The role of hsp90 in heme-dependent activation of apo-neuronal nitric-oxide synthase.

Like other nitric-oxide synthase (NOS) enzymes, neuronal NOS (nNOS) turnover and activity are regulated by the ubiquitous protein chaperone hsp90. We have shown previously that nNOS expressed in Sf9 cells where endogenous heme levels are low is activated from the apo- to the holo-enzyme by addition of exogenous heme to the culture medium, and this activation is inhibited by radicicol, a specific inhibitor of hsp90 (Billecke, S. S., Bender, A. T., Kanelakis, K. C., Murphy, P. J. M., Lowe, E. R., Kamada, Y., Pratt, W. B., and Osawa, Y. (2002) J. Biol. Chem. 278, 15465-15468). In this work, we examine heme binding by apo-nNOS to form the active enzyme in a cell-free system. We show that cytosol from Sf9 cells facilitates heme-dependent apo-nNOS activation by promoting functional heme insertion into the enzyme. Sf9 cytosol also converts the glucocorticoid receptor (GR) to a state where the hydrophobic ligand binding cleft is open to access by steroid. Both cell-free heme activation of purified nNOS and activation of steroid binding activity of the immunopurified GR are inhibited by radicicol treatment of Sf9 cells prior to cytosol preparation, and addition of purified hsp90 to cytosol partially overcomes this inhibition. Although there is an hsp90-dependent machinery in Sf9 cytosol that facilitates heme binding by apo-nNOS, it is clearly different from the machinery that facilitates steroid binding by the GR. hsp90 regulation of apo-nNOS heme activation is very dynamic and requires higher concentrations of radicicol for its inhibition, whereas GR steroid binding is determined by assembly of stable GR.hsp90 heterocomplexes that are formed by a purified five-chaperone machinery that does not activate apo-nNOS.

hsp90 1 has been shown to regulate over 100 signal transduction pathways by controlling the function, trafficking, and turnover of a variety of signaling proteins (reviewed in Ref. 1). The regulation is achieved through the ATP-dependent assembly of signaling protein⅐hsp90 heterocomplexes by the multiprotein hsp90/hsp70-based chaperone machinery (1). The asso-ciation of hsp90 with its "client proteins" is not determined by specific amino acid motifs or secondary modifications (e.g. phosphorylation, acetylation, etc.) in the client protein, and it has not been clear how hsp90 could associate with and regulate such a wide variety of proteins regardless of their structure or sequence. How the chaperone machinery recognizes and interacts with its signaling protein clients is a fundamental mechanistic problem.
In cell-free assays, the two essential components of the machinery, hsp90 and hsp70, have been shown to function individually as chaperones that bind to exposed hydrophobic amino acids in denatured proteins and through rounds of binding and release to facilitate protein refolding (2). However, studies with steroid receptors suggest that this is not the way hsp90 and hsp70 function when they act together as part of an integrated chaperone machinery. hsp90 binds to the ligand binding domain (LBD) of steroid receptors, and the ligand binding activity of some steroid receptors and the aryl hydrocarbon receptor is absolutely hsp90-dependent (1). Thus, when the glucocorticoid receptor (GR) is stripped of its associated hsp90, it immediately loses its ability to bind steroid, and steroid binding activity is regenerated when GR⅐hsp90 heterocomplexes are reformed by the hsp90/hsp70-based multichaperone machinery (3,4). Steroid ligands bind deep in a hydrophobic cleft that appears to be collapsed in the absence of ligand, such that the receptor must change its conformation to allow entry of the ligand (5). The chaperone machinery carries out the ATP-dependent opening of the steroid binding cleft in the GR LBD such that it can be accessed by steroid, and it promotes conformational changes that increase the sensitivity of the GR LBD to attack by thiolderivatizing agents and trypsin (6 -8).
In contrast to the chaperoning action of hsp90 and hsp70 on denatured proteins, there is no evidence that the chaperone machinery acts on a GR that is in any way denatured or in a so-called "nearly native" state. Rather, the machinery acts on a GR where the hydrophobic steroid binding cleft appears to be collapsed such that the LBD is in a native, minimal energy conformation. The initial step in GR⅐hsp90 heterocomplex assembly is the ATP-and hsp40-dependent priming of the GR by hsp70 to form a GR⅐hsp70 complex that can then bind Hop and hsp90 and undergo a second ATP-dependent step in which the steroid binding cleft is opened (3,9). It is the ATP-dependent conformation of hsp70 that initially binds to the GR LBD, rather than the ADP-dependent conformation that favors interaction of hsp70 with hydrophobic peptides (10). A short segment of the GR that lies at the extreme amino terminus of the LBD is required for LBD⅐hsp90 heterocomplex assembly and steroid binding activity (11). This segment lies at the rim of the ligand binding cleft of the receptor (12), and mutational analysis reveals that hsp90 binding requires the presence of the segment but no defined amino acid composition (13).
The work on the mechanism of GR⅐hsp90 heterocomplex assembly led to the notion that the chaperone machinery interacts with the region where the hydrophobic cleft merges with the surface of the receptor. Such regions are a general topologic feature of virtually all proteins in native conformation, and this cleft recognition hypothesis could account for the ability of the chaperone machinery to interact with a variety of proteins in native conformation regardless of sequence or structure (1). The ability of the chaperone machinery to recognize hydrophobic clefts and to open and stabilize them in the open state, even transiently, could facilitate the entry of a number of hydrophobic ligands and prosthetic groups to catalytic centers in the interior of many enzymes. As a start to testing this general proposal, we have examined the role of hsp90 in facilitating the binding of heme by apo-neuronal nitric-oxide synthase (apo-nNOS).
The NOS enzymes are important signaling proteins that function as cytochrome P450-type hemoproteins to catalyze the formation of nitric oxide (NO) and citrulline from L-arginine, O 2 , and NADPH (14). The NOS enzymes are active as homodimers, with each monomer binding tightly 1 eq each of FAD, FMN, tetrahydrobiopterin, and heme (14 -19). The prosthetic heme is the site of oxygen activation, which is required for the metabolism of L-arginine. The heme-deficient monomer of NOS can be partially reconstituted in vitro in the presence of heme, tetrahydrobiopterin, and arginine to form the functional homodimer (20 -22). We have shown previously that a portion of neuronal NOS (nNOS) is bound to hsp90 in vivo and that treatment of mammalian cells with an hsp90 inhibitor to block nNOS⅐hsp90 assembly leads to nNOS degradation via the ubiquitin-proteasome pathway (23). We have also shown that treatment of nNOS-expressing Sf9 cells with the hsp90 inhibitors geldanamycin and radicicol inhibits activation of apo-nNOS activity by exogenous heme (23,24). Sf9 cells have very low levels of endogenous heme, and addition of exogenous heme to the culture medium increases nNOS activity in transfected cells. 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 heme-mediated assembly of the holo-enzyme (24). The ferrous⅐carbonyl complex of nNOS is not formed when hsp90 is inhibited, indicating that functional heme insertion is prevented (24).
To date, we have not been able to demonstrate a role for hsp90 in heme binding and nNOS activation in a cell-free system. Here, we show that cytosol prepared from Sf9 cells facilitates cell-free heme insertion into apo-nNOS and conversion to the active holo-nNOS enzyme. Cytosol from radicicoltreated Sf9 cells is not active in permitting heme activation of apo-nNOS or in converting the GR to the steroid binding state, and in both cases the inhibition is partially reversed by the addition of purified hsp90. However, the hsp90 machinery that dynamically facilitates heme access to its binding cleft in apo-nNOS is different from the hsp90 machinery that assembles stable hsp90 complexes with the GR to facilitate steroid binding access to its hydrophobic ligand binding cleft.

Materials
Untreated rabbit reticulocyte lysate was purchased from Green Hectares (Oregon, WI). [6, H]Dexamethasone (40 Ci/mmol) and 125 I-conjugated goat anti-mouse and anti-rabbit IgGs were obtained from PerkinElmer Life Sciences. Dulbecco's modified Eagle's medium was from Bio-Whittaker (Walkersville, MD). The BuGR2 monoclonal IgG used to immunoblot the GR was from Affinity Bioreagents (Golden, CO), and the FiGR monoclonal IgG used to immunoadsorb the GR was generously provided by Dr. Jack Bodwell (Dartmouth Medical School, Lebanon, NH). Rabbit antiserum against the C terminus of hsp90 used to immunoblot insect hsp90 and hsp70 was kindly provided by Dr.

Methods
Expression and Purification of Apo-nNOS and Holo-nNOS-Recombinant baculovirus containing nNOS cDNA was produced as described previously (23). Sf9 cells were grown in SFM 900 II serum-free medium (Invitrogen) supplemented with 1% Cytomax (Kemp Biotechnology, Rockville, MD) in suspension cultures maintained at 27°C with continuous shaking (150 rpm). To express holo-nNOS, oxyhemoglobin (25 M) was added as a source of heme during the last 24 h of expression. 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, cells were harvested, washed once with ice-cold phosphate-buffered saline (pH 7.4), and ruptured by Dounce homogenization in 1 pellet volume of a homogenization buffer containing 10 mM Hepes, pH 7.4, 10 g/ml trypsin inhibitor, 10 g/ml leupeptin, 2 g/ml aprotinin, 6 mM phenylmethanesulfonyl fluoride, and 10 M tetrahydrobiopterin. The lysates were spun at 100,000 ϫ g for 60 min at 4°C. The supernatants were collected, aliquoted, flash-frozen, and stored at Ϫ70°C. nNOS was purified from 70 ml of cytosol prepared from infected Sf9 cells by adsorption to a column of 2Ј,5Ј-ADP-Sepharose and elution with 10 mM 2Ј-AMP, followed by gel filtration chromatography on Sephacryl S-300 as described previously (23). Fractions containing nNOS were pooled and concentrated to 2 ml. The concentrated enzyme was divided into aliquots and stored at Ϫ70°C.
Activation of Apo-nNOS with Heme in Sf9 Cells-The activation of apo-nNOS in Sf9 cells by the addition of heme-albumin to the medium was performed as described previously (24). In the experiment of  (25,26), for 1 h at 27°C. Cells were ruptured, and nNOS activity was measured as described below. The remainder of the cell suspension (45 ml) was washed twice with 25 ml of SFM 900 II medium, the cells were resuspended in 45 ml of medium, and the incubation was continued without heme. At various times, 5-ml aliquots were treated with heme for 1 h, ruptured, and assayed for nNOS activity.
Cell-free Activation of Apo-nNOS with Heme-Aliquots (25 l) of cytosol from nNOS-expressing Sf9 cells that were grown in heme-free medium were preincubated at 27°C with 3 l of an ATP-regenerating system (50 mM ATP, 100 units/ml creatine phosphokinase, 20 mM magnesium acetate, and 250 mM creatine phosphate in 10 mM Hepes, pH 7.4) in a total volume of 30 l. After 5 min, 1.5 l of heme⅐bovine serum albumin complex (25,26) was added to yield a final heme concentration of 25 M. Aliquots of 3.5 l were removed at various times for assay of nNOS activity. In some experiments, purified apo-nNOS (10 g) was substituted for cytosol from nNOS-expressing Sf9 cells in the reconstitution mixture. Where indicated, 20 l of cytosol prepared from uninfected Sf9 cells pretreated with radicicol or vehicle was added to the reconstitution mixture containing purified apo-nNOS.
DE52 Chromatography of Sf9 Cytosol-Non-infected Sf9 cytosol (12 ml) was adsorbed to a 1.5 ϫ 20-cm column of DE52 equilibrated with HE buffer (10 mM Hepes, 0.1 mM EDTA, pH 7.4); the column was washed with 100 ml of HE buffer, producing a "drop-through" fraction, followed by elution of bound proteins with 100 ml of HEK buffer (HE buffer containing 500 mM KCl) to produce a "retained" fraction. Each fraction was concentrated to 2 ml, and then a portion (6 l) was used in place of cytosol for heme-mediated reconstitution of purified apo-nNOS.
For the preparation of DE52-retained fractions A, B, and C, 30 ml of Sf9 cytosol was adsorbed to a 2.5 ϫ 18-cm column of DE52 equilibrated with HE buffer; the column was washed with 250 ml of HE buffer, and proteins were eluted with a 400-ml gradient of 0 -0.5 M KCl. hsp90 and hsp70 were assayed by Western blotting an aliquot of every other hsp90 and Neuronal Nitric-oxide Synthase fraction using rabbit antiserum directed against the carboxyl terminus of hsp90 that also detects hsp70. Fractions were combined in three pools designated A-C as described by Dittmar et al. (27). Fractions A, B, and C were concentrated to 1.5 ml, and then a portion (2 l) was used in place of cytosol for heme-mediated reconstitution of purified apo-nNOS.
Assay of nNOS Activity-NO formation was assayed by the NOmediated conversion of oxyhemoglobin to methemoglobin. nNOS-containing samples were added 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 NADPHregenerating system consisting of 400 M NADP ϩ , 10 mM glucose 6-phosphate, and 1 unit/ml glucose-6-phosphate 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 with a microtiter plate reader as described previously (28).
P450 Assay-The heme present in nNOS was assessed by measuring the ferrous⅐CO complex. Because the cell cytosols contained compounds that absorbed in the 420-nm region, we partially purified nNOS from the cytosols. Cytosols (8-mg total) were loaded onto a 2Ј,5Ј-ADP-Sepharose column (2.0 cm ϫ 0.8 cm) and eluted with 2Ј-AMP as described above, except that the gel filtration step was omitted. The ferrous carbonyl complex was measured as a difference spectrum and quantified with an extinction coefficient of 91 mM Ϫ1 , as previously described (29).
Expression of Glucocorticoid Receptor in Sf9 Cells-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 post-infection as described by Srinivasan et al. (30). In the experiment of Fig. 4, at 48 h post-infection, cells were treated with Me 2 SO or 20 M radicicol (or geldanamycin) for 20 min followed by washing and resuspension as described for nNOS-infected cells. At various times after washing, cells were harvested, resuspended in 1.5 volume of HEM buffer (10 mM Hepes, pH 7.4, 1 mM EDTA, 20 mM molybdate) with 1 mM phenylmethylsulfonyl fluoride and 1 tablet of Complete-Mini protease inhibitor mixture per 3 ml of buffer (Roche Applied Science), and ruptured by Dounce homogenization. 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.
Immunoadsorption of GR-Receptors were immunoadsorbed from 50-l aliquots of Sf9 cytosol by rotation for 2 h at 4°C with 18 l of protein A-Sepharose precoupled to 9 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 Sf9 lysate, or the DE52-retained fractions of Sf9 lysate, or the purified proteins of the chaperone machinery, immunoadsorbed receptors were stripped of associated hsp90 by incubating the immunopellet for an additional 2 h at 4°C with 350 l of 0.5 M NaCl in TEG buffer. The pellets were then washed once with 1 ml of TEG buffer followed by a second wash with 1 ml of Hepes buffer (10 mM Hepes, pH 7.4).
GR⅐hsp90 Heterocomplex Reconstitution-FiGR immunopellets containing GR stripped of chaperones were incubated with unfractionated rabbit reticulocyte lysate or with the five-protein assembly system (20 g of purified hsp90, 15 g of purified hsp70, 0.6 g of purified human Hop, 6 g of purified p23, 0.125 g of purified YDJ-1 (see Ref. 31 for details)) adjusted to 55 l with HKD buffer (10 mM Hepes, pH 7.4, 100 mM KCl, 5 mM dithiothreitol) containing 20 mM sodium molybdate and 5 l of an ATP-regenerating system. 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 and for receptorassociated proteins.
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 plus 100 nM [ 3 H]dexamethasone. Samples were then washed three times with 1 ml of TEGM buffer and counted by liquid scintillation spectrometry. The steroid binding is expressed as counts/min of [ 3 H]dexamethasone bound/FiGR immunopellet prepared from 50 l of cell cytosol.
For supernatants to be assayed for steroid binding, a 50-l aliquot of supernatant was incubated overnight at 4°C in 50 l HEM buffer with 100 nM [ 3 H]dexamethasone in the absence or presence of a 1000-fold excess of unlabeled dexamethasone. Samples were mixed with dextrancoated charcoal and, after centrifugation, counted by liquid scintillation spectrometry. The steroid binding is expressed as counts/min of [ 3 H]dexamethasone bound/50 l of cell cytosol.
Gel Electrophoresis and Western Blotting-Samples (immune pellets, aliquots of cytosol, and aliquots of DE52 fractions) were resolved on 12% SDS-polyacrylamide gels and transferred to Immobilon-P membranes. The membranes were probed with 0.25 g/ml BuGR for GR or 0.01% rabbit antiserum to the carboxyl terminus of hsp90 for insect hsp90 (this antibody also weakly detects insect hsp70), or 1 g/ml AC88 for rabbit hsp90. The immunoblots were then incubated a second time with the appropriate 125 I-conjugated counterantibody to visualize immunoreactive bands. Immunoblots for nNOS were run on 6% SDS-polyacrylamide gels, transferred to nitrocellulose membranes, and probed with 0.1% anti-nNOS polyclonal antibody as described (24).
In studies where the SDS-resistant dimer was measured, the reaction mixture was quenched with sample buffer supplemented with 100 M BH 4 and 100 M L-arginine. The samples were kept on ice prior to loading for analysis by SDS-PAGE as described above. This method has previously been described by Klatt et al. (32) to prevent the dissociation of nNOS dimers prior to and during electrophoresis.

Radicicol Treatment of Sf9 Cells Inhibits Subsequent
Cellfree nNOS Activation-The major cell-free system used to study hsp90 action on signaling proteins is rabbit reticulocyte lysate (1), but reticulocyte lysate contains very high levels of heme in addition to the necessary machinery for hsp90 heterocomplex assembly. Thus, although we have been able to activate apo-nNOS with reticulocyte lysate, we have not been able to show that the activation is both heme-dependent and hsp90dependent (23). Because Sf9 cells grown in heme-free medium have low levels of endogenous heme, we asked whether concentrated lysate from cells grown in the absence of exogenous heme would support heme-dependent activation of apo-nNOS in vitro.
In the experiment of Fig. 1, Sf9 cells infected with nNOSexpressing baculovirus were treated for 20 min with Me 2 SO vehicle or with the hsp90 inhibitor radicicol. hsp90 is a member of a very limited family of proteins, the GHKL family, which possess a unique binding pocket for ATP (33). The hsp90 inhibitors geldanamycin and radicicol bind to this nucleotide site (34,35) and prevent hsp90 from achieving its ATP-dependent conformation, thus blocking hsp90 action (36). Because the more commonly used inhibitor geldanamycin contains a benzoquinone moiety that causes oxidant production from nNOS, we treated cells with radicicol, which does not have this effect (24). As shown in Fig. 1, cytosol from vehicle-treated cells activated apo-nNOS in the presence of added heme (closed circles) but not in the absence of heme (closed squares). In contrast, cytosol prepared from radicicol-treated cells did not promote heme activation of nNOS (open circles).
The initial step in heme-mediated dimerization of apo-nNOS in vitro is the insertion of heme into the protein and ligation of hsp90 and Neuronal Nitric-oxide Synthase the heme iron to a cysteine (24). In its ferrous carbonyl state, this heme complex has a characteristic absorbance in the 450-nm region. In the experiments of Fig. 2A, nNOS was partially purified by ADP-Sepharose chromatography to eliminate a heme-derived chromophore at 420 nm (23), and adsorption at 450 nm was assayed to determine formation of the ferrous carbonyl complex. As we reported previously, radicicol inhibited functional heme insertion in intact cells incubated with heme (cf. conditions 1 and 2). Also, as shown here, cytosol prepared from radicicol-treated Sf9 cells was impaired in its ability to promote functional heme insertion into apo-nNOS in vitro (cf. conditions 3 and 4). Consistent with heme insertion and activation, the SDS-resistant dimer was formed (Fig. 2B). Radicicol inhibits formation of the SDS-resistant dimer in intact cells (Fig. 2B, cf. lanes 1 and 2) as well as in cytosol prepared from radicicol-treated cells (cf. lanes 3 and 4).
Reversal of Radicicol Inhibition-In these experiments, Sf9 cells were treated with radicicol for a very short time (20 min) prior to incubation with heme in the cell-free nNOS activating assay. As shown in Fig. 3A, Sf9 cells must be treated with high concentrations of radicicol for inhibition of heme activation of apo-nNOS activity in the cell-free activating assay. Lower concentrations of radicicol are sufficient to inhibit the steroid binding activity of GR in cytosols prepared from Sf9 cells (Fig. 3B). This is one of several clear differences we find between the requirements for apo-nNOS heme binding activity and GR steroid binding activity. Because it is well established that the steroid binding activity of the GR is absolutely hsp90-dependent (1), the two systems will be compared in subsequent experiments.
The inhibition by radicicol is slowly reversed in vivo. In Fig.  4, Sf9 cells expressing nNOS were treated for 20 min with 20 M radicicol, the cells were then washed, and the incubation was continued in radicicol-free medium. At various times, heme was added to an aliquot of cells, and nNOS activity was determined 1 h later. Thus, the 0 time value, for example, is for cells that have been incubated with radicicol or vehicle for 20 min, washed, and incubated for 1 h with heme to maximally activate the nNOS. As shown in Fig. 4A, 24 h of incubation in heme-free medium were required for heme-activated nNOS activity to return to control levels. The immunoblot above the graph shows that the levels of nNOS are comparable in vehicletreated and radicicol-treated cells. The same amount of time was required for steroid binding activity to return to control levels in GR-expressing Sf9 cells treated with radicicol ( Fig.  4B). Fig. 4C shows that the slow return of steroid binding activity does not reflect GR degradation and a slow return of receptor to control levels. Rather, the GR is present throughout the recovery time, and when it is incubated with reticulocyte lysate, it assembles into GR⅐hsp90 heterocomplexes and has steroid binding activity. In the case of nNOS, only trace amounts of rabbit hsp90 associate with nNOS after treatment of immunopurified nNOS with reticulocyte lysate (23). This reflects the very dynamic nature of the hsp90 interaction with nNOS versus the GR.
Sf9 Cytosol Promotes Heme-dependent Activation of Purified nNOS-To this point, we have examined the activation of apo-nNOS in cytosol from nNOS-expressing cells. Thus, the short term radicicol treatment of Sf9 cells could have affected either the apo-nNOS activating activity of cytosol or the ability of the apo-nNOS to be subsequently activated. To address this issue, cytosol was prepared from non-infected Sf9 cells that were treated with vehicle or radicicol, and each cytosol was incubated with purified apo-nNOS and heme to determine its apo-nNOS activating activity. As shown in Fig. 5A, vehicle-treated cytosol alone did not alter the activity of apo-nNOS (solid squares) or heme-bound holo-nNOS (solid diamonds); however, vehicle-treated cytosol activated apo-nNOS in the presence of heme (solid circles), whereas radicicol-treated cytosol did not (open circles). In this work, we have added heme as a hemealbumin mixture to yield 25 M heme. This concentration of hsp90 and Neuronal Nitric-oxide Synthase heme is low enough such that it does not significantly activate apo-nNOS by itself over the incubation time of 30 min (Fig. 5A,  open squares). Thus, we have a cell-free system where we have heme-dependent apo-nNOS activation by an activating activity in Sf9 cytosol that is inhibited by short-term treatment of cells with radicicol. This activating activity in Sf9 cytosol also converts immunopurified GR to the steroid binding state, and that activity is inhibited by radicicol treatment of the intact cells (Fig. 5B). As shown in Fig. 5C, a portion of purified apo-nNOS (lane 1) is converted from monomer to SDS-resistant dimer by Sf9 cytosol (lane 3), and radicicol inhibits the formation of this dimer (lane 4).
The Cytosolic Apo-nNOS Activating Activity Is Retained by DE52-The chaperone machinery in reticulocyte lysate that activates GR steroid binding activity is retained by DE52 (27).
In the experiments of Fig. 6, we asked whether the heme-dependent, apo-nNOS activating activity of Sf9 lysate is retained by DE52. It can be seen in Fig. 6A that the apo-nNOS activating activity is retained by DE52 (lane 4) and there is no activity in the cytosolic fraction (drop-through) that does not bind to DE52 (lane 6). In Fig. 6B, the Sf9 cytosol proteins were eluted from DE52 with a salt gradient and pooled into three fractions (A-C), with the boundaries of fraction pool B being set so that it would contain the majority of the insect hsp90 (Fig. 6B,  inset). The amounts of A, B, and C in the assay were adjusted to a level where combined together as A-C they would not activate apo-nNOS (lane 4) but would activate when combined with cytosol from radicicol-treated cells (lane 5). Pool A, which contains some hsp90 and the major portion of the hsp70, and pool B, which contains most of the hsp90 and some hsp70, both had some ability to activate apo-nNOS when added to cytosol from radicicol-treated Sf9 cells (cf. lane 3 with lanes 6 and 7).
Partial Reconstitution of Apo-nNOS Activating Activity with Purified hsp90 -Because an hsp90-containing subfraction of the DE52-retained material in Sf9 cytosol complemented cytosol from radicicol-treated Sf9 cells in supporting heme-dependent apo-nNOS activation (Fig. 6B), we asked whether addition of purified hsp90 would increase apo-nNOS activation. In Fig. 7A, it can be seen that purified hsp90 had no effect by itself in promoting heme activation of apo-nNOS (open squares), but it increased the activating activity of cytosol from radicicol-treated Sf9 cells (solid squares) to a level approaching the activity of cytosol from vehicle-treated cells (solid circles). Purified hsp90 produced a similar activation of GR steroid , the cells were washed twice with medium without radicicol, and incubation was continued in radicicolfree medium. At the indicated times, cell aliquots were incubated for 1 h with heme, and cells were ruptured and assayed for nNOS activity and nNOS protein by immunoblotting aliquots of vehicle-treated (V) and radicicol-treated (R) cytosols. The nNOS activity shows data from one of three similar reversal experiments and the immunoblots were performed twice. B, reversal of inhibition of steroid binding activity. Sf9 cells expressing the GR were treated for 20 min with Me 2 SO (q) or 20 M radicicol (E), washed, and at various times, cell aliquots were removed for assay of cytosolic [ 3 H]dexamethasone binding activity. C, reactivation of steroid binding activity and assembly of hsp90 heterocomplexes with receptors from radicicol-treated cytosols. Receptors were immunoadsorbed from 50-l aliquots of cytosol prepared from GR-expressing Sf9 cells treated with 20 M radicicol, and the immunopellets were stripped of associated chaperones. The stripped receptors were incubated for 20 min at 30°C with an ATP-regenerating system and buffer (Str) or reticulocyte lysate (RL), and GR and hsp90 in the washed immune pellets were resolved by electrophoresis and immunoblotting. Replicate immune pellets were incubated with [ 3 H]dexamethasone, and steroid binding activity was determined. hsp90 and Neuronal Nitric-oxide Synthase binding activity by cytosol from radicicol-treated Sf9 cells (Fig. 7B).
Except for the marked difference in the concentration dependence for radicicol inhibition of heme activation of apo-nNOS and steroid binding by the GR shown in Fig. 3, we see similarities between the two systems. Both purified apo-nNOS and purified GR require an activity that is present in Sf9 cytosol that promotes heme binding or steroid binding, respectively (Fig. 5). This activity is inhibited by radicicol pretreatment of Sf9 cells, and purified hsp90 partially overcomes this inhibition (Fig. 7). Yet the two signaling proteins are very different in the conditions that will support activation. As shown in Fig. 8, apo-nNOS is activated by Sf9 cytosol (Fig. 8A, condition 2) but not the purified five-protein chaperone machinery (Fig. 8A, condition 3), whereas the GR is activated by both (Fig. 8B).

DISCUSSION
Of the three forms of NOS (inducible NOS, neuronal NOS, and endothelial NOS (eNOS)), most of the studies of hsp90 effects on NOS function in cells have been concerned with eNOS. In most of those studies, cells have been exposed to the hsp90 inhibitors geldanamycin and radicicol at concentrations ranging from 10 to 20 M (37-40). In short term treatment protocols that avoid any effect of radicicol on NOS turnover, we find that similar high concentrations are required for inhibiting heme activation of nNOS ( Fig. 3A and Ref. 24). An order of magnitude lower concentration of radicicol is sufficient for inhibiting the steroid binding activity of the GR under the same conditions (Fig. 3B). The GR is the paradigm example of a signaling protein that requires an hsp90 chaperone machinery for binding its hydrophobic steroid ligands (1). Because the concentration dependence for radicicol inhibition of heme binding to apo-nNOS is so much higher than that for steroid binding to GR, it is reasonable, at FIG. 6. DE52 fractionation of Sf9 activity promoting heme activation of purified apo-nNOS. A, the activity is retained by DE52. Sf9 cytosol was submitted to chromatography on a column of DE52, and the fractions that did not adsorb ("drop-through") and those that were retained were pooled and concentrated. Purified apo-nNOS was then incubated for 20 min at 27°C with an ATP-regenerating system, 10 M  7. Effect of purified rabbit hsp90 on activation of nNOS or GR by Sf9 cytosol from radicicol-treated cells. A, activation of apo-nNOS. Purified apo-nNOS was incubated with 25 M heme and cytosol from vehicle-treated Sf9 cells (q), or radicicol-treated cells (E), or cytosol from radicicol-treated cells plus 15 g of purified rabbit hsp90 (f), or with hsp90 alone (Ⅺ), and nNOS activity was assayed. One of three similar experiments is shown. B, activation of the GR. Stripped GR immunopellets were incubated under the same conditions without heme, except that 6 g of p23 was added with the hsp90, and [ 3 H]dexamethasone binding activity was assayed. p23 is required for the assembly of stable GR⅐hsp90 heterocomplexes in immune pellets that can be washed and assayed for steroid binding activity.
FIG. 8. The GR, but not apo-nNOS, is activated by the purified five-protein hsp90/hsp70-based chaperone machinery. A, activation of apo-nNOS. Purified apo-nNOS was incubated for 20 min with 25 M heme, and an ATP-regenerating system in the presence of HKD buffer alone (lane 1), Sf9 cytosol (lane 2), or the purified five-protein system of hsp90, hsp70, Hop, YDJ-1, and p23 (lane 3). At the end of the incubation nNOS activity was assayed. B, activation of the GR. Stripped GR immunopellets were incubated with an ATP-regenerating system and conditions as above. At the end of the incubation, immunopellets were washed and [ 3 H]dexamethasone binding was assayed.
hsp90 and Neuronal Nitric-oxide Synthase first glance, to conclude that the mechanisms of the two inhibitions are different, and only inhibition of GR reflects inhibition of hsp90. However, the difference in concentration dependence may not reflect different targets of radicicol action but differences in the dynamics of client protein⅐hsp90 assembly and the hsp90-based machinery involved.
The unliganded steroid receptors and a number of signaling protein kinases (e.g. Src and Raf) are recovered in cells essentially entirely in heterocomplexes with hsp90, and these heterocomplexes are stable to a number of biochemical manipulations (1). Both eNOS (41) and nNOS (23) have been recovered from cell cytosols as NOS⅐hsp90 heterocomplexes, but in both cases very little of the cytosolic NOS is bound by hsp90. Thus, the NOS enzymes appear to associate with hsp90 in a very dynamic manner that is more or less a "hit-and-run" mode of hsp90 regulation. In contrast, the steroid receptors have evolved less dynamic interactions with a chaperone machinery such that hsp90 is not just involved in cleft opening and inhibition of their turnover but also regulates their function. In these cases, the association with hsp90 has been brought under control and maintains the protein in an inactive state until the appropriate signal is received.
The hsp90-based chaperone machinery that forms dynamic heterocomplexes with nNOS differs from the machinery that forms the far more stable complexes with the GR. In Fig. 5A, we see that Sf9 cytosol contains an activity that facilitates heme activation of purified apo-nNOS that is not present in cytosol from radicicol-treated cells. Fig. 5B shows that the same is true for activating the purified GR to a state that binds steroid. Fig.  7 shows that hsp90 by itself is not sufficient to promote hemedependent activation of apo-nNOS or activation of GR steroid binding activity. Sf9 cytosol contains the ability to promote both processes, and addition of purified rabbit hsp90 to the cytosol from radicicol-treated cells partially restores that activity (Fig. 7). However, it seems the machinery involved in the assembly of stable GR⅐hsp90 heterocomplexes must be different from the machinery facilitating heme activation of apo-nNOS. Indeed, we show here that the purified five-protein system that stably activates GR steroid binding activity has no ability to promote heme activation of apo-nNOS (Fig. 8).
Thus, we propose that a co-chaperone, or co-chaperones, are involved in the dynamic regulation of the heme binding activity of apo-nNOS, making the hsp90 machinery different from that regulating GR steroid binding activity. It is possible, for example, that the hsp90 co-chaperone Aha1 (activator of hsp90 ATPase) (42) is uniquely part of a dynamic machinery. How the machineries differ, we do not know, but the difference between dynamic regulation of apo-nNOS and the more stable regulation of the GR could explain the difference in concentration dependence for radicicol inhibition of the two activities seen in Fig. 3. In the case of the GR, when 90% of the hsp90 is inhibited by radicicol, 90% of the receptors are not functionally complexed by hsp90 and only 10% are able to bind steroid. This may be the case at a level of 5 M radicicol in Fig. 3B. However, the 10% of hsp90 that is not inhibited at any instant may be sufficient for complete activity of the dynamic machinery facilitating heme activation of apo-nNOS. By this analysis, the interaction of radicicol with the single target hsp90 could produce the different concentration dependences for inhibition of heme activation of apo-nNOS and GR steroid binding activity shown in Fig. 3.