Dehydroepiandrosterone activates endothelial cell nitric-oxide synthase by a specific plasma membrane receptor coupled to Galpha(i2,3).

The adrenal steroid dehydroepiandrosterone (DHEA) has no known cellular receptor or unifying mechanism of action, despite evidence suggesting beneficial vascular effects in humans. Based on previous data from our laboratory, we hypothesized that DHEA binds to specific cell-surface receptors to activate intracellular G-proteins and endothelial nitric-oxide synthase (eNOS). We now pharmacologically characterize a putative plasma membrane DHEA receptor and define its associated G-proteins. The [3H]DHEA binding to isolated plasma membranes from bovine aortic endothelial cells was of high affinity (K(d) = 48.7 pm) and saturable (B(max) = 500 fmol/mg protein). Structurally related steroids failed to compete with DHEA for binding. The putative DHEA receptor was functionally coupled to G-proteins, because guanosine 5'-O-(3-thio)triphosphate (GTPgammaS) inhibited [3H]DHEA binding to plasma membranes by 69%, and DHEA increased [35S]GTPgammaS binding by 157%. DHEA stimulated [35S]GTPgammaS binding to Galpha(i2) and Galpha(i3), but not to Galpha(i1) or Galpha(o). Pretreatment of plasma membranes with antibody to Galpha(i2) or Galpha(i3), but not to Galpha(i1), inhibited the DHEA activation of eNOS. Thus, DHEA receptors are expressed on endothelial cell plasma membranes and are coupled to eNOS activity through Galpha(i2) and Galpha(i3). These novel findings should allow us to isolate the putative receptor and reevaluate the physiological role of DHEA activity.

The physiological role of the adrenal steroid dehydroepiandrosterone (DHEA) 1 is not known. There are widespread data suggesting a beneficial effect of DHEA on vascular function. Extensive epidemiologic evidence shows an inverse correlation between circulating DHEA levels and the prevalence of atherosclerotic and cardiovascular diseases (1)(2)(3)(4)(5). There are few human intervention studies focused on vascular outcomes of DHEA administration, and these are not of a size or duration to define whether DHEA therapy has an effect on cardiovascular morbidity or mortality. Available studies do suggest a beneficial effect on atherosclerosis (6). Studies of the short term effect of DHEA on human vascular function, using sophisticated assays of vascular function, are beginning to emerge. Williams et al. (7) showed a significant increase in flow-mediated dilatation and systemic arterial compliance in postmenopausal women taking DHEA for 3 months. DHEA reduces atherosclerosis, decreases the accumulation of cholesterol in aortic and coronary arteries (8,9), and inhibits platelet aggregation (10) in various animal models. DHEA also affects growth factor-induced mitogenesis and proliferation of vascular smooth muscle cells (11)(12)(13). However, the molecular mechanisms by which DHEA acts to protect from atherosclerotic and cardiovascular diseases are still unknown. Furthermore, it is unclear whether the effect on vascular tissues is related to DHEA or to its metabolites, which include estradiol.
Steroid hormones are known to bind specific intracellular receptors, which function as ligand-dependent gene transcription factors (14). However, previous efforts to isolate an intracellular receptor for DHEA have failed (15)(16)(17)(18). In contrast to this classical pathway of steroid hormone action, there are also rapid, plasma membrane-dependent, non-genomic effects of steroids in various tissues, which lead to important physiological responses (19 -24). Plasma membrane-associated receptors are postulated to mediate these non-genomic actions of steroids. Functional plasma membrane binding sites have been identified for several steroids, including estrogen, vitamin D, and progesterone (25)(26)(27)(28). However, besides the receptor for estrogen, no plasma membrane steroid receptor has yet been unequivocally identified and characterized.
We have found that DHEA stimulates nitric oxide (NO) generation within minutes from bovine aortic endothelial cells (BAEC). 2 Furthermore, DHEA conjugated to bovine serum albumin (BSA) had similar effects. These cellular responses to DHEA were specific and inhibited by pertussis toxin (PTX). Taken together, these results led to the hypothesis that DHEA activates NO production in endothelial cells by a specific, plasma membrane, G-protein-coupled receptor. The aim of this study was to determine whether there is a specific plasma membrane DHEA receptor, and to pharmacologically characterize that receptor. We show, for the first time, a specific, high affinity, DHEA binding site on the plasma membrane of BAEC. Specific DHEA binding to this site was saturable and reversible. We propose this distinct site of DHEA action as a novel DHEA receptor. In addition, we show that the putative DHEA receptor was functionally coupled to G-proteins of the G␣ i2, 3 subtypes, which mediate the activation of endothelial nitricoxide synthase (eNOS) by DHEA in BAEC.

EXPERIMENTAL PROCEDURES
Materials-All tissue culture media and reagents were supplied by the Diabetes and Endocrinology Research Center of the University of Iowa. DHEA, DHEAS, and DHEA-17-carboxylmethyl oxime-BSA complexes were obtained from Steraloids (Newport, RI). 16␣-Fluoro-5-androsten-17-one (fluasterone) was a gift from Dr. Arthur Schwartz (Temple University, Philadelphia, PA). Other steroids, chemical reagents, PTX, preimmune rabbit serum, IgG, protein A-Sepharose, DEAE glass fiber filters, and assay kits for plasma membrane and cytosolic markers were from Sigma. Radioisotopes and the chemiluminescence detection kit (ECL) were obtained from Amersham Biosciences. The polyclonal rabbit antisera to G␣ o and G␣ i were supplied by Santa Cruz Biotechnology (Santa Cruz, CA), and antisera specific to G␣ i1 , G␣ i2 , and G␣ i3 were purchased from Calbiochem (San Diego, CA). Whatman GF/B filters were supplied by Fisher. Protein molecular weight markers, nitrocellulose transfer membranes, Dowex 50 WX-8 resin, and protein assay reagents were purchased from Bio-Rad.
Cell Culture-Bovine aortic endothelial cells (BAEC) were kindly provided by Dr. Robert Bar (Veterans Affairs Medical Center, Iowa City, IA) and human umbilical vein endothelial cells (HUVEC) were provided by Dr. Arthur Spector (University of Iowa, Iowa City, IA). Both BAEC and HUVEC were grown in M199 medium supplemented with 20% fetal calf serum, 50 units/ml penicillin, and 0.05 mg/ml streptomycin, and incubated at 37°C in a 5% CO 2 , 95% air environment. The medium was changed every other day until the cells became confluent. Cells were serially passaged, after brief incubation with 0.25% trypsin-EDTA, and used at passages 4 -8 in all experiments.
Isolation of Plasma Membranes-Purified plasma membranes from BAEC were isolated by sucrose gradient centrifugation, as described previously (29). In some experiments, cytosolic fractions were prepared by differential centrifugation (30). Briefly, cells were homogenized with a Dounce homogenizer in HES buffer (20 mM HEPES, 250 mM sucrose, 1 mM EDTA, 5 mM benzamidine, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), and 1 M each of pepstatin A, aprotinin, and leupeptin). Post-nuclear homogenates were collected following centrifugation at 200 ϫ g for 2 min, and the microsomal aggregates were sedimented by centrifugation at 3,000 ϫ g for 15 min. Plasma membranes from HU-VEC, rat liver, heart, and kidney were isolated by sucrose gradient centrifugation (29,31,32). Plasma membrane purity and exclusion of cytosolic fraction were assessed with measurements of alkaline phosphatase, 5Ј-nucleotidase, and lactate dehydrogenase, using commercial kits. Protein was determined by the method of Lowry (33) using BSA as a standard.
Solubilization of Plasma Membrane-Plasma membranes, prepared as described above, were centrifuged for 30 min at 14,000 ϫ g. The pellet was solubilized in buffer (25 mM HEPES, pH 7.4, 100 mM NaCl, 1 mM EDTA, 2 mM MgCl 2 , 10% (v/v) glycerol, 1% (v/v) Triton X-100, 0.1 mM PMSF, 1 M each of aprotinin, leupeptin, and pepstatin A) for 1 h at 4°C, with gentle stirring. The solubilized preparation was centrifuged at 110,000 ϫ g for 45 min at 4°C. The supernatant and remaining pellet were collected for binding assays. Protein concentration was determined, and ϳ65% of the membrane protein preparation was solubilized. S]GTP␥S binding assay was performed at 25°C in a total volume of 250 l. Plasma membranes (8 g of protein) were added in binding buffer containing 10 M GDP and 0.5 nM [ 35 S]GTP␥S (ϳ11,000 cpm, 1,082 Ci/mmol). The incubation was carried out with gentle shaking for 30 min, in the absence or presence of 1 pM to 1 M DHEA. Nonspecific binding was assayed by addition of 10 M GTP␥S and was less than 10% of total binding. After the incubation period, duplicate 100-l aliquots were subjected to rapid filtration under vacuum through Whatman GF/B glass fiber filters, followed by three washes with 1 ml of ice-cold binding buffer. Bound radioactivity was determined by liquid scintillation counting after overnight extraction of the filters in 4 ml of Biosafe N/A scintillation mixture. Specific binding data represent the difference between total binding and nonspecific binding. All assays were performed four times. , and G i3 (1:500)). After 6 h, 50 l of protein A-Sepharose (1 mg/ml) was added to each sample and the mixture incubated overnight at 4°C with gentle mixing. The immunoprecipitates were collected following centrifugation at 12,000 ϫ g for 2 min and washed three times in a buffer containing 50 mM HEPES, pH 7.4, 100 M NaF, 50 mM sodium phosphate, 100 mM NaCl, 1% Triton X-100, and 0.1% SDS. The final pellets were boiled in 0.5% SDS, and the immunoprecipitated [ 35 S]GTP␥S-labeled G-proteins were counted in a scintillation counter. Nonspecific activity was determined in the presence of 50 M unlabeled GTP␥S.

Immunoprecipitation of [ 35 S]GTP␥S-labeled G-proteins-Plasma
Western Blotting-Twenty g of BAEC plasma membrane proteins and molecular weight markers were separated by 12.5% SDS-polyacrylamide gel electrophoresis as described previously (34). The proteins were electrophoretically transferred onto nitrocellulose membranes. The membranes were blocked at 25°C for 1 h with 5% nonfat milk and 0.1% Tween 20 in Tris-buffered saline, and then incubated overnight at 4°C with the following dilutions of specifically reacting rabbit polyclonal antisera: G␣ o (1:400), G␣ i1 (1:1,000), G␣ i2 (1:1,000), and G␣ i3 (1:1,000). The membranes were washed four times with Tris-buffered saline buffer containing 0.1% Tween 20 and then blocked for 10 min, prior to the addition of donkey anti-rabbit horseradish peroxidase secondary antibody (1:10,000). Chemiluminescence reaction was performed before exposure to radiographic film.
NO Synthase Activation-Endothelial NOS activity was determined by measuring the conversion of L-[ 3 H]arginine to L-[ 3 H]citrulline, as previously described (34 -36), with minor modifications. Briefly, purified plasma membranes were reconstituted in Hepes-buffered saline solution buffer (135 mM NaCl, 1.2 mM CaCl 2 , 1.2 mM MgCl 2 , 5.0 mM KOH, 10 mM HEPES, 10 mM glucose, pH 7.4) supplemented with 100 g/ml bacitracin, 0.1 mM PMSF, and 1 M each of pepstatin A, leupeptin, and aprotinin. Plasma membranes (8 g in a 100-l volume) were incubated for 15 min at 37°C with 1.5 Ci/ml L-[ 3 H]arginine in the presence or absence of DHEA (1 nM), and citrulline production was measured. Studies were also performed in the presence of added calmodulin (5 units/ml), ␤-NADPH (2 mM/liter), tetrahydrobiopterin (2 M/liter), flavin adenine dinucleotide (10 M/liter), flavin mononucleotide (10 M/liter), and CaCl 2 in excess of EDTA (36). For PTX experiments, the BAEC were pre-incubated in 100 ng/ml PTX for 6 h prior to isolation of the plasma membranes. In selected studies, plasma membranes were treated with DHEA (1 nM) or 17␤-estradiol (10 nM) in the absence or presence of the estrogen antagonist ICI 182,780 (1 M) for 15 min, without added calmodulin or enzyme cofactors. The reaction was terminated by the addition of HEPES buffer, pH 5.5, containing EDTA (5 mM/liter), and solutions were immediately applied to a column containing 2 ml of Dowex 50 WX-8 resin (Na ϩ form), added as a 1:1 slurry in water. L-[ 3 H]Citrulline was then eluted with 2 ml of HEPES buffer. A sample of eluate in Biosafe N/A scintillation mixture was counted by liquid scintillation. Nonspecific activity was determined by the addition of excess L-arginine (10 M) and represented ϳ10% of total activity. Endothelial NOS activation by the known agonist bradykinin (1 M) was also tested, as a positive control. In individual experiments, each treatment condition was determined in duplicate and all findings were repeated in four independent studies.
Statistical Analysis-Binding data of [ 3 H]DHEA were analyzed by nonlinear regression using the Prizm GraphPad program (GraphPad Software, San Diego, CA). Affinity (K d ) and maximum binding capacity of [ 3 H]DHEA were calculated from nonlinear curve fitting. Results were analyzed using a one-site model because this was superior to other models tested. All other data were subjected to one-way analysis of variance using the General Linear Model procedure of SAS ® , and significant differences were subjected to Duncan's multiple comparison test at 5% probability. All values in each study were derived from at least three separate experiments and expressed as mean Ϯ S.E.

RESULTS
Plasma Membrane Purity-Plasma membrane fractions from BAEC were isolated by homogenization and sucrose gradient centrifugation as previously described (29). The activities of plasma membrane-specific enzymes, 5Ј-nucleotidase and alkaline phosphatase, were 13-15-fold higher in the plasma membrane preparations than in the post-nuclear homogenate ( Table I). The activity of the cytosolic marker, lactate dehydrogenase, was negligible in the plasma membrane preparations compared with that in the postnuclear homogenate fractions. These data confirm that the plasma membrane samples had little contamination with non-membrane proteins.

Binding of [ 3 H]DHEA to BAEC Plasma Membranes-
The association kinetic data showed that the binding of [ 3 H]DHEA to plasma membrane fractions occurred rapidly at 4°C and pH 7.4 ( Fig. 1), reaching a plateau at 5 min of incubation and remaining essentially stable for 90 min. To determine the affinity constant, we incubated each sample (8 g of protein in 250 l) with increasing concentrations of [ 3 H]DHEA, from 5 pM to 2 nM in the absence or presence of 10 M unlabeled DHEA. The saturation binding curve and Scatchard analysis of the data indicated a single high affinity membrane binding site, with a K d ϭ 48.7 Ϯ 4.6 pM and B max ϭ 500.3 Ϯ 8.7 fmol/mg protein (Fig. 2).
To determine whether high affinity DHEA binding sites exist in tissues other than the vascular endothelium, we isolated plasma membranes from rat liver, kidney, and heart. There was no specific binding detected in rat kidney (Fig. 3). The binding activity was higher in heart than in liver, but was only 39.8 and 23.5%, respectively, of the specific binding activity observed in BAEC plasma membranes. Binding to the heart and liver plasma membranes reached saturation at 45 min of incubation and remained constant to 60 min (data not shown). These results suggest that the high affinity binding sites for DHEA are expressed in a differential tissue-specific manner and may not be expressed in all tissues. Furthermore, HUVEC plasma membranes express approximately the same degree of  Fig. 5A, the specific binding of [ 3 H]DHEA to plasma membranes was reduced by 69% after preincubation with GTP␥S for 30 min. These data showed that the binding of [ 3 H]DHEA to membranes was sensitive to GTP␥S, which indicates coupling of membrane-bound DHEA receptors to G-proteins. Consistent with our previous findings, pre-incubation of the plasma membranes with PTX inhibited specific [ 3 H]DHEA binding by 85% (Fig. 5B).
Effect of DHEA on [ 35 S]GTP␥S Binding-Ligand-regulated GTP␥S binding to G-proteins is a sensitive method for examining the coupling of G-proteins to a receptor (37). To confirm that the putative DHEA receptors were coupled to G-proteins, we examined the effect of DHEA on [ 35 S]GTP␥S binding to G-proteins in plasma membranes. As shown in Fig. 6A 6B). Specific binding declined to zero by 2 h of incubation (data not shown). The stimulation of [ 35 S]GTP␥S binding was de-pendent on DHEA concentration between 1 pM and 1 M, with a maximal increase of 72% at 10 nM DHEA and an EC 50 of 43 pM (Fig. 6C). There was no significant difference in DHEAstimulated [ 35 S]GTP␥S binding between the 0.1, 1, and 10 nM DHEA concentrations. Pre-incubation of the plasma membranes with PTX reversed the DHEA-stimulated increase in [ 35 S]GTP␥S binding (Fig. 6D).
We solubilized membrane proteins with the nondenaturing detergent, Triton X-100 (1%), to investigate whether soluble DHEA receptors retain ligand binding affinity and functional coupling to G-proteins. Solubilization with Triton X-100 yields ϳ65% of membrane-bound proteins. In binding studies with soluble membrane proteins, DHEA actively and specifically bound to detergent-solubilized receptor, as illustrated in Fig.  7A. Furthermore, DHEA enhanced the [ 35 S]GTP␥S binding to G-proteins in the solubilized protein extract (Fig. 7B), indicating that intact and functional receptor-G-protein complex was solubilized. Taken together, these results provide evidence that the plasma membranes have putative DHEA receptors, which are coupled to G-proteins.  (Fig. 8). However, DHEA did not affect [ 35 S]GTP␥S binding to G␣ i1 and G␣ o . In DHEA-treated plasma membranes, G␣ i2 and G␣ i3 accounted for approximately 53 and 42%, respectively, of total stimulated ␣-subunit GTP binding activities. Immunoblotting analysis revealed that G␣ i2 and G␣ i3 predominated in the BAEC membranes, with little G␣ i1 (Fig. 9). There was no G␣ o expressed in the plasma membranes. Taken to- gether, these findings provide the first evidence that putative DHEA plasma membrane receptors are selectively coupled to G␣ i2 and G␣ i3 .

Immunoprecipitation of [ 35 S]GTP␥S-labeled G-proteins-We
Endothelial NO Synthase Activation by DHEA-We have recently demonstrated that DHEA (0.1-10 nM) maximally stimulated NO production in intact endothelial cells. 2 Based on the immunoprecipitation studies above, we hypothesized that the putative DHEA receptors activate eNOS through G␣ i2 and G␣ i3 , to which they are coupled. To test this hypothesis, we measured eNOS activity in purified plasma membranes, in response to DHEA and in the presence of G␣ i subtype antibodies. In the absence of added calmodulin and cofactors, basal eNOS activity was low but detectable in membranes (1.37 pmol of citrulline/mg of protein). Under the same conditions, DHEA (1 nM) and the known eNOS agonist bradykinin (1 M) increased eNOS activity by 66 and 65%, respectively, compared with basal levels (Fig. 10A). The addition of calmodulin, calcium, and co-factors greatly increased the basal activity of eNOS. DHEA (1 nM), but not bradykinin, enhanced this activity by 42% (Fig. 10B). In the presence of calmodulin, calcium, and eNOS co-factors, DHEA stimulated eNOS activity in a concentration-dependent manner between 1 pM and 1 M, with a maximal increase of 78.3 Ϯ 9.3% at 10 nM DHEA and an EC 50 of 87 pM (Fig. 10C). Pre-incubation of the BAEC with PTX reversed the DHEA-stimulated increase in eNOS activity (Fig.  10D).
To determine the G␣ i subtype coupling of the putative receptor and eNOS, we preincubated solubilized plasma membranes with G␣ i1 , G␣ i2, or G␣ i3 antibodies or nonimmune IgG. Basal eNOS activity was not altered by nonimmune IgG. Endothelial NOS activation by DHEA was inhibited by 78.3% in membranes preincubated with G␣ i2 antibody and by 97.8% with G␣ i3 antibody, in comparison with the nonimmune IgG control. Activation of eNOS was completely blocked with antibody directed against all three G i proteins (Fig. 11). Nonimmune IgG or antibody to G␣ i1 had no effect on eNOS activation by DHEA. In addition, none of these antibodies prevented eNOS activation by bradykinin (data not shown).
Estradiol has been shown to stimulate NO production in endothelial cells (35). To determine whether the effect of DHEA on eNOS was mediated by an estrogen receptor, we evaluated the effect of the estrogen receptor antagonist ICI 182,780 in our assay. 17␤-Estradiol (10 nM) increased eNOS activity in plasma membranes by 77.3 Ϯ 8.2% (Fig. 12). ICI 182,780 attenuated this effect (44.2 Ϯ 1.2% increase), as previously observed by others (36). In contrast, ICI 182,780 did not inhibit eNOS activation by DHEA (71.3 Ϯ 6.6% (DHEA) versus 69.9 Ϯ 10.6% (DHEA ϩ ICI 182,780)) (Fig. 12). This confirms the data seen in our displacement studies above and suggests that DHEA does not interact with the vascular endothelial, plasma membrane, estrogen receptor. DISCUSSION Based on our previous studies, 2 we hypothesized that DHEA binds to specific receptors on the surface of endothelial cells to activate intracellular G-proteins and eNOS. To test this hypothesis, we used assays of radioligand binding, ligand-dependent G-protein activation, and eNOS activity. The principal findings of this study are that BAEC plasma membranes contained a specific, saturable, high affinity DHEA binding site. The binding site for DHEA had features of a G-protein-coupled receptor. DHEA binding was linked to activation of G-proteins (specifically G␣ i2 and G␣ i3 ) and was inhibited by incubation with nonhydrolyzable GTP analogs or PTX. Binding of DHEA to solubilized plasma membrane preparations was associated with activation of eNOS within those solubilized fragments, and specific G␣ i2 and G␣ i3 antibodies inhibited this activation. Taken together, we show for the first time strong evidence for a high affinity DHEA receptor in endothelial cells. Furthermore, we show that this putative receptor is expressed on plasma membranes, is G-protein-coupled, and activates endothelial cell NOS. This new information will form the basis for efforts to isolate a specific DHEA receptor. The findings may also allow us to develop a unifying mechanism for the physiological actions of DHEA.
The [ 3 H]DHEA binding activity in plasma membranes was differentially expressed, in a tissue-specific manner. Plasma membranes from rat liver and heart expressed ϳ30% of the specific DHEA binding of BAEC membranes, whereas rat kidney membranes had no detectable binding. These results, demonstrating quantitative differences in expression of functional binding sites in different cellular systems, may be relevant in determining the physiological function of DHEA. Furthermore, the high level of expression in HUVEC suggests that the hu-man cardiovascular system may be an important target of the membrane-initiated effects of DHEA.
The results from ligand competition experiments demonstrated that the binding of [ 3 H]DHEA to plasma membranes was highly specific. Closely related steroid structures, DHEAS, androstenedione, 17␣-hydroxypregnenalone, testosterone, and 17␤-estradiol, did not compete with [ 3 H]DHEA for binding at any concentrations tested. Fluasterone (38) had a low affinity for the plasma membrane DHEA binding site, displacing 30% of the [ 3 H]DHEA binding at a concentration of 1 M.
DHEA and DHEAS differ only in the substitution of a sulfate for the 3␤-hydroxyl group. The absence of DHEAS competition of DHEA binding suggests that the 3-position of the A ring may be an important component of the functional group for this receptor. Therefore, it seems that similar absences of 3␤-hydroxyl group in androstenedione, testosterone, and fluasterone may account for their inability to compete for the DHEA binding site. However, the inability of 17␣-hydroxypregnenolone to displace the binding of [ 3 H]DHEA cannot be explained on the basis of this structural difference at the 3-position. Other structural differences between DHEA and these steroids must also contribute to the specificity of DHEA binding to the plasma membranes. These differences include the specific side chain positions and structures, the position of the double bond structures in the A or B rings, and the resulting three-dimensional shape of the steroid, all of which are important for plasma membrane binding (39). In previous studies, 2 we have found that DHEA-17-carboxylmethyl oxime and BSA-conjugated DHEA-17-carboxylmethyl oxime compete with [ 3 H]DHEA for plasma membrane binding to BAEC. These agents are also as potent as native DHEA in stimulating NO production from BAEC. This suggests that receptor binding of the ligand tolerates large added groups at the 17-position. This property may be useful in designing affinity ligands to isolate the putative receptor.
Agonist-stimulated [ 35 S]GTP␥S incorporation into G-protein ␣-subunits has been widely used as evidence of G-protein coupling to receptors. Using this method, we found that DHEA stimulated [ 35 S]GTP␥S incorporation into G-protein ␣-subunits in a concentration-and time-dependent manner. Specifically, DHEA maximally stimulated [ 35 S]GTP␥S binding at 10 nM, although the differences in effect between 0.1 and 10 nM DHEA did not reach statistical significance. At these concentrations DHEA consistently induced the highest production of NO from BAEC 2 in previous studies. Activation of eNOS in the presence of calcium, calmodulin, and eNOS co-factors was also maximal at 10 nM DHEA (Fig. 10C). Furthermore, the K d for DHEA binding (48.7 pM) is close to the EC 50 for DHEA-induced [ 35 S]GTP␥S binding (43 pM) and the EC 50 for eNOS activity (87 pM), suggesting that these processes are functionally linked. The DHEA-induced binding of GTP␥S and activity of eNOS showed similar biphasic dose-response curves, with a decreased responsiveness at DHEA concentrations greater than 10 nM. The similarity of the concentration dependence curves for GTP␥S binding and eNOS activity again support a link between these two processes. The DHEA-stimulated binding of [ 35 S]GTP␥S increased over time, reaching a maximum at 15 min and then declining to zero at 2 h of incubation. This suggested that DHEA catalytically increased the rate of GTP␥S binding at relatively early time points, but did not alter the maximal nucleotide binding at equilibrium. These results are consistent with other studies of G-protein-coupled receptors showing ligand-dependent induction of GTP binding to G-proteins (40 -43). Furthermore, we found that the addition of GTP␥S potently inhibited [ 3 H]DHEA binding. This again confirmed the functional interaction of receptor and G-protein (44), because binding of GTP or its analogues to the G-proteins reduces the agonist affinity of the G-protein-coupled receptor by uncoupling the G-protein from its receptor (45). This inhibitory effect of GTP, or its nonhydrolyzable analogues, on the high affinity binding of agonist to its plasma membrane receptor is observed for most other G-protein-coupled receptors (46 -49). Therefore, these analyses of G-protein coupling provided strong evidence that the binding sites for [ 3 H]DHEA were responsible for G-protein activation.
The G i subfamily of G-proteins consists of at least six different subtypes, including four PTX-sensitive isoforms (G o , G i1 , G i2 , G i3 ). The immunoblotting study showed that there were detectable amounts of G␣ i1 and significant levels of G␣ i2 and G␣ i3 , but G o protein was absent in BAEC, in agreement with previous observations (50 -52). Using the agonist-stimulated [ 35 S]GTP␥S binding assay and antibodies raised against distinct epitopes of G␣ i1 , G␣ i2 , G␣ i3 , and G␣ o subunits, we determined that only G␣ i2 and G␣ i3 were coupled to the putative DHEA membrane receptor. Consistent with these data, we showed that binding of DHEA or [ 35 S]GTP␥S to plasma membranes was inhibited by PTX.
DHEA caused acute activation of eNOS in isolated plasma membranes from endothelial cells. These findings demonstrate that DHEA directly acts on the plasma membrane to activate eNOS and stimulate NO production, as we previously observed in the intact BAEC. 2 The attenuation of DHEA-induced eNOS activity by G␣ i2 and G␣ i3 antisera in isolated plasma membranes confirms our previous data that these two isoforms of G␣ i mediate the DHEA stimulation of eNOS. However, these experiments do not determine whether the ␣ or ␤␥ G-proteins are specifically responsible for the signaling to eNOS. The mechanism of activation of eNOS by DHEA is unknown. One possibility is that DHEA activates a tyrosine kinase such as Src by a G␣ i -dependent (53) or G␤␥-dependent (54) mechanism. This may then activate eNOS through a mechanism involving phosphoinositide 3-kinase and Akt kinase (28,55). We are continuing to evaluate this hypothesis. It is also noteworthy that the Triton-solubilized membrane preparations retained high affinity binding to [ 3 H]DHEA and functional association with G-proteins. Thus, it seems likely that the solubilized receptors and related G-proteins were still preserved in a functionally stable configuration with this solubilization protocol. This property should provide a major advantage for the subsequent affinity purification of DHEA receptors.
Although animal and human studies have demonstrated the beneficial effects of DHEA on the cardiovascular system (1-4, 8 -13), the mechanism of DHEA action is not known. Nitric oxide is a potent vasodilator and has other beneficial effects in the vasculature, including inhibition of monocyte chemotaxis and adhesion to endothelial cells (56,57), regulation of smooth muscle cell proliferation (58), as well as inhibition of platelet aggregation (59). By showing an effect of physiological concentrations of DHEA on endothelial cell NO production, our results propose a mechanistic basis for the physiological actions of DHEA in the vascular system. How the effects that we demonstrate are associated with the epidemiological studies showing an inverse correlation of cardiovascular disease and DHEA levels remains to be studied.
Both DHEA and its sulfated metabolite, DHEAS, have been shown to have multiple actions in different tissues and organs. Many of these actions are reported at pharmacological concentrations, or with more prolonged incubations than we have used in our studies (60). In these situations the effects of lipophilic steroids on the plasma membrane (61)(62)(63) or the direct effects of intracellular steroid (64) may be of greater importance than in our study. Furthermore, many of the effects of high circulating levels of DHEA in vivo cannot be differentiated from the effects of DHEA metabolites, including estradiol or testosterone. Our data do not negate any of the mechanisms proposed for pharmacologic concentrations of DHEA, but we do propose this high affinity ligand-receptor interaction as a potential mechanism for physiological concentrations of DHEA. The circulating plasma concentration of DHEA peaks at ϳ16 nM in the third decade. The steroid is also bound to plasma components, including proteins, e.g. albumin and high density lipoprotein (65,66). Thus, the circulating free hormone concentration would be lower than the total DHEA levels. The cellular effects that we show, between 0.1 and 10 nM, are well within the range of physiological concentrations in humans.
There are extensive data showing that steroid hormones interact with intracellular receptors to modulate nuclear gene transcription and subsequent protein synthesis (14). In keeping with this paradigm of steroid hormone action, specific intracellular DHEA binding activities have been demonstrated by many (15)(16)(17)(18), although not all (67), research groups. Specific high affinity DHEA binding has been found in cytosolic, nuclear, and whole cell assays in hepatocytes, melanoma, and lymphoid cells of various species. Although the expression levels of these putative DHEA receptors are similar to our results, the dissociation constants are 50 -1,400 times higher than in the present study (Table II). At this time, researchers have been unable to isolate proteins supporting these intracellular DHEA binding activities. Thus, there is no known intracellular steroid hormone receptor for DHEA. Furthermore, we have not been able to identify a DHEA binding site in BAEC cytosolic protein under the conditions of our assay. 2 DHEA has also been shown to interact with known intracellular steroid hormone receptors. Nephew et al. (68) showed interaction of DHEA with a cloned human estrogen receptor in a yeast system. However, the K d for binding was 15.8 Ϯ 3.4 M, well above the potential circulating DHEA concentration. Thus, it is unlikely that the physiological effects of DHEA are mediated by the intracellular estrogen receptor. Besides their classical "genomic" effects, characterized by latency of action (69), rapid and non-genomic effects of steroids have been widely recognized and characterized (69 -72). These membrane-initiated steroid actions appear to be transmitted by specific membrane-bound receptors, which subsequently activate intracellular signaling to regulate cellular function. Our results suggest that DHEA acts at a plasma membrane site to activate intracellular signaling.
Our discovery of a putative plasma membrane receptor for DHEA does not negate the possibility of also finding an intracellular receptor for this hormone. All of the major steroid hormones, where plasma membrane receptors have been proposed, have well characterized intracellular receptors. A coordinated cellular mechanism of action, with initial rapid, membrane-dependent activity and subsequent longer term, genomic activity, has been postulated for mineralocorticoids and other steroid hormones (73)(74)(75).
The plasma membrane estrogen receptor (ER) has been described and characterized. The membrane receptor appears to be structurally identical to the intracellular ER␣ and -␤ (76). Ligand binding of the receptor activates G␣ i proteins and eNOS (77). In these ways the plasma membrane ER resembles the DHEA receptor that we describe. However, estradiol does not inhibit binding of DHEA to the DHEA binding site. Furthermore, the ER antagonist, ICI 182,780, blocks the action of estradiol on eNOS in plasma membrane preparations, but does not alter the effect of DHEA in the same system. We have also found that tamoxifen blocks the effect of estradiol to stimulate NO production in cultured BAEC, but does not inhibit the action of DHEA, 2 again suggesting that these steroids are not functioning through the same receptor. Because the ER does not have the classical seven-transmembrane-spanning structure of a G-protein-coupled receptor, Wyckoff et al. (77) have hypothesized that the ER may be interacting with another "classical" G-protein-coupled receptor to initiate its signaling. A similar mechanism may be postulated for the DHEA receptor. The similarity of the activities of estradiol and DHEA raises the possibility that both the DHEA receptor and the ER may signal by interacting with a common G-protein-coupled receptor. It will be important to determine whether the DHEA receptor is interacting directly with G-proteins or is interacting with a G-protein-coupled receptor. It will also be important to determine whether the cell surface-initiated activities of DHEA and estradiol are entirely the same in endothelial cells or whether they have some differences.
The significance of our findings is that they establish a mechanistic basis for the physiological actions of DHEA, although the detailed signaling pathways involved in receptormediated DHEA action remain to be elucidated. Based on our results, we are now in a position to isolate the putative plasma membrane receptor for DHEA. The data on vascular endothelial effects of DHEA raise the possibility of developing nonmetabolized DHEA receptor ligands to evaluate their effects on vascular function in vivo. Furthermore, our results are another example of the plasma membrane-initiated, non-genomic activity of steroid hormones. This property has been demonstrated for many other steroids. However, the finding with DHEA is novel, because it has no known intracellular receptor, unlike other steroids with partially or fully characterized membrane receptors. These findings expand the repertoire of signaling pathways that have been associated with plasma membraneinitiated actions of steroids and with DHEA.
In summary, we present evidence for a putative specific DHEA plasma membrane receptor on bovine aortic endothelial cells. This receptor is functionally coupled to the G i protein subfamily and primarily to G␣ i2 and G␣ i3 subtypes. Activation of these G-proteins mediates the effect of DHEA on eNOS. This finding has a fundamental impact on our interpretations of the biological actions of DHEA. The signal transduction pathway associated with the ligand-receptor interaction requires further investigation. The receptor purification, characterization, and cloning are necessary to fully understand the physiological significance of DHEA.