(Received for publication, August 29, 1996, and in revised form, December 23, 1996)
From the Department of Biochemistry, Kanazawa
University School of Medicine, Kanazawa 920, Japan, the
§ Discovery Research Laboratory 3, Kissei Pharmaceutical Co.
Ltd., Hotaka 399-83, Japan, the ¶ Institute of Biological
Science, Mitsui Pharmaceuticals Inc, Mobara 297, Japan, and the
Life Science Laboratory, Mitsui Toatsu Chemicals Inc.,
Mobara 297, Japan
This study was undertaken to determine whether and how advanced glycation end products (AGE), senescent macroproteins accumulated in various tissues under hyperglycemic states, cause angiogenesis, the principal vascular derangement in diabetic microangiopathy. We first prepared AGE-bovine serum albumin (BSA) and anti-AGE antiserum using AGE-RNase A. Then AGE-BSA was administered to human skin microvascular endothelial cells in culture, and their growth was examined. The AGE-BSA, but not nonglycated BSA, was found to induce a statistically significant increase in the number of viable endothelial cells as well as their synthesis of DNA. The increase in DNA synthesis by AGE-BSA was abolished by anti-AGE antibodies. AGE-BSA also stimulated the tube formation of endothelial cells on Matrigel. We obtained the following evidence that it is vascular endothelial growth factor (VEGF) that mainly mediates the angiogenic activities of AGE. (1) Quantitative reverse transcription-polymerase chain reaction analysis of poly(A)+ RNA from microvascular endothelial cells revealed that AGE-BSA up-regulated the levels of mRNAs for the secretory forms of VEGF in time- and dose-dependent manners, while endothelial cell expression of the genes encoding the two VEGF receptors, kinase insert domain-containing receptor and fms-like tyrosine kinase 1, remained unchanged by the AGE treatment. Immunoprecipitation analysis revealed that AGE-BSA did increase de novo synthesis of VEGF. (2) Monoclonal antibody against human VEGF completely neutralized both the AGE-induced DNA synthesis and tube formation of the endothelial cells. The results suggest that AGE can elicit angiogenesis through the induction of autocrine vascular VEGF, thereby playing an active part in the development and progression of diabetic microangiopathies.
Glucose and other reducing sugars can react nonenzymatically with the amino groups of proteins to form reversible Schiff bases and, then, Amadori products. These early glycation products undergo further complex reactions such as rearrangement, dehydration, and condensation to become irreversibly cross-linked, heterogeneous fluorescent derivatives termed advanced glycation end products (AGE)1 (1). The formation and accumulation of AGE in various tissues have been known to progress during normal aging and at an extremely accelerated rate in diabetes mellitus. This has been implicated in the development of diabetic micro- and macro-vascular complications (1), which may account for the disabilities and high mortality rate in patients with this disease (2).
Microvessels are composed of only two types of cells, endothelial cells and pericytes, and have been known to show both functional and structural abnormalities during prolonged diabetic exposure, resulting in the deleterious effects on the organs that they supply (3-5). Using pericyte-endothelial cell co-culture systems, we have shown previously that pericytes can not only regulate the growth but also preserve the prostacyclin-producing ability and protect against lipid peroxide-induced injury of endothelial cells (6). This has provided a basis for understanding how diabetic retinopathy develops consequent to "pericyte loss," the earliest histopathological hallmark in diabetic retinopathy (5, 7).
Recently, we have found that AGE exert a growth inhibitory effect and a cell type-specific immediate toxicity on pericytes through interactions with their receptor for AGE (RAGE), a cell surface receptor belonging to the immunoglobulin superfamily (8), and have proposed a novel mechanism for pericyte loss (9). The AGE-induced, RAGE-mediated decrease in pericyte number would then indirectly cause angiogenesis (6, 9).
In the present study, we investigated the effects of AGE on the growth and tube formation of human skin microvascular endothelial cells, the key steps of angiogenesis. We demonstrate that AGE exert angiogenic activities directly on microvascular endothelial cells and that autocrine vascular endothelial growth factor (VEGF) is the major mediator of the AGE-driven angiogenesis.
Bovine serum albumin (BSA) was purchased from
Boehringer Mannheim GmbH (Mannheim, Germany). Bovine pancreatic RNase
A, bovine hemoglobin (Hb),
N-tosyl-lysine methyl ester and
phenylmethylsulfonyl fluoride were from Sigma.
Heparin-Sepharose CL-4B was from Pharmacia LKB (Uppsala, Sweden).
Horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG was from
BioMakor (Rehovot, Israel). [3H]Thymidine and
[
-32P]ATP were from DuPont NEN. Reverse transcriptase
and T4 polynucleotide kinase were from Takara (Kyoto, Japan).
Hybond-N+ nylon membrane was from Amersham Corp.
(Buckinghamshire, United Kingdom). Matrigel was from Collaborative
Research (Bedford, MA).
BSA
(fraction V, fatty acid-free, free endotoxin) was incubated with 0.5 M glucose at 37 °C for 6 weeks under sterile conditions in the presence of 1.5 mM phenylmethylsulfonyl fluoride,
0.5 mM EDTA, 100 units/ml penicillin, and 40 µg/ml
gentamycin (9-11). After unincorporated sugars were removed by
dialysis against phosphate-buffered saline, glucose-modified high
molecular weight materials were purified by heparin-Sepharose CL-4B
column chromatography and used as AGE-BSA. Control nonglycated BSA was
incubated in the same conditions except for the absence of glucose. The
concentration of AGE-BSA was determined by the method of Bradford (12).
AGE-RNase A and AGE-Hb were prepared according to the method of Makita
et al. (13). For reducing AGE-Hb, NaBH4 was
employed as described by Horiuchi et al. (14).
1-Deoxy-1-propylamino-D-fructose, an Amadori compound, was
synthesized by the method of Micheel and Hagemann (15).
N-Tosyl-N
-carboxymethyllysine
methyl ester (CM-TsLME) was synthesized from N
-tosyl-lysine methyl ester by the
method of Ahmed et al. (16) with minor modifications.
N-Carboxymethylated BSA (CM-BSA) was prepared according to
the method of Reddy et al. (17).
1 mg of AGE-RNase A was emulsified in 50% Freund complete adjuvant and injected intradermally into rabbits. Two weeks later, a booster with the same amount of AGE-RNase A was administered, followed by nine additional booster injections, with one given every 2-3 weeks. Ten days after the final injection, the antiserum was obtained.
Enzyme-linked Immunosorbent Assay (ELISA)In the noncompetitive ELISA system, wells of 96-well microtiter plates were coated with increasing amounts of Hb, AGE-Hb, and reduced AGE-Hb. After washing and blocking, the wells were incubated with 100 µl of anti-AGE antiserum (1:4000) for 2 h and then with 100 µl of HRP-conjugated goat anti-rabbit IgG (1:2000) for 30 min. Finally, 100 µl of substrate tetramethylbenzidine solution was added into each well. After 10-15 min, the absorbance at 450 nm was measured. In the competitive ELISA system, procedures similar to the noncompetitive ELISA were used except for the following two points. Wells were first coated with reduced 100 ng/ml AGE-Hb solution as absorbent antigens, and then 50 µl of test samples were added as a competitor together with 50 µl of anti-AGE antiserum (1:2000) into each well.
CellsEndothelial cells from human skin microvessels were maintained in E-BM medium supplemented with 5% fetal bovine serum, 0.4% bovine brain extracts, 10 ng/ml human epidermal growth factor, and 1 µg/ml hydrocortisone according to the supplier instructions (Clonetics Corp., San Diego, CA). Cells at 5-10 passages were used for the experiments. AGE treatment was carried out in a medium lacking epidermal growth factor and hydrocortisone.
Measurement of Cell GrowthEndothelial cells cultured for various time periods in the presence or absence of AGE-BSA were dislodged with trypsin, and counted by the dye exclusion method (18). [3H]Thymidine incorporation was determined as described previously (19). For determining the effects of anti-AGE antiserum on endothelial cell growth, the antiserum was added to the medium at 1% (v/v) together with or without 50 µg/ml AGE-BSA, after which cells were incubated for 24 h and [3H]thymidine incorporation was measured.
Primers and ProbesOligonucleotide primers and probes for
quantitative reverse transcription-polymerase chain reactions (RT-PCR)
were synthesized by a Perkin-Elmer 392 DNA synthesizer (Foster City,
CA) and purified as described previously (20). Primer sequences and
internal oligonucleotide probes for detecting VEGF, fms-like
tyrosine kinase 1 (flt 1), kinase insert domain-containing
receptor (kdr), and -actin mRNA were the same as
described in Ref. 20.
Poly(A)+ RNAs were isolated (21) from cells treated with or without AGE-BSA for the various time periods and analyzed by RT-PCR as described previously (22). 6-µl aliquots of each RT-PCR reaction mixture were electrophoresed on a 2% agarose gel and transferred to a Hybond-N+ nylon membrane, and the membrane was hybridized with the respective 32P-end-labeled probes (20). The amounts of poly(A)+ RNA templates (30 ng) and cycle numbers (30 cycles) for amplification were chosen in quantitative ranges where reactions proceeded linearly, which had been determined by plotting signal intensities as functions of the template amounts and cycle numbers (20). Signal intensities of hybridized bands were measured by a Fujix BAS 1000 Image analyzer (Fuji Photo Film Co. Ltd., Hamamatsu, Japan).
Metabolic Labeling and Immunoprecipitation of VEGFSubconfluent cultures of endothelial cells were incubated in the presence or absence of 50 µg/ml AGE-BSA for 2 h and further incubated for 6 h at 37 °C in methionine-free RPMI 1640 medium/complete RPMI 1640 medium (9:1) containing 0.2 mCi/ml [35S]methionine with or without AGE-BSA. Immunoprecipitation was carried out as described previously (20). The samples were analyzed on a 15% SDS-polyacrylamide gel under reducing conditions.
Preparation of Monoclonal Antibody (MoAb) against Human VEGFPeripheral blood lymphocytes from healthy volunteers were immortalized by Epstein-Barr virus, and the resultant Epstein-Barr virus-transformants were screened (23) for their ability to produce antibody molecules that bind to recombinant human VEGF165 (Peprotech, Rocky Hill, NJ). The candidate cells were then fused with SHM-D33 human myeloma cells (American Type Culture Collection, Rockville, MD). A cell line that consistently produced an IgM-MoAb reactive to VEGF was cloned and designated BL-2. Then the BL-2 MoAb was purified from the serum-free supernatant by 50% ammonium sulfate precipitation and by Sephadex G-200 gel chromatography. The specific binding of MoAb BL-2 to VEGF was confirmed by immunoblotting.2
Tube Formation in VitroWells of 24-well culture cluster dishes (Coster 3524, Cambridge, MA) were coated with Matrigel solution (250 µl/well) and then allowed to solidify for at least 1 h at 37 °C (24, 25). Endothelial cells (4 × 104 cells/well) were then seeded on Matrigel with or without 10 µg/ml MoAb BL-2. After 30 min, the cells were treated with or without 50 µg/ml AGE-BSA for 6 h, 4 microscopic fields selected at random were photographed, and the lengths of tube-like structures were measured with microcomputer-assisted NIH Image (Version 1.56).
AGE-BSA was prepared by
incubating BSA with glucose and then purified by heparin-Sepharose
CL-4B column chromatography. Fig. 1 shows its
electrophoretic profile on reducing SDS-polyacrylamide gel
electrophoresis. Control nonglycated BSA migrated to the position at 68 kDa. On the other hand, the purified materials migrated much more
slowly, yielding a broad band larger than 68 kDa. This indicated that
covalently linked adducts were formed nonenzymatically on BSA without
discernible degradation. As shown in Table I, the
purified materials also exhibited spectrophotometric features characteristic to AGE (26, 27). The peak fluorescence was noted at 440 nm with excitation at 370 nm, and its intensity was increased 10-fold
in comparison with nonglycated BSA. Chromogen products also appeared in
the purified materials, whereas they were barely detectable in
nonglycated BSA. Based on these observations, the purified materials
were used as AGE-BSA.
|
As a tool to evaluate
AGE and their biological effects, an antiserum was raised against
AGE-RNase A. The reactivity of this antiserum to several AGE-modified
proteins, Amadori compounds, and carboxymethyllysine derivatives was
examined. Since it was possible that AGE-modified proteins might
contain early glycation products, the immunoreactivity of the antiserum
to reduced AGE-Hb was also tested because the early glycation products
are known to be converted to glucitol-lysine by reduction with
NaBH4 (14, 28). As shown in Fig.
2A, there was no difference in the
immunoreactivity in the noncompetitive ELISA between AGE-Hb and reduced
AGE-Hb. Further, as shown in Fig. 2B, the antiserum binding
to AGE-Hb was not competed for by
1-deoxy-1-propylamino-D-fructose in the competitive ELISA,
indicating that the antiserum does not recognize Amadori compounds. We
next tested whether the antiserum reacted to CM-TsLME and CM-BSA
in the competitive ELISA. As shown in Fig. 2, B and
C, these glycoxidative products were found to partially inhibit the antiserum binding to reduced AGE-Hb, whereas AGE-Hb and
AGE-RNase A fully inhibited its binding (Fig. 2D). Further, the antiserum reactivity to AGE-BSA was dependent on the duration of
incubation of BSA with glucose (Fig. 2E). These results
suggested that the antiserum could recongnize AGE structures common to
various AGE preparations.
Stimulation of the Growth of Microvascular Endothelial Cells by AGE
Endothelial cells obtained from human skin microvessels were
cultured in the presence or absence of AGE-BSA, and the viable cell
number was determined at days 1, 2, and 3 after the AGE addition. As
shown in Fig. 3, AGE-BSA was found to increase the
viable microvascular endothelial cell number in a
dose-dependent manner; at 50 µg/ml AGE, there was about a
40% increase in viable cell number. Moreover, AGE-BSA significantly
increased DNA synthesis in microvascular endothelial cells to 130%
(p < 0.01, Fig. 4). However,
nonglycated BSA induced no change in either the cell number or DNA
synthesis.
Neutralization of the AGE-induced DNA Synthesis by Anti-AGE Antibody
To evaluate the specificity of the AGE-BSA effect on endothelial cell growth, we examined the effects of the antiserum against AGE-RNase A on AGE-induced DNA synthesis. As shown in Fig. 4, the anti-AGE-RNase A antiserum was found to completely neutralize the AGE-induced synthesis of endothelial cell DNA at 1%, while the same concentration of the antiserum did not affect DNA synthesis in endothelial cells not exposed to AGE-BSA.
Microvascular Endothelial Cells Express mRNA for Secretory Forms of VEGF in Response to AGEPoly(A)+ RNAs were isolated from microvascular endothelial cells treated with various concentrations of AGE-BSA for various time periods, and analyzed by a quantitative RT-PCR technique to determine the effects of AGE on the expression of the VEGF gene. It has been reported that there are four alternatively spliced products from the single VEGF gene, VEGF121, VEGF165, VEGF189, and VEGF206 (29, 30). Since Northern blot analysis cannot clearly discriminate the four mRNA species, we employed a more sensitive RT-PCR technique as described previously (20). In this experiment, 486- and 618-base pairs (bp)-long cDNA products would be amplified from mRNAs for VEGF121 and VEGF165, respectively (20).
As shown in Fig. 5, A and B,
microvascular endothelial cells were expressing mRNAs for
VEGF121 and VEGF165, the secretory forms of
VEGF. When the endothelial cells were exposed to AGE-BSA, the level of
VEGF mRNAs was found to be significantly increased in a time- and
dose-dependent manner. The VEGF mRNA level began to
increase at 2 h, and reached a maximum at 4 h in the presence of 50 µg/ml AGE-BSA; the peak value was 3-fold higher than the basal
level when standardized with the signal intensities of -actin mRNA as an internal control. Maximal stimulation was achieved at
50-100 µg/ml AGE. However, the larger alternatively spliced products
coding for VEGF189 and VEGF206 were not
detected in microvascular endothelial cells regardless of the presence
or absence of AGE-BSA.
To confirm whether AGE-BSA increased the synthesis of VEGF proteins, we performed immunoprecipitation using the anti-VEGF monoclonal antibody (20) from lysates of the cells that had been treated with or without AGE-BSA. As shown in Fig. 5C, 35S-labeled proteins that migrated to the positions of 18 and 22 kDa, corresponding to VEGF121 and VEGF165, respectively, were immunoprecipitated, and the amounts of these proteins were found to be increased to about 2.5-fold by the AGE treatment.
VEGF Receptor Expressions in Microvascular Endothelial CellsVEGF exerts its biological actions through its specific receptors, KDR and Flt 1 (31, 32). We then determined the types of VEGF receptors expressed in microvascular endothelial cells and whether their expressions could be altered by AGE-BSA. As shown in Fig. 5B, both kdr and flt 1 mRNA were detected in microvascular endothelial cells, and the content of kdr mRNA was more abundant than that of flt 1 mRNA. In contrast to VEGF mRNAs, the levels of the two types of receptors were essentially unchanged by the exposure to AGE-BSA.
Neutralization of the AGE-Induced DNA Synthesis of Microvascular Endothelial Cells by MoAb against Human VEGFWe next investigated whether vascular VEGF may have a
functional role in the AGE action on endothelial cells. Microvascular endothelial cells were preincubated with various concentrations of MoAb
BL-2 for 30 min, then exposed to 50 µg/ml AGE-BSA for 24 h in
the presence of the antibody, and assayed for
[3H]thymidine incorporation. As shown in Fig.
6, MoAb BL-2 was found to significantly diminish the
AGE-induced increase in DNA synthesis in a dose-dependent
manner; at 10 µg/ml, a complete reversal was obtained. The MoAb alone
did not affect DNA synthesis in endothelial cells not exposed to
AGE.
AGE Induction of Tube Formation of Microvascular Endothelial Cells and Its Inhibition by Anti-VEGF MoAb
The process of angiogenesis
has been assumed to be completed by the formation of microvascular
tubes (19). In vitro assays for tube formation of
endothelial cells have been developed and used to study this crucial
step of angiogenesis. Accordingly, we examined whether AGE affect
in vitro tube formation of microvascular endothelial cells.
For this, we employed an on-gel assay system using Matrigel, in which
endothelial cells take only several hours to associate with each
together and form microtubes. Microvascular endothelial cells were
seeded on Matrigel with or without AGE-BSA, and tube formations were
judged after 6 h. As shown in Fig. 7A, AGE-BSA was found to double the length of the tubes of endothelial cells formed on Matrigel; and, the AGE-induced tube formation was
inhibited by BL-2 MoAb as was the AGE-induced DNA synthesis of
endothelial cells. The MoAb per se did not affect the tube formation. Fig. 7B shows typical micrographs; with 10 µg/ml BL-2 MoAb, AGE-induced tube formation was markedly
inhibited.
In the present study, we have demonstrated for the first time that AGE, nonenzymatically glycated protein derivatives formed under hyperglycemia, stimulate the growth and tube formation of human microvascular endothelial cells, the key steps of angiogenesis which take place in this very cell type (33, 34). The present findings have extended our preliminary work employing endothelial cells from a larger vessel, i.e. the umbilical vein. Though modestly, AGE-BSA caused a consistent increase in cell number and in DNA synthesis of both umbilical and microvasular endothelial cells, the same AGE-BSA concentration (50 µg/ml) giving the maximal effect in both cases (35). This concentration of AGE was comparable with that of the in vivo situation in diabetes; Makita et al. reported (13) that human serum AGE levels were elevated more than 2-fold in diabetic patients (about 25 µg/ml) and almost 8-fold in diabetic patients on hemodialysis (about 80 µg/ml) in comparison with that in normal patients.
That it was AGE moieties that elicited the angiogenic activity was evidenced as follows. First, the AGE-BSA employed exhibited the biochemical hallmarks of AGE (Fig. 1 and Table I). Second, control nonglycated BSA made no change (data not shown). Third, a newly developed antiserum against AGE-RNase A could neutralize the growth-promoting effect of AGE-BSA (Fig. 4). Although this antiserum was partially reactive to carboxymethyllysine, it would seem unlikely that the AGE effect could be accounted for by such glycoxidative biproducts of the Maillard reaction, which might be present in trace amounts in the AGE-BSA preparation, because authentic carboxymethylated BSA added to the culture medium at concentrations from 1 to 100 µg/ml failed to stimulate the endothelial cell synthesis of DNA (data not shown). We speculate that the AGE actions on microvascular endothelial cells may require their binding to RAGE, the AGE-specific receptor, as is the case with bovine retinal pericytes (9), human umbilical vein endothelial cells (35), and human pancreatic cancer cells (36).
The present study has also demonstrated that the angiogenic activity of
AGE is mainly mediated by autocrine VEGF synthesized by micrcovascular
endothelial cells per se. mRNAs for VEGF121 and VEGF165 are present in microvascular endothelial cells,
and their levels are up-regulated by AGE in both dose- and
time-dependent manners (Fig. 5, A and
B). AGE did increase de novo synthesis of VEGF in
endothelial cells (Fig. 5C). mRNAs for the two VEGF receptors, kdr and flt 1, were also detected in
human skin microvascular endothelial cells, their relative abundance
being kdr flt 1. However, their levels were
essentially unaltered when exposed to AGE (Fig. 5B). This
suggests that the ligand expression should be the rate-limiting step in
the putative autocrine action of VEGF. In effect, the neutralization
experiments established the functional role of VEGF in the AGE-induced
endothelial cell growth and tube formation. MoAb against human VEGF
could completely inhibit the AGE-induced tube formation as well as the
DNA synthesis of microvascular endothelial cells (Figs. 6 and 7). Since
the basal growth or tube formation in unexposed cells was not affected
by the same concentration of the antibody, the antibody-induced
inhibition is not likely the result of its toxic or nonspecific
effects. These results thus indicate that autocrine VEGF is the main
mediator of the AGE-driven angiogenesis in vitro.
In light of the present findings, together with the previous
observations, we can now posit an overall scheme concerning the roles
of AGE in the development of diabetic microangiopathy (Fig. 8). First, AGE act on pericytes, the microvascular
constituent that encircles the endothelium. Through interactions with
RAGE, AGE decrease the number of this cell type (9), leading to
pericyte dropout, which would in turn relieve the restriction on
endothelial cell replication and facilitate angiogenesis. The resultant
cessation of pericyte-endothelial cell interactions would impair
prostacyclin production (6), which would cause thrombogenesis. Second,
AGE act on endothelial cells, which serve as a barrier between
circulating blood and parenchyma and produced various vasoactive
substances. As shown in this paper, one consequence is angiogenesis
that is probably mediated by autocrine VEGF. In addition, the
prostacyclin-synthesizing ability of microvascular endothelial cells
might be directly inhibited by AGE, as in umbilical endothelial cells
(35). Again, the consequence of this would be thrombogenesis.
Diminished circulation or microthrombus formation may occur in such
lesions, giving rise to hypoxia, the major factor triggering VEGF
expression in both endothelial cells and pericytes (20, 37-40). In
such circumstances, angiogenesis would further proceed, which may
eventually lead to the clinical expression of diabetic
microangiopathies, exemplified by proliferative retinopathy. According
to this model, procedures that can halt those events, e.g.
inhibition of AGE formation,3 AGE
absorption by immobilized antibodies, antisense RAGE DNA (35, 36),
prostacyclin analogues, or anti-VEGF neutralizing antibodies would then
theoretically help circumvent the development and progression of
diabetic microangiopathies.
Although this view basically stands on the in vitro experiments, in support are several in vivo or clinical observations. Hammes et al. (41, 42) reported that aminoguanidine, an inhibitor of AGE formation, prevented AGE accumulation at branching sites of precapillary arterioles and that this inhibitor could diminish pericyte dropout and inhibit abnormal endothelial cell proliferation in streptozotocin-induced diabetic rats. Dolhofer-Bliesener et al. (43) have shown that, in human diabetic subjects, the serum level of AGE was associated with the state of late complications, particularly in cases with retinopathy. Wautier et al. (44) reported that infusion of diabetic red blood cells into normal rats can induce vascular hyperpermeability, which was completely inhibited by anti-RAGE IgG; they also demonstrated that an antioxidant, probucol, can similarly reverse the red blood cell transfer-induced vascular permeability, suggesting a role of AGE-RAGE interactions and the involvement of an oxidant stress-sensitive pathway in the development of hyperpermeability. Recent clinical studies at different institutions have established that intraocular concentrations of VEGF correlated with active neovascularization (45, 46). Further, the increased permeability of retinal capillaries and the breakdown of the blood-retinal barrier have been shown to be as an early event in both human diabetic subjects and streptozotocin diabetic rats followed by new vessel formation (47, 48). The barrier breakdown would help ensure pericyte access to AGE. Moreover, it is reported that the barrier breakdown could be protected by the treatment of aminoguanidine, suggesting that AGE themselves may be directly involved in the barrier breakdown (48).
We thank Professor Kari Alitalo, Helsinki University and Dr. Ulf Eriksson, Ludwig Institute for Cancer Research for kindly providing VEGF-B and VEGF-C, Kurabo Industries, Ltd., Osaka for human capillary endothelial cells, Shin-ichi Matsudaira and Reiko Kitamura for assistance, and Brent Bell for reading the manuscript.