Heat Shock Protein 90 Mediates the Balance of Nitric Oxide and Superoxide Anion from Endothelial Nitric-oxide Synthase*

The balance of nitric oxide (·NO) and superoxide anion (O⨪2) plays an important role in vascular biology. The association of heat shock protein 90 (Hsp90) with endothelial nitric-oxide synthase (eNOS) is a critical step in the mechanisms by which eNOS generates ·NO. As eNOS is capable of generating both ·NO and O⨪2, we hypothesized that Hsp90 might also mediate eNOS-dependent O⨪2 production. To test this hypothesis, bovine coronary endothelial cells (BCEC) were pretreated with geldanamycin (GA, 10 μg/ml; 17.8 μm) and then stimulated with the calcium ionophore,A23187 (5 μm). GA significantly decreasedA23187-stimulated eNOS-dependent nitrite production (p < 0.001, n = 4) and significantly increased A23187-stimulated eNOS-dependent O⨪2production (p < 0.001, n = 8).A23187 increased phospho-eNOS(Ser-1179) levels by >1.6-fold over vehicle (V)-treated levels. Pretreatment with GA by itself or with A23187 increased phospho-eNOS levels. In unstimulated V-treated BCEC cultures low amounts of Hsp90 were found to associate with eNOS. Pretreatment with GA and/or A23187 increased the association of Hsp90 with eNOS. These data show that Hsp90 is essential for eNOS-dependent ·NO production and that inhibition of ATP-dependent conformational changes in Hsp90 uncouples eNOS activity and increases eNOS-dependent O⨪2production.

0.001, n ‫؍‬ 8). A23187 increased phospho-eNOS(Ser-1179) levels by >1.6-fold over vehicle (V)-treated levels. Pretreatment with GA by itself or with A23187 increased phospho-eNOS levels. In unstimulated V-treated BCEC cultures low amounts of Hsp90 were found to associate with eNOS. Pretreatment with GA and/or A23187 increased the association of Hsp90 with eNOS. These data show that Hsp90 is essential for eNOS-dependent ⅐NO production and that inhibition of ATP-dependent conformational changes in Hsp90 uncouples eNOS activity and increases eNOS-dependent O 2 . production.
Nitric oxide (⅐NO) and superoxide anion (O 2 . ) play opposing roles in vascular biology. Nitric oxide generation is increased greatly when Hsp90 associates with eNOS 1 in endothelial cell cultures (1,2). A decrease in the amount of Hsp90 co-precipitating with eNOS is associated with a decrease in ⅐NO production by pulmonary artery endothelial cells exposed to prolonged periods of hypoxia (3). Geldanamycin (GA) is an ansamycin antibiotic that binds to the ATP binding site of Hsp90, thereby inhibiting the ATP/ADP cycle required for the interaction with client proteins such as eNOS (2)(3)(4). GA has been used to demonstrate that ⅐NO production in mesentary arteries and rat aortas depends on Hsp90 activity, implying that factors adversely affecting this interaction between Hsp90 and eNOS may be one of the mechanisms for portal hypertension and increased vascular tone (2,4). Taken together, these reports indicate that Hsp90 is critical for eNOS generation of ⅐NO.
Emerging evidence suggests that under pathological conditions eNOS may also generate O 2 . (5)(6)(7). Electron spin resonance studies clearly demonstrate that purified recombinant eNOS generates O 2 . when activated by calmodulin (CaM) in the absence of its substrate, L-arginine, or the essential cofactor, tetrahydrobiopterin (8,9). As Hsp90 is important for mediating eNOS-dependent ⅐NO generation we wondered whether Hsp90 also mediates eNOS O 2 . generation. Our findings indicate that inhibiting Hsp90 function uncouples eNOS activity to increase eNOS-dependent O 2 . generation.
Endothelial Cell Culture-Bovine coronary endothelial cells (BCEC) were provided by William B. Campbell (Milwaukee, WI). BCEC were cultured in RPMI 1640 media containing 20% fetal bovine serum, antibiotics, mycotics, rhFGF (10 ng/ml), and heparin (5 units/ml Measurements of Nitrite-After the third wash, V-treated, L-NMAtreated, GA-treated, and GA-L-NMA-treated test cultures in 6-well plates were incubated with 0.75 ml HBSS containing A23187 (5 M) and L-arginine (25 M) for 30 min. Nitrite was quantified using Sievers NOA analyzer as described (10). Each experiment was performed in triplicate; nitrites were analyzed in duplicate or triplicate, and cell protein for each well was determined in duplicate.
Measurements of Superoxide Anion-After the final HBSS wash, the test groups in 6-well dishes (V, L-NAME, GA, and GAϩL-NAME) were incubated with 1 ml of HBSS containing ferricytochrome c (50 M) and A23187 (5 M) with and without L-NAME (1 mM) for 30 min. Superoxide anion production was calculated from the absorbance of ferricytochrome c at 550 nm. The release of O 2 . was calculated from the molar extinction from independent wells incubated with HBSS containing ferricytochrome c and SOD, 1000 units/ml). Each experiment was performed in triplicate, and the cell protein from each well was analyzed in duplicate.

Detection of Endothelial Superoxide Anion Generation in Isolated Carotid
Arteries-Canine carotid arteries were obtained from adult mongrel dogs. The vessel was removed, transferred to a physiologic saline solution and adventitia was cleared. The artery was sectioned into segments at least 2 cm long and placed in RPMI 1640 media. After two washings in RMPI 1640 (to remove adherent blood cells), vessels were placed in organ culture. Artery segments were incubated with V or GA as above, washed free of vehicle and GA, and then incubated with hydroethidine (10 M) for 30 min. Hydroethidine is taken up by cells and in the presence of O 2 . is converted to ethidine, which intercalates with nuclear DNA. The degree of fluorescence is proportional to the amount of O 2 . present (11). In situ fluorescence was assessed using confocal microscopy. Western Analysis, Immunoprecipitation, and Immunoblotting-Phospho-eNOS(Ser-1177) and eNOS levels were determined using the manufacturer's protocol. Briefly, confluent BCEC cultures in 60-mm dishes were pretreated with GA (10 g/ml) for 30 min. The cultures were washed three times with HBSS and then stimulated with A23187 (5 M) in HBSS for 10 min at 37°C with gentle horizontal rotation (50 rpm). After incubation, the HBSS solutions were removed by aspiration, and cell proteins were harvested in 200 l of SDS sample buffer. Aliquots (50 l) were heated (95°C, 5 min) and stored on ice until loaded (20 l/lane) on 7% SDS-PAGE. The proteins were transferred to nitrocellulose membranes and blotted with anti-phospho-eNOS(Ser-1177) and anti-eNOS (9D10, Zymed Laboratories Inc.) overnight at 4°C. Bands were visualized using the appropriate horseradish peroxidase (HRP)-linked secondary antibodies and ECL reagents.
Co-precipitation studies for determining the interaction of Hsp90 with eNOS were performed on confluent BCEC cultures in 100-mm dishes. The four test groups were V, GA, VϩA23187, and GAϩA23187. After preincubation with GA, the test groups were washed three times with 5 ml of HBSS and then incubated at 37°C in a tissue culture incubator in 5 ml of HBSS containing L-arginine (10 M) with and without A23187 (5 M). After 10 min, the HBSS solutions were removed, and the cells were lysed in modified RIPA buffer as described (2). The samples were sonicated two times at a power setting of 1.25-1.5 for 30 s on a Fisher Scientific Dismembrater (Model 550) fitted with an Ultrasonic Convertor (Model No. CL4 with a frequency of 20 kHz) probe. For this procedure it is important to adjust the power settings of the unit to 5-10% of maximum power output because higher power outputs disrupt eNOS interactions with other cell proteins. Cell debris was removed from the cell lysates, and eNOS was immunoprecipitated as described (2). The immunoprecipitated proteins were separated by SDS-PAGE (7.5-15%) and transblotted onto nitrocellulose. The membrane was blocked with 5% nonfat milk in TBS-Tween (0.1%) and then immunoblotted for eNOS and Hsp90 using 9D10 (33-4600, 1:1000) from Zymed Laboratories Inc., and H38220 (1:1000) from Transduction Laboratories, respectively. Bands were visualized with the appropriate anti-immunoglobulin-HRP conjugate from Sigma using ECL reagents and Kodak X-OMAT film. The autoradiograms were imaged with Adobe PhotoShop v5.5 and UMAX Magicscan v4.4 software, and the relative band densities were quantified using NIH Image 1.62.
Statistical Analysis-Nitrite and superoxide anion data were exam-ined by one-way analysis of variance to determine whether the variances and means were significantly different. The Newman-Kuel posthoc test was employed to determine the level of significance between means (**, p Ͻ 0.001 and *, p Ͻ 0.01).

RESULTS
Geldanamycin (GA) significantly inhibited A23187-stimulated nitrite production by BCEC cultures incubated in HBSS containing low concentrations of L-arginine (25 M) (Fig. 1A). In contrast, A23187 stimulation of GA-treated BCEC cultures significantly increased the release of O 2 . compared with vehicle (V)-treated BCEC cultures (Fig. 1B). L-NAME significantly reduced O 2 . release from GA-treated cells (Fig. 1B) (5,6). When the production of O 2 . was assayed on cells that were pretreated with sepiapterin, the effects of GA and L-NAME (Fig. 1C) were essentially the same as that obtained from non-sepiapterin-treated cells (Fig. 1B). On the basis that L-NAME is a well recognized substrate analog inhibitor of eNOS and an ineffective agent for blocking relaxation in large vessels from eNOS knock-out mice (12) Fig. 2, left). In contrast, the endothelium of GA-treated carotid arteries exhibit marked increases in ethidine staining (Fig. 2,  right). Deeper optical sectioning of the arteries was unable to detect ethidine staining in smooth muscle cells of the vessel wall (data not shown). These data suggest that GA increases vascular O 2 . generation by an endothelium-dependent mechanism. Serine phosphorylation of eNOS at Ser-1177 (human) and Ser-1179 (bovine) has been shown to correlate directly with increased ⅐NO production (13). Treatment of BCEC cultures with A23187 increased phospho-eNOS(Ser-1179) levels compared with the levels seen in V-treated cultures (Fig. 3A, top  panel). Pretreatment with GA alone increased phospho-eNOS levels (Ͼ2.5-fold) compared with V-treated cultures. A23187 stimulation of GA-treated cultures also increased phospho-eNOS. These data indicate that GA increases serine phosphorylation of eNOS, a marker of increased eNOS activity (2,14). The observation that A23187 simultaneously increases L-NAME-inhibitable O 2 . generation and eNOS phosphorylation in GA-treated cultures suggests that when Hsp90 is bound to eNOS in the ADP conformation, enzyme activity is uncoupled. Although the phosphorylation of eNOS stimulated by growth factors or shear stress has been shown to correlate directly with ⅐NO production (13), if phospho-eNOS levels are detected in the absence of ⅐NO, perhaps the phosphorylation of eNOS reflects uncoupled eNOS activity and enhanced electron flux in order to generate O 2 . as was shown previously (14).
Hsp90 serves many functions within cells (see Ref. 15 for review). It assists in protein folding by first binding and then changing the conformation of the Hsp90 bound to the client proteins. In this respect, GA has been a useful tool for distinguishing between the steps of association and conforma-Hsp90 Mediates the Balance of ⅐NO and O 2 . from eNOS tion in protecting enzyme function (16). To begin to understand how each step plays a role in preserving eNOS function co-precipitation studies of eNOS were performed on V-treated, V-treated ϩ A23187, GA-treated, and GA-treated ϩ A23187stimulated test groups. None of these treatments affected eNOS levels (Fig. 3B, upper panel). Independently GA and A23187 slightly increased the levels of Hsp90 that co-precipitated with eNOS (Fig. 3B, lower panel, lanes 2 and 3,  respectively). In combination however, they markedly increased the amount of Hsp90 that could be co-immunoprecipitated with eNOS (Fig. 3B, lower panel, lane 4). These observations are consistent with the report showing GA increases Hsp90 binding to heat-denatured luciferase but not necessarily enzyme activity (16 . generation in the endothelium but not smooth muscle cells. Such specificity is consistent with the observation that Hsp90 and eNOS are found to co-localize in the endothelium, and that their interaction is critical for mediating vasorelaxation and vascular tone (1,2,4). Increased phospho-eNOS(Ser-1179) levels in A23187-stimulated cultures pretreated with GA suggest that eNOS-dependent O 2 . generation and ⅐NO production may share common signaling mechanisms. The marked increase in the amount of Hsp90 that can be co-precipitated with eNOS in A23187-stimulated GA-pretreated cultures indicates that the ability of Hsp90 to change conformation is essential to the mechanisms by which Hsp90 increases ⅐NO and limits O 2 . production by eNOS. To our knowledge this is the first demonstration where the ATP/ADP state of Hsp90 influences different enzyme activities of a single protein.
By measuring the effects of GA on nitrite and O 2 . production in the absence and presence of well recognized eNOS inhibitors we were able to determine how GA altered eNOS function. What we observed was that when GA decreased A23187-stimulated nitrite production it also increased A23187-stimulated O 2 . generation by a mechanism that could be inhibited by L-NAME, which is now interpreted as an eNOS-dependent mechanism. In the present study, the finding that A23187 markedly increases phospho-eNOS in GA-treated cultures is entirely consistent with the idea that the flow of electrons through eNOS was increased (14). However when pretreating with GA, the increase in electron flow resulted in increased O 2 . generation by an activated eNOS. Superoxide anion can be generated by eNOS at either the NADPH reductase domain or the arginine oxidase domain (8,(17)(18)(19)(20). Xenobiotics such as paraquat, lucigenin, and adriamycin as well as excess FAD and FMN increase O 2 . generation via redox-cycling with the NADPH reductase domain of eNOS (8,(17)(18)(19)(20). "Uncoupled" eNOS activity develops when eNOS fails to "couple" activated oxygen to arginine metabolism. When this happens, the activated oxygen is released from the heme site as    (25,26).
On the basis that GA is a specific inhibitor of Hsp90 activity and BH4 supplementation did not inhibit the effects of GA on eNOS-dependent O 2 . generation, we hypothesize that the way Hsp90 interacts with eNOS plays an important role in mediating the balance of ⅐NO and O 2 . generation by eNOS. What these data seem to indicate is that when high levels of phospho-eNOS are detected in cells or tissues that generate low concentrations of ⅐NO, citrulline or cGMP eNOS activity may be uncoupled. On the basis of known stoichiometry for arginine metabolism by NOS (27) it is clear that for each ⅐NO made by coupled activity, two molecules of O 2 . can be generated by uncoupled activity (21). The ability of an enzyme to generate vasoactive radicals with opposing physiological properties may actually be an advantage in that mechanisms mediating vasorelaxation and vasoconstriction could be integrated into a single control point. The importance of uncoupled eNOS activity to vascular physiology remains unclear at this time. Future studies aimed at understanding how vasoactive agents alter Hsp90 activity in relation to eNOS generation of ⅐NO and O 2 .
are required to determine the extent to which such a mechanism plays a role in vasorelaxation and vascular tone (1).
In conclusion, Hsp90 modulates eNOS product formation. When Hsp90 is bound to eNOS and can change conformation, eNOS generates ⅐NO upon stimulation. When Hsp90 is bound to eNOS and conformational changes are restricted or impaired, eNOS generates O 2 . upon stimulation. Thus, Hsp90 modulation of eNOS production formation may play an important role in vascular physiology as well as atherosclerosis, hypertension, and diabetes. Such observations begin to explain some of the hypertensive and anti-angiogenic effects of inhibiting Hsp90 activity.