Full text of DOI:10.1007/BF02882392

Vol. 44 No. 5
SCIENCE IN CHINA (Series C)
October 2001
Screening of human vascular endothelial growth factor
(VEGF) receptor Flt-1 domain and study on its biological
activity
MA Li (马 骊)1, ZHANG Zhiqing (张智清)2, WANG Xiaoning (王小宁)1,
SUN Dajun (孙大军)3, ZHOU Xiaoming (周小明)2, CHEN Aijun (陈爱君)2
& YAO Lihong (姚立红)2
1. Institute of Molecular Immunology, The First Military Medical University, Guangzhou 510515, China;
2. National Laboratory of Molecular Virology and Genetic Engineering, Beijing 100052, China;
3. Department of Vascular Surgery, The Third Teaching Hospital of Norman Bethune University of Medical Science,
Changchun 130031, China
Correspondence should be addressed to Zhang Zhiqing (email: zhangzq@public3.bta.net.cn) and Wang Xiaoning
(email: xnwang@hotmail.com)
Received March 16, 2001
Abstract Four human vascular endothelial growth factor receptor Flt-1 cDNA fragments contain-
ing extracellular domain loops 2, 1-2, 2-3 and 1-3 respectively were amplified from human placen-
tal cDNA library by PCR and used for screening ligand binding domains by yeast two-hybrid sys-
tem. The result showed that, not only loop 1-3, but also the smaller fragment loop 2-3 could bind to
hVEGF165. Recombinant expression plasmids pPIC9K/Flt-1(1-3) and pPIC9K/Flt-1(2-3) were con-
structed and transformed to Pichia. pastoris host strain GS115, cultured in flasks, and expressed
under the induction of 1% methanol. The expressed product existed in supernatant in the form of
soluble molecules and contained more than 60% of total protein after being induced for 4d. After
being purified by CM-Sepharose FF and Sephacryl S-100 chromatography, its purity reached
above 90%. Biological assay in vitro showed that the binding capacity of expressed soluble
Flt-1(2-3) to hVEGF165 and its inhibiting effect on the proliferation of human umbilical veins endo-
thelial cells (HUVEC) stimulated with hVEGF165 were close to those of sFlt-1(1-3). Animal test
showed that sFlt-1(2-3) could inhibit the formation of regenerate blood vessels stimulated with
hVEGF165 significantly.
Keywords: vascular endothelial growth factor (VEGF), receptor, yeast two-hybrid, Pichia. pastoris, gene expression.
The abnormal expression of vascular endothelial growth factor (VEGF) is closely correlated
with the occurrence of many diseases. High expressions of VEGF and its receptor have been
found in the tissues of many solid tumors, such as stomach, kidney, colon and bladder cancer.
VEGF can promote the proliferation and diffusion of tumor cells by accelerating the growth of
regenerate blood vessels. The study on diabetic retinopathy showed that a large quantity of VEGF
receptors with high affinity on vascular endothelial cell of retina could cause the proliferation of
blood vessels and result in blindness. The VEGF existing at a high level in the synovium of pa-
tients with rheumatoid arthritis can cause the formation of synovial pannus and destruction of ar-
ticular cartilage[1].

498
SCIENCE IN CHINA (Series C)
Vol. 44
Flt-1 is the VEGF receptor with a high affinity and a whole length of 1338 amino acids. As a
family member of type III tyrosine protein kinase receptor, it consists of extracellular, transmem-
brane and intracellular domains. The region for binding to ligand is located in the extracellular
domains consisting of 7 Ig-like loops[2]. Soluble Flt-1 (sFlt-1) can block the biological activity of
VEGF by competitive binding to it with the receptors on the surface of vascular endothelial cells.
In this way, the growth and metastasis of tumor, as well as the formations of regenerate blood
vessels in retina and articular synovium are inhibited. So sFlt-1 has a potential value in clinical
application.
So far all the study work on the ligand binding domain of Flt-1 is based on extracellular test.
It has been reported that loop 1-3 at N-terminal of Flt-1 extracellular domain showed no
significant difference from the whole Flt-1 molecule[3] in binding capacity to VEGF, and loop 2
might play an important role in the binding of Flt-1 to its ligand[4]. In order to develop an sFlt-1
domain with the binding capacity to VEGF and a low molecular weight, the smallest ligand
binding domain of Flt-1 was screened using yeast two-hybrid system in this paper. On the basis of
this, the screened Flt-1 domain genes were expressed in Pichia. pastoris system, and the in vitro
and in vivo biological activities of expressed products were analyzed at protein level.
1
Materials and methods
1.1 Cell line and main agents
HUVEC strains were stored in our laboratory. Human placental cDNA library was a gift from
Dr. Wang Yanfang in Institute of Virology, Chinese Academy of Preventive Medicine. Yeast
two-hybrid system was manufactured by Clontech Corp. Pichia. pastoris expression system was
purchased from Invitrogen Corp. Yeast medium pepton and YNB (W/O) were manufactured by
Difco Corp. hVEGF165 was purchased from Jingmei Bioengineering Corp. CM-Sepharose FF and
Sephacryl S-100 chromatography media were manufactured by Pharmacia Corp. 125I-hVEGF165
with a specific radioactivity of 31.05—47.5 Ci/g, was manufactured by Amersham Corp. MTT
was manufactured by SIGMA Corp.
1.2 Amplification of target gene fragment
Primers were synthesized in Shanghai Shenggong Bioengineering Co. Ltd. According to the
cDNA sequence of Flt-1 gene extracellular domain, the following 4 primers were designed:
Upstream primer P1: 5′-TAGAATTCATGTCTAGTTCAGGTTCAA-3′
Upstream primer P2: 5′-TAGAATTCGGTAGACCTTTCGTAGAG-3′
Downstream primer P3: 5′-TAGGATCCTATTATGTGCTTATTTGGAC-3′
Downstream primer P4: 5′-TAGGATCCTATTAATGTTTCACAGTGAT-3′
Of the primers, P2-P3, P1-P3, P2-P4 and P1-P4 were used for the amplification of loops 2,
1-2, 2-3 and 1-3 respectively. P1 and P2 were introduced to EcoR I site, and P3 and P4 were in-
troduced to BamH I site. The gene fragments were amplified by PCR (94℃ 45 s, 68℃ 45 s, 72℃

No. 5
SCREENING OF HUMAN VEGF RECEPTOR Flt-1 DOMAIN
499
60 s, 30 cycles, and extended at 72℃ for 10 min) using human placental cDNA library as tem-
plate.
1.3 Construction of recombinant plasmid for yeast two-hybrid system
The 4 kinds of PCR products were inserted into pGEM-T vector and transformed into host
strain E. coli DH5α. The plasmid was extracted and identified by DNA sequencing, then digested
with EcoR I and BamH I. The corresponding fragments were recovered and inserted into the same
sites of pGAD424 respectively. Plasmid pBV220/hVEGF165 was digested with EcoR I and BamH
I, then a 495 bp fragment was recovered and inserted into the same site of pGBT9.
1.4 Cotransformation of a pair of plasmids to yeast cell
100 μL of competence yeast SFY526 prepared by the method described in ref. [5] was mixed
with a pair of plasmid DNA (the content of each plasmid was 5—10 μg) and, after ice-bath for 5
min in 0.2 cm electro-transformation cuvette, transformed under the condition of 1500 V, 25 μF,
200 Ω using Bio-Rad Gene Pulser electro-transformer, then 1.0 mL of sorbitol cooled by ice-bath
was added immediately. The mixture was spread to SD/-Leu-Trp (6.7 g/L YNB, 20 g/L D-glucose,
15 g/L agar, 20 mg/L L-adenine, 20 mg/L L-His, 30 mg/L L-Lys) selective culture plate and incu-
bated at 30℃ for 2—3 d.
1.5 Filter paper analysis of β-galactosidase activity
After the transformed colonies were cultured to the sizes of 1—2 mm, a single colony was
spread to a piece of aseptic filter paper on SD/-Leu-Trp selective culture plate and further incu-
bated at 30℃ for 1—2 d. The filter paper was frozen in liquid nitrogen and thawed at room tem-
perature, then placed onto another piece of filter paper presoaked in Z buffer/X-gal liquid [100 mL
Z buffer (16.1 g/L Na2HPO4·7H2O, 5.5 g/L Na2HPO4·2H2O, 0.75 g/L KCl, 0.246 g/L
MgSO4·7H2O, pH 7.0), 0.27 mL β-mercaptoethanol, 1.67 mL X-gal stock solution] and incu-
bated at 30℃ until blue colonies appeared.
1.6 Construction of recombinant plasmid of Pichia. pastoris system
After the pGEM-T/Flt-1 identified by DNA sequencing was digested with EcoR I and Not I,
the fragments with the length of 948 and 621 bp were recovered and inserted into the same sites of
pPIC9K for constructing recombinant plasmid pPIC9K/Flt-1(1-3) and pPIC9K/Flt-1(2-3) respec-
tively.
1.7 Transformation of linear plasmid into yeast cells
80 μL of competence yeast GS115 prepared by the method described in ref. [5] was mixed
with 5—10 μg of Bgl II linear plasmid and, after ice-bath for 5 min in 0.2 cm electro-transforma-
tion cup, transformed under the condition of 1500 V, 25 μF, 200 Ω using Bio-Rad Gene Pulser
electro-transformer, then 1.0 mL of sorbitol cooled by ice-bath was added immediately. The mix-

500
SCIENCE IN CHINA (Series C)
Vol. 44
ture was spread to MD/-His (13.4 g/L YNB, 4×10−4 g/L D-biotin, 20 g/L D-glucose, 15 g/L agar)
selective culture plate and incubated at 30℃ for 2—3 d. The growth of transformant was ob-
served.
1.8 Induction of expression
A single colony was inoculated in 100 mL of BMGY medium (10 g/L yeast extract, 20 g/L
pepton, 100 mmol/L potassium phosphate, pH 6.0, 13.4 g/L YNB, 4×10−4 g/L D-biotin, 1% glyc-
erol) and cultured at 30℃ to a A600 value of about 4.0. Bacterial strains were collected by cen-
trifugation, resuspended in 20 mL of BMMY medium (10 g/L yeast extract, 20 g/L pepton, 100
mmol/L potassium phosphate, pH 6.0, 13.4 g/L YNB, 4×10−4 g/L D-biotin, 1% methanol) and
cultured for 4 d, then the supernatant was separated by centrifugation and analyzed by SDS-PAGE.
Methanol was supplied to a final concentration of 1% every day during the culture.
1.9 Purification of recombinant protein
The supernatant was dialyzed and equilibrated in 20 mmol/L sodium acetate (pH 5.0), then
purified by CM-Sepharose FF chromatography and eluted by 0—1.0 mol/L NaCl linear gradient
elution. The eluted recombinant protein was desalted by dialysis, concentrated by ultrafiltration,
and further purified by Sephacryl S-100 chromatography, then eluted with 20 mmol/L sodium
acetate (pH 5.0). The component of each elution peak was analyzed by SDS-PAGE.
1.10 Assay of biological activity in vitro
1.10.1 Solid phase competition binding experiment. By the method described in ref. [4],
ELISA plate was coated with Flt-1(1 μg per well) at 4℃ for 18 h and washed with TBS (20
mmol/L Tris·HCl, 150 mmol/L NaCl, pH 7.5) for 3 times, then blocked with TBS containing 2%
bovine serum albumin at 37℃ for 1 h. After the 125I-hVEGF165 with a constant concentration (10
μg/L) and hVEGF165 with a gradually increasing concentration were added, the plate was incu-
bated at 37℃ for 2 h. The bound protein was dissolved with 0.3 mol/L NaOH and 1% SDS, and
its radioactivity was calculated by γ-counter.
1.10.2 Cell-based binding assay. By the method described in ref. [4], HUVEC was inoculated
to a 24-well plate (5×104 per well) and incubated at 37℃ for 72 h, then washed with binding
buffer (DMEM, 25 mol/L HEPES, 10% bovine serum albumin, pH 7.4) for 3 times. After the
125I-hVEGF165 with a constant concentration (1 μg/L) and Flt-1 with a gradually increasing con-
centration were added, the plate was placed at 4℃ for 2 h. The bound protein was dissolved with
0.3 mol/L NaOH and 1% SDS, and the radioactivity of it was calculated by γ-counter.
1.10.3 3H-Thymidine incorporation assay. By the method described in ref. [4], HUVEC was
inoculated to a 48-well plate (1×104 per well) and incubated at 37℃ overnight. After the me-
dium was replaced by the 1640 basal culture medium containing 1.7% FCS, the plate was further

No. 5
SCREENING OF HUMAN VEGF RECEPTOR Flt-1 DOMAIN
501
incubated at 37℃ for 24 h, then 5 μg/L hVEGF165 and the Flt-1 were added with a gradually
3
increasing concentration and incubated at 37℃ for 18 h. 1 μCi H-TdR was added, and the plate
was further incubated at 37℃ for 6 h. The cell monolayer was fixed with methanol, washed with
5% trichloroacetic acid and dissolved with 0.3 mol/L NaOH, then its radioactivity was calculated
by liquid scintillation.
1.11 Biological activity assay in vivo
Thirty adult male New Zealand white rabbits with an average body weight of 3.5 kg were
narcotized with Ketamine (50 mg/kg i.m.) and xylazine (5 mg/kg i.m.), and the left hind legs of
them were shaved and sterilized. Deep femoral artery and the branches of superficial femoral ar-
tery were amputated and ligated, total and superficial femoral arteries were resected. After opera-
tion, the animals were treated with Cefazolin (5 mg/kg i.m.) for 4 d. The animals were randomly
divided into 3 groups and intramuscularly injected with various drugs in both left and right hind
leg 10 d after operation. Negative control, positive control and experimental groups were injected
with 3 mL 0.9% normal saline + 1000 U heparin, 1 mg hVEGF165 + 3 mL 0.9% normal saline
+ 1000 U heparin and 1 mg of hVEGF165 + 1 mg of sFlt-1(2-3) + 3 mL 0.9% normal saline + 1000
U heparin respectively. One rabbit of each group was selected out and examined by angiography
10 d after treatment. After the animals were narcotized, 4F catheters were inserted into the sites 3
cm over abdominal aorta through left common carotid artery. 76% compound meglucamine dia-
trizoate was injected at a rate of 3 mL/s. Angiogram was taken 4 s after injection.
1.12 Statistical analysis
The results were analyzed by Student t test (n = 6).
2
Results
2.1 Identification of ligand binding domain of Flt-1
Each of the constructed pGAD424/plasmid was paired with pGBT9/hVEGF165 and trans-
formed to yeast strain SFY526, and the β-galactosidase activity of positive colonies was analyzed
by filter paper method. After coloration with X-gal, the colonies co-transformed with pTD1 and
pVA3 (positive control) turned blue only 15—30 min later, and those with pGAD424/Flt-1(1-3)
and pGBT9/hVEGF165 turned blue 2 h later, pGAD424/Flt-1(2-3) and pGBT9/hVEGF165 4 h later,
pGAD424/Flt-1(1-2) and pGBT9/hVEGF165 10 h later. However, no color reaction was observed
in the colonies co-transformed with pGAD424/Flt-1(2) and pGBT9/hVEGF165 even 30 h later. In
addition, none of the colonies transformed with single plasmid showed color reaction (fig.1). It
proved that Flt-1 loops 1-3, 2-3 and 1-2 could bind to VEGF. In the order of binding capacity, the
loops are 1-3, 2-3 and 1-2. Only loop 2 showed no binding capacity.
2.2 Expression and purification of loops 1-3 and 2-3
After being induced by 1% methanol for 4 d, the expressed Flt-1(1-3) and Flt-1(2-3) existed

502
SCIENCE IN CHINA (Series C)
Vol. 44
in supernatant in the form of soluble molecules and contained more than 60% of total protein. Af-
ter being purified by CM-Sepharose FF and Sephacryl S-100 chromatography, their purity reached
above 90%.
2.3 Biological activities of sFlt-1(1-3) and sFlt-1(2-3)
2.3.1 Solid phase competition binding experiment. With the increasing concentration of
hVEGF165, the 125I-hVEGF165 bound to the two kinds of sFlt-1 showed a dose-dependent de-
crease. It indicated that both sFlt-1(2-3) and sFlt-1(1-3) could bind to hVEGF165. No significant
difference was observed between the ED50 values of them (p>0.05). It proved that the binding
capacity of sFlt-1(2-3) was close to that of sFlt-1(1-3) (fig. 2).
Fig. 1. Filter paper analysis of interaction of VEGF and
Fig. 2. Solid phase competition binding experiment of
Flt-1s. 1, pTD1 and pVA3; 2, pGAD424/Flt-1(1-2) and
pGBT9/hVEGF165; 3, pGAD424/Flt-1(2-3) and pGBT9/
Flt-1s with hVEGF165. ◆, sFlt-1(1-3); ▲, sFlt-1(2-3).
hVEGF165; 4, pGAD424/Flt-1(1-3) and pGBT9/hVEGF165
5, pGAD424/Flt-1(2) and pGBT9/hVEGF165; 6, pGBT9
and pGAD424; 7, pGAD424/Flt-1(1-2); 8, pGAD424/
Flt-1(1-2); 9, pGAD424/Flt-1(2-3); 10, pGAD424/Flt-1
;
(1-3); 11, pGBT9/hVEGF165
.
2.3.2 Cell-based binding assay. With the increasing concentrations of the two kinds of sFlt-1,
the 125I-hVEGF165 bound to the Flt-1 receptors on the surface of HUVEC showed a
dose-dependent decrease. It indicated that both sFlt-1(2-3) and sFlt-1(1-3) could competitively
inhibit the binding of hVEGF165 to the Flt-1 receptors on the surface of HUVEC. No significant
difference was observed between the ED50 values of the two kinds of sFlt-1 (p>0.05). It showed
that the binding capacity of sFlt-1(2-3) was close to that of sFlt-1(1-3) (see fig. 3).
2.3.3 3H-thymidine incorporation assay. With the increasing concentrations of the two kinds of
3
sFlt-1, the incorporation of H-TdR showed a dose-dependent decrease. It indicated that both the
sFlt-1 could inhibit the DNA synthesis of HUVEC stimulated with hVEGF165. However, no sig-
nificant difference was observed between the ED50 values of the two kinds of sFlt-1 (p＞0.05 ) . It

No. 5
SCREENING OF HUMAN VEGF RECEPTOR Flt-1 DOMAIN
503
showed that the inhibiting effect of sFlt-1(2-3) was close to that of sFlt-1(1-3) (see fig. 4).
Fig. 4. Inhibiting effect of sFlt-1s on HUVEC proliferation
stimulated with hVEGF165 (3H-thymidine incorporation as-
say). ◆, sFlt-1(2-3); ▲, sFlt-1(1-3).
Fig. 3. Cell-based binding experiment of Flt-1s with hVEGF165
▲, sFlt-1(1-3); ◆, sFlt-1(2.3).
.
2.4 Activity test in animal model of acquired limb artery occlusion
Angiogram showed no significant collateral circulation in negative control group. The quan-
tity of regenerated blood vessels in positive control group 10 d after administration was signifi-
cantly larger than that in negative control group. Most of the regenerated blood vessels aggregated
in “intermediate region”. The diameters of former arteriole of the group were also significantly
larger than those of negative control. The arteriole were linearly extended to the distal portion and
connected to the aorta there. The formation of collateral circulation in experimental group was
significantly decreased. It proved that sFlt-1(2-3) could inhibit the formation of regenerated blood
vessels stimulated with hVEGF165 (fig. 5).
Fig. 5. Inhibiting effect of sFlt-1(2-3) on the formation of regenerate blood vessels stimulated by hVEGF165
(a) Negative control group; (b) positive control group; (c) sFlt-1(2-3) experimental group.
.
3
Discussion
VEGF and its receptor play special roles in stimulating the proliferation of HUVEC, so

504
SCIENCE IN CHINA (Series C)
Vol. 44
blocking the binding of VEGF to its receptor has been an important goal in inhibiting vasoforma-
tion. Suramin can selectively bind to heparintropic VEGF and inhibit it from binding to its recep-
tor. In this way, the proliferation and migration of HUVEC stimulated with VEGF are inhibited. At
present, the phase I clinical trial of the drug has been carried out. Many studies proved that murine
or human monoclonal antibodies to VEGF could inhibit the growth and metastasis of tumors in
—
vivo of animals[6 9]. Milauer et al.[10] proved that tumor growth in nude mice was inhibited by us-
ing a retrovirus encoding dominant-negative Flk-1 mutant to infect the endothelial target cells.
Recently, the application of soluble receptor sKDR provided a novel route for blocking the bind-
ing of VEGF to endogenous receptors[11]
.
However, natural soluble receptors with large molecular weights often show obvious adverse
reactions along with good curative effects. The soluble domain receptors prepared by gene engi-
neering technique show small molecular weight, mild adverse reaction and low immunogenicity
and are easy to take effect in vivo. So it is of great value to develop a soluble domain with high
biological activity and low molecular weight. The domain should be able to bind to VEGF and
suitable for large-scale production.
It has been reported that the binding capacity of Flt-1(1-3) fragment to VEGF was close to
that of the whole Flt-1 molecule. In this paper, the Flt-1 domain for ligand binding was further
studied using yeast two hybridization system. It showed that the binding capacity of Flt-1(2-3)
was close to that of Flt-1(1-3). The result, which has not been reported, provided a basis for de-
veloping new VEGF domain receptors. Yeast two-hybrid system is characterized by short experi-
mental cycle and high screening flux and easy to handle. It provided a novel route for primary
screening of ligand binding domain with various affinities. The established technique platform can
also be used for screening the domain of other receptors, such as CTLA4.
The screened Flt-1 domain was expressed in Pichia. pastoris and purified, then detected for
the activity of blocking the binding of VEGF and its receptor. Both sFlt-1 (2-3) and sFlt-1(1-3)
showed strong blocking activities, and no significant difference was observed between them. The
result was in agreement with that obtained by using double hybridization system, more important
still, sFlt-1(2-3) could significantly inhibit the formation of regenerated blood vessels in vivo. The
study laid a foundation of application of sFlt-1(2-3).
In addition, the soluble VEGF domain receptor sFlt-1(2-3) with biological activity was
highly expressed in Pichia. pastoris system. It provides an industrialization foundation for devel-
oping a novel drug for treating vasoformation related diseases, such as tumor and diabetic reti-
nopathy.
Acknowledgements This work was supported by National High Technology Programs of China (Grant No.
102-08-01-03) and Natural Science Fund Program of Guangdong Province (Grant No.001098).
References
1. Folkman, J., Angiogenesis in cancer, vascular, rheumatoid and other disease, Nat. Med., 1995, 1: 27—31.

No. 5
SCREENING OF HUMAN VEGF RECEPTOR Flt-1 DOMAIN
505
2. Pan, X., Cao, G. W., Ke, C. W. et al., Structure and function of vascular endothelial growth factor and its related proteins,
Acta Biochimica et Biophysica Sinica, 1999, 31(4): 357—361.
3. Barleon, B., Totzke, F., Herzog, C. et al., Mapping of sites for ligand binding and receptor dimerization at the extracellular
domain of the vascular endothelial growth factor receptor Flt-1, J. Biol. Chem., 1997, 272(16): 10382—10388.
4. Davis-Smyth, T., Chen, H., Park, J. et al., The second immunoglobulin-like domain of the VEGF tyrosine kinase receptor
Flt-1 determines ligand binding and may initiate a signal transduction cascade, EMBO, 1996, 15(18): 4919—4927.
5. Scorer, C. A., Clare, J. J., Mccombie, W. R. et al., Rapid selection using G418 of high copy number transformants of
Pichia Pastoris for high-level foreign gene expression, Bio/Technology, 1994, 12: 181—184.
6. Kim, K. J., Li, B., Winer, J. et al., Inhibition of vascular endothelial growth factor-induced angiogenesis suppresses tumor
growth in vivo, Nature, 1993, 362: 841—844.
7. Torpe, P., Antibody-directed targeting of tumor vasculature, Proc. Am. Assoc. Cancer Res., 1996, 37: 668—671.
8. Presta, L. G., Chen, H., O’Connor, S. J. et al., Humanization of an anti-vascular endothelial growth factor monoclonal an-
tibody for the therapy of solid tumors and the disorders, Cancer Research, 1997, 57: 4593—4599.
9. Borgstrom, P., Bourdon, M. A., Hillan, K. J. et al., Neutralizing anti-vascular endothelial growth factor antibody com-
pletely inhabits angiogenesis and growth of human prostate carcinoma micro tumors in vivo, Prostate, 1998, 35: 1—10.
10. Millauer, B., Loghi, M., Plate, K. et al., Dominant-negative inhibition of Flk-1 suppresses the growth of many tumor types
in vivo, Cancer Res., 1996, 56: 1615—1620.
11. Lin, P., Sanker, S., Shan, S. Q. et al., Inhibition of tumor growth by targeting tumor endothelium using a soluble vascular
endothelial growth factor receptor, Cell Growth & Differentiation, 1998, 9: 49—58.