Inhibition of Endothelial Nitric-oxide Synthase by Ceruloplasmin*

The plasma copper protein ceruloplasmin (CP) was found to inhibit endothelial nitric-oxide synthase activation in cultured endothelial cells, in line with previous evidence showing that the endothelium-dependent vasorelaxation of the aorta is impaired by physiological concentrations of ceruloplasmin. The data presented here indicate a direct relationship between the extent of inhibition of agonist-triggered endothelial nitric oxide synthase activation and CP-induced enrichment of the copper content of endothelial cells. Copper discharged by CP was mainly localized in the soluble fraction of cells. The subcellular distribution of the metal seems to be of relevance to the inhibitory effect of CP, because it was mimicked by copper chelates, like copper-histidine, able to selectively enrich the cytosolic fraction of cells, but not by copper salts, which preferentially located the metal to the particulate fraction.

Ceruloplasmin (CP) 1 is a copper-containing glycoprotein that is found in the plasma of all vertebrates, in which it carries approximately 90% of plasma copper (1). Each CP molecule tightly binds six copper atoms to the three different copper binding sites that characterize blue oxidases (2,3). Nevertheless, as indicated by considerable experimental evidence, it is particularly prone to transfer its copper atoms to tissues (4 -6) delivering copper to intracellular copper proteins (7,8). However, recent studies on aceruloplasminemic patients (9 -11) indicate that this protein has no essential role in copper transport, whereas it plays a primary role in iron homeostasis, possibly through its ferroxidase activity (12).
Ceruloplasmin is an acute-phase reactant; its concentration in the plasma increases up to 3-fold during pregnancy and during multiple pathological processes including trauma and inflammation (13). Recent attention has focused on the role that this protein may have in the function of the vascular system in health and disease. CP has been detected in human atherosclerotic lesions (14,15), and it has been shown to oxidize low density lipoproteins in the presence of vascular cells (16). A copper binding site labile to Chelex treatment has been proposed to be responsible for the oxidative damage to low density lipoproteins (17). On the other hand, we have shown that CP, at physiological concentrations, inhibits the endothelium-dependent relaxation of rabbit aorta induced by agonists and that this effect is not due to a trapping of NO by the copper sites (18).
Vasodilation requires a (NO)-cGMP transduction pathway between endothelium and smooth muscle cells (19,20). Endothelial NO synthase (eNOS) is a constitutive enzyme that converts L-arginine into NO and citrulline (21) with a relatively low basal activity (22,23). After agonist stimulation evoking an increase in the [Ca 2ϩ ] i concentration, Ca 2ϩ -bound calmodulin disrupts the inhibitory eNOS-caveolin-1 interactions (24,25), thereby allowing conformational changes within eNOS, leading to the activated form that produces NO (26,27). The agonist bradykinin (Bk) has a distinct role among vasodilators because it has also been shown that its receptor physically associates with eNOS (28).
In this context, it should be recalled that a regulatory role for dietary copper in the control of vascular functions has been assessed by numerous studies focused mainly on copper deficiency-induced defects of vessels and on the impairment of NO-mediated vasodilation under copper restriction (reviewed in Ref. 29). On the contrary, an ability of copper ions (Cu 2ϩ ) to relax vessels seems well established. It has been shown that copper enhances the relaxation of precontracted aortic rings evoked by the calcium ionophore A23187 and sodium nitroprusside (30), and it also elevates intracellular cGMP levels and induces relaxation of pulmonary arterial rings (31). More recently, it has been demonstrated that copper induces the activation of eNOS in cultured endothelial cells (32), an event supposed to occur in vivo in hypercupremic states induced by Cu 2ϩ released by ceruloplasmin. This result is apparently in contrast with our previous findings (18), which are in better accordance with the inhibitory effects exerted by divalent metal ions on eNOS (32,33) as well as on neuronal nitric oxide synthase (34), another constitutive enzyme, when activity measurements are performed on crude cell extracts or on the purified enzyme (35).
To clarify the mechanism underlying the inhibitory effect of CP on the relaxation of rabbit aortic vessels (18), we tested whether this protein could affect the agonist-induced activation of eNOS in cultured OAECs. Here we show that exogenously added CP reversibly reduces the formation of cGMP, nitrite, and citrulline but not Ca 2ϩ mobilization in endothelial cells stimulated with agonists, with a time course consistent with that of copper delivery to cells, and we show that copper bound to histidine, but not free ionic copper, mimics the effect of CP. Altogether, the results are consistent with a mechanism whereby intracellular, CP-derived copper inhibits eNOS activation by agonists.

EXPERIMENTAL PROCEDURES
Materials-All reagents were purchased from Sigma Italia (Milan, Italy) unless otherwise noted and were used without further purification. Radioactive chemicals were from Amersham Pharmacia Biotech (Milan, Italy).
Cell Culture-OAECs were harvested from the internal surface of aortas according to a previously described procedure (36) using collagenase XI and grown in a culture medium containing DMEM (Life Technologies, Inc.) supplemented with 10% FBS (Life Technologies, Inc.), 30 g/ml endothelial cell growth supplement (Sigma Italia), 100 units/ml penicillin, 100 g/ml streptomycin, and 50 g/ml gentamicin (Life Technologies, Inc.) at 37°C under an atmosphere of 5% CO 2 in air. Their identity was verified by their morphological features and immunofluorescence staining with antibodies to factor VIII. The confluent monolayers were subcultured by conventional trypsinization. For the present study, cells were used at the third to sixth passages. Stimulation by agonists was carried out with confluent cells in either modified Hanks' balanced salt solution (m-HBSS), serum-free DMEM, or DMEM supplemented with 10% FBS (DMEM/FBS). To remove possible traces of thrombin, all CP samples were treated with benzamidine-Sepharose (Amersham Pharmacia Biotech, Uppsala, Sweden) immediately before addition to cells.
Preparation of Ceruloplasmin-Sheep and human CP were purified as described previously (2,37) and further purified by mono Q fast protein liquid chromatography to remove traces of prothrombin. The resulting proteins were Ͼ99% pure as judged by spectroscopic and electrophoretic analyses. In some experiments, purified CP samples were treated with Chelex 100 (38). Apoceruloplasmin (ApoCP) was prepared as described previously (39). ApoCP had a residual oxidase activity of ϳ10% with respect to the native protein, consistent with the presence of ϳ10% unremoved copper. Neutralization of CP with anti-CP antibody was carried out by preincubating CP on ice for 30 min with a 2-fold excess of a polyclonal specific anti-sheep ceruloplasmin antibody (generous gift of Dr. Marmocchi) before the treatment of cells.
cGMP Assay-To determine [cGMP] i levels in cultured OAECs, cells were preincubated in complete DMEM for 30 min at 37°C under 5% CO 2 with 1 mM isobutylmethylxanthine, a phosphodiesterase inhibitor, before treatment with effectors and/or agonists. At the end of the stimulation with the agonist (10 min), cells were washed with PBS and lysed in 10 mM Tris/HCl, pH 7.4, with 0.5% Triton X-100. The lysate was collected in 1% perchloric acid and centrifuged at 10,000 ϫ g for 10 min at 4°C. Precipitated proteins were determined with either the biuret method (43) or the bicinchoninic acid reagent (Pierce), whereas the supernatants were lyophilized and assayed for cGMP by radioimmunoassay (Amersham Pharmacia Biotech). Each experiment was performed in duplicate. NO x Assay-Formation of nitrites in the medium was quantitated fluorimetrically with 2,3-diaminonaphthalene according to Misko (40). Briefly, confluent cells in 100-mm dishes were stimulated in DMEM/ FBS without phenol red, and the medium (2 ml) was collected and centrifuged at 10,000 ϫ g for 10 min. 250 l of a solution of 2,3diaminonaphthalene (5 mg/100 ml in 0.62 N HCl) were added to the supernatant, and the mixture was incubated for 10 min at 20°C in the dark. After the addition of 250 l of 1.4 M NaOH, samples were filtered through a 0.45-m cellulose acetate filter (Iwaki, Japan), and the fluorescence was measured with ex ϭ 375 nm and em ϭ 426 nm on a Perkin Elmer LS50-B spectrofluorimeter. A standard curve was constructed with a known concentration of sodium nitrite in the same medium. To measure total NO x (nitrite ϩ nitrate), nitrates were first converted to nitrite by incubation for 15 min at 37°C with 50 milliunits/ml nitrate reductase from Aspergillus sp. in the presence of 30 M NADPH. Residual NADPH was oxidized by incubation for 5 min at 37°C with lactic dehydrogenase (5 milliunits/ml) in the presence of 0.3 mM pyruvate. Measurements of total NO x were carried out using m-HBSS as the cell medium because DMEM contains micromolar levels of nitrate.
NO Synthase Activity-eNOS activity in intact OAECs was assayed by measuring the intracellular formation of [ 3 H]citrulline using the method of Hu and El-Fakahany (41), modified as follows. Confluent cells grown in 6-well dishes were loaded with [ 3 H]arginine by a 15-min incubation at 37°C with DMEM/FBS containing 2 mM glutamine, 1 mM citrulline, and [ 3 H]arginine (10 Ci/ml, 5 Ci/well, and 1 Ci/40 nmol total arginine). After an additional 15-min incubation with or without CP, cells were treated with agonists, washed twice with PBS containing 1 mM arginine, and lysed in cold absolute methanol. Proteins were assayed after quantitative recovery from the wells with 0.1 M NaOH. The methanol extracts were dried in a Jouan RC 10.22 Speedvac, and pellets were redissolved in 500 l of H 2 O. Chromatographic separation of [ 3 H]citrulline from [ 3 H]arginine was achieved by batch incubation of the mixture for 10 min with 100 l of a Dowex 50WX8 slurry under continuous mild agitation. Supernatants were counted for radioactivity to evaluate intracellular formation of [ 3 H]citrulline. Values were normalized for the protein content of each well.
The conversion of [ 3 H]arginine into [ 3 H]citrulline was also used to measure eNOS activity in OAEC homogenates, essentially as described by Hecker et al. (42). Cells were detached by gently scraping the culture dishes and resuspended in 50 mM Tris/HCl, pH 7.4, containing 120 mM NaCl, 15 mM KCl, 5 mM MgCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 g/liter leupeptin, 10 g/liter pepstatin, 10 g/liter aprotinin, and 20 mM CHAPS. The suspension was homogenized in ice with a mini-potter apparatus and then centrifuged at 10,000 ϫ g for 10 min at 4°C. 20 -50-l aliquots of the supernatants were assayed for eNOS activity by the addition of 50 mM Tris/HCl, pH 7.4, containing 1 mM NADPH, 1.25 mM CaCl 2 , 1 mM dithiothreitol, 1 mM EDTA, 15 M 6R-tetrahydrobiopterin, 1 M FAD, 1 M FMN, 0.1 M calmodulin, 10 M arginine, and 5 Ci/ml [ 3 H]arginine (total volume, 150 l). After incubation at 37°C for 60 min, the reaction was stopped with 5 volumes of 20 mM Hepes, pH 5.5, containing 10 mM EDTA. The mixture was loaded on a 1-ml column of Dowex 50WX8 pre-equilibrated in the latter buffer. The eluate was counted for radioactivity to evaluate the extent of [ 3 H]citrulline formation. The effect of copper was assessed by adding increasing concentrations of CuCl 2 to the homogenate 5 min before adding the cofactors.
Western Blotting-Cell lysates were separated by SDS-polyacrylamide gel electrophoresis. The proteins were electrophoretically transferred onto polyvinylidene difluoride membranes, immunodetection was carried out with either anti-eNOS (1:1000; Transduction Laboratories) or anti-CP and visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech).
Copper Transport Experiments-Cells were incubated with the indicated amounts of CP or copper-histidine, prepared by adding 1 mol of CuCl 2 to 3 mol of histidine in water, or CuCl 2 , in DMEM/FBS, DMEM or m-HBSS. At the end of the incubation, cells were washed three times with PBS containing 1 mM EDTA and either lysed in 10 mM Tris/HCl buffer, pH 7.4, with 0.5% Triton X-100 to determine total copper content or collected by scraping and homogenized by 30 strokes in a mini-Potter apparatus in ice-cold homogenization buffer (50 mM Tris/HCl buffer, pH 7.4, containing 120 mM NaCl, 15 mM KCl, 5 mM MgCl 2 , 1 mM EDTA, 10 g/liter leupeptin, 10 g/liter pepstatin A, 10 g/liter aprotinin, and 1 mM phenylmethylsulfonyl fluoride) and centrifuged for 60 min at 100,000 ϫ g. The pellets were resuspended in 50 mM Tris/HCl buffer, pH 7.4, containing 1% Triton X-100. Copper content was determined on aliquots of lysates, homogenates, supernatants, and pellets after digestion with HNO 3 by flameless atomic absorption (Perkin Elmer 3030 spectrometer equipped with graphite furnace). The data were normalized by dividing picomoles Cu by mg proteins. Proteins were determined using either the biuret method (43) or the bicinchoninic acid reagent (Pierce).
Calcium Mobilization Assay-Cells were grown at confluence in chamber slides (Lab-Tek; Nunc, Naperville, IL) and treated with 4 M fura 2-acetoxymethyl ester for 45 min at 37°C to monitor variations of cytosolic free Ca 2ϩ concentrations (44). Emission fluorescence at 510 nm was measured upon excitation at 340 nm and 380 nm and expressed as F 340 /F 380 .

RESULTS AND DISCUSSION
Effects of CP on eNOS Activation-Nitric-oxide synthase is activated in endothelial cells by several agonists including bradykinin and acetylcholine. Because this leads to NO-dependent activation of endothelial guanylate cyclase (45), the rise of the intracellular cyclic GMP concentration, [cGMP] i , was used as an index of eNOS activity. Untreated OAECs have a basal level of cGMP that was not affected by CP (Fig. 1A). Bk induced an approximately 6-fold increase of [cGMP] i that was abolished by inhibition of eNOS by 1 mM L-NAME. When added to cells 15 min before stimulation, CP had a strong, dose-dependent inhibitory effect on the agonist-induced increase of cGMP levels, with a maximal effect at 10 M concentration. The inhibition was already evident at 1 M CP and required the native form, because copper-free CP had a small effect, the magnitude of which was accounted for by the presence of approximately 10% residual active holoprotein (see "Experimental Procedures"). Treatment of CP samples with Chelex 100 (38) to remove loosely bound copper before the addition to cells had no effects on the results shown. Neutralization by a specific antibody substantially relieved the effect of 10 M CP. By itself, the antibody affected neither the basal nor the bradykinin-stimulated levels of cGMP.
To examine whether CP altered [cGMP] i through NO-independent routes rather than affecting eNOS catalytic activity, the eNOS activity was monitored using the conversion of [ 3 H]arginine to [ 3 H]citrulline as a measure of NO synthase activity (Fig. 1B). Stimulation of untreated cells by Bk produced a 3-fold enhancement of citrulline production, which was abolished by L-NAME. Preincubation of cells with CP before stimulation substantially reduced Bk-stimulated citrulline pro-duction in the same manner as that observed for cGMP production, indicating that the lower levels of [cGMP] i attained by cells stimulated in the presence of extracellular CP were indeed related to an inhibition of eNOS activity. Consistent with this finding, CP strongly hindered the release of NO in the medium. OAECs stimulated in m-HBSS for 10 min with 1 M bradykinin produced ϳ130 pmol/mg protein NO x (nitrite ϩ nitrate), with over 80% of NO x accounted for by nitrite. Preincubation of cells with 10 M CP for 15 min nearly abolished NO x production. Human CP was also effective in inhibiting the agonist-induced eNOS activation, being ϳ80% as active as the ovine enzyme.
The effect of CP was independent of the signaling pathway responsible for eNOS activation. As revealed by [cGMP] i (Fig.  1C) and [ 3 H]citrulline (Fig. 1D), CP exerted a strong inhibitory effect on the response of OAECs to acetylcholine or ADP and on the non-receptor-dependent response to the Ca 2ϩ ionophore A23187. Altogether, the data reported in Fig. 1, A-D, show that the effect of CP is independent of the agonist used to stimulate cells and of the product of eNOS activation that is monitored. The fact that similar end points are obtained with all tested agonists suggests that the effect of CP is exerted on the common target of all agonists, i.e. on eNOS itself.
In agreement with the observation that aortic rings recover the capability to relax after the removal of CP (18), these cells were again found to be responsive to agonists upon the removal of CP and subsequent incubation in fresh medium. However, the results reported in Fig. 1E show that cells that had been exposed to 10 M CP need 30 min to recover full response to the agonist, either Bk or acetylcholine.
The extent of inhibition of the agonist-induced activation of eNOS depended on the time of exposure of cells to CP. In Fig.  2, the levels of [cGMP] i , intracellular [ 3 H]citrulline, and nitrite in the medium are shown for cells stimulated with bradykinin at different times after the addition of 10 M CP to the incu-bation medium. It should be noted that the indicated times actually encompassed the 10-min time period required for stimulation. It is evident that: (i) the first 15-min preincubation with CP caused an essentially complete (nearly 80%) inhibition of eNOS activity that remained at this level for more than ϳ60 min, and (ii) a small inhibition occurred when CP was added simultaneously to bradykinin, i.e. zero time of preincubation. Moreover, no significant inhibition was found when CP was added 2 min after the agonist. This experiment was critical because it showed that CP had to be in contact with cells before the agonist in order to exert its effect and was nearly ineffective in suppressing the response of cells once a response had been evoked. It should be noted that the kinetics of formation of citrulline and nitrite (i.e. NO) are similar; however, they differ at short times from that of cGMP. The lag phase observed in the latter case could be explained when we consider that the formation of cGMP requires the secondary activation of guanylate cyclase. Because agonists activate eNOS through enhancing the free cytosolic calcium concentration, the next question was whether CP could interfere with calcium fluxes in endothelial cells. As indicated by a representative experiment (Fig. 3), the F 340 /F 380 fluorescence intensity ratio, which increases upon elevation of [Ca 2ϩ ] i , varied after stimulation of OAECs by bradykinin, independently of prior exposure of cells to CP. Similar results were also obtained with A23187, with the ionophore inducing only a bigger variation of the fluorescence intensity ratio (data not shown).
Copper Delivery by CP to Endothelial Cells-It has been shown repeatedly that CP interacts directly with the membrane surface of cells of many tissues, including heart and aorta (46), and that such interactions lead to a transmembrane transport of copper but not of the protein moiety (47). An enhanced metal exportation generally follows the increase of the intracellular concentration of copper in normal cells (48,49).
To assess the possible effect on eNOS of a transient increase in copper levels, the delivery of copper to OAECs by CP was studied. CP affected the copper content of OAECs in a time-dependent manner (Fig. 4). OAECs exposed to 10 M CP accumulated copper for at least 60 min, with an almost 7-fold increase, after 60 min, with respect to that of untreated cells (Fig. 4A). Copper uptake by a number of cell types has been the object of intensive investigation, which has also shown that the metal does not easily redistribute among cellular components when cells are homogenized (50). Therefore, analyses of total membranes and cytosolic fractions of treated OAECs were performed and demonstrated that copper was mostly accumulated in the cytosol (Fig. 4B). In contrast, the amount of CP taken by cells or bound to cell membranes, as detected by immunoblot analyses of cell lysates, was at any time approximately 2 orders of magnitude less than the amount of copper (results not shown). As expected, it was also found that the higher the amount of copper extracted by the cells, the lower the amount of cGMP formed upon stimulation, with an apparently linear relationship (Fig. 4C).
When the copper content was measured on lysates of the cells exposed to CP for 30 min and then washed and left to equilibrate with the medium, the copper level was found to decrease within 30 min to the original level of the untreated cells upon removal of CP (Fig. 4D). These results are consistent with the transient increase in intracellular copper induced by CP as the origin of the inhibitory effect and with the copper efflux mechanism at the basis of cell recovery after CP removal.
Effect of Copper Complexes-If an increase in intracellular copper is required to suppress eNOS activation by agonists, then any copper-donating complex should have the inhibitory effect. Cultured cells are able to take up copper from whatever source is offered, including copper salts and copper-amino acid chelates, although with different transport properties, i.e. kinetics, preference, and path of entry (51).
To this purpose, copper bound to histidine, a complex of physiological relevance for copper transport (50,52), and copper chloride were tested for their ability to donate copper to OAECs and to impair the bradykinin-induced activation of eNOS. The metal concentrations of the copper salt and of the copper bound to histidine were adjusted to bring the amount of copper to the same level of 1-10 M CP. Incubations were performed in serum-free media, either DMEM or m-HBSS, to avoid possible interference by serum on the copper status. Copper-histidine was assayed in DMEM to maintain a large excess of histidine, whereas the copper salt was assayed in m-HBSS both to avoid the formation of amino acid chelates and to prevent its calcium-dependent activating effects on eNOS (32).
Control experiments demonstrated that all effects of CP were reproduced under these conditions. As shown in Fig. 5A, the response of cells to the agonist varied depending on the medium; nevertheless, CP was able to exert its effect regardless of the composition of the medium. When cells were incubated for 30 min with copper chloride or copper-histidine (50 M), there was no overall difference in the uptake of copper with respect to CP (data not shown). Copper bound to histidine reduced the Bk-stimulated production of cGMP in a dose-dependent manner to nearly the same extent as CP (Fig. 5A). On the contrary, CuCl 2 did not exert an inhibitory effect on the agonist-induced response or have any activating effect without the agonist, as expected from the absence of extracellular calcium. The same behavior was observed when the production of [ 3 H]citrulline was monitored (data not shown).
Although copper chloride and the copper-histidine complex appeared to be as efficient as CP in promoting copper enrich- ment of the cells, only the latter complex reproduced the subcellular distribution observed with CP (Fig. 5B). This fact can be rationalized on the basis of the chemical difference between the two copper sources; copper-histidine is a very stable complex (53) likely recognized by a specific carrier on the cell surface. In the case of CuCl 2 , the copper content was much higher in the total membranes than in the soluble fraction when normalized for protein amount, indicating that in order to be active in inhibiting the cell response to agonists, copper has to be localized in the soluble pool.
Recent studies have shown that Cu 2ϩ assayed in the range of 10 Ϫ8 to 10 Ϫ4 M inhibits eNOS activity in crude cell extracts of cultured pulmonary endothelial cells in a manner consistent with the binding of copper to the enzyme, with one copper atom per enzyme dimer giving the best fit of the experimental data (32). Accordingly, we have observed an inhibitory effect of CuCl 2 on eNOS activity in a cell-free assay, with half inhibition at 40 M (Fig. 5C). If eNOS in intact endothelial cells is as freely accessible to copper binding as it appears to be in cellfree extracts, a reasonable conclusion is that CP-or histidinederived copper interacts with resting eNOS in such a way as to hinder its activation by agonists. Consistent with this mechanism is the observation that an estimate of the cytosolic concentrations of copper in Cu-loaded cells after exposure to 10 M CP or 50 M copper-histidine gives values in the range of hundreds of micromoles/liter. It should be noted that at a high concentration of copper, the tightly regulated network of chaperones of the intracellular traffic can be bypassed entirely (54).
Possible in Vivo Implications-The concentrations at which CP exerts its inhibition are within the range found in the plasma, in which the protein is present at 1-5 M, and under particular physiological (i.e. pregnancy) and pathological (i.e. inflammation) conditions, the concentrations can be as high as 10 -15 M (55). Considering that NO synthesized by eNOS and released by endothelial cells regulates the vascular tone in normal vessels (19,56), one would expect aceruloplasminemic patients to be hypotensive. This has not been reported and is unlikely to occur due to redundancy of the copper transport systems (57). In other words, the lack of circulating ceruloplasmin does not disrupt copper metabolism (9), and this demonstrates that different carriers can substitute for ceruloplasmin. Here we show that in eNOS inhibition, copper-histidine is nearly as efficient as CP.
In view of its dose dependence, the inhibitory effect of CP on eNOS is likely to work in vivo under conditions of high CP levels. As a matter of fact, a hypertensive state can develop during pregnancy, and the involvement of CP in it has been taken into consideration (58,59). Moreover, epidemiological studies have clearly shown that elevated serum copper levels are associated with increased risk of cardiovascular mortality (60 -62). In this context, our findings would suggest a role of CP in the control of endothelial NOS activation processes and may provide a regulatory role for copper entering endothelial cells.