom VEGFs, we performed receptor-ligand binding experiments using surface plasmon resonance assay. As shown in Fig. 3a, human VEGF 165 associated with the two immobilized VEGF receptors, Flt-1-IgG and KDR-IgG, with high affinity in a rapid and time-dependent manner. The calculated association (k a ) and dissociation (k diss ) rate constants and dissociation constants (K d ) were similar to previous reports (Table I). Vammin also exhibited a high affinity of binding to immobilized KDR with a K d of 4.11 ϫ 10 Ϫ10 M (Fig. 3b) but no affinity of binding to immobilized Flt-1 (Fig. 3a). The binding parameters of snake venom VEGFs are summarized in Table I. VR-1 and HF also interacted with KDR with high affinity similar to VEGF 165 but not with Flt-1. In addition, we tested the binding ability of snake venom VEGFs to other VEGF receptors, Flt-4 (VEGF receptor-3) (25, 26) and neuropilin-1 (27); however, we could detect no specific binding ability to either receptor (Fig. 3, c and d). These findings strongly indi-

TABLE I Kinetic parameters for binding of snake venom VEGFs to immobilized
VEGF receptors Association and dissociation of various concentrations of snake venom VEGFs to immobilized VEGF receptors. The data obtained were analyzed to calculate association (k a ) and dissociation (k diss ) rate constants and maximal rate (R max ) using BIAevaluation 3.1. K d values were calculated from the ratio k diss /k a . Similar results were obtained from four independent experiments. Note that VR-1 and HF had essentially the same binding parameters to all receptors. cate that snake venom VEGFs have unique and highly specific binding properties clearly distinct from VEGF 165 . VEGF-E, a new member of the VEGF family encoded by Parapoxvirus, has also been reported to associate with KDR but not Flt-1 (28 -30). Previous studies (28 -30) have revealed that the biological activity of VEGF-E is almost identical to or somewhat weaker than VEGF 165 with respect to mitogenic activity of endothelial cells and vascular permeability activity.
In this study, we demonstrated that snake venom VEGFs specifically bound only KDR with essentially an equal affinity to VEGF 165 (Fig. 3) and exhibited a more potent biological effect than VEGF 165 (Fig. 2, a and d). Recently, it has been reported that five viral VEGFs bind KDR 2.6-fold less avidly than mouse VEGF 164 (corresponding to human VEGF 165 ), although they have comparable biological activity to VEGF 164 (31). Flt-1 is thought to be involved in the inhibition of the biological effect of VEGF through several mechanisms (32,33). Taking these factors into account, the potent effects of snake venom VEGFs are hypothesized to reflect their action directly through KDR without the requirement of Flt-1, because the entire loss of Flt-1 binding ability does not lead to the loss of binding activity to KDR. In addition, some viral VEGFs are shown to interact with neuropilin-1, although their binding abilities exhibit significant variation (30,31). From this point of view, snake venom VEGFs can classified as a new member of the VEGF family, which has highly specific receptor selectivity to KDR.
Most of the important residues for receptor binding are highly conserved in snake venom VEGFs, but some critical residues involved in Flt-1 binding are altered to other amino acid residues. First, Tyr 25 in VEGF 165 (corresponding to the 13th amino acid in vammin) is replaced by Ala. Recent studies showed that an alanine mutation at this position results in an ϳ20-fold decrease in Flt-1 binding without affecting KDR binding (34,35). Second, three amino acids (His 86 , Gln 89 , and Ile 91 ) in the third loop of VEGF 165 , which are known to be involved in the complex formation with Flt-1 (36), are replaced by Arg 74 , Ser 77 , and Lys 79 in vammin, respectively. Third, one of three negatively charged residues in the second loop (Asp 63 , Glu 64 , and Glu 67 of VEGF 165 ), reported to be critical for Flt-1 binding (37), are changed to a positively charged residue (Lys 55 ) in snake venom VEGFs. In total, 5 critical residues involved in Flt-1 binding are altered in snake venom VEGFs. We postulate that it is through these variations between mammalian and snake venom VEGFs that snake venom VEGFs acquired their unique and specific receptor selectivity.
In conclusion, we report the characterization of VEGF molecules from snake venoms. Snake venom VEGFs exhibited highly specific binding to KDR with essentially an equal ability to VEGF 165 but did not bind to other VEGF receptors at all, and the binding to KDR alone resulted in potent biological activity. As both proteins are comparatively rich in venom (0.4 -0.6% of total venom protein), they make very useful and attractive tools for studying vascular biology, such as determining specific signal transduction of KDR. During the course of our attempt to determine the mechanism of potent activity of VEGFs derived from snake venoms, the discovery of two new snake venom VEGFs, namely svVEGF and ICPP, was reported by two groups (Fig. 1) (38,39). Snake venom VEGFs form the sixth group of the VEGF family, and we propose the new group name "VEGF-F" or alternatively "F-VEGF" for snake venom VEGFs. We anticipate that further analysis will elucidate the binding mechanism of these receptor-selective VEGFs and facilitate the design of new pharmaceutical treatments based on novel receptor specific ligands.