Stimulation of endothelin-1 gene expression by insulin in endothelial cells.

The present study characterized the regulation of the genetic expression of the vasoactive peptide endothelin-1 (ET-1) by insulin in bovine aortic endothelial cells. By RNA blot analysis, insulin (1.67 x 10(-8) M) increased ET-1 mRNA levels by 2.3-fold over the basal within 10 min and attained a maximum (5.3-fold increase) in 2 h. Dose-response studies showed that a maximum effect of insulin was reached at 1.67 x 10(-8) M although a significant increase can be observed at 1.66 x 10(-9) M. Radioligand receptor studies indicated that the affinity constant for insulin receptors on endothelial cells correlated closely with the dose response observed for ET-1 mRNA. The ET-1 mRNA half-life was estimated with actinomycin D studies to be 20 min in control cells and was not affected by insulin treatment. Moreover, the effects of phorbol 12-myristate 13-acetate (PMA) and insulin were additive in the induction of ET-1 gene expression. When protein kinase C in the bovine aortic endothelial cells was down-regulated by preincubation with 8 x 10(-7) M PMA for 24 or 48 h, insulin was still able to increase ET-1 mRNA levels whereas PMA was ineffective. Using a chloramphenicol acetyltransferase (CAT) fusion plasmid containing the CAT gene and the 5'-flanking region of the ET-1 gene (Lee, M. E., Bloch, K. D., Clifford, J. A., and Quertermous, T. (1990) J. Biol. Chem. 265, 10446-10450), we observed that 1.67 x 10(-8) M insulin increased CAT enzyme activity and mRNA levels. The insulin dose-response curve observed for CAT activity correlated with that observed for ET-1 mRNA levels. These results suggest that insulin stimulates expression of the ET-1 gene at the transcriptional level via its own receptors. This effect is mediated mostly through a protein kinase C-independent pathway, suggesting the existence of an insulin-responsive element in the ET-1 gene 5'-flanking sequence.

Stimulation of Endothelin-1 Gene Expression by Insulin in Endothelial Cells* (Received for publication, March 21, 1991) F. Javier Oliver$ §, Guadalupe de la Rubiatll, Edward P. FeenerS that the affinity constant for insulin receptors on endothelial cells correlated closely with the dose response observed for ET-1 mRNA. The ET-1 mRNA half-life was estimated with actinomycin D studies to be 20 min in control cells and was not affected by insulin treatment. Moreover, the effects of phorbol 12-myristate 13-acetate (PMA) and insulin were additive in the induction of ET-1 gene expression. When protein kinase C in the bovine aortic endothelial cells was downregulated by preincubation with 8 X M PMA for 24 or 48 h, insulin was still able to increase ET-1 mRNA levels whereas PMA was ineffective. Using a chloramphenicol acetyltransferase (CAT) fusion plasmid containing the CAT gene and the 5'-flanking region of the ET-1 gene (Lee, M. E., Bloch 265, 10446-10450), we observed that 1.67 X lo-' M insulin increased CAT enzyme activity and mRNA levels. The insulin dose-response curve observed for CAT activity correlated with that observed for ET-1 mRNA levels. These results suggest that insulin stimulates expression of the ET-1 gene at the transcriptional level via its own receptors. This effect is mediated mostly through a protein kinase C-independent pathway, suggesting the existence of an insulin-responsive element in the ET-1 gene 5"flanking sequence.
* This work was supported in part by National Institutes of Health Grant EYO-5110, by the Lions Club International, and by Diabetes Endocrinology Research Center Grant. DK36836. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Recipients of a postdoctoral fellowship from the Spanish Ministry of Education and Science/Fulbright. Insulin resistance and diabetic states have been associated with hypertension, but the biochemical mechanism for their linkage is unclear. Since hyperinsulinemia is often described in both of these clinical states, a direct effect of insulin on the arterial wall has been implicated (1). In vascular cells, insulin receptors have been found on both endothelial and smooth muscle cells (1). Biologically, these cells are also responsive to insulin, with both metabolic and growth effects (1).
In this report we have characterized the effects of insulin on the expression of endothelin-1 (ET-11,' a potent vasoconstrictor peptide (2), in aortic endothelial cells. ET-1 was first reported in the supernatant of cultured porcine aortic endothelial cells (Z), although recently it has been isolated from other sources (3)(4)(5). There are three types of endothelins (ET-1, ET-2, and ET-3) with very similar structure. Although all three endothelins have 21 amino acids wit.h two disulfide intrachain bonds, each has different biological activities and potencies (6). Human ET-1 is synthesized as a 212-amino acid prepropeptide and converted into endothelin after two endopeptidase cleavages (2). The sequence of the human preproendothelin-1 and its chromosomal assignment have been reported (7). Several agents including the platelet-derived polypeptide transforming growth factor-P (8), thrombin (2), and phorbol ester (9) have been reported to regulate the expression of ET-1. In addition, cis-regulatory elements in the 5'-flanking region including the AP-ljc-jun-responsive element (lo), the nuclear factor NF-1 binding motif (IO), and the TATC (or GATA) motif (11) have been described.
In this study we have focused on the stimulatory effects of insulin on the expression of the ET-1 gene at the mRNA levels and showed that this effect of insulin is on transcription of the ET-1 gene.

MATERIALS AND METHODS
Cell Culture-Bovine aortic endothelial cells (BAE) were isolated as described previously (1) and cultured in Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% plasma donor-derived horse serum (PDHS) (GIBCO) in fibronectin-coated 100-mm plates (Nunc, Roskilde, Denmark). Cells were passed every 4-5 days, and confluent dishes of passages 4-12 were used for all the experiments. Two days before the assay, the medium was changed to 2% PDHS-supplemented DMEM, and all serum was removed 3 h before the addition of phorbol 12-myristate 13-acetate (PMA), human in- Extraction and Analysis of RNA-Total cellular RNA was extracted by the method of Chomczymski and Sacchi (12) using the commercial preparation RNAzol (Biotech, Inc., Houston, TX). For cell harvesting, medium was aspirated, and cells were washed once with phosphate-buffered saline, p H 7.4. RNAzol was added immediately, and cells were scraped off with a rubber policeman. The solution was transferred to 15-ml polyethylene tubes and treated with 0.10 volume of chloroform, shaken vigorously, and incubated for 15 min at 0 "C for 15 min. This suspension was centrifuged a t 10,000 X g for 15 min at 4 "C. The resulting aqueous phase was treated with 1 volume of chilled 2-propanol and incubated for 1 h at -20 "C. Precipitated RNA was collected by centrifugation at 10,000 X g for 15 min a t 4 "C, and the pellet was washed twice with 80% ethanol, dried under vacuum, and resuspended in diethyl pyrocarbonate-treated water. RNA to be used for S1 analysis was ethanol precipitated once more and then resuspended in diethyl pyrocarbonate-treated water.
Partial Purification and PKC Assay-Isolation of cytosolic and membranous fractions was performed as described previously (13). Briefly, cells were scraped from the dish in buffer A (20 mM Tris, pH 7.5, 2 mM EDTA, 0.5 mM EGTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 0.33 M sucrose, and 25 pg/ml leupeptin) and homogenized with a tight fitting Dounce homogenizer. The postnuclear supernatant was centrifuged a t 100,000 X g for 60 min, and the soluble fraction was defined as the cytosolic extract. The pellet was resuspended in buffer B (buffer A without sucrose) containing 1% Triton X-100, incubated a t 4 "C for 30 min, and recentrifuged. The soluble fraction was defined as the membranous extract. Both cytosolic and membranous fractions were bound to DE52cellulose, washed with buffer B, and eluted off DE52 with buffer B containing 200 mM NaCl.
Protein kinase C activity was measured as described previously (13) and defined as the Caz+, phosphatidylserine and diacylglycerolstimulated transfer of 32P from [Y-~*P]ATP into the octapeptide (RKRTLRRL) corresponding to residues 651-658 of the epidermal growth factor receptor. Protein determination was performed according to the method of Bradford (14).
Zmmunoblotting-PKC isoform-specific polyclonal antibodies were developed against synthetic peptides corresponding to residues 310-334 for PKC-a and 645-673 for PKC-@, in collaboration with Dr. William Heath, Lilly. These antibodies have been characterized previously (15).
Proteins from cytosolic and membranous DE52 fractions were separated on 7.5% SDS-polyacrylamide gel electrophoresis (10 pg/ lane) and transferred to Schleicher & Schuell nitrocellulose. Nitrocellulose was blocked with 3% bovine serum albumin in rinse buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% Triton X-100) at 4 'C for 16 h. Nitrocellulose paper was incubated with 2 pg/ ml affinity-purified antibody against a or p2 at 4 "C for 16 h. Blots were then incubated with '251-protein A (Amersham Corp.) a t 4 "C overnight and washed four times with rinse buffer containing 1.0 M NaCI. Results were visualized by autoradiography.
Plasmids Constructs and CAT Assay-Reporter plasmid was constructed by cloning the procaryotic chloramphenicol acetyltransferase (CAT) gene downstream of a 4.4-kilobase (kb) ET-1 promoter fragment (16). The @-galactosidase gene was employed in control plasmids driven by the Rous sarcoma virus long terminal repeat. Detailed information with respect to plasmid constructs has been published (16).
BAE cells were transfected with 20 pg of the ET-1 CAT fusion construct plasmid by the calcium phosphate method as previously described (16). To correct for variability in transfection efficiency, 10 pg of pRSV @-galactosidase plasmid DNA was co-transfected in all experiments. Cell extract preparation and the CAT and @-galactosidase assay methods have been published elsewhere (17,18). The ratio of CAT activity to @-galactosidase activity in each sample served as a measure of normalized CAT activity. No effect of insulin was observed on the /%galactosidase activity. For each series of experiments, the normalized CAT activity of each sample was divided by that of a control construct, and the quotient was expressed in relative CAT units.
SI Analysis-The procedure followed for S1 analysis of the CAT mRNA has been described previously (19,20), except that RNA was prepared by the RNAzol procedure (see above). Briefly, 20 pg of RNA was probed for CAT mRNA with a single-stranded DNA probe uniformly labeled with ["PIdCTP (3,000 Ci/mmol) during synthesis from M13 phage DNA. After overnight hybridization in solution a t 42 "C samples were treated with S1 nuclease and run in a 6% polyacrylamide denaturing gel. The gel was autoradiographed a t -70 "C, and results were quantified by densitometry.

RESULTS
Stimulatory Effect of Insulin on ET-1 Gene Expression-Endothelial cells from macro-and microvessels have been shown to be responsive to insulin or contain functional insulin receptors (1). The presence of a functional insulin receptor in our BAE cell preparation was characterized by insulin receptor binding studies using 1z51-insulin, which showed 7.9% binding/mg of protein, in keeping with previously published reports. Insulin at 1.25 x lo-' M displaced 50% of the specific binding, showing high affinity receptors (1,21).
When BAE cells were incubated in the presence of insulin, time-dependent and dose-response increases in the steadystate levels of ET-1 mRNA were observed (Figs. 1 and 2). After hybridization with the ET-1 probe, nitrocellulose membranes were washed and rehybridized with a mouse a-tubulin cDNA probe. The signal for a-tubulin obtained by densitometry was used to normalize the values obtained for ET-1 mRNA. Fig. 1A shows the time course of insulin's effect on ET-1 (2.3-kb band) and a-tubulin (2.1 kb) mRNA. Insulin at 1.67 X lo-' M increased ET-1 mRNA by 2.3-fold within 10 min and 5.3-fold within 2 h (Fig. 1B). After 24 h the stimulatory effect of insulin was still significant (data not shown).
A dose-response curve for insulin was determined by measuring the expression of ET-1 mRNA in BAE cells with a range of insulin concentrations (Fig. 2, A

Effect of Insulin and ['MA on E T -] Grnr Gxprrxsion in
Control and PKC-doclln-rcgulatrd RAE Crll.s-1t has heen proposed that act.ivation of I'KC may he involved in the regulation of the ET-1 gene expression in endothelial cells since there is a n AI'-1 consensus motif in the promoter region. In addition, phorhol esters have heen shown to have a stimulatory effect. on mRNA levels (9, 10). Thus, we decided to investigate whether PKC has a role mediating the effect of insulin on the expression of the ET-1 gene. PKC activity was down-regulated in RAE cells after 24 h of incuhation with 800 nM {'MA ( Fig. 4 and Tahle I). Tahle I shows that PKC activity after PMA treatment was decreased hy 83";' in the membrane-associated activity. No increase in PKC activities was observed when stimulation was repeated with 160 nM PMA. Western hlot. from hoth control and down-regulated BAE cells was performed. Among PKC isoforms (r, /j,, /j2, and 7 , only (Y and 13, were identified (Fig. 5 ) . In down-regulated BAE cells, hoth PKC isoforms were dramatically decreased by immunohlot, corresponding to the decrease in activity levels. When control cells were incuhated in the presence of 100 ng/ml insulin (1.67 X IO-" M ) or 160 nM PMA for 30 min, increases of 2.1 0.1 and 3.3 k 0.9 fold, respectively, of E T -1 mRNA levels were ohserved. When hoth agents were combined, an additive effect was detected (5.5 l.:i-fold increase). In PKC down-regulated cells, only insulin was able to increase ET-I mRNA expression although the effect was smaller than A . ..   Total cellular RNA was harvested 48 h after transfection, and CAT mRNA was measured by S1 nuclease protection assay. HAE cells were exposed to 1.67 X lo-" M insulin or vehicle for 3 h. The prohe was generated using an insert from the adenovirus EZ promoter linked to the CAT gene within M l 3 m p l 8 (20). The probe was produced hy liberation from M I 3 hy EroRI digestion is 367 nucleotides in length, 150 of which would he protected by CAT mRNA (indicated hy the arrow). The results shown are representative of two independent S1 protection assays.

Kb
lation of the secretion of these factors could be critical in the maintenance of the tonus required in various physiological situations (2). In this report we have found that insulin is a potent regulator of ET-1 gene expression in aortic endothelial cells. This effect of insulin on ET-1 mRNA is probably mediated by insulin receptors, which have been reported on both macro-and microvascular cells with structure and actions similar to receptors found in other peripheral cells such as fibroblasts and hepatocytes (1,21). These receptors contain functional tyrosine kinase activity and have been shown to mediate metabolic effects in capillary endot.helial cells. The concentrat,ion range observed in the dose-dependent increase in ET-1 transcription correlated closely with the affinity constant of the insulin receptor, suggesting that this action of insulin is mediated by binding to its own receptor. It is very unlikely that insulin is exerting this effect via the insulinlike growth factor-1 receptor, in view of the low affinity of the insulin-like growth factor-1 receptor for insulin (21). Furthermore, studies using specific antibodies to the insulin receptor were also able to enhance ET-1 gene expression (data not shown). Recause of our previous finding that endothelial cells from capillaries are more responsive to insulin than those from macrovessels ( l ) , we would predict that insulin should also be able to regulate ET-1 expression in microvessel endothelial cells, which have been reported to produce ET-1 (23). Our data showed that the addition of insulin did not alter the ET-1 mRNA half-life indicating that changes in mRNA levels may be a result of an increase in transcription rate. This possibility is firmly supported by the results of transient transfection studies in endothelial cells using constructs which utilized the ET-1 promoter linked to the CAT reporter gene. Insulin increased CAT expression to a similar extent and in a similar dose-response manner as found by Northern blot analysis for ET-I mKNA. These findings strongly suggest that the insulin agonist effect is mediated at the transcription level and not by prolonging mRNA half-life. This effect of insulin is not unusual since this hormone has been reported to regulate the transcription rate of genes such as glucokinase, pyruvate kinase, amylase, glyceraldehyde-:~-phosphate dehydrogenase phosphoenolpyruvate carboxykinase, c-/os, and cmyc (24)(25)(26)(27)(28)(29)(30)(31)(32). In some cases, such as amylase and c-/os activation, the effect of insulin was reported to occur in less than 30 min (28,29), similar to the rapid effect seen here with ET-1. The existence of cis-acting regulatory sequences mediating the effect of insulin on the expression of several genes have been shown previously. The best characterized mechanism of insulin regulation of gene expression is that associated with the gluconeogenic enzyme phosphoenolpyruvate carhoxvkinase (30)(31)(32). Recently, a 5"specific insulin-responsive element (IRE) has been recognized to direct this hormonal effect on phosphoenolpyruvate carboxykinase gene expression, and a specific DNA-binding protein has heen implicated (31 ).
A different IRE has been characterized in association with the glyceraldehyde-%phosphate dehydrogenase gene Rased on an analysis of nucleotide sequence critical for formation of an insulin-sensitive protein-DNA complex, the authors proposed a minimal structural IRE. They found that 8 consecutive bases of the motif are perfectly conserved in the promoter region of other insulin-responsive genes. In the *5'flanking region of the ET-I gene at base pairs -2988 to 2979 we have found a sequence, CCCGCATCTC, that matches 9 bases of the proposed IRE (the underlined A is a C in the GADPH IRE). The ET-1 gene contains a second sequence at base pairs -136 t o -145 (GGCCTCGCCT) which has been found in association with other genes regulated by insulin ( 2 7 , 30,36). The functional meaning of this second motif is unknown. It is possible that one of these sequences functions as a cis-acting element which mediates the insulin responsiveness of the ET-1 gene. This hypothesis is currently under study in our laboratory.
T h e act,ion of insulin appears to differ from the effect of PMA in several aspects. As reported previously.
we have confirmed that. PMA, an activator of protein kinase C, will also increase ET-1 mRNA levels in steady state (9, IO). Our actinomycin D experiments suggest that the effect of PMA may be to increase the stability of the mRNA although more studies are needed to support this speculation. However, it is also possible that PMA could be increasing transcription rate as well since previous reports describing the promoter region for ET-1 have identified an 8-base pair AI'-I consensus sequence. Such sequences have been suggested to mediate responsiveness to phorbol esters, probably by activating I'KC (34, 3 5 ) . This mechanism, nevertheless, is probably not the main pathway by which insulin activates the expression of ET-1 gene, since reducing the activities of I'KC in the entlo-

Effect of
Insulin on ET-1 Expression thelia1 cells did not prevent the effect of insulin although it made the cells unresponsive to PMA. Immunoblotting studies have provided further supportive evidence that a majority of the isoenzymes have been removed. Similar conclusions were drawn by Stumpo and Blackshear (28) regarding the transducing pathway for the effect of insulin on the response of c-10s in PKC-depleted 3T3-Ll fibroblasts and adipocytes, in which they showed a rapid accumulation of this mRNA species in PKC-deficient cells in response to insulin. The data from the down-regulation experiments need to be evaluated with care since PKC activities have not been completely removed and PMA may not have been completely worked out in the restimulation studies. However, the data are clear with respect to the fact that some of the effect of insulin is not mediated in PKC activation. This conclusion is based on the fact that the down-regulated endothelial cells are still responsive to insulin but are refractory to a supermaximal concentration of PMA.
Clinically, the finding that insulin at physiological levels can enhance ET-1 expression could be important for the understanding of the relationship between hyperinsulinemia and vascular disorders in diabetic patients such as hypertension and acceleration of atherosclerosis (37). Recently, Yamaguchi et al. (23) reported that endothelial cells exposed to elevated glucose produced a greater amount of ET-1. The mechanism of this effect of glucose was not identified, but it could be mediated by PKC. Activation of PKC has been shown previously to occur in endothelial cells exposed to high glucose levels (15). Thus, it is tempting to speculate that a combination of hyperinsulinemia and hyperglycemia will increase ET-1 expression in a paracrine fashion in the diabetic state. The resultant increase in ET-1 expression may have a role in causing some of the vascular abnormalities found in diabetic patients.