hsp90 Is Required for Heme Binding and Activation of Apo-Neuronal Nitric-oxide Synthase

It is established that neuronal NO synthase (nNOS) is associated with the chaperone hsp90, although the functional role for this interaction has not been defined. We have discovered that inhibition of hsp90 by radicicol or geldanamycin nearly prevents the heme-mediated activation and assembly of heme-deficient apo-nNOS in insect cells. This effect is concentration-dependent with over 75% inhibition achieved at 20 μm radicicol. The ferrous carbonyl complex of nNOS is not formed when hsp90 is inhibited, indicating that functional heme insertion is prevented. We propose that the hsp90-based chaperone machinery facilitates functional heme entry into apo-nNOS by the opening of the hydrophobic heme-binding cleft in the protein. Previously, it has been reported that the hsp90 inhibitor geldanamycin uncouples endothelial NOS activity and increases endothelial NOS-dependent O 2 ⨪ production. Geldanamycin is an ansamycin benzoquinone, and we show here that it causes oxidant production from nNOS in insect cells as well as with the purified protein. At a concentration of 20 μm, geldanamycin causes a 3-fold increase in NADPH oxidation and hydrogen peroxide formation from purified nNOS, whereas the non-quinone hsp90 inhibitor radicicol had no effect. Thus, consistent with the known propensity of other quinones, geldanamycin directly redox cycles with nNOS by a process independent of any action on hsp90, cautioning against the use of geldanamycin as a specific inhibitor of hsp90 in redox-active systems.

It is established that neuronal NO synthase (nNOS) is associated with the chaperone hsp90, although the functional role for this interaction has not been defined. We have discovered that inhibition of hsp90 by radicicol or geldanamycin nearly prevents the heme-mediated activation and assembly of heme-deficient apo-nNOS in insect cells. This effect is concentration-dependent with over 75% inhibition achieved at 20 M radicicol. The ferrous carbonyl complex of nNOS is not formed when hsp90 is inhibited, indicating that functional heme insertion is prevented. We propose that the hsp90-based chaperone machinery facilitates functional heme entry into apo-nNOS by the opening of the hydrophobic hemebinding cleft in the protein. Previously, it has been reported that the hsp90 inhibitor geldanamycin uncouples endothelial NOS activity and increases endothelial NOSdependent O 2 . production. Geldanamycin is an ansamycin benzoquinone, and we show here that it causes oxidant production from nNOS in insect cells as well as with the purified protein. At a concentration of 20 M, geldanamycin causes a 3-fold increase in NADPH oxidation and hydrogen peroxide formation from purified nNOS, whereas the non-quinone hsp90 inhibitor radicicol had no effect. Thus, consistent with the known propensity of other quinones, geldanamycin directly redox cycles with nNOS by a process independent of any action on hsp90, cautioning against the use of geldanamycin as a specific inhibitor of hsp90 in redox-active systems.
The endothelial and neuronal isoforms of nitric-oxide synthase (NOS) 1 have been reported to exist in heterocomplexes with hsp90 (1,2). These proteins join a list of numerous other signaling proteins, including steroid receptors, some transcrip-tion factors, and a variety of protein kinases, that are associated with and regulated by hsp90 (for a review, see Ref. 3). These signaling protein⅐hsp90 heterocomplexes are assembled in an ATP-dependent process by a five-protein system in which hsp90 and hsp70 are essential assembly components and Hop, hsp40, and p23 function as non-essential co-chaperones (4). One of the most studied hsp90-bound proteins is the glucocorticoid receptor (GR), which must be associated with hsp90 to have steroid binding activity (5,6). Hsp90 binds directly to the ligand-binding domain of the GR (3), and biochemical data (7) coupled with data from GR mutants (8,9) support the notion (6) that formation of a complex with hsp90 opens up a hydrophobic pocket in the ligand-binding domain to access by steroid. We have proposed a similar model for neuronal NOS (nNOS) in which the hsp90-based chaperone machinery acts in vivo to open the heme-binding cleft in heme-deficient apo-nNOS to access by heme (2).
In contrast to the observations with steroid receptors and nNOS, it has been proposed that hsp90 regulates endothelial NOS (eNOS) through an allosteric mechanism. Garcia-Cardeñ a et al. (1) were able to demonstrate direct activation of purified eNOS catalytic activity by purified hsp90 in the absence of ATP, hsp70, and the co-chaperones of the hsp90-based chaperone machinery. The proposed allosteric regulation of eNOS by hsp90 would be a unique mode of regulation in that other hsp90-regulated proteins require the ATP-and hsp70-dependent, multiprotein hsp90 chaperone system for their regulation. More recently, the same laboratory (10) has shown that geldanamycin, a specific inhibitor of hsp90, causes eNOS to produce superoxide in cells and tissues. From this, it was concluded that hsp90 is essential for eNOS-dependent NO production and that inhibition of hsp90 leads to uncoupling of eNOS and increased superoxide production due to conformational changes in hsp90 and eNOS. Other reports suggest that hsp90 facilitates the calmodulin-assisted dissociation of eNOS from caveolin, thereby opposing the inhibitory effect of caveolin on eNOS activity (11), and that eNOS-bound hsp90 recruits the protein kinase Akt, resulting in eNOS phosphorylation and sustained activation of the enzyme (12).
In the case of nNOS, it has been shown that geldanamycin causes an increased turnover of nNOS and a loss in nNOS activity, implying that hsp90 is important for the stability and function of the enzyme (2). Song et al. (13,14) have shown that hsp90 in the absence of ATP, hsp70, or co-chaperones enhances the affinity of calmodulin binding to purified nNOS by ϳ10fold. In our previous study (2), we were unable to show hsp90mediated activation of the holo-nNOS, but we did show that geldanamycin treatment of Sf9 cells expressing nNOS inhibits heme-mediated activation of heme-deficient apo-nNOS. Thus, in analogy with the well studied steroid receptor model, we proposed that the hsp90-based chaperone machinery plays a role in opening up the hydrophobic-binding cleft in apo-nNOS to facilitate the entry of heme into its binding site in the interior of the enzyme.
In this study, we show that the hsp90-inhibitors, geldanamycin and radicicol, cause a profound inhibition of heme-mediated activation of apo-nNOS activity in the insect overexpression system. Inhibition of hsp90 is accompanied by a marked inhibition in heme binding and by an inability to form the dimeric state of nNOS, implying a role for hsp90 in hememediated assembly of the holo-enzyme. In the course of these studies, we have discovered that geldanamycin, an ansamycin benzoquinone, is capable of redox cycling with nNOS to produce reactive oxygen species in a manner similar to that characterized for other quinones (15). In contrast to geldanamycin, radicicol, a specific hsp90 inhibitor that is not a quinone, blocks heme binding by nNOS but does not cause the formation of reactive oxygen species. Thus, in contrast to the conclusion of Pritchard et al. (10) in their work with eNOS, hsp90 does not regulate oxidant generation from nNOS. This observation provides a caution against the use of geldanamycin as a mechanistic probe for hsp90 actions in redox-active systems in which drug-mediated oxidants may be generated.

Materials
Untreated rabbit reticulocyte lysate was from Green Hectares (Oregon, WI). [6, H]triamcinolone acetonide (38 Ci/mmol) and 125 I-conjugated goat anti-mouse and anti-rabbit IgGs were obtained from PerkinElmer Life Sciences. (6R)-5,6,7,8-Tetrahydro-L-biopterin was purchased from Dr. Schirck's Laboratory (Jona, Switzerland). Protein A-Sepharose, iron protoporphyrin IX, L-arginine, myoglobin (horse heart), glucose-6-phosphate, glucose-6-phosphate dehydrogenase, calmodulin, catalase, radicicol, NADPH, and NADP ϩ were purchased from Sigma. The affinity-purified rabbit IgG against brain NOS used for immunoblotting nNOS was from Transduction Laboratories (Lexington, KY). The cDNA for rat neuronal NOS was kindly provided by Dr. Solomon Snyder (The Johns Hopkins Medical School, Baltimore, MD). Hybridoma cells producing the FiGR monoclonal IgG against the GR were generously provided by Dr. Jack Bodwell (Dartmouth Medical School). The baculovirus for expression of the mouse GR was generously provided by Dr. Edwin Sanchez (Medical College of Ohio, Toledo, OH). Geldanamycin was obtained from the Drug Synthesis and Chemistry Branch, NCI National Institutes of Health (Bethesda, MD).

Methods
Expression of nNOS or GR in Sf9 Cells-Recombinant baculovirus containing nNOS cDNA was produced as described previously (2). Sf9 cells were grown in SFM 900 II serum-free medium (Invitrogen) supplemented with Cytomax (Kemp Biotechnology, Rockville, MD) in suspension cultures maintained at 27°C with continuous shaking (150 rpm). Cultures (2 ϫ 10 6 cells/ml) were infected in log phase of growth with recombinant baculovirus at a multiplicity of infection of 1.0. After 48 h, aliquots (3.0 ml) of the suspensions were treated with heme (24 M), which was added as an albumin conjugate (16). In some samples, the cells were treated with geldanamycin, radicicol, or Me 2 SO vehicle 20 min before the addition of heme. The final Me 2 SO concentration was less than 0.4%. Cells were harvested, washed twice with 10 ml of ice-cold phosphate-buffered saline, and suspended in 350 l of 10 mM Hepes, pH 7.5, containing 320 mM sucrose, 100 M EDTA, 1.0 mM dithiothreitol, 10 g/ml trypsin inhibitor, 10 g/ml leupeptin, 2 g/ml aprotinin, and 6 mM phenylmethanesulfonyl fluoride. The cells were ruptured by Dounce homogenization, and the lysates were spun at 16,000 ϫ g for 10 min at 4°C. The supernatant was collected, aliquoted, flash-frozen, and stored at Ϫ80°C.
Sf9 cultures infected in log phase of growth with recombinant baculovirus containing the mouse GR cDNA at a multiplicity of infection of 3.0 were supplemented with 0.1% glucose at infection and 24 h postinfection as described by Srinivasan et al. (17). Cells were harvested, washed in Hanks' buffered saline solution, resuspended in 1.5 volumes of buffer (10 mM Hepes, pH 7.4, 1 mM EDTA, 20 mM molybdate, 1 mM phenylmethylsulfonyl fluoride) with 1 tablet of Complete-Mini protease inhibitor mixture per 3 ml of buffer, and ruptured by Dounce homoge-nization. The lysate was then centrifuged at 100,000 ϫ g for 30 min, and the supernatant was collected, aliquoted, flash-frozen, and stored at Ϫ70°C.
Purification of nNOS-Insect cells were infected as above except that 25 M oxyhemoglobin was added instead of heme. Lysates from infected Sf9 cells (8 ϫ 10 9 ) were centrifuged at 100,000 ϫ g for 1 h. The supernatant fraction was loaded onto a 2Ј5Ј-ADP-Sepharose column, and the nNOS was affinity-purified as described previously except that 10 mM 2ЈAMP in high salt buffer was used to elute the protein (18). The nNOS-containing fractions were combined and loaded onto a Sephacryl S-300 HR gel filtration column (2.6 ϫ 100 cm, Amersham Biosciences) equilibrated with 50 mM Tris-HCl, pH 7.4, containing 100 mM NaCl, 10% glycerol, 0.1 mM EDTA, and 0.1 mM dithiothreitol. The column was eluted at a flow rate of 1.0 ml/min, and 1.3 ml-fractions were collected and analyzed for protein content and NOS activity. Fractions containing nNOS were pooled and concentrated with the use of a Centriplus concentrator (Millipore Corporation, Bedford, MA). The concentrated enzyme was stored at Ϫ80°C.
Assay of nNOS Activity-NO formation was assayed by the NOmediated conversion of oxyhemoglobin to methemoglobin. Conversion of oxyhemoglobin was assayed by adding 15 l of insect cell supernatant to an assay solution containing 100 M CaCl 2 , 100 M L-arginine, 100 M tetrahydrobiopterin, 100 units/ml catalase, 10 g/ml calmodulin, 25 M oxyhemoglobin, and an NADPH-regenerating system consisting of 400 M NADP ϩ , 10 mM glucose-6-phosphate, and 1 unit/ml glucose-6phosphate dehydrogenase, expressed as final concentrations, in a total volume of 180 l of 50 mM potassium phosphate, pH 7.4. The mixture was incubated at 37°C and the rate of oxidation of oxyhemoglobin was monitored by measuring the absorbance at 401-411 nm with a microtiter plate reader as described (19). In studies in which the NADPH oxidation was measured, purified nNOS (5 g/ml) was added to a mixture containing 0.2 mM CaCl 2 , 10 g/ml calmodulin, 10 M tetrahydrobiopterin, 1 mM dithiothreitol, and 0.2 mM NADPH and the desired amount of geldanamycin or radicicol in a total volume of 180 l of 20 mM Hepes, pH 7.4, at 37°C. The NADPH oxidation was measured by the loss in absorbance at 340 nm, and the initial rate of oxidation was determined from the linear portion of the plot. In some studies, an aliquot (75 l) of the mixtures used for NADPH oxidation measurements was taken, and the amount of hydrogen peroxide present was determined by the thiocyanate method (20). In some samples, catalase (10 units/ml) was added.
Heme Assay-The amount of heme was determined by HPLC as described previously (21). Samples were injected onto an HPLC column (C4 Vydac, 5 m, 0.21 ϫ 15 cm) equilibrated with solvent A (0.1% trifluoroacetic acid) at a flow rate of 0.3 ml/min. A linear gradient was run to 75% solvent B (0.1% trifluoroacetic acid in acetonitrile) over 30 min and then to 100% solvent B over the next 5 min. Absorbance at 220 and 400 nm was monitored. Myoglobin (horse heart) was used as a standard.
The heme present in nNOS was assessed by measuring the ferrous carbonyl complex. Since the cell cytosols from heme-treated cells contained compounds that absorbed in the 420-nm region, we partially purified the cytosols for nNOS. Cytosols (20 mg of protein) were loaded onto a 2Ј5Ј-ADP-Sepharose column (1.5 cm ϫ 0.9 cm) and purified as described above. The ferrous carbonyl complex was measured as a difference spectrum and quantified with an extinction coefficient of 91 mM Ϫ1 as described previously (22).

SDS-Polyacrylamide Gel Electrophoresis and Western
Blotting-The insect cytosols were added to an equal volume of sample buffer containing 5% SDS, 20% glycerol, 100 mM dithiothreitol, 100 M tetrahydrobiopterin, 100 M L-arginine, and 0.02% bromphenol blue in 125 mM Tris-HCl, pH 6.8. Samples were kept ice-cold prior to loading on gels. This method has been described previously by Klatt et al. (23) to prevent the dissociation of nNOS dimers prior to and during electrophoresis. Samples (50 g of protein) were subjected to electrophoresis on 6% SDS-polyacrylamide gels (10 ϫ 8 cm) and transferred to nitrocellulose membranes (0.2 m, Bio-Rad). The nNOS protein was visualized by Western blotting using 0.1% anti-nNOS polyclonal antibody from Transduction Laboratories. An anti-rabbit IgG conjugated to peroxidase (Roche Molecular Biochemicals) was used as secondary antibody at a concentration of 0.01%. For electrophoresis under non-denaturing conditions, 25 l of insect cytosol was mixed with 50 l of detergent-free buffer (312 mM Tris-HCl, pH 6.8, 50% glycerol, 0.05% bromphenol), and proteins were resolved on a 7.5% polyacrylamide gel followed by Western blotting.
Glucocorticoid Receptor Heterocomplex Reconstitution-Receptors were immunoadsorbed from 50-l aliquots of Sf9 cytosol by rotation for 2 h at 4°C with 14 l of protein A-Sepharose precoupled to 7 l of FiGR ascites suspended in 200 l of TEG buffer (10 mM TES, pH 7.6, 50 mM NaCl, 4 mM EDTA, 10% glycerol). Prior to incubation with rabbit reticulocyte lysate, immunoadsorbed receptors were stripped of associated hsp90 by incubating the immunopellet for an additional 2 h at 4°C with 300 l of 0.5 M NaCl in TEG. The pellets were then washed once with 1 ml of TEG followed by a second wash with 1 ml of Hepes buffer (10 mM Hepes, pH 7.35). FiGR immunopellets containing GR stripped of chaperones were incubated with 50 l of rabbit reticulocyte and 5 l of an ATP-regenerating system (50 mM ATP, 250 mM creatine phosphate, 20 mM magnesium acetate, and 100 units/ml creatine phosphokinase). The assay mixtures were incubated for 20 min at 30°C with suspension of the pellets by shaking the tubes every 2 min. At the end of the incubation, the pellets were washed twice with 1 ml of ice-cold TEGM buffer (TEG with 20 mM sodium molybdate) and assayed for steroid binding capacity.
Assay of Steroid Binding Capacity-Immune pellets to be assayed for steroid binding were incubated overnight at 4°C in 50 l of HEM buffer (10 mM Hepes, pH 7.4, 1 mM EDTA, 20 mM molybdate) plus 50 nM [ 3 H]triamcinolone acetonide. Samples were then washed three times with 1 ml of TEGM and counted by liquid scintillation spectrometry.

Geldanamycin and Radicicol Inhibit Heme-mediated Activation of Apo-nNOS in Sf9
Cells-Sf9 cells have very low levels of endogenous heme, and the addition of exogenous heme to the culture medium causes an increase in nNOS activity in infected cells (2). In the experiment shown in Fig. 1, heme was added to the medium of Sf9 cells expressing rat nNOS, and cells were harvested at various times for assay of nNOS activity by the conversion of oxyhemoglobin to methemoglobin. The addition of heme (closed squares) increased nNOS activity ϳ5-fold with respect to the activity in the absence of exogenous heme (open squares). This exogenous heme-dependent increase in nNOS activity is inhibited by ϳ60% by either 20 M geldanamycin (open circles) or 20 M radicicol (closed circles). Both compounds bind to the nucleotide-binding site of hsp90 and inhibit hsp90 function (24 -26). As shown in Fig. 2A, geldanamycin (open squares) and radicicol (closed squares) inhibited the heme-mediated activation of nNOS with the same concentration dependence, and nearly complete inhibition was achieved at concentrations above 30 M.
We have compared the inhibition of nNOS activation in Sf9 cells with the potencies of geldanamycin and radicicol in inhibiting the cell-free activation of the glucocorticoid receptor to a steroid binding state, a well established model of an hsp90-dependent process (3). As shown in Fig. 2B, activation of the glucocorticoid receptor to a steroid binding state by the hsp90/ hsp70-based chaperone system is inhibited most potently by radicicol (closed squares) with complete inhibition at 1 M. Geldanamycin (open squares) is slightly less potent, producing nearly complete inhibition at 5 M. The concentration dependence of this in vitro inhibition is similar to that reported for in vitro inhibition of hsp90 ATPase activity by geldanamycin and radicicol (26). The higher concentrations required for inhibition of nNOS activation in Sf9 cells are similar to those found in the recent report by Pritchard et al. (10), who used 18 M geldanamycin to cause eNOS-dependent superoxide production in bovine coronary endothelial cells.

Both Geldanamycin and Radicicol Inhibit Heme-initiated Dimerization of Apo-NOS in Sf9
Cells-The inactive, hemedeficient, monomeric apo-nNOS must form a heme-bound homodimer in order for the enzyme to produce nitric oxide (27). Under in vivo conditions, this assembly process gives rise to a very stable dimeric state of the enzyme that can be observed on low temperature SDS-PAGE gels as an SDS-resistant dimer (23). As shown in Fig. 3A, in Sf9 cells, the monomeric form of the enzyme is produced in the absence of exogenous heme (lane 1). After the addition of exogenous heme for 30 min, the nNOS exists, in large part, as an SDS-resistant dimer (lane 2). When the cells are treated with heme in the presence of 20 or 40 M geldanamycin, the SDS-resistant dimer is substantially decreased (lanes 3 and 4), and radicicol decreases the amount of SDS-resistant dimer to a similar extent at the same concentrations ( lanes 5 and 6).
The dimeric form of nNOS can also be measured under non-denaturing conditions in which detergent is omitted. Under these conditions, it is possible to see a dimeric form of nNOS that is not SDS-resistant (28). As shown in Fig. 3B Radicicol Hinders Functional Heme Insertion into Apo-nNOS-The initial step in the heme-mediated dimerization of apo-nNOS in vitro is the insertion of heme into the protein and ligation of the heme iron to a cysteine (28). This heme complex has a characteristic absorbance in the 450-nm region in its ferrous carbonyl state. Although this complex can be readily detected in Sf9 cytosols, a heme-derived chromophore at 420 nm interferes with the quantification of the P450. To minimize the contribution of this 420-nm absorbing compound, we have partially purified the nNOS by the use of a 2Ј5Ј-ADP-Sepharose affinity column and then measured the P450 spectrum (Fig. 4,  inset). As shown in Fig. 4, the amount of P450 was quantified and compared with the cytosolic nNOS activity. The addition of heme (Heme) to Sf9 cells expressing apo-nNOS increased nNOS activity and P450 content ϳ5-fold when compared with that found when heme was not added (Unt). When the cells were treated with heme in the presence of 20 or 40 M radicicol (Heme ϩ 20 M RAD or Heme ϩ 40 M RAD), the nNOS activity and P450 content remained at nearly the level of the untreated cells (Unt). The addition of 40 M radicicol after activation of apo-nNOS with heme did not cause a decrease in nNOS activity (data not shown). Thus, it appears that radicicol does not act to inhibit the active form of nNOS. As will be described in more detail below, geldanamycin caused the oxidative destruction of heme and therefore was not used in these studies.
Geldanamycin, but Not Radicicol, Causes Oxidant Production from nNOS in Sf9 Cells-To determine whether heme availability could be a mechanism for the reduction in dimeric nNOS observed after treatment with the hsp90 inhibitors, we measured the amount of heme in the Sf9 cells by reverse phase HPLC. Fig. 5A shows the HPLC profiles at 400 nm cytosolic samples prepared from Sf9 cells. Treatment of Sf9 cells overexpressing nNOS with exogenous heme and subsequent analysis of the cell cytosol gave a major peak at 22 min corresponding to heme (upper panel). This represents an ϳ7-fold increase in heme over that found in Sf9 cells that have not been treated with exogenous heme (data not shown). Treatment of cells with exogenous heme and 20 M geldanamycin causes ϳ80% decrease in the peak corresponding to heme (middle panel), whereas 20 M radicicol does not affect the amount of heme (lower panel). The peak areas for heme were quantified, as shown in Fig. 5B. The loss of heme is dependent on the concentration of geldanamycin (open squares), whereas radicicol had no substantial effect (closed squares). In that both compounds are hsp90 inhibitors, the heme loss appears unrelated to hsp90. The decrease in cytosolic heme may contribute to the inhibition of heme-mediated activation of Sf9 cell nNOS seen with geldanamycin treatment, but an effect on heme cannot explain any inhibition of nNOS activation by radicicol.
We further examined the heme levels in the medium to determine whether heme availability is a determinant of geldanamycin action. As shown in Fig. 6A on the 400-nm profile (solid lines), the heme in the medium of radicicol-treated cells (lower panel) was not different from that of untreated cells (upper panel). However, 20 M geldanamycin (middle panel) causes a 70% decrease in heme as well as the formation of a new peak on the 400-nm profile (solid line) that co-elutes with albumin, which is seen on the 220-nm profile (dashed line). This new heme product likely represents an altered heme product that is irreversibly bound to albumin. Large amounts of albumin are present in the medium because it was added to stabilize the heme as a heme⅐protein conjugate. The formation of altered heme products occurs when hemoproteins, including heme⅐albumin conjugates (29), react with reactive intermediates, such as hydrogen peroxide (21,30). As shown in Fig. 6B, this notion that hydrogen peroxide is involved in the formation of the altered heme product is supported by the finding that catalase prevents the heme loss (cf. GA with GA ϩ CAT) as well as the formation of the altered heme adduct (data not shown). Hydrogen peroxide treatment of the medium containing heme causes the loss of heme (H 2 O 2 ) and the formation of the altered heme adduct (data not shown). Thus, hydrogen peroxide is formed and causes the alteration of the heme in the medium. Moreover, hydrogen peroxide formation requires nNOS, as no heme loss was observed in non-infected cells treated with geldanamycin (cf. NI with NI ϩ GA). Thus, nNOS expression and geldanamycin are both essential for the oxidant production.
Geldanamycin, but Not Radicicol, Causes Enhanced NADPH Oxidation and Hydrogen Peroxide Formation from Purified nNOS-Geldanamycin is a benzoquinone ansamycin antibiotic (24 -26), and based on the known propensity of nNOS to redox cycle with quinones (15), we directly tested whether geldanamycin could redox cycle and produce hydrogen peroxide from purified nNOS. It is noteworthy that radicicol, although a macrocyclic antibiotic, is not structurally related to geldanamycin and does not contain a quinone moiety (25,26). As shown in Fig. 7, geldanamycin (GA) causes a concentration-dependent increase in the rate of NADPH oxidation (solid bars) that, in part, leads to the formation of detectable hydrogen peroxide (hatched bars). Radicicol at 40 M did not increase the rate of NADPH oxidation or production of hydrogen peroxide. As a control, no hydrogen peroxide could be detected in the presence of catalase (GA ϩ CAT). We next asked whether substrates or inhibitors that act on the heme active site of nNOS could modulate the redox cycling of geldanamycin. Fig. 8 shows that N G -nitro-L-arginine, a slowly reversible inhibitor that binds to the oxygenase site of nNOS, does not inhibit the NADPH oxidation or hydrogen peroxide formation due to geldanamycin (cf. GA ϩ NNA with GA). L-arginine, the natural substrate for the enzyme that binds to the oxygenase domain, also had little effect on redox cycling due to geldanamycin (GA ϩ ARG). In addition, the heme-deficient apo-nNOS was able to catalyze NADPH oxidation and hydrogen peroxide formation (GA ϩ APO). Thus, it appears that the flavin domain, and not the heme domain, mediates the electron transfer reactions that lead to redox cycling of geldanamycin. The flavin domain of nNOS has been reported to be responsible for redox cycling of a variety of other quinones (15). DISCUSSION In this study, we have shown that geldanamycin and radicicol, both inhibitors of hsp90, inhibit the heme-mediated activation of heme-deficient apo-nNOS in insect cells. From indirect observations, we have suggested previously that the hsp90 machinery may facilitate functional heme insertion into nNOS (2). As evident by measurements of the nNOS cytochrome P450 complex, we prove here that the lack of activation in the presence of radicicol is due to the inability to insert a functionally active heme prosthetic group into apo-nNOS. This in turn prevents assembly of monomeric apo-nNOS to the enzymatically active homodimeric form. Thus, we conclude that the hsp90-based chaperone machinery plays a role in facilitating functional heme insertion.
This notion of the insertion of a hydrophobic heme into nNOS is consistent with the action of the hsp90-based chaperone machinery on the GR, where the hydrophobic ligand-binding cleft is opened to allow access by steroid (7)(8)(9). In a similar manner, the chaperone machinery may favor the opening of the heme-binding cleft and facilitate heme entry. The presence of hydrophobic clefts is a universal feature of all properly folded proteins, and the ability of the hsp90-based chaperone machinery to recognize the general topological regions where clefts merge with the surface of a protein is consistent with the notion that the cleft opening and facilitated binding of hydrophobic compounds are primary functions of the chaperone machinery. It is likely that the hsp90-based chaperone machinery facilitates binding of a number of hydrophobic ligands and prosthetic groups through stabilization of a partially unfolded state of native acceptor proteins.
The role of hsp90 in heme insertion is also consistent with the known association of hsp90 with nNOS in vitro and in vivo as well as the enhanced degradation of nNOS seen after inhibition of hsp90 in cells (2). nNOS is known to be proteolyzed by a ubiquitin-proteasomal system, and the monomeric apo-nNOS is preferentially ubiquitinated (31). Inhibition of hsp90 and decreased heme insertion into apo-nNOS would favor the monomeric form and thus enhance the proteolysis and turnover of nNOS in cells.
In the course of our study, we discovered that geldanamycin, unlike radicicol, mediates the formation of oxidants from nNOS that cause the oxidative alteration of heme. The oxidant ap-  pears to be hydrogen peroxide since catalase prevents against heme alteration, and similar heme products are formed after treatment with exogenous hydrogen peroxide. Treatment of endothelial cells with geldanamycin causes eNOS-dependent oxidant production, leading Pritchard et al. (10) to conclude that inhibition of ATP-dependent conformational changes in hsp90 uncouples eNOS activity to increase O 2 . generation by the enzyme. In essence, it was concluded that the ADP-versus ATP-dependent conformation of eNOS-bound hsp90 determined O 2 . versus NO production. In the case of nNOS, we have found that geldanamycin-mediated oxidant production is a direct action on nNOS independent of hsp90. This is based on the finding that geldanamycin redox cycles and forms oxidants with purified nNOS and that radicicol, which is an equally potent hsp90 inhibitor, does not cause oxidant production in cells or with purified nNOS. The redox cycling of geldanamycin, an ansamycin benzoquinone, is consistent with the known ability of nNOS to redox cycle other quinones via the flavin redox center (15). Thus, the redox cycling of geldanamycin precludes the use of this inhibitor in the study of hsp90 effects in redoxactive systems.