Vascular endothelial growth factor induces heparin-binding epidermal growth factor-like growth factor in vascular endothelial cells.

Although several cytokines and growth factors have been shown to regulate vascular endothelial growth factor (VEGF) production, little is known about how VEGF may regulate growth factors that have known mitogenic and chemotactic actions on mesenchymal cells (which are involved in the maturation of the angiogenic process). We investigated the effect of VEGF on heparin-binding epidermal growth factor-like growth factor (HB-EGF) expression in human umbilical vein endothelial cells. HB-EGF mRNA was induced by 8-fold after 2 h of VEGF stimulation, and it returned to base line within 6 h. VEGF did not alter the half-life of HB-EGF mRNA (55 min). Nuclear run-on experiments showed a 4.9-fold increase in HB-EGF gene transcription within 2 h of VEGF stimulation, and Western analysis demonstrated an associated increase in cellular HB-EGF protein. We found that platelet-derived growth factor-BB (PDGF-BB) mRNA was also induced 3-fold after 5 h of VEGF stimulation, whereas neither endothelin 1 nor transforming growth factor-beta1 was regulated by VEGF. Finally, conditioned medium from VEGF-stimulated endothelial cells produced an increase in DNA synthesis in vascular smooth muscle cells, and this effect was blocked by a neutralizing antibody to PDGF. The induction of HB-EGF and PDGF-BB expression in endothelial cells may represent the mechanism by which VEGF recruits mesenchymal cells to form the medial and adventitial layers of arterioles and venules during the course of angiogenesis.

Heparin-binding epidermal growth factor-like growth factor (HB-EGF) is a 22-kDa protein that was originally purified from the conditioned medium of human macrophage-like U-937 cells (22). The COOH terminus of HB-EGF contains six cysteine residues with spacing characteristic of members of the EGF family, and HB-EGF shares 40% sequence identity with EGF. HB-EGF binds to the EGF receptor and triggers its downstream signal transduction pathways (23). HB-EGF is mitogenic and chemotactic for several primary cell types, such as vascular smooth muscle cells, keratinocytes, fibroblasts, and cardiomyocytes, but not for vascular endothelial cells. Vascular endothelial cells do express HB-EGF, however, and this expression can be induced by TNF-␣ and shear stress (24 -27).
HB-EGF is induced at the site of blastocyst implantation in the mouse uterus (28), a site of intensive angiogenesis. Upregulation of HB-EGF expression also occurs in other pathological states that require angiogenesis, such as wound healing, and in human atherosclerotic plaques (29,30). Because VEGF is a critical factor in the development of new blood vessels, we speculated that VEGF may induce growth factors in vascular endothelial cells, which then act in a paracrine fashion to promote angiogenesis. We studied the regulation by VEGF of HB-EGF, PDGF-BB, ET1, and TGF-␤1 in human umbilical vein endothelial cells (HUVEC) and human coronary artery endothelial cells (HCAEC). HB-EGF and PDGF-BB mRNA were up-regulated by VEGF in distinct temporal patterns. Moreover, the increase in HB-EGF mRNA was regulated transcriptionally and translated to an increase in accumulation of cellular HB-EGF protein.

EXPERIMENTAL PROCEDURES
Cell Culture and mRNA Isolation-Primary culture HUVEC, HCAEC, and HASMC were obtained from Clonetics (San Diego, CA). HUVEC and HCAEC were grown in M199 (JRH Biosciences, Lenexa, KS) supplemented with 20% fetal calf serum (HyClone Laboratories, Logan, UT), 100 units/ml penicillin, 100 g/ml streptomycin, 100 g/ml heparin sulfate (Sigma), and 50 g/ml endothelial cell growth supplement (Collaborative Biomedical Products, Bedford, MA). HASMC were grown in M199 supplemented with 20% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were passaged every 3-5 days, and cells from passages 6 -8 were used in the experiments described here. After the cells had grown to 80 -90% confluence, they were placed in M199 medium supplemented with 5% fetal calf serum for 1 h before the experiments. Recombinant human VEGF (Collaborative Biomedical Products) was dissolved in PBS (JRH Biosciences) and stored at Ϫ80°C until use. Total RNA from cells in culture was prepared by guanidinium isothiocyanate extraction and centrifugation through cesium chloride.
Probes for Northern Analysis-An HB-EGF cDNA fragment from HUVEC RNA was amplified by the reverse transcription-polymerase chain reaction (31). The human PDGF-BB probe was obtained similarly by reverse transcription-polymerase chain reaction using forward (5Ј-CGTCTGGTCAGCGCCGAGGGG-3Ј) and reverse (5Ј-CGTCTTGTCAT-GCGTGTGCTT-3Ј) primers. The ET1 probe was obtained as described by Bloch et al. (32). The TGF-␤1 probe was a gift from G. S. Hotamisligil (Harvard School of Public Health, Boston, MA).
Northern Analysis-Total RNA (10 g) from cells in culture was fractionated on 1.3% formaldehyde-agarose gels and transferred to nitrocellulose filters. The filters were hybridized with a randomprimed 32 P-labeled HB-EGF cDNA probe as described (31). In some experiments, filters were also hybridized with PDGF-BB, ET1, and TGF-␤1 probes. The hybridized filters were then washed in 30 mM sodium chloride, 3 mM sodium citrate, and 0.1% sodium dodecyl sulfate at 55°C and autoradiographed with Kodak XAR film at Ϫ80°C for 12-24 h or stored on phosphor screens for 6 -8 h. To correct for differences in RNA loading, the filters were rehybridized with a radiolabeled 18 S oligonucleotide (5Ј-ACGGTATCT-GATCGTCTTCGAACC-3Ј) probe. The filters were scanned and radioactivity was measured on a PhosphorImager running ImageQuant software (Molecular Dynamics, Sunnyvale, CA).
Nuclear Run-on Analysis-Subconfluent HUVEC were not stimulated (control) or stimulated with VEGF for 2 h. The cells were subsequently lysed, and nuclei were isolated as described by Perrella et al. (25). Nuclear suspension (200 l) was incubated with 0.5 mM each of CTP, ATP, and GTP and with 125 Ci of 32 P-labeled UTP (3,000 Ci/mmol, NEN Life Science Products). The samples were extracted with phenol/chloroform, precipitated, and resuspended at equal counts/ min/ml in hybridization buffer (1.4 ϫ 10 6 cpm/ml). Radiolabeled run-on transcripts were hybridized (40°C) to denatured probes (HB-EGF and 18 S, 1 g each) dot-blotted on nitrocellulose filters. The filters were then washed and scanned on a PhosphorImager running ImageQuant software (25).
Western Analysis-Subconfluent HUVEC were treated with VEGF or PBS for 2 h as described under "Cell Culture and mRNA Isolation." Cells were lysed in radioimmune precipitation buffer (50 mM Tris-Cl, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, 1 mM phenylmethyl sulfonyl fluoride, 1 mM NaF, 1 mM Na 3 VO 4 , 1 g/ml aprotinin, and 1 g/ml leupeptin) and disrupted by passage through a 22-gauge needle five times. Cell debris was removed by centrifugation at 14,000 ϫ g, and the supernatant was used as cell lysate for Western blotting. A Bradford-based assay kit (Bio-Rad) was used to measure protein concentrations. Equal amounts of protein from each time point were fractionated by 15% Tricine/sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to an Immobilon-P membrane. Western blotting was carried out using an ECL kit as described by the manufacturer (Amersham Life Science). HB-EGF was detected by chemiluminescence with a polyclonal primary antibody to HB-EGF (0.1 g/ml, R&D Systems) and a horseradish peroxidase-conjugated secondary antibody to goat Ig. The film was scanned into Adobe Photoshop 3.0, and the signal density was measured with the NIH Image software.
Migration Assay-Cell migration was measured in a 48-well micro-Boyden chamber apparatus (Neuroprobe, Cabin John, MD) as described (33). Confluent HASMC were made quiescent by incubation in Dulbecco's modified Eagle's medium with 0.4% fetal calf serum for 72 h before the assay. HB-EGF and PBS (control) were diluted in Dulbecco's modified Eagle's medium with 0.25% bovine serum albumin and loaded into the lower wells of the Boyden chamber in triplicate. The wells were subsequently covered with a polyvinylpyrrolidone-free filter with 8-m pores (Nucleopore, Palo Alto, CA) coated with type I collagen (Vitrogen, Collagen Corp., Palo Alto, CA). Cells were washed four times in PBS and trypsinized (0.01% trypsin, 0.11 mM EDTA) for the minimum period of time required to obtain a mononuclear suspension. The cells were washed twice in Dulbecco's modified Eagle's medium, 0.25% bovine serum albumin and resuspended at a density of 1 million cells/ml. Cells (50,000 cells in 50 l) were loaded into the upper wells of the Boyden chamber. The chambers were incubated for 4 h at 37°C in an atmosphere of 95% O 2 , 5% CO 2 . At the end of the incubation, cells that had attached to the filter were fixed and stained in Dif Quick (American Hospital Supply, McGaw Park, IL). Cells that had migrated to the lower side of the filter were counted under a microscope at 20ϫ magnification. Chemotaxis was calculated as the difference between the number of cells that had migrated in the presence of the chemoattractant and the number that had migrated in its absence.
[ 3 H]Thymidine Incorporation-Primary cultured rat aortic smooth muscle cells (RASMC) (25) were plated on 24-well plates in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, penicillin (100 units/ml), streptomycin (100 g/ml), and 25 mM Hepes (pH 7.4) and incubated for 24 h. Cells were made quiescent by incubation in Dulbecco's modified Eagle's medium plus 0.4% calf serum for 72 h before the experiments were performed. HUVEC were grown as described, and placed in M199 supplemented with 2% fetal calf serum for 1 h before stimulation with 20 ng/ml VEGF. After 2 h of VEGF treatment, the conditioned medium was harvested from HUVEC and used to stimulate quiescent RASMC (final proportion of 33% conditioned medium and 67% RASMC medium). Prior to its addition to RASMC, HUVEC conditioned medium was treated with either no antibody, anti-human PDGF polyclonal antibody (8 g/ml, R&D Systems) or normal goat IgG (8 g/ml, R&D Systems) for 1 h at 25°C. Conditioned medium was applied to the RASMC for 24 h. During the last 2 h of this incubation period, the cells were labeled with [ 3 H]thymidine (NEN Life Science Products) at 1 Ci/ml. Labeled cells were washed with phosphate-buffered saline, fixed in cold 10% trichloroacetic acid, and washed with 95% ethanol. Incorporated [ 3 H]thymidine was extracted in 0.2 N NaOH and measured in a liquid scintillation counter.

Regulation of HB-EGF mRNA by VEGF in Vascular
Endothelial Cells-Northern blot analysis was performed with total RNA from HUVEC, and blots were hybridized with a human HB-EGF cDNA probe. In response to stimulation with 10 ng/ml VEGF (Fig. 1A), HB-EGF mRNA levels increased rapidly to a maximum at 2 h. After 6 h, the HB-EGF mRNA level returned to base line. VEGF also induced HB-EGF mRNA in HUVEC in a dose-dependent manner (Fig. 1B). To confirm that this response was not limited to HUVEC, we performed the same experiment in endothelial cells from another source, HCAEC. As in HUVEC, HB-EGF mRNA was induced by 10 ng/ml VEGF in HCAEC (data not shown).
Induction of HB-EGF mRNA by VEGF in HUVEC Does Not Require New Protein Synthesis-To determine whether induction of HB-EGF mRNA required protein synthesis de novo, we treated HUVEC with the protein synthesis inhibitor anisomycin (100 M) for 1 h before adding VEGF. This dose of anisomycin completely inhibited leucine uptake in HUVEC (data not shown). Anisomycin did not prevent the induction of HB-EGF mRNA by VEGF (Fig. 2), suggesting that this process does not require new protein synthesis.
Transcriptional Regulation of HB-EGF by VEGF-We measured mRNA stability and rate of transcription to elucidate the mechanism by which VEGF up-regulates HB-EGF mRNA. The half-life of HB-EGF mRNA was determined in the presence of the transcription inhibitor actinomycin D. HUVEC were stimulated with VEGF (10 ng/ml) or vehicle (PBS) for 2 h. HB-EGF mRNA was then measured 0, 1, and 2 h after administration of actinomycin D (5 g/ml). The half-life of HB-EGF mRNA at base line was approximately 55 min (Fig. 3A); it was not altered by exposure to VEGF. Nuclear run-on experiments were then performed to assess the rate of transcription. VEGF increased transcription of the HB-EGF gene in HUVEC by 4.9-fold (Fig.  3B). These experiments demonstrate that the increase in HB-EGF mRNA levels after VEGF stimulation was the result of an increase in HB-EGF gene transcription.
Effect of VEGF on HB-EGF Protein-To determine whether the increase in HB-EGF mRNA in HUVEC was associated with an increase in HB-EGF protein, we performed Western blot analysis with a polyclonal antibody specific for recombinant human HB-EGF. The level of HB-EGF protein increased in HUVEC stimulated with VEGF for 2 h (Fig. 4A). This observation demonstrates that the effect of VEGF on HB-EGF mRNA in HUVEC is associated with an increase in HB-EGF protein.
Stimulation of HASMC Migration by HB-EGF-Because the FIG. 2. Effect of anisomycin on HB-EGF mRNA in HUVEC. HUVEC were exposed to PBS control, VEGF (10 ng/ml), anisomycin control (100 M), or a combination of anisomycin (100 M) and VEGF (10 ng/ml). Anisomycin was applied to cells 1 h before VEGF was applied. Total RNA was extracted from the cells after 2 h of VEGF stimulation, and Northern blot analysis was performed as described for Fig. 1A.

FIG. 3. Effect of VEGF on HB-EGF mRNA half-life and rate of transcription in HUVEC.
A, HUVEC were exposed to PBS (control) or VEGF (10 ng/ml) for 2 h before addition of actinomycin D (5 g/ml). Total RNA was extracted from the cells at the indicated times after actinomycin D administration. Northern blot analyses were performed as in Fig. 1. The corrected density was then plotted as a percentage of the 0-h value against time (in log scale). B, subconfluent HUVEC were not stimulated (PBS, control) or stimulated with VEGF (10 ng/ml) for 2 h. Nuclei were isolated, and in vitro transcription was allowed to resume in the presence of [␣-32 P]UTP. Equal amounts of 32 P-labeled, in vitro-transcribed RNA probes from each group were hybridized to 1 g of denatured HB-EGF and 18 S cDNA.

FIG. 1. Time course and dose response of HB-EGF mRNA induction by VEGF in HUVEC.
A, HUVEC were treated with vehicle (PBS, white bars) or VEGF (10 ng/ml, black bars), and total RNA was extracted from the cells at the indicated times and subjected to Northern analysis. The HB-EGF signal density was divided by that of 18 S, and the corrected density was then plotted as a percentage of control. B, HUVEC were incubated for 2 h with the indicated concentrations of VEGF, and total RNA was extracted from the cells and analyzed as in A.
maturation of new blood vessels requires cell migration and investment of capillary tubules with pericytes and smooth muscle cells, we measured the chemotactic effect of HB-EGF on vascular smooth muscle cells. HB-EGF induced a significant increase in HASMC migration (Fig. 4B).
Effect of VEGF on Other Growth Factors-To see if VEGF increased the expression of other factors known to affect angiogenesis (34,35), we studied its effect on the mRNA levels of PDGF-BB, ET1, and TGF-␤1 in HUVEC. Northern blot analysis showed that PDGF-BB mRNA increased as early as 2 h, and its induction reached 3-fold after 5 h of VEGF stimulation. TGF-␤1 and ET1 mRNA levels were not altered by VEGF (Fig. 5).
Effect of Conditioned Medium from VEGF-stimulated HU-VEC on RASMC-To further delineate the role of endotheliumderived growth factors on vascular smooth muscle cells, studies were performed using conditioned medium from HUVEC stimulated with VEGF. Compared with medium from vehicletreated cells, medium from VEGF-stimulated HUVEC produced a significant increase in RASMC DNA synthesis (Fig. 6). Incubating the conditioned medium with a neutralizing antibody to PDGF prevented this increase in DNA synthesis. When control antiserum, normal goat IgG, was used in the same concentration as the PDGF antibody, the induction of DNA synthesis was not abrogated. DISCUSSION A series of sequential events promote angiogenesis. These include (a) degradation of the vascular basement membrane and the interstitial matrix by endothelial cells, (b) endothelial cell migration and proliferation, (c) formation of new capillary tubes, and (d) formation of new basement membrane with investment of the tubule by pericytes and smooth muscle cells (1,36). It has been established that VEGF is an endothelial cell-specific, multifunctional growth factor that plays a major role in the initiation of angiogenesis by acting directly as a mitogenic and a chemotactic factor, and indirectly by promoting remodeling of the extracellular matrix (16,(37)(38)(39)(40). Although pericytes and smooth muscle cells are essential in the late phase of angiogenesis, there is no clear explanation of how these cell types are recruited into the process by angiogenic factors such as VEGF (39,(41)(42)(43).
Suri et al. (44) have reported recently that mice lacking the TIE2 tyrosine kinase ligand angiopoietin-1 exhibit abnormal vascular architecture, where the principal defect is a failure to recruit smooth muscle and pericyte precursors. Additionally, Lindahl et al. (45)  angiogenesis. We demonstrate here in endothelial cells that VEGF induces HB-EGF and PDGF-BB, which have mitogenic and chemotactic effects on pericytes and smooth muscle cells. This demonstration links VEGF to early (endothelial cell recruitment) and late (smooth muscle cell migration and investment) stages of angiogenesis. Since the transcriptional mechanisms downstream of VEGF have not been described, this represents a first step toward defining those mechanisms.
The induction of VEGF and its receptors (flk-1 and flt) in human atherosclerotic lesions and in animal models of arterial injury has also been described (37,47,48). Lazarous et al. (49) reported recently that VEGF given postoperatively in a canine model of left circumflex coronary artery occlusion and iliofemoral artery denudation induced further accumulation of neointima without inducing collateral blood vessel formation in the coronary circulation. These findings imply that VEGF may play a paracrine role in exacerbating neointima formation and vessel occlusion, both by inducing neovascularization in lesions and by stimulating the release of smooth muscle cell mitogens and chemoattractants, as suggested by Li et al. (50). Our demonstration that VEGF induces the vascular smooth muscle cell mitogenic and chemotactic growth factors HB-EGF and PDGF-BB elucidates a mechanism by which this atherogenic process occurs.
We have demonstrated that VEGF induces expression of angiogenic growth factors in vascular endothelial cells. VEGF up-regulates HB-EGF and PDGF-BB but not TGF-␤1 or ET1. Induction occurs both in HUVEC and HCAEC (data not shown), suggesting that VEGF has this effect in vascular endothelial cells of various origins. HB-EGF and PDGF-BB are both chemotactic agents for vascular smooth muscle cells, and conditioned medium from endothelial cells treated with VEGF produces an increase in DNA synthesis in vascular smooth muscle cells. This growth-promoting effect of conditioned medium can be prevented by a neutralizing antibody to PDGF. Taken together, our data suggest that HB-EGF and PDGF-BB provide a critical endothelial cell-derived signal for the recruitment of pericytes and smooth muscle cells during angiogenesis, which is consistent with the model proposed by Folkman and D'Amore (46). These findings may provide important insight into the process of new blood vessel formation and maturation.