INTRODUCTION

Angiogenesis is a multistep process that results in the formation of new capillaries, usually by sprouting from pre-existing small blood vessels (1). The impetus for vascular remodeling arises during both physiologic and pathophysiologic events. These include embryonic development, wound healing or tissue regeneration, and tumor growth (2). In each of these processes, the invasion of the perivascular extracellular matrix and the migration and proliferation of endothelial cells (EC)¹ occur as steps in angiogenesis and result in part from the action of EC-directed growth factors. The normally quiescent microvasculature has a turnover time for endothelial cells extending to years. However, when angiogenesis is enacted by EC growth factors, proliferation of EC can be accomplished in as short as 5 days (1). Therefore, the modulation of specific growth factor production is important to the invasion properties, proliferation, and migration of EC. Several vascular growth factors have been implicated as playing an important role in these phases of angiogenesis, including acidic and basic fibroblast growth factors (bFGF), angiogenin, and platelet-derived growth factor (1, 3).

Recently, a peptide growth/permeability factor has been identified that is produced predominantly in vascular smooth muscle cells (VSMC) and binds to the specific receptors Flt1 and Flk1 (KDR), expressed mainly on EC (4). This protein, vascular endothelial cell growth factor (VEGF), has been implicated in the stimulation of normal angiogenesis and the increased capillary formation that characterizes proliferative diabetic retinopathy (5). VEGF is also considered to be critical for the angiogenesis that underlies tumor metastasis (6, 7). VEGF triggers EC proliferation, probably after binding the KDR receptor (8-10) and enacting tyrosine phosphorylation of numerous cytosolic proteins, some of which ultimately signal to the nuclear growth program (11). VEGF production is stimulated under a variety of circumstances, most notably by tissue hypoxia (12, 13), but the regulation of this protein by vascular factors or proteins is incompletely understood.

One family of vasoactive peptides, the endothelins (ET), are produced primarily in EC, but also in VSMC (14), and are known to directly stimulate EC proliferation in vitro, modulated through the endothelin B-type (ETB) receptor (15). Another family, the natriuretic peptides, are produced in both the heart and EC (16, 17) and have been described to inhibit the proliferation of EC in culture (18). We investigated whether ET and ANP might regulate the production of VEGF as a potentially important mechanism by which these peptides could modulate important steps for the process of angiogenesis. We also sought to understand which receptors could mediate vasoactive peptide regulation of VEGF production and what signaling mechanisms are involved. We report that members of the ET family that are produced in the vasculature, ET-1 and ET-3, are potent stimulators of VEGF production. More important, members of the natriuretic family, ANP and CNP, substantially inhibit VEGF production and subsequent key steps of angiogenesis in two models that simulate the in vivo interactions of VEGF with EC.

MATERIALS AND METHODS

Bovine aortic EC cultures were prepared as described previously (19, 20). Bovine aortic EC were plated into 100-mm culture dishes as primary cultures. After confluency, the cells were replated in various sized culture dishes depending on the experimental protocol. The cells displayed the typical morphologic characteristics of endothelial cells, and virtually all cells showed positive fluorescence with an antibody to factor VIII. For human umbilical vein smooth muscle cells (hUVSMC), the preparation and plating of cells were similar except that fetal umbilical veins were obtained immediately after birth and perfused with iced saline, and lumens were stripped of the endothelial layer by 30 min of incubation in collagenase-containing medium. This was followed by rinsing and additional exposure to collagenase for 2 h to obtain smooth muscle cells (21). The cells had the typical appearance of smooth muscle cells and did not stain positive for Von Willebrand's factor, but were found to stain for -actin.

hUVSMC were synchronized in the absence of serum for 24 h and then incubated in methionine-free Dulbecco's modified Eagle's medium with dialyzed 10% fetal bovine serum for 1 h prior to experimentation (19, 20). The cells were then incubated with 250 µCi of [³⁵S]methionine in the presence or absence of ET-1 or ET-3 (1 or 100 nM) or ANP, CNP, or C-ANP-(4-23) (1 or 100 nM) for various times over 6 h. In some wells, BQ-123, a specific endothelin A-type (ETA) receptor antagonist, or IRL-1038 (both from Peninsula Laboratories, Inc.), a specific ETB receptor antagonist, was added (22). Other dishes of cells were placed in an anaerobic chamber (Gas Pack system, Becton Dickinson), which was purged with 95% N₂ and 5% CO₂ and sealed with an O₂-consuming palladium catalyst. This created hypoxic conditions of PO₂ at 35 mm Hg. The control cells were subjected to normoxia (atmospheric air and 5% CO₂; PO₂ at 150 mm Hg). The media from the experimental conditions were saved, and the cells were washed and then lysed in buffer for 1 h at 4 °C. The lysate and secretion medium were precleared, and labeled VEGF protein was immunoprecipitated using polyclonal antibody to VEGF (Santa Cruz Biotechnology, Santa Cruz, CA). Antibody that was preabsorbed with recombinant VEGF (10⁷ M; Merck) for 12 h prior to the immunoprecipitation procedure served as a specificity control. The immunoprecipitated protein was solubilized in reducing SDS sample buffer, denatured, and resolved by 10% SDS-polyacrylamide gel electrophoresis. The gel was subjected to fluorography and then autoradiography for 1-2 days. Each translation experiment was performed at least three times.

The extracted RNA was hybridized with a ³²P-labeled cRNA probe, made from a template of a full-length human cDNA for VEGF (kindly provided by Dr. Richard Kendall, Merck) (23). The cDNA was in an EcoRI-EcoRI orientation in pGEM33zf and was linearized with NcoI; antisense and sense cRNA probes were promoted using SP6 and T₇ RNA polymerases, respectively. A transcript of ~400 bases was protected. Solution hybridization, S1 nuclease digestion, and electrophoretic separation were carried out as described previously (19, 20). A ³²P-labeled cRNA for -actin (Ambion Inc.) served as an RNA loading standardization probe. Autoradiographic bands were compared by laser densitometry (Pharmacia Biotech Inc.). Sense probes produced no hybridization, and the studies were repeated three times.

VSMC were grown to 40-50% confluence on 60-mm Petri dishes. Cells were then transiently transfected as described previously (24) with 1 µg of a fusion plasmid containing 1.7 kb of 5-flanking rat VEGF promoter sequence in the luciferase-containing expression vector pXP2 (kindly provided by Dr. Andrew Levy, Harvard Medical School) (25). A separate similar construct, but with known minimal activity containing only the proximal 100 bases of the promoter region, was transfected as a low activity control (25). To control for transfection efficiency, cells were cotransfected with 0.1 µg of pRL-SV40 expressing the Renilla luciferase (Promega), and VEGF/Luc data were accordingly normalized. Cells were incubated with LipofectAMINE reagent (8 µl/plate; Life Technologies, Inc.) and DNA complexes for 5 h at 37 °C, followed by recovery for 12 h in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Prior to experimental treatment, cells were synchronized for 24 h in serum-free Dulbecco's modified Eagle's medium. Cells were then treated with hypoxia, ET-1 or ET-3 (100 nM) ± ANP (100 nM), or ANP alone for 6 h at 37 °C, followed by harvesting. Some dishes of cells were incubated with ET-1 and the MAP kinase kinase (MEK) inhibitor PD 98059 (20 µM) (26) or the relatively specific protein kinase C (PKC) inhibitor calphostin C (50 nM). In addition, some cells were cotransfected with 1.7-kb VEGF/Luc and a MAP kinase (Erk2) protein expression vector (pCMV5erk) (24, 27). Cell extracts were briefly microcentrifuged, and supernatants were assayed for luciferase activity by the dual luciferase reporter assay system (Promega). Experiments were carried out four times, each condition in triplicate in each experiment, and all data were combined.

hUVSMC were first incubated in the presence or absence of ET-1 or ET-3 and in the presence or absence of ANP for 4 h, and the secretion medium was then aspirated and frozen for subsequent use. Some of the ET-1-incubated VSMC medium was then absorbed with antibody (1:1000 dilution) to VEGF or bFGF (Santa Cruz Biotechnology) or ET-1 (Peninsula Laboratories, Inc.) for 2 h prior to use with EC. Subconfluent EC were synchronized by incubation for 24 h in serum-free medium. The cells were then incubated for 20 h in the absence or presence of the various experimental secretion media from the VSMC experiments. This was followed by the addition of 0.5 µCi of [³H]thymidine for an additional 4 h as described previously (28). Cells were then washed, incubated for 10 min with 10% trichloroacetic acid at 4 °C to precipitate the nuclear incorporated thymidine, washed twice, and lysed with 0.2 M NaOH, and the lysates were neutralized with 0.2 N HCl and counted in a liquid scintillation -counter.

To understand the intracellular signal that is necessary for ET modulation of VEGF protein production, VSMC were incubated with ET-1 or ET-3 for 4 h in the presence or absence of the PKC inhibitor calphostin C. VEGF protein synthesis was carried out as described above, and the intracellular lysates were then separated by SDS-polyacrylamide gel electrophoresis and characterized.

To assess whether the guanylate cyclase receptors and cGMP mediate the actions of ANP or CNP, the cells were incubated with ET and the NP with or without the compound LY 53853, an antagonist of ANP-stimulated cGMP generation (29). VEGF protein synthesis was again determined. The generation of cGMP in whole cells was determined in response to the NP. VSMC cultures in 12-well plates were synchronized for 24 h at 37 °C in culture medium without fetal bovine serum. The cells (in quadruplicate wells/condition) were preincubated for 5 min with 0.5 mM 3-isobutyl-1-methylxanthine and then incubated for an additional 5 min with ANP or CNP (100 nM). The reactions were stopped on ice with immediate aspiration of the medium, followed by three washes of the cells with cold Hanks' balanced saline solution. Cyclic nucleotides were extracted and assayed as described previously (29).

VSMC secretion medium was obtained from cells incubated with ET-1, ANP, or ET plus ANP for 4 h as described for the proliferation studies. To examine the "paracrine" chemoattractant effects of the VSMC incubation medium on EC invasion of matrix, a standard invasion assay was used (30). The medium from each VSMC experimental condition was added to the bottom of 24-well Falcon TC companion plates. Matrigel invasion chambers and control inserts (Becton Dickinson) were placed inside each of the wells; the chamber contained a thin Matrigel layer on top of an 8-µm membrane pore, and EC were seeded at a density of 10⁵ cells/ml into the chamber on top of the membrane, where they were incubated for 12 h at 37 °C. After incubation, the non-invading cells adhering to the upper surface of the membrane were scraped away, and invasion was detected after staining by counting the cells under the microscope in the center fields of the underlayer of the membrane. The 12 h of incubation precluded cell proliferation as contributing to the differences in invasion number since the doubling time of these cells is ~24 h. Triplicate membranes were assessed per VSMC media experimental condition, and the data were normalized for invasion through a control membrane. The normalized data were compared with invasion of EC through membranes incubated in the presence of control VSMC medium (no peptides added to the incubation medium). Under some conditions, the incubation medium from ET-1-incubated VSMC was preabsorbed with VEGF, bFGF, or ET-1 antibody as described for the proliferation studies.

Protein bands from the translation studies were compared by laser densitometry. RNA comparisons were quantified by laser densitometry of autoradiographs, and data were normalized for RNA loading by creating a ratio of the density of the experimental RNA hybridized with the VEGF probe divided by the same amount of RNA hybridized with -actin. A ratio was then established by comparing normalized experimental RNA with normalized control RNA that was extracted from untreated endothelial cells. A value of 1 was arbitrarily assigned to the control. This resulted in values expressing the relative densities of the experimental conditions compared with the control. Data from multiple experiments for the thymidine incorporation or cGMP generation studies were combined, and the different conditions were compared by analysis of variance; a multiple range test (Scheffe's F test) was used for significant F values (p < 0.05), and the Statview statistical computer program was used (Abacus Concepts, Berkeley, CA). Transcription and invasion data were similarly analyzed.

RESULTS

ET-1 or ET-3 caused a significant increase in the production of new VEGF protein in VSMC (Fig. 1A). Based on data from three experiments, ET-3 stimulated this growth factor's production, beginning at 1 h and maximally stimulating VEGF protein production by 2.5 ± 0.2-fold above base line at 4 h. ET-1 also stimulated the production of VEGF by 1 h of incubation and maximally by 3.5 ± 0.4-fold above control levels at 6 h of incubation. Stimulation by the ET peptides was comparable in magnitude and time course to the stimulation of VEGF production produced by hypoxia, a recognized potent and important stimulus of VEGF production (Fig. 1A). The predominant form of this protein produced in the cell and isolated from the intracellular lysate was compatible with VEGF165. This isoform is a 45-kDa homodimeric glycoprotein, which should migrate on gel in its reduced (monomeric) form at ~20-23 kDa (31), depending upon the degree of glycosylation by the cell. VEGF165 is the most abundant form detected in VSMC and is known to be one of two forms secreted by these cells (31). The other isoforms of VEGF (VEGF206, VEGF189, and VEGF121) (32) were not seen in our experiments using the cell lysate. Antibody that was first preabsorbed with this growth factor failed to precipitate the band (Fig. 1B). The ability of ET-1 and ET-3 to stimulate VEGF production was concentration-related (Fig. 1B). The ET peptides also caused a comparable, dose-related stimulation of VEGF165 secretion into the cell incubation medium (Fig. 1C). More important, ANP inhibited the ET-stimulated production and secretion of VEGF (Fig. 1 (B and C) and text below) in a concentration-dependent fashion.

We then investigated the ET receptors involved in the regulation of VEGF production. After 6 h of incubation, the stimulation by ET-1 was reversed 91% by the ETA receptor antagonist BQ-123 (Fig. 2, lane 5 versus lane 13), indicating that ET-1 stimulated VEGF production after binding the ETA receptor. In contrast, ET-3 stimulation of this growth factor was antagonized 93% by IRL-1038 (lanes 9 and 14), but not by BQ-123 (lane 15), indicating that the ETB receptor mediated these effects. These results also indicate that either of the two known human ET receptor subtypes has the capability to signal to the program for VEGF production.

We then further examined the interaction between the NP and ET and found that two naturally occurring and one synthetic NP inhibited ET-stimulated VEGF production (Fig. 2). ANP, CNP, and C-ANP-(4-23) also inhibited the hypoxia-augmented VEGF protein synthesis. The effects were potent, with ANP causing a 77 ± 6% maximal inhibition of ET- stimulated VEGF synthesis and 88 ± 3% inhibition of hypoxic action. Each of the three NP were approximately equipotent in their effects, within 4% of each other based upon combined data from three studies.

ET signals through several G proteins to multiple effectors, resulting in a variety of cell physiologic effects (14). Several important functions of ET appear to involve protein kinase C. Based on three experiments, we found that the ability of ET-1 or ET-3 to stimulate VEGF protein synthesis was reversed 48 and 32%, respectively, by co-incubation with calphostin C, a PKC inhibitor (33) (Fig. 3, lane 6 versus lane 11 and lane 12 versus lane 17). This indicates that signaling through this serine/threonine kinase is an important mediator for ET-induced VEGF production, but also that PKC-independent pathways or PKC isoforms that are not inhibited by calphostin C are probably involved.

We also determined whether the effects of ANP or CNP are mediated through cGMP generation, indicating involvement of the guanylate cyclase A and B receptors, respectively. Inhibition by ANP or CNP of ET-stimulated VEGF averaged 64 ± 5%, based on combined data. We found that co-incubation of the NP with 1 µM LY 53853 had little effect on ANP or CNP inhibition of ET-stimulated VEGF protein synthesis (Fig. 3, lanes 6-10 and 12-16). This suggests that the generation of cGMP does not mediate the inhibitory effects of ANP or CNP. LY 53853 at this concentration was capable, however, of dose-dependently and significantly inhibiting the ANP- or CNP-stimulated production of cGMP. The natriuretic peptides caused a 250-300% stimulation of cGMP generation above control levels (Table I). This was maximally inhibited by 1 µM LY 53853: 57 and 54% for ANP and CNP, respectively. Although the natriuretic peptide clearance receptor ligand C-ANP-(4-23) was equipotent to ANP or CNP in inhibiting VEGF production (Fig. 2), this compound did not generate cGMP in limited studies and as we have previously shown (20).

Table I. Effect of LY 53853 on cGMP generation in cultured hUVSMC in response to ANP or CNP Cells were cultured in the presence of 3-isobutyl-1-methylxanthine and the natriuretic peptides ± LY 53853 for 5 min; the reaction was stopped; and cGMP was extracted and measured (see "Materials and Methods"). Data are the means ± S.E. from two experiments combined (n = six observations/condition) and are representative of a third. cGMP fmol/ml Control 1027  ± 31 LY 53853 (106 M) 899  ± 4 LY 53853 (107 M) 1015  ± 6 LY 53853 (108 M) 1078  ± 4 ANP 53853 (107 M) 3929  ± 76a   +LY 53853 (106 M) 2265  ± 23b   +LY 53853 (107 M) 2502  ± 16b   +LY 53853 (108 M) 3104  ± 19b CNP (107 M) 3449  ± 46a   +LY 53853 (106 M) 2142  ± 09b   +LY 53853 (107 M) 2316  ± 59b   +LY 53853 (108 M) 2695  ± 19b a p < 0.05 for control versus ANP or CNP by analysis of variance plus post hoc test (Scheffe's). b p < 0.05 for ANP or CNP versus natriuretic peptide plus LY 53853. C-ANP-Schefee's (4-23) did not stimulate cGMP generation in one experiment and as we have previously shown (20).

The ET peptides and hypoxia each stimulated VEGF gene expression (mRNA) (Fig. 4). Incubation of VSMC for 6 h with ET-1 caused a 3.3 ± 0.4-fold increase in VEGF mRNA above control expression. ET-3 caused a 2.4 ± 0.3-fold stimulation above control levels at this time. Each of the natriuretic peptides reversed both these stimulations by 74-85%. In addition, hypoxia stimulated VEGF mRNA by 4.5 ± 0.5-fold, and all three natriuretic peptide forms caused as much as a 73% inhibition of hypoxia-stimulated VEGF mRNA; C-ANP-(4-23) was slightly more potent than CNP in this regard (73 versus 62% inhibition). The ETA and ETB receptor antagonists again reversed the respective ET-1 and ET-3 stimulations of VEGF mRNA.

We then determined whether the effects of the vasoactive peptides on VEGF gene expression were predominantly transcriptional. We found that hypoxia, ET-1, or ET-3 stimulated the activity of the 1.7-kb VEGF/Luc reporter by 3-4 times that of this reporter activity in the absence of peptide under normoxic conditions (control) (Fig. 5). It is already established that hypoxia has both transcriptional and post-transcriptional effects on VEGF synthesis in several cell and tumor types (13, 34), and our results indicate that the ET peptides are comparably potent for inducing VEGF transcription. As representative of the natriuretic peptides, ANP inhibited ET-1- or ET-3-induced VEGF transcription by 71 and 76%, respectively, and hypoxia-stimulated VEGF transcription by 64%. These data indicate that the vasoactive peptides modulate VEGF gene transcription and that this substantially accounts for the changes in steady-state mRNA levels.

We then carried out studies to provide insight into the signaling mechanisms by which the vasoactive peptides modulate VEGF transcription. Consistent with our VEGF protein synthesis data, we found that inhibition of PKC with calphostin C reversed the ET-1 stimulation of VEGF transcription by 76% (Fig. 5). We also postulated that ET-induced MAP kinase (Erk) activity contributed to this action since we previously reported that ET stimulates Erk activity (35) and that ET stimulation of egr-1 and bFGF transcription is dependent on activating this proline-directed kinase (24). In the studies reported here, we found that cotransfection of pCMV5erk, an Erk2 expression vector, with the VEGF reporter construct increased VEGF transcription. Furthermore, we found that the ability of ET-1 to stimulate VEGF transcription was reversed 82% by PD 90859, an inhibitor of MAP kinase kinase (MEK) activity, an enzyme that is the only consensus activator of Erk.

We used an in vitro model of the potential in vivo paracrine interactions between VSMC and EC to determine whether ET and ANP could modulate VEGF production and the subsequent proliferation of EC. We incubated VSMC with 100 nM ET-1 or ET-3 in the presence or absence of 100 nM ANP. The secretion medium from the ET-1-treated VSMC caused a 103 ± 5% increase above control levels in DNA synthesis in EC when used as the incubation medium for EC over 24 h (Fig. 6A). The medium from ET-3-treated VSMC stimulated EC proliferation above control levels by 53 ± 4% more than control VSMC medium. In contrast, co- incubation of VSMC with ANP reduced the ET-1 stimulation effect by 52 ± 3%, and the natriuretic peptide also inhibited ET-3-induced stimulation by 58 ± 3%.

We then determined the nature of the EC-stimulating activity contained in the VSMC secretion medium, augmented by ET and inhibited by ANP. The EC stimulatory effects of the ET-1-incubated VSMC secretion medium were inhibited 71 ± 4% by preincubating the VSMC medium for 2 h with a specific antibody to VEGF (Fig. 6B), which presumably bound much of the VEGF secreted into the medium. In contrast, incubations of the same VSMC medium with antibody to either bFGF or ET-1 had little effect on the EC-proliferating activity. VEGF antibody, but not ET-3 or bFGF antibody, also inhibited the EC-stimulating activity of the ET-3-incubated VSMC medium by 59 ± 5%. At a moderately greater concentration of VEGF antibody, we found that 95% of the VEGF secreted by VSMC in response to ET-1 was removed, as determined by radioimmunoassay, and that this reversed the ability of the same medium to simulate EC proliferation by 89% (data not shown). This indicates that the substance in the VSMC secretion medium from ET-incubated cells that was largely responsible for the increased EC proliferation was VEGF. The in vivo translation studies previously showed that the VSMC secretion medium contained secreted VEGF165 protein, increased in response to ET-1 or ET-3. We also found that neither ET nor ANP affected ¹²⁵I-VEGF binding to EC, as a possible mechanism of action.

The effects of the vasoactive peptides on influencing VEGF secretion from VSMC and the subsequent invasion by EC of matrix protein are seen in Table II. The VSMC medium from ET-1-exposed cells resulted in a doubling of EC number migrating to the underside of the membrane through the Matrigel, compared with the effect of control medium. Co-incubation with ET plus ANP medium resulted in 62% fewer cells migrating, compared with ET-1 alone medium. The invasion stimulus secreted by VSMC in response to ET-1 was VEGF since medium that was preabsorbed with VEGF antibody for 2 h prior to exposure to EC resulted in 80% fewer cells invading the matrix. In contrast, the same medium incubated with bFGF antibody resulted in a small reduction of invasion; similar results were seen with antibody to ET-1 (Table II). These data indicate that an important measure of VEGF-related angiogenesis can be positively or negatively modulated by ET or ANP, respectively, resulting from the regulated production of this growth factor in VSMC.

Table II. Effects of culture medium from hUVSMC exposed to peptides on the invasion of matrix by EC Data are from two experiments combined (n = six wells/experimental condition). EC were layered on Matrigel membranes in inserts, and the inserts were then placed in wells containing the incubation medium from VSMC exposed to peptides and incubated for 12 h. EC invading the matrix to the bottom of the membrane were counted in triplicate in the center fields. Control VSMC medium is Dulbecco's modified Eagle's medium exposed to VSMC in the absence of peptides prior to incubation with EC. Control is the same medium, but never exposed to VSMC. No. of cells on membrane Control VSMC medium 95  ± 3 ANP (100 nM) 90  ± 4 ET-1 (100 nM) 196  ± 7a ET-1 + ANP 133  ± 6b ET-1 + VEGF antibody 116  ± 9b ET-1 + bFGF antibody 174  ± 7 ET-1 + ET-1 antibody 186  ± 7 a p < 0.05 for control VSMC versus peptide or growth factor. b p < 0.05 for peptide or growth factor versus peptide/growth factor plus ANP or plus VEGF antibody.

DISCUSSION

VEGF stimulates critical events in angiogenesis including EC proliferation and EC migration into and invasion of the extracellular matrix (1). Therefore, the regulation of its synthesis is a critical event in new capillary formation. In these studies, we showed that the two members of the ET family that circulate in plasma, ET-1 and ET-3, fairly comparably stimulate VEGF production in VSMC and are equipotent to hypoxia, an important known stimulus of VEGF production. ET-1 may have a paracrine effect on VEGF production in vivo since ~75% of ET-1 secretion from EC is directed toward VSMC and away from the vascular lumen (36). More important, we report that several members of the NP family inhibit the stimulatory effects of the ET or hypoxia on VEGF synthesis. These results provide a mechanism by which the NP can act as anti-angiogenesis peptides and identify the NP as the first proteins reported to inhibit VEGF production.

We also showed that ET stimulates the secretion of the VEGF165 isoform from cultured VSMC and that this secreted growth factor is largely responsible for subsequent EC proliferation and invasion of matrix, providing a model of how these interactions could occur in vivo. The actions of ET-1 were mediated predominantly through the ETA receptor since the specific antagonist BQ-123 reversed the stimulation of VEGF production. ET-1 can also directly act as a mitogen for EC in vitro, and this results from binding to the ETB receptor, triggering undetermined signaling mechanisms (15). We also found that ET-3 stimulates VEGF production, through the ETB receptor, based upon reversal by a specific receptor antagonist. Thus, both ET receptors can participate in the regulation of VEGF production. There are several examples where both known ET receptors can signal to the same cell physiologic event; for instance, both the ETA and ETB receptors can activate the MAP kinase (Erk) and proliferation pathway in mesangial cells (37).

The regulation by vascular factors of VEGF production and subsequent angiogenesis are important. Previous studies have identified that the growth factors platelet-derived growth factor, transforming growth factor-1, and bFGF can each stimulate the production of VEGF (38, 39). Cytokines, such as interleukin-1, can induce VEGF transcripts in VSMC (40); nitric oxide can stimulate VEGF mRNA in the lung (41); and E-series prostaglandins can stimulate VEGF production in osteoblasts (42). Hypoxia is one of the best characterized stimuli for the induction of VEGF production by a variety of tumor cells, as well as VSMC, both in vitro and in vivo (12, 13, 39). Since the endothelins are equipotent to hypoxia in stimulating VEGF production in cultured VSMC, we propose that ET may play an important comparable role in the in vivo vasculature. Hypoxia is also a potent and very rapidly acting stimulus for the production of ET-1 in endothelial cells (43, 44), and hence, in devitalized tissues, acute or chronic hypoxia may stimulate VEGF production through both direct and indirect effects, the latter involving ET secretion. Also, ET has been implicated as playing an important role in the growth of solid neoplasms (45, 46), and the growth and metastases of tumors are known to be strongly dependent on neovascularization (47). It is reasonable to hypothesize that the tumor-promoting effects of ET could be mediated in part through the stimulation of VEGF in the existing tumor or vasculature, resulting in angiogenesis.

Recently, a second member of the human VEGF family, VPR, has been cloned, but it has only 32% homology on an amino acid basis to VEGF (48, 49). This protein does not bind to Flt1, may bind Flk1, and does bind the VEGF receptor (Flt4). VPR (also known as VEGF-C) can trigger the migration and mitogenesis of some human endothelial cells (48, 49). The protein we have described is unlikely to be VPR since the antibody we used does not identify VPR as it is raised against dissimilar amino acids. Furthermore, our gene expression data closely parallel the protein production results, and the cRNA for VEGF is specific for only this transcript within the family. Similar considerations hold for a recently identified third VEGF, VEGF-B, which in the mouse is 43% identical to mouse VEGF (50).

We found that the ability of either ET-1 or ET-3 to stimulate VEGF production was partially dependent on a PKC-related mechanism. Based upon previous studies from a number of investigators, both ET receptors couple to PKC activation through a G_q-related hydrolysis of phospholipase C, and this leads to various cell physiologic effects (reviewed in Ref. 14). PKC phosphorylation of substrate proteins like Raf-1 (51) can transduce VEGF transcription since phorbol ester can stimulate the transcription of the human VEGF gene, probably through AP-1 DNA-binding sites on the VEGF promoter (52). Hypoxia stimulates VEGF mRNA through a c-Src- and Raf-1-mediated mechanism (13), although hypoxia also has post-transcriptional effects (34). PKC, c-Src, and Raf-1 are kinases that ET can also activate, and as mentioned, Raf-1 is a substrate for PKC. It has not been determined in any model what are the signaling events downstream of PKC or Raf that lead to VEGF transcription. We provide the first evidence that MAP kinase (Erk2) can augment the transcription of this gene, and we further show that the stimulatory effect of ET-1 on VEGF transcription is critically dependent on the activation of Erk. Therefore, we propose that one important pathway for the effects of ET is the stimulation of PKC, leading to the activation of MAP kinase. This would presumably occur through activation of Raf-1 by PKC (51). Recently, we showed that ANP can inhibit ET-stimulated PKC and Erk activity (35), indicating one means by which ANP probably inhibits ET-induced VEGF. Additional signaling pathways through Erk are likely to contribute to the complete mechanism by which ET stimulates and ANP inhibits VEGF production.

ET might also play a role in the action of VEGF. Friedlander et al. (53) recently showed that VEGF-induced angiogenesis is strongly dependent on _v5 integrin expression and signaling in EC. This results from integrin interactions with endothelial adhesion molecules, such as E-selectin and vascular cell adhesion molecule-1 (54). ET-1 has been shown to induce neutrophil adhesion to endothelial cells through an integrin-related mechanism (55), and ET-1 can stimulate E-selectin, vascular cell adhesion molecule-1, and intercellular adhesion molecule-1 production in cerebral microvascular endothelial cells (56). Therefore, if these mechanisms are pervasive throughout the vasculature, ET-1 would be expected to aid VEGF-induced angiogenesis through these interactions in vivo.

We found that the natriuretic peptides strongly inhibit the stimulation of VEGF transcription and protein production by either hypoxia or ET. This suggests that the NP have both direct (18, 57) and indirect effects as antiproliferation factors for EC, the latter potentially mediated through modulating VEGF synthesis. Normally, ANP is secreted from the heart and circulates in plasma (16), whereas CNP is synthesized in the endothelial cell (17) and has been postulated to play an important paracrine role as a growth inhibitor for VSMC (58, 59). Our findings indicate that both locally produced and circulating natriuretic peptides can inhibit VEGF production. We also found that ANP inhibited the ET-stimulated secretion of VEGF and the subsequent EC proliferation induced by this growth factor in our model of "in vivo" vascular cell interaction. Enhanced VEGF production, induced by ET, also led to increased invasion by EC of the extracellular matrix, and this was inhibited by ANP. This latter finding probably resulted from less VEGF production stimulated in the presence of ANP since medium from cells incubated with ANP alone did not affect invasion. The ability of the NP to inhibit VEGF production could lead to strategies to prevent the development of diabetic proliferative retinopathy (5) or tumor metastasis (6), both of which are dependent on VEGF-induced angiogenesis. This is important since there are no currently available specific receptor antagonists of VEGF action. Interestingly, VEGF has recently been shown to inhibit CNP production in EC, suggesting a paracrine interaction between VSMC and EC that modulates vascular cell proliferation (60).

The receptor that mediates the approximately equipotent actions of the various NP in inhibiting VEGF production is of interest. ANP activates the guanylate cyclase A receptor, whereas CNP binds the guanylate cyclase B receptor; both peptides also bind to the natriuretic peptide clearance receptor, which is abundantly expressed on cultured VSMC (61). Most known physiologic effects of the NP appear to occur after binding the guanylate cyclase receptors. We propose that the NP act through the clearance receptor to modulate the in vitro production of VEGF. This is based upon finding that 1) C-ANP-(4-23), which binds only to the natriuretic peptide clearance receptor and does not generate cGMP at the concentrations used in these studies, inhibited ET-stimulated VEGF; 2) C-ANP-(4-23) was equipotent to ANP or CNP in inhibiting VEGF production, and the only common receptor for these peptides is the natriuretic peptide clearance receptor; and 3) the effects of ANP or CNP were not reversed by the cGMP inhibitor LY 53853. We determined that the LY compound caused an ~55% inhibition of ANP- or CNP-stimulated cGMP generation in VSMC. This indicates that this compound significantly inhibits NP activation of the guanylate cyclase receptors, yet does not reverse the effects of the NP on VEGF production. Furthermore, we previously showed that this same concentration of LY 53853 completely inhibited the ability of ANP to stimulate CNP secretion from cultured vascular cells (62), yet here had no effect on VEGF production. These findings indicate that both types of NP receptors, known to be present on VSMC, can mediate distinct actions of ANP.

In summary, important vasoactive peptides regulate the production of VEGF and thereby modify VEGF-induced EC proliferation and invasion. Through these actions, ET or the NP may have novel physiologic or therapeutic functions in vascular remodeling and angiogenesis that underlie disease or wound-healing processes.