Direct Interaction of Endothelial Nitric-oxide Synthase and Caveolin-1 Inhibits Synthase Activity*

Endothelial nitric-oxide synthase (eNOS) and caveolin-1 are associated within endothelial plasmalemmal caveolae. It is not known, however, whether eNOS and caveolin-1 interact directly or indirectly or whether the interaction affects eNOS activity. To answer these questions, we have cloned the bovine caveolin-1 cDNA and have investigated the eNOS-caveolin-1 interaction in an in vitrobinding assay system using glutathione S-transferase (GST)-caveolin-1 fusion proteins and baculovirus-expressed bovine eNOS. We have also mapped the domains involved in the interaction using anin vivo yeast two-hybrid system. Results obtained using both in vitro and in vivo protein interaction assays show that both N- and C-terminal cytosolic domains of caveolin-1 interact directly with the eNOS oxygenase domain. Interaction of eNOS with GST-caveolin-1 fusion proteins significantly inhibits enzyme catalytic activity. A synthetic peptide corresponding to caveolin-1 residues 82–101 also potently and reversibly inhibits eNOS activity by interfering with the interaction of the enzyme with Ca2+/calmodulin (CaM). Regulation of eNOS in endothelial cells, therefore, may involve not only positive allosteric regulation by Ca2+/CaM, but also negative allosteric regulation by caveolin-1.

Plasmalemmal caveolae are small membrane invaginations present in most cells of higher eukaryotes. These membrane specializations appear to function both as endocytotic carriers and as signal transduction organizing centers. In the latter case, caveolae compartmentalize a subset of signal-transducing molecules in membrane microdomains at the cell surface (1,2). A major structural protein of caveolae is caveolin, a 21-24-kDa integral membrane protein that occurs in three homologous, but distinct isoforms termed caveolins-1, -2, and -3 (3). Fulllength caveolin-1 contains three domains: a 101-residue Nterminal domain, a 33-residue membrane-spanning region, and a 44-residue C-terminal domain. The N-and C-terminal domains of caveolin-1 face the cytoplasm suggesting that the membrane-spanning region forms a hairpin loop within the membrane (4 -6). A cytosolic membrane-proximal subdomain of the N-terminal domain (residues 82-101) interacts directly with G␣ subunits, Ha-Ras, and Src family tyrosine kinases (7)(8)(9). Interaction of these signaling proteins with this caveolin-1 scaffolding domain serves to sequester the proteins in caveolae and to inhibit or suppress their catalytic activities.
Another important signaling protein known to be localized in caveolae is endothelial nitric-oxide synthase (eNOS) 1 (10,11). Production of NO by eNOS in endothelial caveolae appears to play a key role in modulating vascular tone, platelet aggregation, leukocyte adhesion, vascular smooth muscle cell proliferation, and vascular lesion formation (12,13). eNOS has a bidomain structure consisting of an N-terminal oxygenase domain and a C-terminal reductase domain (14). Located between the oxygenase and reductase domains is a Ca 2ϩ /calmodulin (CaM)-binding region (15). Association of eNOS and caveolin-1 in cultured bovine endothelial cells has been demonstrated previously in coimmunoprecipitation experiments (16,17). It is not known, however, whether eNOS and caveolin-1 interact directly or indirectly (i.e. through an adaptor protein). Also unidentified are the interacting domains, if any, in the two proteins. Most importantly, the functional consequences of caveolin-1 binding on eNOS catalytic activity have not been determined. Each of these questions with regard to the eNOScaveolin-1 protein-protein interaction has been addressed in the present study.

EXPERIMENTAL PROCEDURES
Materials-The glutathione S-transferase (GST)-fusion protein cloning vector, pGEX-4T-1, CaM-Sepharose 4B, and anti-GST polyclonal antibody were obtained from Pharmacia Biotech Inc. Sf9 insect cells were purchased from Pharmingen (San Diego, CA) and maintained in serum-supplemented Hink's TNM-FH media from Mediatech, Inc. (Herndon, VA). Monoclonal antibody to eNOS (clone 3) was purchased from Transduction Laboratories (Lexington, KY). L-[ 14 C]Arginine and ECL reagents came from Amersham Corp. pGBT9 (DNA binding domain hybrid cloning vector), pGAD424 (activation domain hybrid cloning vector), and Saccharomyces cerevisiae SFY526 were obtained from CLONTECH (Palo Alto, CA). Oligonucleotide primers for PCR, 5Ј-RACE kit, and TriZOL reagent were purchased from Life Technologies Inc. TA Cloning kit was obtained from Invitrogen (Carlsbad, CA). Protein assay kit was purchased from Bio-Rad. Bovine CaM came from Sigma. Synthetic peptides were obtained from Research Genetics, Inc. (Huntsville, AL) and were Ͼ95% pure as determined by high performance liquid chromatography.
Cloning of the cDNA Encoding Bovine Caveolin-1-Total RNA from cultured bovine aortic endothelial cells was isolated with TriZOL reagent and subjected to reverse transcription-polymerase chain reaction. The upstream primer for PCR was based on the first 20 nucleotides of the caveolin-1 coding sequence that are identical in sequences previously cloned from dog, mouse, and human (18 -20). The downstream primer used was an oligo(dT) 17 oligonucleotide. PCR amplification produced a 568-base pair fragment that was subcloned into the TA cloning vector and sequenced in the Molecular Biology Core Facility of the Medical College of Georgia. Three independent PCR reactions produced identical nucleotide sequences ruling out the possibility of PCR-associated nucleotide incorporation errors. To confirm that the first 20 nucleotides of the bovine coding sequence are identical to the sequence of the * This work was supported by National Institutes of Health Grant HL57201 (to R. C. V.) and by grants-in-aid from the Georgia Affiliate of the American Heart Association (to R. C. V.) and the American Heart Association National Center (to R. C. V.). 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.
‡ To whom correspondence should be addressed: Vascular Biology Center, Medical College of Georgia, Augusta, Georgia 30912-2500. Tel.: 706-721-2576; Fax: 706-721-8555; E-mail: rvenema@mail.mcg.edu. upstream primer, 5Ј-RACE (rapid amplification of cDNA ends) was performed with a 5Ј-RACE kit (Life Technologies Inc.). The sequence obtained by 5Ј-RACE was identical to that obtained in the initial PCR. The nucleotide sequence has been submitted to the GenBank TM /EMBL Data Bank with accession number U86639.
Construction and Purification of GST-Caveolin-1 Fusion Proteins-cDNA constructs encoding GST-caveolin-1 fusion proteins were created by subcloning into the GST-fusion protein cloning vector, pGEX-4T-1. Caveolin-1 cDNA sequences encoding full-length caveolin-1 (residues 1-178) and caveolin-1 residues 1-60, 1-101, 102-134, and 135-178 were generated by PCR amplification of the full-length bovine sequence cloned into the TA cloning vector. Primers for PCR were designed to incorporate 5Ј EcoRI and SalI restriction sites for subcloning. The cDNAs encoding the fusion proteins were sequenced to confirm the creation of in-frame fusions devoid of PCR-associated nucleotide incorporation errors. Fusion proteins and a GST-nonfusion protein were expressed in Escherichia coli and purified by affinity chromatography on glutathione-agarose as described by Frangioni and Neel (21).
Expression and Purification of eNOS in a Baculovirus System-Bovine eNOS was expressed in a baculovirus/Sf9 insect cell system and purified to Ͼ95% homogeneity as described previously (15,22). eNOS was purified in buffers containing 2 mM EGTA, and purified enzyme was completely dependent on exogenous CaM for activity.
Interaction of Recombinant Baculovirus-expressed eNOS with GST-Caveolin-1 Fusion Proteins-GST or GST-caveolin-1 fusion proteins (100 pmol each, quantitated by Bio-Rad protein assay) prebound to glutathione-agarose beads were washed three times in buffer containing 50 mM Tris-HCl, pH 7.4, 20% glycerol, and the following protease inhibitors: 1% phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, 10 g/ml pepstatin A, and 5 g/ml aprotinin. Equimolar amounts of fusion proteins in each condition were confirmed by immunoblotting with anti-GST antibody. Washed beads were incubated overnight (with shaking at 4°C) in 1 ml of the above buffer containing 100 pmol of bovine eNOS, expressed and purified from a baculovirus system as described previously (22). Following the overnight binding reaction, beads were washed six times in 1 ml of 50 mM Hepes, pH 7.5, 120 mM NaCl, 1 mM EDTA, 0.5% CHAPS plus the protease inhibitors listed above. Bound proteins were eluted with 100 l of 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 1% Triton X-100, 100 mM reduced glutathione, plus protease inhibitors. Eluted proteins were separated on SDS-polyacrylamide gels, transferred to nitrocellulose, and immunoblotted with anti-eNOS monoclonal antibody as described previously (23). Binding experiments were also performed with GST or GST-caveolin-1 fusion proteins and purified bovine CaM. In one set of experiments the binding assay was carried out exactly as described above for eNOS. In another set of experiments, EDTA was omitted from all buffers and replaced with 2 mM CaCl 2 .
Interaction of eNOS and Caveolin-1 Domains in a Yeast Two-hybrid System-Construction of bovine eNOS oxygenase domain (residues 1-505) and reductase domain (residues 506 -1205) hybrids with the GAL4 DNA binding and activation domains has been described previously (24). Bovine caveolin-1 hybrids encoding residues 1-178, 1-101, and 135-178 were created by subcloning into the shuttle/expression vectors, pGBT9 and pGAD424 (CLONTECH). DNA sequences for subcloning were generated by PCR amplification of the full-length bovine caveolin-1 cDNA in the TA cloning vector. PCR primers were designed to incorporate 5Ј EcoRI and SalI restriction sites into the amplified products to facilitate subcloning. All DNA constructs were sequenced to confirm the creation of an in-frame hybrid devoid of PCR-associated nucleotide incorporation errors. Transformation of yeast and colony lift filter assay of ␤-galactosidase activity has been described previously (24). All DNA-binding domain and activation domain fusion constructs were confirmed not to reconstitute GAL4 activity by themselves, and all activation domain fusion constructs were confirmed not to activate transcription when combined with the unrelated pLAM 5Ј (human lamin C 66 -230 in pGBT9).
Determination of the Effects of GST-Caveolin-1 Fusion Proteins and Synthetic Peptides on eNOS Activity-GST or GST-caveolin-1 fusion proteins were eluted from glutathione-agarose beads with reduced glutathione, and 300 pmol of each protein (quantitated by Bio-Rad protein assay) was incubated for 5 min at 37°C with purified, baculovirusexpressed bovine eNOS (100 pmol) in 50 mM Tris-HCl, pH 7.5, buffer. eNOS activity was then determined by monitoring the rate of formation of L-[ 14 (22). Experiments testing the effects of synthetic peptides were carried out using the same protocol and various concentrations of peptides.
CaM-Sepharose Chromatography-Purified eNOS (1 g) was incubated with or without synthetic peptides (10 M) in 50 mM Tris-HCl, pH 7.5, buffer for 5 min at 37°C and then subjected to CaM-Sepharose chromatography as described previously (22). The amount of eNOS eluted from the CaM-Sepharose column was quantitated by immunoblotting with anti-eNOS antibody as described previously (23).

RESULTS AND DISCUSSION
eNOS and caveolin-1 are known to be associated in cultured bovine aortic and lung microvascular endothelial cells (16,17). It is not known, however, whether they interact directly or indirectly (i.e. through an adaptor protein). To answer this question for purified, baculovirus-expressed bovine eNOS, we have isolated and sequenced the cDNA encoding bovine caveolin-1. Analysis of the deduced amino acid sequence indicates that bovine caveolin-1 shares 94, 96, 97, and 86% identity with the human, murine, canine, and chicken sequences, respectively (18 -20, 25) (Fig. 1). To determine whether eNOS interacts directly with caveolin-1, we expressed full-length bovine caveolin-1 as a GST-fusion protein in E. coli. In addition, to determine which domains of caveolin-1 are involved in eNOS binding, we also expressed GST-fusion proteins of caveolin-1 residues 1-60, 1-101 (N-terminal cytoplasmic domain), 102-134 (membrane-spanning domain), and 135-178 (C-terminal cytoplasmic domain). The fusion proteins and a GST-nonfusion protein were purified by affinity chromatography on glutathione-agarose. The GST-caveolin-1 fusions or GST alone prebound to agarose beads were then used in in vitro binding assays with recombinant bovine eNOS, expressed and purified from a baculovirus system (22). Beads were incubated with eNOS at 4°C overnight and extensively washed, and bound proteins were eluted with reduced glutathione. Eluted proteins were separated on SDS-polyacrylamide gels, transferred to nitrocellulose, and immunoblotted with anti-eNOS antibody. As shown in Fig. 2, eNOS bound specifically to the full-length GST-caveolin-1 fusion protein but not to GST alone, demonstrating that eNOS and caveolin-1 interact directly. Furthermore, eNOS bound specifically to GST-fusions containing only the N-terminal caveolin-1 cytoplasmic domain (residues 1-101) or only the C-terminal caveolin-1 cytoplasmic domain (residues 135-178). In contrast, GST-fusions containing only caveolin-1 residues 1-60 or the caveolin-1 membrane-spanning domain (residues 102-134) did not bind to eNOS. The eNOS-caveolin-1 association thus appears to involve binding of eNOS to both cytoplasmic tails of caveolin-1. Furthermore, either cytoplasmic domain of caveolin-1 by itself is sufficient to mediate the eNOS binding.
To verify the conclusions reached based on the in vitro binding assays and to determine whether caveolin-1 binds to either the eNOS oxygenase domain or the eNOS reductase domain, or both, we have also investigated the eNOS-caveolin-1 interaction in a yeast two-hybrid system. Hybrid cDNA constructs were prepared that encoded full-length bovine caveolin-1 (residues 1-178), the caveolin-1 N-terminal cytoplasmic domain (residues 1-101), the caveolin-1 C-terminal cytoplasmic domain (135-178), the bovine eNOS oxygenase domain (1-505), and the bovine eNOS reductase domain (506 -1205) fused to either the GAL4 DNA binding domain or activation domain. Various pairwise combinations of the plasmid constructs were used to cotransform the yeast strain, SFY526. Interactions of hybrid proteins were assessed by colony lift filter assay of ␤-galactosidase reporter gene transcription. As shown in Table  I, caveolin-1 interacted with itself in the two-hybrid system through both N-and C-terminal cytoplasmic domains, as has been demonstrated previously in in vitro binding assays with GST-fusion proteins (26,27). Furthermore, both N-and Cterminal domains of caveolin-1 interacted with eNOS in the two-hybrid system, confirming the results obtained in the GSTfusion protein binding assays. Caveolin-1 interactions were restricted to the eNOS oxygenase domain and did not occur with the eNOS reductase domain.
To determine whether interaction of eNOS with caveolin-1 alters nitric-oxide synthase activity, we incubated equal quantities of purified, baculovirus-expressed eNOS with equimolar quantities of the GST alone, GST-caveolin 1-60, GST-caveolin 102-134, GST-caveolin 1-178, GST-caveolin 1-101, and GSTcaveolin 135-178 fusion proteins. eNOS activity was then determined by arginine-to-citrulline conversion assay in the presence of excess cofactors, Ca 2ϩ , and CaM. As shown in Fig. 3, the full-length caveolin-1 fusion protein inhibited eNOS activity by about 60%. Furthermore, either of the caveolin-1 cytoplasmic domains appears to be sufficient to mediate eNOS inhibition because the GST-caveolin 1-101 and GST-caveolin 135-178 fusion proteins also inhibited enzyme activity by about 60%. In contrast, the GST-caveolin 1-60 and GST-caveolin 102-134 fusion proteins were without effect on activity. To confirm that inhibition was due to binding of the fusion proteins to eNOS rather than to CaM, we also performed in vitro binding assays with CaM and GST-caveolin-1 fusion proteins under the same conditions used in the eNOS binding studies. In addition, we performed in vitro binding assays of eNOS and CaM in which 1 mM EDTA was omitted from all buffers and replaced by 2 mM CaCl 2 . In both sets of experiments, no CaM binding to any of the GST-caveolin-1 fusion proteins was detected.
Inhibition by GST-caveolin 1-101 but not by GST-caveolin 1-60 suggests that the inhibitory region of the N-terminal cytoplasmic domain may correspond to the caveolin-1 scaffolding domain (residues 82-101) previously shown to inhibit G␣ subunits, Ha-Ras, and Src family tyrosine kinases (7)(8)(9). To test this hypothesis we prepared synthetic peptides corresponding to caveolin-1 residues 61-81 and 82-101. As shown in Fig. 4, the 82-101 peptide potently inhibited eNOS activity (IC 50 ϳ1 M). Complete inhibition was observed at a 10 M concentration of peptide. A 10 M concentration of the 61-81 peptide, on the other hand, actually increased activity by about 30%. To determine whether inhibition was due to an effect of the 82-101 peptide on the eNOS interaction with Ca 2ϩ /CaM, we preincubated eNOS with and without the 61-81 and 82-101 peptides (10 M) and then subjected the enzyme to CaM-Sepharose chromatography. Enzyme was allowed to bind to the column in the presence of 2 mM CaCl 2 and was eluted with 2 mM EGTA. The amount of enzyme eluted in each condition was quantitated by immunoblotting with monoclonal anti-eNOS antibody. As shown in Fig. 5A, the 82-101 peptide reduced binding of eNOS to CaM-Sepharose by Ͼ90% (as determined by  cav 102-134), plus GST alone were expressed in E. coli and purified by affinity binding to glutathioneagarose beads. Proteins prebound to beads were incubated with purified baculovirus-expressed recombinant eNOS. Following binding, extensive washing, and elution with reduced glutathione, proteins were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose, and immunoblotted with anti-eNOS antibody. The result shown is a representative immunoblot from three separate experiments.

TABLE I
Interactions between eNOS and caveolin-1 in a yeast two-hybrid system Pairwise combinations of hybrid plasmids were used to cotransform yeast cells. Cotransformants were assayed for ␤-galactosidase activity by the colony lift filter assay method using 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside as substrate. Similar results were obtained in five separate transformations. Caveolin-1 Inhibits Endothelial NOS densitometry of immunoblots). In contrast, the 61-81 peptide had no effect on CaM binding. Furthermore, peptide inhibition of eNOS was reversible by increasing the molar excess of Ca 2ϩ / CaM by 10-fold. eNOS was preincubated for 5 min at 37°C with the 82-101 peptide (10 M). Enzyme activity was then determined in the presence of the standard molar excess concentration of Ca 2ϩ /CaM (1.25 M) routinely used in the arginine-to-citrulline conversion assay. Activity was confirmed to be completely inhibited in the presence of 10 M peptide and 1.25 M Ca 2ϩ /CaM. eNOS was then incubated for an additional 5 min at 37°C with either 1.25 M or 12.5 M Ca 2ϩ /CaM. eNOS activity was then redetermined. As shown in Fig. 5B, increasing the Ca 2ϩ /CaM concentration by 10-fold completely reversed the inhibitory effects of the 82-101 peptide.
In summary, the results of the present study provide several important new insights into the eNOS-caveolin-1 interaction. First, eNOS and caveolin-1 interact directly rather than indirectly. Second, interaction involves both the N-and C-terminal cytoplasmic domains of caveolin-1 and is thus fundamentally different from the interaction of caveolin-1 with G␣ subunits, Ha-Ras, and Src family tyrosine kinases. Third, the caveolin-1 interaction with eNOS involves only the eNOS oxygenase domain and not the eNOS reductase domain. Fourth, interaction of eNOS with caveolin-1 significantly inhibits eNOS catalytic activity. Finally, inhibition appears to be due to interference with the eNOS interaction with Ca 2ϩ /CaM. Regulation of eNOS activity in endothelial cells, therefore, may involve not only positive allosteric regulation by Ca 2ϩ /CaM, but also negative allosteric regulation by caveolin-1. It is conceivable that interaction of eNOS with caveolin-1 provides a mechanism to deactivate the enzyme subsequent to its activation by agoniststimulated elevation of intracellular Ca 2ϩ . Protein-protein interactions between eNOS and caveolin-1, however, are probably not sufficient to mediate membrane attachment. Fatty acylation of eNOS by myristate and palmitate appears to also be required. This requirement has been demonstrated in previous studies of an eNOS myristoylation-deficient mutant in which glycine 2 has been mutated to an alanine. The mutant enzyme is neither myristoylated nor palmitoylated and, as a result, is not membrane-associated (22).