Reciprocal Regulation of Endothelial Nitric-oxide Synthase by Ca2+-Calmodulin and Caveolin*

The endothelial nitric-oxide synthase (eNOS) is a key determinant of vascular homeostasis. Like all known nitric-oxide synthases, eNOS enzyme activity is dependent on Ca2+-calmodulin. eNOS is dynamically targeted to specialized cell surface signal-transducing domains termed plasmalemmal caveolae and interacts with caveolin, an integral membrane protein that comprises a key structural component of caveolae. We have previously reported that the association between eNOS and caveolin is quantitative and tissue-specific (Feron, O., Belhassen, L., Kobzick, L., Smith, T. W., Kelly, R. A., and Michel, T. (1996) J. Biol. Chem.271, 22810–22814). We now report that in endothelial cells the interaction between eNOS and caveolin is importantly regulated by Ca2+-calmodulin. Addition of calmodulin disrupts the heteromeric complex formed between eNOS and caveolin in a Ca2+-dependent fashion. In addition, overexpression of caveolin markedly attenuates eNOS enzyme activity, but this inhibition is reversed by purified calmodulin. Caveolin overexpression does not affect the activity of the other NOS isoforms, suggesting eNOS-specific inhibition of NO synthase by caveolin. We propose a model of reciprocal regulation of eNOS in endothelial cells wherein the inhibitory eNOS-caveolin complex is disrupted by binding of Ca2+-calmodulin to eNOS, leading to enzyme activation. These findings may have broad implications for the regulation of Ca2+-dependent signal transduction in plasmalemmal caveolae.

The endothelial nitric-oxide synthase (eNOS) is a key determinant of vascular homeostasis. Like all known nitric-oxide synthases, eNOS enzyme activity is dependent on Ca 2؉ -calmodulin. eNOS is dynamically targeted to specialized cell surface signal-transducing domains termed plasmalemmal caveolae and interacts with caveolin, an integral membrane protein that comprises a key structural component of caveolae. We have previously reported that the association between eNOS and caveolin is quantitative and tissue-specific ( . We now report that in endothelial cells the interaction between eNOS and caveolin is importantly regulated by Ca 2؉ -calmodulin. Addition of calmodulin disrupts the heteromeric complex formed between eNOS and caveolin in a Ca 2؉ -dependent fashion. In addition, overexpression of caveolin markedly attenuates eNOS enzyme activity, but this inhibition is reversed by purified calmodulin. Caveolin overexpression does not affect the activity of the other NOS isoforms, suggesting eNOSspecific inhibition of NO synthase by caveolin. We propose a model of reciprocal regulation of eNOS in endothelial cells wherein the inhibitory eNOS-caveolin complex is disrupted by binding of Ca 2؉ -calmodulin to eNOS, leading to enzyme activation. These findings may have broad implications for the regulation of Ca 2؉ -dependent signal transduction in plasmalemmal caveolae. Nitric oxide is a ubiquitous molecule implicated in diverse biological processes and is synthesized in mammalian cells by a family of Ca 2ϩ -calmodulin-dependent nitric-oxide synthase (NOS) 1 enzymes (1-3). Endothelium-derived nitric oxide (NO), formed by the endothelial isoform of nitric-oxide synthase (eNOS), serves as an important determinant of blood pressure and platelet aggregation. In endothelial cells, increases in intracellular Ca 2ϩ elicited by diverse extracellular signals lead to activation of eNOS. The three known mammalian nitric-oxide synthases share similar overall Ca 2ϩ -calmodulin-dependent catalytic pathways. However, the eNOS enzyme is unique among the three known NOS isoforms in being localized to the specialized cell surface signal-transducing domains termed plasmalemmal caveolae (4,5).
Plasmalemmal caveolae are small invaginations in the plasma membrane that may serve as sites for the sequestration of signaling proteins (6, 7) including receptors, G proteins, and protein kinases, as well as eNOS. The principal protein in caveolae is the integral membrane protein caveolin, an oligomeric protein that serves as a structural "scaffold" within caveolae (8). eNOS can be quantitatively immunoprecipitated by antibodies directed against caveolin; conversely, eNOS antiserum also immunoprecipitates caveolin (5), although it has not yet been established whether the interaction between these two proteins is direct. Moreover, a functional role of the eNOScaveolin association, beyond its postulated role in subcellular targeting of the enzyme, remains to be determined. We document in this report that the interaction between eNOS and caveolin may be regulated by Ca 2ϩ -calmodulin, and we show that caveolin can specifically inhibit eNOS enzyme activity.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-A plasmid construct encoding eNOS has been described previously (9,10). cDNA clones encoding iNOS (11) and nNOS (12) were kindly provided by Carl Nathan (Cornell University Medical College) and Solomon Snyder (Johns Hopkins University), respectively. Caveolin-1 cDNA (13) in the eukaryotic expression vector pCB-7 was obtained from Michael Lisanti (Whitehead Institute). An irrelevant plasmid encoding ␤-galactosidase was used as a control to maintain identical amounts of DNA in each transfection.
Cell Culture and Transfection-Cultures of bovine aortic endothelial cells (studied between passages 4 and 10) and COS-7 cells were performed as described previously (9,14). COS-7 cells were transfected with 10 g of total plasmid DNA in 100-mm cell culture plates using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's protocol.
Immunoprecipitation, SDS-PAGE, and Immunoblotting-Endothelial cells were lysed and solubilized either with: 1) a previously described Ca 2ϩ -free CHAPS buffer (5,14) supplemented with 1 mM EGTA/1 mM EDTA or 2) an otherwise identical CHAPS buffer containing 1 mM CaCl 2 and no EDTA/EGTA. CHAPS-solubilized bovine aortic endothelial cell lysates were incubated either with a polyclonal caveolin antibody (Transduction Laboratories) at a final concentration of 4 g/ml or with a previously characterized polyclonal antiserum directed against eNOS (9) used at a final dilution of 1:100. Immunoprecipitated complexes were then recovered, denatured in Laemmli sample buffer, separated on 12% SDS-PAGE, and transferred to a polyvinylidene difluoride membrane (5). Monoclonal antibodies directed against eNOS or caveolin-1 (Transduction Laboratories) were then used to detect * This work was supported by awards (to T. M.) from the National Institutes of Health, the American Heart Association, and the Burroughs Wellcome Fund. 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.
This report is dedicated to the memory of Professor Thomas W. Smith.
§ Bugher-American Heart Association Fellow in Cardiovascular Molecular Biology. eNOS and caveolin-1 using chemiluminescence, as described previously (5). Calmodulin was detected using a previously characterized monoclonal calmodulin antibody (15). Expression of immunoblotted proteins was quantitated by laser densitometric analysis of x-ray films following chemiluminescence detection.
NOS Enzymatic Assays-NO synthase activity in lysates prepared from transfected COS-7 cell was determined by measuring conversion of [ 3 H]L-arginine to [ 3 H]L-citrulline as described previously (9,14). In some experiments, this NO synthase activity assay was performed in washed membrane fractions prepared from endothelial cell lysates as described previously (9), except that calmodulin concentrations were varied by the addition of purified bovine brain calmodulin (Sigma) as indicated.

RESULTS AND DISCUSSION
In exploring the factors governing eNOS-caveolin association, we discovered that the co-immunoprecipitation of eNOS and caveolin was markedly attenuated when endothelial cell lysates were prepared in buffers containing excess free Ca 2ϩ (Fig. 1). Under no conditions did the presence of Ca 2ϩ affect the recovery of caveolin or eNOS from endothelial cell lysates when the proteins were directly immunoprecipitated by their cognate antibody. Only co-immunoprecipitation was abrogated by Ca 2ϩ , as shown in Fig. 1.
We next investigated the role of Ca 2ϩ in modulating the association of eNOS and calmodulin. This is particularly important because calmodulin, a ubiquitous Ca 2ϩ -binding protein, plays a central role in nitric-oxide synthase catalysis (16). Co-immunoprecipitation experiments using eNOS antiserum to explore eNOS-calmodulin interactions in endothelial cell lysates are shown in Fig. 2. We could detect co-immunoprecipitation of eNOS and calmodulin only when free Ca 2ϩ was present. This result is in striking contrast to the loss of eNOScaveolin co-immunoprecipitation observed in the presence of Ca 2ϩ (Fig. 1). Taken together, these data suggest that Ca 2ϩ differentially modulates the association of eNOS with caveolin versus calmodulin in endothelial cell lysates.
We next used antibodies against caveolin to co-immunoprecipitate eNOS in Ca 2ϩ -free buffers and then washed the immune complexes extensively to remove residual Ca 2ϩ and calmodulin. Subsequent addition of either Ca 2ϩ or calmodulin alone had no effect on the subsequent recovery of co-immunoprecipitated eNOS from the eNOS-caveolin complex (Fig. 3,  upper panel). However, when Ca 2ϩ and calmodulin were added together, eNOS was entirely lost from the caveolin immune complex. This "lost" eNOS could be completely recovered in the supernatant of the immune complex (Fig. 3, middle panel), indicating that the eNOS molecule had been released and not degraded following treatment of the caveolin-eNOS complex with Ca 2ϩ plus calmodulin. None of these treatments affected the recovery of caveolin itself from the immune complex (Fig. 3,  lower panel). The extensive washing of these immune complexes, required to remove the endogenous calmodulin, likely led to the loss of some eNOS (because the affinity of the eNOScaveolin interaction is undoubtedly less than the affinity of the antibody for caveolin), leading to the detection of a relatively faint but highly reproducible (n ϭ 4) signal for eNOS released from these complexes by the combination of Ca 2ϩ plus calmodulin.
Diverse experimental approaches have shown that agonist activation of eNOS in endothelial cells is dependent on Ca 2ϩcalmodulin (1, 2). The striking effects of Ca 2ϩ -calmodulin on the interactions of eNOS and caveolin therefore suggested to us that caveolin may influence eNOS enzyme activity. Indeed, caveolin has recently been shown in vitro to interact with other signaling proteins, including H-ras and G protein ␣ subunits, preferentially associating with the "inactive" forms of these proteins (17), but the cellular regulation of these interactions is less well understood. We explored the functional consequences of the interaction between eNOS and caveolin using transient transfection experiments in COS-7 cells and analyzed eNOS enzyme activity in transfected cells by assaying the conversion of [ 3 H]L-arginine to [ 3 H]L-citrulline in cell lysates, as shown in Fig. 4. In three separate experiments, each conducted in triplicate, we found that the co-transfection of a plasmid cDNA construct encoding caveolin with eNOS cDNA led to a marked attenuation of NOS activity (3.4 Ϯ 0.3 versus 1.6 Ϯ 0.1 pmol citrulline/min⅐mg protein in the absence or the presence of caveolin co-expression, respectively; see Fig. 4A). There was no change in the abundance of eNOS protein associated with caveolin co-transfection, as assessed in immunoblots of these cellular lysates analyzed in each experiment (data not shown). Importantly, caveolin co-transfection failed to attenuate the enzyme activity expressed by transfected iNOS or nNOS cDNA, shown in Fig. 4B. As for eNOS, the enzyme activity of iNOS and nNOS is calmodulin-dependent (although important differences in the Ca 2ϩ dependence of the different NOS isoforms have been noted). In further contrast to eNOS, the other NOS isoforms are not targeted to caveolae. To explore the specificity of the inhibitory effect of caveolin co-expression on eNOS enzyme activity, we performed activity assays in the presence of varying concentrations of purified calmodulin added to washed membrane fractions prepared from transfected COS-7 cells. As shown in Fig. 4C, in cells transfected with eNOS cDNA alone, there is a robust NOS activity even in the absence of added calmodulin (presumably due the presence of endogenous calmodulin in these membranes); enzyme activity increases only slightly with the addition of exogenous calmodulin. By contrast, caveolin co-expression markedly inhibits eNOS activity (by Ͼ90%) in the absence of added calmodulin; addition of increasing concentrations of exogenous calmodulin relieves this enzyme inhibition in a dose-dependent fashion, documenting that the caveolin inhibitory effect may be specifically overcome by purified calmodulin.
Taken together, these studies suggest that the interaction between eNOS and caveolin is dynamically and specifically regulated by Ca 2ϩ -calmodulin and may serve as an important point of control in NO-dependent signaling. A direct interaction of caveolin with calmodulin appears unlikely to us because there was no influence of caveolin on the activity of other calmodulin-binding proteins (iNOS and nNOS) closely related to eNOS. This hypothesis is consistent with our failure to detect co-immunoprecipitation of calmodulin with caveolin ( Fig. 2), under conditions in which eNOS was shown to associate with either one or the other protein. Furthermore, the amino acid sequence of caveolin isoforms show no obvious sequence or structural homologies to the known NOS isoforms (10 -12) nor to any known calmodulin-binding protein sequences. Caveolin can attenuate the tyrosine kinase activity of c-src, an enzyme that bears no structural homology to eNOS, and is not known to be regulated by calmodulin (13). We speculate that there is a common higher order structure assumed by the inactive conformation of diverse caveolae-targeted sig-naling proteins that forms the basis for their common interaction with caveolin.
The targeting of eNOS to caveolae is likely to facilitate the interactions of eNOS with other co-localized signaling and regulatory molecules (3). Formation of an inhibitory eNOS-caveolin heteromeric complex may serve to ensure the latency of the NO signal until calcium-mobilizing extracellular stimuli destabilize this complex and activate the enzyme. This close control of enzyme activity may be particularly important for eNOS in caveolae, where calmodulin, the enzyme's key allosteric activator, also is localized (4) and where even subtle increases in intracellular Ca 2ϩ could thus lead to enzyme activation if the interactions of caveolin with eNOS were not keeping the sys- FIG. 3. Release of eNOS from caveolin immune complexes by Ca 2؉ -calmodulin and its recovery. Endothelial cell lysates were solubilized in Ca 2ϩ -free (1 mM EDTA, 1 mM EGTA) CHAPS buffer, and caveolin was immunoprecipitated as described in the text. Immune complexes were precipitated by the addition of protein G-Sepharose (16). The immune complexes bound to protein G-Sepharose beads were washed ten times in CHAPS buffer, and the beads were then equally distributed in four separate aliquots and incubated for 1 h at 4°C as L-citrulline nitric-oxide synthase activity assay performed in lysates of transfected COS-7 cells as described above. A, COS-7 cells were transfected with an expression plasmid encoding eNOS, in a co-transfection either with caveolin (cav) cDNA (as noted) or with an equivalent quantity of plasmid DNA encoding an irrelevant protein (␤-galactosidase). This experiment was conducted in triplicate three times with equivalent results. B, transfection of COS-7 cells with eNOS, iNOS, or nNOS cDNAs without (black bars) or with (shaded bars) caveolin cDNA were performed as described for A, with NOS enzyme activity measured in cell lysates and normalized to the activity seen in the absence of caveolin. Maximal NOS enzyme activities (in the absence of caveolin) in different experiments averaged 4.2 Ϯ 2.1 pmol citrulline formed/min⅐mg protein. The experiment shown was repeated three times in triplicate with identical results. C, NOS enzyme activity was measured in washed membranes prepared from COS-7 cells transfected with eNOS cDNA either without (black bars) or with (hatched bars) caveolin cDNA. NOS activity was assayed under conditions identical to those described above except that varying concentrations of purified calmodulin were added as noted on the abscissa. This experiment was performed twice in duplicate, yielding equivalent results. tem in check. Because nitric oxide has cytotoxic as well as signaling functions (1, 2), attenuation of basal enzyme "leakiness" by caveolin may be of particular importance. The reciprocal regulation of eNOS by caveolin and calmodulin may represent a novel mechanism for the concerted control of NO production in the vascular wall.