A set of loop-1 and -3 structures in the novel vascular endothelial growth factor (VEGF) family member, VEGF-ENZ-7, is essential for the activation of VEGFR-2 signaling.

The vascular endothelial growth factor (VEGF) family plays important roles in angiogenesis and vascular permeability. Novel members of the VEGF family encoded in the Orf virus genome, VEGF-E, function as potent angiogenic factors by specifically binding and activating VEGFR-2 (KDR). VEGF-E is about 45% homologous to VEGF-A at amino acid levels, however, the amino acid residues in VEGF-A crucial for the VEGFR-2-binding are not conserved in VEGF-E. To understand the molecular basis of the biological activity of VEGF-E, we have functionally mapped residues important for interaction of VEGF-E with VEGFR-2 by exchanging the domains between VEGF-E(NZ-7) and PlGF, which binds only to VEGFR-1 (Flt-1). Exchange on the amino- and carboxyl-terminal regions had no suppressive effect on biological activity. However, exchange on either the loop-1 or -3 region of VEGF-E(NZ-7) significantly reduced activities. On the other hand, introduction of the loop-1 and -3 of VEGF-E(NZ-7) to placenta growth factor rescued the biological activities. The chimera between VEGF-A and VEGF-E(NZ-7) gave essentially the same results. These findings strongly suggest that a common rule exists for VEGFR-2 ligands (VEGF-E(NZ-7) and VEGF-A) that they build up the binding structure for VEGFR-2 through the appropriate interaction between loop-1 and -3 regions.


The vascular endothelial growth factor (VEGF) family plays important roles in angiogenesis and vascular permeability. Novel members of the VEGF family encoded in the Orf virus genome, VEGF-E, function as potent angiogenic factors by specifically binding and activating VEGFR-2 (KDR). VEGF-E is about 45% homologous to VEGF-A at amino acid levels, however, the amino acid residues in VEGF-A crucial for the VEGFR-2-binding are not conserved in VEGF-E. To understand the molecular basis of the biological activity of VEGF-E, we have functionally mapped residues important for interaction of VEGF-E with VEGFR-2 by exchanging the domains between VEGF-E NZ-7 and PlGF, which binds only to VEGFR-1 (Flt-1). Exchange on the amino-and carboxylterminal regions had no suppressive effect on biological activity. However, exchange on either the loop-1 or -3 region of VEGF-E NZ-7 significantly reduced activities.
On the other hand, introduction of the loop-1 and -3 of VEGF-E NZ-7 to placenta growth factor rescued the biological activities. The chimera between VEGF-A and VEGF-E NZ-7 gave essentially the same results. These findings strongly suggest that a common rule exists for VEGFR-2 ligands (VEGF-E NZ-7 and VEGF-A) that they build up the binding structure for VEGFR-2 through the appropriate interaction between loop-1 and -3 regions.
Vascular endothelial growth factor-A (VEGF-A) 1 plays a pivotal role in vasculogenesis, angiogenesis, and differentiation of hemangioblasts to hematopoietic precursor cells in embryogenesis. VEGF-A is also known to be closely involved in a variety of pathological conditions such as tumor angiogenesis and diabetic retinopathy (1)(2)(3).
VEGF-A is a member of the PDGF superfamily because of their structural similarities. VEGF is found to be a dimeric glycoprotein of M r 34,000 -42,000, and have conserved eight cysteine residues in each monomer. These cysteine residues construct a particular folding consisting of two intermolecular and three intramolecular disulfide bonds that generate three loop-like structures, loop-1, -2, and -3. VEGF-A exerts its biological activity by interacting with receptor-type tyrosine kinases, VEGFR-1 (Flt-1) and VEGFR-2 (KDR/Flk-1) (4 -7). Homozygous loss of the VEGFR-1 or VEGFR-2 genes resulted in embryonic lethality between days 8.5 and 9.5, indicating that these VEGF receptors play important roles in vasculogenesis and angiogenesis (8,9). The different phenotypes of these VEGFR-mice suggest that VEGFR-2 is the major positive signal transducer, whereas VEGFR-1 has a negative regulatory role in angiogenesis at early embryogenesis.
The VEGF family in vertebrate genomes includes VEGF-A, PlGF (placenta growth factor), VEGF-B, -C, and -D. PlGF and VEGF-B specifically bind to VEGFR-1 (10 -13), whereas VEGF-C and -D bind and activate VEGFR-3 (Flt-4), regulating lymphangiogenesis as well as angiogenesis in the middle stage of embryogenesis (14 -18). Amino acids in VEGF-A essential for binding with VEGFR-2 have been studied by the alanine scanning method. Keyt et al. (19) have reported that Arg-82, Lys-84, and His-86 are indispensable for the interaction between VEGF-A and VEGFR-2.
Recently, we have shown that the VEGF-E NZ-7 protein, a novel member of the VEGF family could bind specifically to VEGFR-2, activate the receptor, and promote the growth of endothelial cells in vitro and in vivo at a transient condition (20). VEGF-E genes were originally found as an open reading frame in the genome of the NZ-7, NZ-2, and D1701 strains of parapoxvirus, Orf virus (21).
These three genes were structurally very similar to each other compared with VEGF family proteins, and were designated as VEGF-E NZ-7 , ORFV2-VEGF/VEGF-E NZ-7 , and VEGF-E D1701 (20,22,23). Orf virus causes contagious pustular dermatitis in sheep, goats and, sometimes, humans. Histologically, the lesions are highly vascularized and edematous with proliferation of endothelial cells and inflammatory cells. In addition to VEGF-E NZ-7 , other two VEGF-E members were also shown to specifically bind to VEGFR-2 but not to other receptors.
VEGF-E NZ-7 has a high affinity to VEGFR-2 at similar levels as VEGF-A, and efficiently competes to VEGF-A (20). This indicates that the binding pocket on VEGFR-2 for VEGF-A and VEGF-E NZ-7 is significantly overlapped to each other. Interestingly, however, three basic amino acids on VEGF-A essential for the VEGFR-2-binding are not conserved in VEGF-E NZ-7 , and these basic amino acids were changed to hydrophobic or non-charged ones, Val, Gly, and Ser, respectively. These results indicate that the local structure built up by these basic amino acids in VEGF-A are not always required for the ligands that bind to VEGFR-2.
To elucidate a novel rule for the ligand-binding to VEGFR-2, we carried out a series of domain-exchange analysis between VEGF-E and PlGF, or VEGF-E and VEGF-A. Our results clearly indicate that an intimate relationship between the loop-1 and -3 of VEGF-E NZ-7 as well as VEGF-A is crucial for the formation of the three-dimensional structure important for the high-affinity binding to VEGFR-2.
Polyclonal and Monoclonal Antibodies-Polyclonal antisera against VEGF-E NZ-7 were raised in rabbits using a 20-amino acid sequence of the carboxyl terminus as antigen (20). Anti-human VEGFR-2 antiserum (B2) was prepared previously (24). A monoclonal antibody specific to phosphotyrosine (PY20) was obtained from ICN Biochemicals (Costa Mesa, CA). Secondary antibodies conjugated to horseradish peroxidase were purchased from Amersham Biosciences. Monoclonal neutralizing antibody to VEGF-A 165 was purchased from R&D Systems (Minneapolis, MN).
Construction of VEGF-E NZ-7 Chimeric Mutant-The synthetic cDNA encoding VEGF-E NZ-7 was cloned to the BamHI and EcoRI restriction sites of pUC18 (20), and PlGF cDNA was cloned to pUC18 (13). At first, sequences encoding six histidines were introduced by using a double stranded oligonucleotide encoding amino acid residues Cys-130 to Arg-148 of VEGF-E NZ-7 and six histidines followed by stop codon. This oligomer also had the cohesive end for DraIII at the NH 2 terminus and for EcoRI at the COOH terminus, and it was ligated to the 3-kb DraIII-EcoRI fragment of the plasmid that contains cDNA of VEGF-E NZ-7 cloned to pUC18 (pUCE-his). With the same manner, histidine residues FIG. 1. Alignment of the amino acid sequence for the VEGF family and three-dimensional structure of the dimeric molecule of VEGF-A. A, the amino acid sequences of VEGF-E (VEGF-E NZ-7 , VEGF-E NZ-2 /VEGF-OR-FV NZ-2 , VEGF-E D1701 ) and human VEGF-A, -B, -C, and -D are shown. The amino acid residues of VEGF-A critical for interacting with VEGFR-2 are shaded in orange. The numbers in red and blue boxes represent the amino acid residues of VEGF-A and VEGF-E NZ-7 , respectively. The numbers and thin lines in black represent the regions of VEGF-E NZ-7 that were replaced with corresponding amino acid residues of human PlGF and/or VEGF-A. The double black line with number 27 indicates a unique amino acid stretch only found in VEGF-E NZ-7 . Red character, the eight cysteine residues conserved in all VEGF family members. Blue character, the amino acid residues identical to VEGF-E NZ-7 . Black line, the core region that is conserved in all VEGF family. Red bar, loop regions 1, 2, and 3 of VEGF-A. Green and blue bars, the ␣-helix regions and ␤-strands of VEGF-A, respectively. Signal peptide sequences at the NH 2 -terminal region of the VEGF family were omitted in this figure. Ϫ, no amino acids. B, dimer structure of VEGF-A. One monomer of the VEGF-A dimer is shown in red and another in green. The VEGF-A molecule forms three intramolecular and two intermolecular disulfide bonds. The disulfide bonds and cysteine residues of a homodimer are shown in yellow in one monomer and in red in another. The numbers in black indicate the loop structure. The amino acid residues of VEGF-A important for interacting VEGFR-2 are shaded in red. This figure was adapted from Muller et al. (30,31).
were introduced into the COOH terminus of PlGF-1 cDNA. The double stranded synthetic oligonucleotide encoding Arg-131 to Arg-149 of PlGF-1 was followed by six histidine residues and a stop codon. This oligomer had cohesive ends for the BsmI site at the NH 2 terminus and EcoRI site at the COOH terminus. This oligomer was cloned to a 3-kb BsmI-EcoRI fragment of the plasmid (pUCP-his) with full-length cDNA of PlGF-I cloned to BamHI and the EcoRI site of pUC18.
The constructs of chimera proteins were produced by exchanging variable regions among VEGF-E NZ-7 , VEGF-A, and PlGF. They were achieved by ligating a series of digested fragments of pUCP-his and/or pUCE-his with synthetic double stranded oligonucleotides and/or the fragments produced by the PCR technique. The details of each plasmid construction are available on request.
Each construct contains the following amino acid sequence: A, a variety of constructs for the replacement of amino-and carboxyl-terminal regions in VEGF-E NZ-7 (orange) with the corresponding regions of PlGF (green). a, VEGFR-2 (KDR) autophosphorylation was measured by using NIH3T3-KDR cells. The cells were stimulated with chimera mutant proteins (4 -120 ng/ml), lysed, and then subjected to SDS-PAGE for Western blotting with an anti-phosphotyrosine antibody (␣-PY) or with an anti-VEGFR-2 antibody (␣-KDR IK-5). The activity was indicated by ϩ. ϩϩϩ, equal to wild type VEGF-E NZ-7 ; ϩϩ, 2-3-fold weaker than wild type VEGF-E NZ-7 ; ϩ, 3-10-fold weaker than wild type VEGF-E NZ-7 ; Ϫ, phosphorylation was not detected in this concentration range. b, proliferation of HUVEC. Quiescent HUVEC were stimulated with 1, 10, and 100 ng/ml chimera mutant proteins. After 4 days, cells were stained and the cell number was determined by averaging the counting of five spots to wild type VEGF-E NZ-7 . ϩϩ, 2-fold weaker than wild type VEGF-E NZ-7 at 100 ng/ml; ϩ, 4-fold weaker than wild type VEGF-E NZ-7 ; Ϫ, HUVEC proliferation was not detected. B, binding activities of the chimeric proteins between VEGF-E NZ-7 and PlGF to VEGFR-2. The activities were measured by competition experiments using 125 I-VEGF-A and VEGFR-2 binding systems (see "Materials and Methods"). C, the activities of chimeric proteins for stimulation of HUVEC growth.
Expression of Mutant Proteins and Purification-The full-length coding regions of chimera mutants were subcloned into the BamHI-EcoRI site on the multicloning region of pVL1393. These transfer vector DNAs were used for co-transfection into Sf9 cells along with the linearized baculovirus DNA "BaculoGold" by liposome transfection. The recombinant viruses were amplified at 3-day intervals. Sf9 cells grown in serum-free medium EX-CELL 400 were used to produce chimera proteins. For a large scale preparation of proteins, Sf9 cells were infected with viruses at a multiplicity of infection of about 10. Three days after infection, the culture media were harvested and centrifuged to remove debris. The supernatants were resolved by SDS-PAGE on a 15% gel followed by Western blotting.
One-hundred ml of the supernatant of cells infected with chimera recombinant virus was collected, concentrated, and dialyzed to 22.5 mM sodium phosphate buffer, pH 8.0, containing 375 mM NaCl and 62.5 mM imidazole. After dialysis and filtration, glycerol was added for a final concentration at 20%, giving rise to the final buffer with 20 mM sodium phosphate, pH 8.0, 300 mM NaCl, and 50 mM imidazole. The nickelnitrilotriacetic acid beads (Qiagen, Germany) were applied to the dialyzed sample and rotated for 3 h at 4°C. The nickel-nitrilotriacetic acid containing sample was loaded onto a 5-ml column, and washed with washing buffer (50 mM imidazole, 20 mM sodium phosphate buffer, 300 mM NaCl, and 20% glycerol). The chimera proteins were eluted with elution buffer (250 mM imidazole, 20 mM sodium phosphate buffer, pH 8.0, 300 mM NaCl, and 20% glycerol). The positive fractions were collected and further concentrated by Microcon (Millipore). For detection of VEGF-E protein, aliquots of fractions were analyzed by Western blotting using anti-His antibody (Qiagen, Germany) and Coomassie staining. The purity of chimera proteins was above 80%.
Binding Competition Assays-Binding competition assay was performed as previously described (25,26) using receptor protein immobilized to solid phase. Preparation of the receptor protein consisting of the extracellular region of VEGFR-2 tagged with the Fc portion of IgG was described previously (25). Aliquots (50 l) of the Fc-tagged receptor (0.3 g/ml) in phosphate-buffered saline were attached to 96-well plates, Immunon 2 (Dynex Technologies, Inc., Chantilly, VA), overnight at 4°C. The plates were washed twice and blocked with binding buffer (1% bovine serum albumin in phosphate-buffered saline) for 30 min at 25°C. Competition activity was examined by incubating the KDRcoated plates with a fixed concentration of 125 I-VEGF-A and increasing concentrations of non-radiolabeled chimera mutant proteins for 3 h at 25°C. The wells were washed three times with binding buffer, then the bound 125 I-labeled proteins were quantified in a ␥-counter. All experiments were performed in duplicate.
Human Endothelial Cell Proliferation Assay-Human umbilical vein endothelial cells (HUVEC) (Morinaga, Tokyo) were grown in HuMedia-EG2 (Kurabo, Tokyo) and used for endothelial cell growth assay. HU-VEC were seeded at 4,000 cells/well on 24-well collagen-coated plates (CELLTIGHT, Sumitomo Bakelite Co., Ltd., Tokyo) in medium with free-growth factor and low concentration serum (0.2%). Four h after plating, mutant proteins containing medium were added. Two days after, medium and mutant proteins were replaced with fresh sample. At day 4, cells were stained with the crystal violet. The cell numbers were determined by averaging the counting of five spots randomly chosen at each well using a Coulter counter.
VEGFR-2 Autophosphorylation Assay-For in vivo phosphorylation, NIH3T3-KDR (VEGFR-2) cells were grown to semiconfluence and stimulated with a variety of ligands at 37°C for 5 min. The cells were washed in ice-cold phosphate-buffered saline with 0.1 mM Na 3 VO 4 twice and lysed in 1% Triton X-100 lysis buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 1.5 mM MgCl 2 , 2% aprotinin, 1 mM phenylmethylsulfonyl fluoride, 50 mM sodium fluoride, 10 mM Na 4 P 2 O 7 , and 2 mM Na 3 VO 4 ). The lysates were clarified by centrifugation (15,000 rpm for 10 min). Protein concentrations were measured using a Bio-Rad protein assay and the same amounts of protein of each sample were used for analysis. For immunoblotting, the cell lysates were subjected to 7.5% SDS-PAGE and transferred to a nitrocellulose sheet. The blots were incubated with a blocking solution (5% bovine serum albumin containing washing buffer (20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.3% Tween 20)) and probed with the primary antibody diluted in blocking solution. The signal was visualized using peroxidase-conjugated secondary antibodies and exchanged chemiluminescence (ECL, Amersham) according to the manufacturer's instructions.
Tubular Formation Assay-The angiogenesis kit was purchased from Kurabo (Tokyo) and the tubular formation assay was demonstrated as per the manufacturer's instructions. In this system, human endothelial cells and fibroblasts were co-cultured in 24-well plates at 37°C, 5% CO 2 . At days 1, 4, 7, and 9, medium was exchanged with fresh medium, which contains chimera proteins and the excess amount of neutralizing antibody for VEGF-A. At day 11, incubation was terminated, fixed with 70% ethanol, and immunostained for CD31/PECAM-1. For immunostaining, mouse anti-human CD31, alkaline phosphataseconjugated goat anti-mouse IgG and 5-bromo-4-chloro-3-indolyl phosphate were purchased from Kurabo (Tokyo). The cells were incubated with mouse anti-human CD31 for 60 min at 37°C in blocking buffer (1% bovine serum albumin in phosphate-buffered saline) and washed with blocking buffer. Secondary, alkaline phosphatase-conjugated goat antimouse IgG was added to the cells followed with incubation for 60 min at 37°C. After washing, 5-bromo-4-chloro-3-indolyl phosphate was used to develop a color to visualize endothelial cells. Tubules were analyzed under a bright-field microscope and total length of branching in a fixed area (mm/mm 2 ) of the randomly chosen 5 spots per well.

Production and Dimer Formation of VEGF-E NZ-7 Mutants-
Alanine scanning analysis of VEGF-A has shown that basic amino acids at 82, 84, and 86 as well as other amino acids in loop-3 are essential for VEGFR-2 binding, however, the VEGF-E family that binds to VEGFR-2 at a similar affinity as VEGF-A does not conserve these amino acids (Fig. 1). To examine which region(s) in VEGF-E is crucial for the binding and activation of VEGFR-2, we constructed a series of chimeric mutants by exchanging the variable region of VEGF-E NZ-7 with that of PlGF and/or VEGF-A. The carboxyl terminus of all chimeric mutants had an insertion of six histidines as a tag sequence for the convenience of protein purification. This His tag did not suppress the activity of VEGF-E NZ-7 .
VEGF-E NZ-7 was exchanged with PlGF from the NH 2 -terminal, and the exchanging region was extended toward the COOH-terminal, serially (Figs. 2 and 3). Secondary, exchanging was performed from the COOH terminus toward serially. Loop-wise exchanging on loop-1, -2, and -3 was produced by both PlGF and VEGF-A amino acid sequences (Fig. 4A). The site-directed mutagenesis was also performed by exchanging only 4 amino acid residues (Fig. 4B). Finally, loop-1 and -3 regions of VEGF-E NZ-7 were introduced to PlGF as the gain of function mutants (Fig. 5) (see "Materials and Methods").
The molecular size of mutant proteins in non-reducing conditions were in the range of M r 40,000 to 45,000, and 20,000 to 25,000 in reducing conditions, consistent with wild type VEGF-E NZ-7 and PlGF (Fig. 3, data not shown). These results indicate that all mutants did not show any disruption in dimerization. affinity of each mutant to receptor, a binding assay was carried out by using the extracellular domain of VEGFR-2 tagged with the Fc portion of IgG (see "Materials and Methods"). Because VEGF-E NZ-7 could compete with VEGF-A in binding to VEGFR-2, chimeric proteins were tested for their ability to compete with 125 I-labeled VEGF-A 165 for interaction to VEGFR-2-Fc immobilized onto 96-well plates.
When VEGF-E NZ-7 was exchanged with PlGF from the NH 2terminal toward COOH-terminal serially ( Fig. 2A), 34 amino acids of the NH 2 -terminal region in VEGF-E NZ-7 were replaceable without any reduction of affinity to VEGFR-2 (Fig. 2,  chimera 1). An extension of the PlGF region to the first conserved cysteine residue resulted in a minor reduction of the affinity (Fig. 2B, 2). However, as shown by chimeras 3, 4, and 5, the PlGF amino acid sequence close to and over loop-1 resulted in the loss of affinity similar to intact PlGF (Fig. 2B).
Secondary exchanging was performed from the COOH terminus toward the NH 2 terminus serially ( Fig. 2A). Among the mutants 6, 7, and 8, chimera 6 with 18 amino acids of the COOH-terminal exchange from the 8th conserved cysteine to the COOH-terminal end remained its affinity. However, the replacement over loop-3 such as chimeras 7 and 8 had lost their binding ability. Taken together, NH 2 and COOH termini were not critical for VEGF-E NZ-7 to interact with VEGFR-2.
To confirm this result, a chimera with simultaneous exchanging of both NH 2 and COOH termini were evaluated ( Fig.  2A, chimera 9), and showed no significant suppression in its affinity. These results may imply that at least loop-1 and -3 are very critical for VEGF-E NZ-7 , whereas the importance of loop-2 remained unclear. (27) reported that several amino acids in loop-3 are essential for the binding of VEGF-A to VEGFR-2 (27). This result suggests that loop-3 is a direct binding site for VEGFR-2 in VEGF-A. To analyze the importance of each loop region of VEGF-E NZ-7 , a variety of chimera mutants with loop region-specific replacement were tested (Figs. 4 and 5). These exchanged regions do not contain ␤-strands that flank the loop site. Therefore, the basic architecture of these chimeric proteins would not be disrupted.
We found that the loop-2 region in VEGF-E NZ-7 is exchangeable to that of PlGF or VEGF-A (Fig. 4, chimera 12 and 13), whereas loop-1 and -3 specific exchanges to PlGF caused a significant defect (Fig. 4, chimera 11, 15, and 17). These results indicate that the important regions for VEGF-E NZ-7 to interact to VEGFR-2 are not only loop-3 but also loop-1.
Along with these PlGF replacements, the VEGF-A region was also introduced into loop regions of VEGF-E NZ-7 as a reference. Surprisingly, independent introduction of loops of VEGF-A, which is a strong ligand for VEGFR-2, also resulted in significant reduction in biochemical and biological activities (Fig. 4, chimeras 10, 14, and 16). Therefore, both loop-1 and -3 regions are equally important in VEGF-A as well as in VEGF-E NZ-7 . This result implies that VEGF-E NZ-7 and VEGF-A may have a common mechanism in the interaction to VEGFR-2, by having critical regions in loop-1 and -3.
Loop-1 but Not the Adjacent Short Sequences Are Important for Biological Function-More detailed mutations were introduced to identify important amino acid residues. The region for loop-1 exchange was divided into 3 parts, which were composed of YLGE, ESTN, and LQYN. These residues were changed to PlGF counterparts, DVVS, SEVE, and HMFS (Figs. 1A and 4B). Among these mutants, a chimera with ESTN exchanged with SEVE severely lost the affinity, and exchange of YLGE to DVVS had minor loss in affinity, whereas no effect in another exchange mutant (chimeras 20, 19, and 21, respectively). These results suggest that the affinity reduction conferred by the exchange of the loop-1 region in chimera 11 was mainly caused by the replacement of ESTN on VEGF-E NZ-7 to SEVE in PlGF. In the loop-1 of VEGF-E NZ-7 , accumulation of negatively charged residues might have disrupted the binding action of VEGF-E NZ-7 . In addition, regions upstream from loop-1 (chimera 22 and 23) and a region downstream of loop-3 (number 26) were replaceable to the corresponding regions of PlGF, however, short stretches upstream from loop-3 were not (numbers 24 and 25).
VEGF-E NZ-7 have its specific insertion-like stretch with 8 amino acid residues in the loop-3 region (Fig. 1A). This stretch FIG. 5. Functional analysis of VEGF-E NZ-7 /PlGF chimera mutant proteins, 28 -33. A, the construct map of chimera mutants and their activities for VEGFR-2 autophosphorylation and HUVEC growth (see Fig. 2, legend). B, binding activities of the chimeric proteins to VEGFR-2. C, the activities of chimeric proteins for stimulation of HUVEC growth.
was not found in other VEGF-E members or in the VEGF family. Interestingly, the flanking region of this stretch showed similarities in amino acid sequence to other VEGF-E genes. This stretch was deleted to test its significance for the VEGF-E NZ-7 protein. The deletion resulted in a complete loss of biological activities, indicating that this short stretch is indispensable for VEGF-E NZ-7 (Fig. 4B, chimera 27).

Cooperation between a Proper Set of Large Loop-1 and Large Loop-3 Is Crucial for Restoration of Biological Activity in
VEGF-E NZ-7 -To examine the regions in VEGF-E NZ-7 necessary and sufficient for the activation of KDR/VEGFR-2, we further constructed chimeric molecules between VEGF-E NZ-7 and PlGF. Under the background of the PlGF sequence, the loop-1-containing sequence and the loop-3-containing sequence in VEGF-E NZ-7 were introduced to the corresponding regions (Fig. 5A, 31-33). For a control experiment, under the background of VEGF-E NZ-7 , the loop-1 and -3 regions in VEGF-A were replaced to the corresponding regions (Fig. 5A, 28 -30). As shown in chimera 33, both loop-1 and -3-containing regions were found to be required for the activation of VEGFR-2 kinase.
Although VEGF-E NZ-7 and VEGF-A are able to stimulate VEGFR-2, to our surprise, a single replacement of either loop-1 or -3 in VEGF-A to the corresponding region in VEGF-E NZ-7 strongly suppressed the biological activity (Fig. 5, chimeras 28  and 29). Both loop-1 and -3 of VEGF-A are required for a partial recovery of the activation of VEGFR-2 (chimera 30). Taken together, these results strongly suggest that an appropriate pair of loop-1 and -3 is essential for the construction of threedimensional structure for the binding and activation of VEGFR-2.
The Binding Ability of Mutants to VEGFR-2 Correlates Well with the Activity of Receptor Autophosphorylation Assay and HUVEC Proliferation Assay-These mutant proteins were tested for their ability to induce autophosphorylation of VEGFR-2 using a cell line, NIH3T3-KDR (see "Materials and Methods") ( Fig. 6, data not shown). The ability was highly correlated with the affinity of mutant proteins to the receptor (Figs. 2, 4, and 5). The mutants with high affinity, as the wild type VEGF-E NZ-7 protein to receptor, demonstrated a strong activity in induction of autophosphorylation of KDR/VEGFR-2, whereas the mutants with no affinity did not in a concentration range up to 150 ng/ml in the final medium. These results indicate that the critical regions of VEGF-E NZ-7 for binding to VEGFR-2 are also important for inducing autophosphorylation of VEGFR-2.
VEGF-A-induced signal transduction for the proliferation of endothelial cells is mainly mediated by VEGFR-2. To evaluate the relationship of the abilities to induce autophosphorylation of VEGFR-2 and to induce a proliferation of endothelial cells, mutant proteins were tested for their activity to induce proliferation of HUVEC. As expected, mutant chimeras 1, 2, 6, 9, 12, 13, 19, 21-23, 26, 30 (weak), and 33 with high affinity for VEGFR-2 could lead the proliferation of HUVEC, whereas mutants with low receptor affinity facilitated weak mitogenesis (Figs. 2, 4, and 5). Those mutants with no affinity failed to show such activity, with one exception, number 18. Mutant number 18 did not show detectable affinity to VEGFR-2, but it partially induced mitogenesis of endothelial cells. These results suggest that the affinity strength of mutant proteins mostly correlates with their activity to induce mitogenesis of HUVEC.
Evaluation of Mutants in Tubular Formation Assay-VEGF-A is known to stimulate the endothelial cells to form a tubule-like structure in vitro and in vivo. The chimeric mutants of VEGF-E NZ-7 were tested for their activity for tubular formation in a recently developed co-culture system between HUVEC and human diploid fibroblasts (see "Materials and Methods") ( Fig. 7). To decrease a background tube formation in this system detectable without any exogenous angiogenic factors, we added anti-VEGF-A neutralizing antibody into culture medium.
At first, we demonstrated that VEGF-E NZ-7 could induce the tubular formation in vitro (Fig. 7A). Next we examined tubular formation by chimera mutants. Representative results and quantitative analysis are shown in Fig. 7, B and C. The strength of chimera proteins for tubular-forming activity was correlated with the affinity to the receptor, and with the abilities to autophosphorylate the receptor and induce the proliferation of endothelial cells. Interestingly, the tubules induced by VEGF-E NZ-7 were morphologically distinguished from that of VEGF-A, where VEGF-E NZ-7 -induced tubules were slightly thicker. Chimera number 18, which showed only weak HUVEC proliferation without the significant affinity to receptor and the receptor autophosphorylation ability in vitro could induce the tubular formation as effectively as VEGF-A and VEGF-E NZ-7 . The morphology induced by number 18 was closely related to that of tubules induced rather by VEGF-A than VEGF-E NZ-7 . This result suggests that this tubular formation system is more sensitive than other assays such as the binding assay, receptor autophosphorylation assay, and HUVEC proliferation assay, so that the activity of chimera 18 was detected. In conclusion, both loop-1 and -3-containing regions in VEGF-E NZ-7 were found to be required for the binding and activation of VEGFR-2 in a cooperative manner (Fig. 8). DISCUSSION In this study, the important regions of VEGF-E NZ-7 for binding to VEGFR-2 were analyzed by using domain exchanging with PlGF and site-directed mutagenesis. PlGF does not bind to VEGFR-2 but only to VEGFR-1 (12,13,28). VEGF-E NZ-7 and PlGF have a similar amino acid composition, and they also completely conserve the critical cysteine-knot motif that was composed of eight cysteine residues (see Fig. 1B). Recently, the crystal structure study for PlGF (29) revealed that PlGF actually conserves the tertially structure with VEGF (30,31) and PDGF (32). Thus, the VEGF-E NZ-7 /PlGF chimera mutants were assumed to fold appropriately in a similar manner and build up tertiary structure. We confirmed that these chimeric mutants conserve the dimerization property in post-translational modification (Fig. 3).
In previous reports, the receptor interacting amino acids for PDGF-B and VEGF-A were identified. For PDGF, site-directed mutagenesis revealed that Arg-27 and Ile-30 were critical amino acid residues in interacting to PDGF receptor (33). These amino acid residues were found in the loop-1 region of PDGF-B. The alanine scanning mutagenesis of VEGF-A demonstrated important amino acid residues for binding to receptors. For VEGFR-1 binding, negatively charged residues Asp-63, Glu-64, and Glu-67 were critical, and they were found in the loop-2 region (27). For VEGFR-2 binding, mutation in positively charged residues, Arg-82, Lys-84, and His-86 resulted in severe reduction in receptor binding, and these basic amino acid residues are located in the loop-3 region (27). Because these three basic amino acids are not conserved and changed to hydrophobic ones in the VEGF-E family, we suggest that these basic amino acids in VEGF-A are not directly interacting with VEGFR-2 but are important for keeping the proper tertially structure within the VEGF-A molecule.
In another report (31), supporting the previous result, critical amino acids for binding VEGFR-2 were found clustered at ␣-helix, loop-1, -2, and -3 regions. They were designated "hot spots." The majority of these hot spot amino acid residues are located in loop-1 and -3 regions. In addition, VEGF-A neutralizing antibody and peptides, which inhibit the interaction of VEGF-A to VEGFR-2, have been reported. Interestingly, the epitopes for these inhibitors are located at loop-1 and -3, thus, they might modify the structure of loop-1 and/or loop-3 regions (34,35).
We found that PlGF could replace the amino and carboxyl terminus of VEGF-E NZ-7 without any changes in biochemical and biological activities. This result indicates that the amino and carboxyl termini in VEGF-E NZ-7 are not involved directly in binding with VEGFR-2. As shown in mutants 3-5, 7, and 8 ( Fig. 2), extending the exchange region into the core region and over loop-1 and -3 regions resulted in a significant defect in biological and biochemical activities. These results suggest that the regions containing loop-1 and -3 are critical for VEGF-E NZ-7 to be functionally intact. As expected, exchanging the narrowed regions of VEGF-E NZ-7 only containing loops-1 and -3 (mutant FIG. 7. Analysis of tubular formation by chimera mutant proteins. VEGF-E chimera proteins were tested for stimulatory activity of endothelial tubular formation using the HUVEC-fibroblast co-culture system (see "Materials and Methods"). At days 1, 4, 7, and 9, medium was replaced with fresh medium that contains angiogenic factor and neutralizing antibody for VEGF-A (500 ng/ml). At day 11, incubation was terminated, fixed, and immunostained with ␣-CD31/PECAM-1 antibody for endothelial cells (arrows). The co-cultured fibroblasts were not stained but slightly giving rise to the yellow background. A, VEGF-E NZ-7 (50 ng/ml) was tested for tubular formation activity. a, the supplied medium induced tubular formation. b, tubular formation was completely inhibited by an excess amount of neutralizing antibody (500 ng/ml) for human VEGF-A. c and d shows tubular formation by VEGF-E NZ-7 without and with an excess amount of neutralizing antibody for human VEGF-A, respectively. It indicates that VEGFR-2 activation only by VEGF-E NZ-7 is sufficient for tubular formation. B, chimera mutants (50 ng/ml) were tested for tubular formation. The number in each panel indicates the serial number of chimera mutants. 9, 12, and 18 show the tubular formation activity as effective as wild type VEGF-E NZ-7 . C, quantitative analysis of tubular formation with mutant VEGF-E NZ-7 (see "Materials and Methods"). Control, VEGF-A (50 ng/ ml); Ab, anti-VEGF-A antibody (500 ng/ml); E, VEGF-E NZ-7 (50 ng/ml); number, each chimera mutant VEGF-E NZ-7 (50 ng/ml).
FIG. 8. Schematic representation of exchangeable regions and non-exchangeable regions for VEGF-E NZ-7 to interact with VEGFR-2. Exchangeable regions are shown in red, unexchangeable regions obtained from the chimera study in blue, and the regions conserved in VEGF-E NZ-7 and PlGF in green. Regions exchangeable with PlGF in a detailed chimeric analysis (Fig. 4B) are shown in weak red color. Exchangeable regions include the NH 2 terminus, COOH terminus, loop-2, ␣-helix, and part of loop-1 regions. Non-exchangeable regions include the middle part of loop-1 and the broad region associated with loop-3. By introducing these "non-exchangeable regions" of PlGF to VEGF-E NZ-7 significantly reduces activities for receptor interaction and endothelial proliferation. Note that the conserved regions were not mutated. These results suggest that loop-1 and -3 play the important roles in providing the binding determinant for VEGFR-2.
10 and 11 for loop-1, 14 -17 for loop-3) also produced functionally defective proteins, whereas the loop-2 exchange did not reduce activity (mutants 12 and 13). Mutant proteins with low affinity to VEGFR-2 could not induce autophosphorylation of VEGFR-2, proliferation of HUVEC, or tubular formation by endothelial cells except for chimera 18. These results suggest that the affinity strength to the receptor basically reflect their effectiveness in biological response.
In addition to the replacement by PlGF, VEGF-A regions were also introduced to VEGF-E NZ-7 in a loop-specific manner, as a reference (Figs. 4A and 5A). Surprisingly, these mutants with VEGF-A loops also had significant reduction in biochemical and biological activities. It could suggest that a single loop of VEGF-A is not enough to be biologically functional. A pair of loop-1 and -3 of VEGF-E NZ-7 or VEGF-A may be required to build up the receptor-binding determinant for VEGFR-2. Mutants 10 and 14 did not bind to VEGFR-2 and lost biological activity. Mutant 18, which is a combination of numbers 10 and 14 with regions of loops-1 and -3 of VEGF-A onto VEGF-E NZ-7 , had a tubular formation activity, whereas it did not show activities in other VEGFR-2-associated analysis at detectable levels. Chimera 18 molecule might interact with VEGFR-2 better in a two-dimensional co-culture system that seems closer to the in vivo situation, compared with the one-dimensional culture of HUVEC.
As reported previously, regional exchange of the VEGF-A loop-3 region with that of PlGF resulted with significant reduction in VEGFR-2 binding and proliferation of endothelial cells (36,37). However, its activity in inducing vascular permeability was still functional (37). It was proposed that the determinant of VEGF-A to facilitate vascular permeability could be different from that to bind to VEGFR-2 and induce endothelial cell proliferation. It is also possible that the Miles assay, an assay for activity to facilitate vascular permeability could be very sensitive compared with other assays, so that activity of that mutant was detectable. The construction pattern of two chimera proteins in this study, numbers 15 and 17, are similar to such loop-3 exchanged mutants of VEGF-A. Interestingly, these VEGF-E NZ-7 chimeric proteins lost the activity to facilitate vascular permeability, 2 indicating that the loop-3 region of VEGF-E NZ-7 was critical in this particular biological activity. This result suggests that some of the biochemical properties of VEGF-A and VEGF-E could be different, especially in the aspect of the mechanisms for facilitating vascular permeability.
The chimera proteins that could be clinically applied as angiogenic stimulators include numbers 1, 6, 9, and 33. In keeping with the strong activities as wild type VEGF-E NZ-7 , they have human PlGF amino acid residues at the amino and/or carboxyl termini instead of viral amino acid residues of VEGF-E NZ-7 . Amino-or carboxyl-terminal peptide sequences are generally highly immunogenic. Therefore, these "humanized VEGF-E family" are expected to be less immunogenic compared with the wild type VEGF-E NZ-7 and clinically more useful.
The important region of VEGF-E NZ-7 for interacting with VEGFR-2 was identified as loop-1 and -3. Loop-1 and -3 may be playing an important role in presenting the binding determinant of VEGF-E NZ-7 for VEGFR-2. According to these results, a heterodimer in which one molecule binds only VEGFR-2 but the other does not bind either VEGFR-1 or VEGFR-2 could be designed and synthesized. Such peptides may be very useful as a VEGFR-2 antagonist for clinical application to inhibit the pathological angiogenesis in which VEGFR-2 plays the major role as a signal transducer.