Neuronal Nitric-oxide Synthase Is Regulated by the hsp90-based Chaperone System in Vivo *

It is established that the multiprotein heat shock protein 90 (hsp90)-based chaperone system acts on the ligand binding domain of the glucocorticoid receptor (GR) to form a GR·hsp90 heterocomplex and to convert the receptor ligand binding domain to the steroid-binding state. Treatment of cells with the hsp90 inhibitor geldanamycin inactivates steroid binding activity and increases the rate of GR turnover. We show here that a portion of neuronal nitric-oxide synthase (nNOS) exists as a molybdate-stabilized nNOS·hsp90 heterocomplex in the cytosolic fraction of human embryonic kidney 293 cells stably transfected with rat nNOS. Treatment of human embryonic kidney 293 cells with geldanamycin both decreases nNOS catalytic activity and increases the rate of nNOS turnover. Similarly, geldanamycin treatment of nNOS-expressing Sf9 cells partially inhibits nNOS activation by exogenous heme. Like the GR, purified heme-free apo-nNOS is activated by the DE52-retained fraction of rabbit reticulocyte lysate, which also assembles nNOS·hsp90 heterocomplexes. However, in contrast to the GR, heterocomplex assembly with hsp90 is not required for increased heme binding and nNOS activation in this cell-free system. We propose that, in vivo, where access by free heme is limited, the complete hsp90-based chaperone machinery is required for sustained opening of the heme binding cleft and nNOS activation, but in the heme-containing cell-free nNOS-activating system transient opening of the heme binding cleft without hsp90 is sufficient to facilitate heme binding.

1 and 2). These heterocomplexes with hsp90 are formed by a multicomponent chaperone machinery consisting of hsp90, hsp70, Hop, hsp40, p23, and probably also the hsp70-interacting protein Hip and the GrpE-like protein BAG-1 (for review, see Ref. 3 and references therein). As first shown for the glucocorticoid receptor (GR) (4) and then for some other steroid receptors and the dioxin (Ah) receptor, association of the ligand binding domain (LBD) with hsp90 is required for the high affinity ligand binding conformation (1,2). Complexing of the GR with hsp90 also opens up both thiol moieties (5) and trypsin cleavage sites (6,7) in the LBD to attack by a thiol-derivatizing agent and the protease. These direct data, coupled with recent genetic observations (8), support the notion (9, 10) that the hsp90-based chaperone machinery directs an ATP-dependent partial unfolding of the receptor LBD, thus making the hydrophobic steroid-binding pocket accessible to steroid.
The problem of providing access of ligands to hydrophobic binding sites situated in the interior of properly folded proteins is not unique to steroid and dioxin receptors. To test whether the hsp90-based chaperone machinery may play a more general role in opening up hydrophobic binding clefts, we have asked whether this system facilitates the entry of heme into its binding site in the interior of neuronal nitric-oxide synthase (nNOS). The nitric-oxide synthases are cytochrome P450-type hemoproteins that catalyze the formation of nitric oxide (NO) and citrulline from L-arginine, O 2 , and NADPH (11,12). NOS is active as a homodimer (13,14), with each monomer binding tightly 1 eq each of FAD, FMN, tetrahydrobiopterin, and heme (13,(15)(16)(17). The prosthetic heme is the site of oxygen activation, which is required for the metabolism of L-arginine (15). 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 (18,19). This is unlike the microsomal P450 cytochromes, which have not been successfully reconstituted with heme in vitro. We present here evidence that the activity of nNOS is regulated by the hsp90based chaperone system and that hsp90 may facilitate heme binding in vivo. Since covalent alterations of the prosthetic heme and inactivation of P450 cytochromes are caused by drugs and xenobiotics as well as endogenous compounds (20,21), reconstitution of NOS activity by heme may be an important process in regulation of the active enzyme.
In the course of our studies, Garcia-Cardeñ a et al. (22) published that the endothelial isoform of NOS (eNOS) is activated by hsp90, and they were able to demonstrate direct activation of purified eNOS catalytic activity by purified hsp90 in the absence of added heme and other cofactors and without the co-chaperones of the hsp90-based chaperone machinery. On the basis of these in vitro results, they suggested that hsp90 may act as an allosteric modulator of eNOS. Although Garcia-Gardeñ a et al. (22) found that hsp90 was not co-immunoprecipitated with nNOS from lysates of cerebellum, we show here that * This investigation was supported in part by National Institutes of Health Grants ES08365 (to Y. O.) and DK31573 (to W. B. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ a portion of nNOS in HEK 293 cytosol is recovered as a native nNOS⅐hsp90 heterocomplex and that nNOS activity is reduced by the hsp90 inhibitor geldanamycin. As with the GR, the DE52-retained fraction of reticulocyte lysate can assemble heterocomplexes between hsp90 and purified apo-nNOS. This cellfree system also activates apo-nNOS to the specific activity of holo-nNOS, and activation is accompanied by increased heme association with the enzyme. However, in contrast to activation of GR steroid binding activity, activation of apo-nNOS heme binding activity in this cell-free system does not require heterocomplex assembly with hsp90. We speculate that the hsp90based chaperone machinery activates nNOS in vivo by facilitating heme entry into its binding site in the enzyme interior. However, in the cell-free system where abundant heme is readily available, transient opening of the heme binding cleft by components of a reticulocyte protein folding/unfolding system may be sufficient for promoting heme entry and nNOS activation without a requirement for hsp90.

Materials
Untreated rabbit reticulocyte lysate was from Green Hectares (Oregon, WI). 125 I-Conjugated goat anti-mouse and anti-rabbit IgGs were from NEN Life Science Products. The AC88 monoclonal IgG against hsp90 was from StressGen Biotechnologies (Victoria, British Columbia, Canada), and rabbit antiserum (23) that reacts with both hsp90 and hsp70 was kindly provided by Dr. Ettore Appella (National Cancer Institute). The affinity purified rabbit IgG against brain NOS used for immunoblotting nNOS was from Transduction Laboratories (Lexington, KY). The BuGR2 monoclonal IgG antibody against the GR was from Affinity Bioreagents (Neshanic Station, NJ). The rabbit antiserum used to immunoprecipitate nNOS was raised against rat neuronal NOS and was the generous gift of Dr. Lance Pohl (NHLBI, Bethesda). The antibody was affinity purified prior to use. The cDNA for rat neuronal NOS was kindly provided by Dr. Solomon Snyder (Johns Hopkins Medical School, Baltimore).

Methods
Cell Culture and Cytosol Preparation-Human embryonic kidney (HEK) 293 cells stably transfected with rat neuronal NOS by Bredt et al. (24) were obtained from Dr. Bettie Sue Masters (University of Texas Health Science Center, San Antonio). HEK 293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with calf serum and G418 (0.5 mg/ml) as described previously (25), except that the L-arginine concentration was 100 M. HEK cells were harvested, washed with PBS, and homogenized with a Tenbroeck ground glass homogenizer in 1 ml (per 10 8 cells) of HE buffer (10 mM Hepes, pH 7.4, 1 mM EDTA) with 20 mM sodium molybdate, 10 g/ml trypsin inhibitor, 100 M leupeptin, 0.5 M pepstatin A, 2 g/ml aprotinin, and 3 mM phenylmethylsulfonyl fluoride. Homogenates were centrifuged for 10 min at 16,000 ϫ g, with the supernatant being the cytosol fraction. L cell cytosol for immunoadsorption of GR was prepared as described previously (4).
Expression of nNOS in Sf9 Cells-The nNOS cDNA derived from pBluescript plasmid was subcloned into the EcoRI and NotI sites of the transfer vector PVL1393 as follows. The non-coding region from NotI (position 4699) to NotI site on pBluescript was excised, and the remaining cDNA was religated to remove the 3Ј EcoRI site. Primers were synthesized (ACTGAATTCACCATGGAAGAGAACACGTTTGG and GCTTTTCATCGTGGGGTCAATGG) to generate a polymerase chain reaction product corresponding to position 346 -906 as well as introducing an EcoRI site. The sequence of the polymerase chain reaction product was confirmed by the dideoxy chain termination method. The polymerase chain reaction product was subcloned into the EcoRI to BclI sites. The resulting clone was directionally cloned into PVL1393. Recombinant baculovirus was cloned by standard under-agarose plaque assays (26).
Sf9 cells were grown in SFM 900 II serum-free medium (Life Technologies, Inc.) supplemented with Cytomax (Kemp Biotechnology, Rockville, MD) in suspension cultures maintained at 27°C with continuous shaking (150 rpm). Cultures were infected in log phase of growth with recombinant baculovirus at a multiplicity of infection of 1.0. In some cases, after 48 h, heme (24 M) was added as an albumin conjugate, prepared as described previously (27). Cells were harvested and sus-pended in 1 volume of 10 mM Hepes, pH 7.5, containing 320 mM sucrose, 100 M EDTA, 0.1 mM dithiothreitol, 10 g/ml trypsin inhibitor, 100 M leupeptin, 0.5 M pepstatin A, 2 g/ml aprotinin, 3 mM phenylmethanesulfonyl fluoride, and 10 M tetrahydrobiopterin, and the suspended cells were ruptured by Dounce homogenization.
Immunoadsorption and Western Blotting-Native nNOS⅐hsp90 heterocomplexes were immunoadsorbed from 300 l of HEK 293 cytosol for 2 h at 4°C with 15 l of anti-nNOS IgG and subsequently incubated with 8-l pellets of protein A-Sepharose. Immune pellets were washed three times with 1 ml of ice-cold TEG buffer (10 mM TES, pH 7.6, 50 mM NaCl, 4 mM EDTA, 10% glycerol) containing both 20 mM sodium molybdate (TEGM) and 0.4% Triton X-100. In Fig. 1A, some of the samples were washed with radioimmune precipitation assay (RIPA) buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate). Immune pellets were boiled in SDS sample buffer with 10% ␤-mercaptoethanol, and proteins were resolved on 7% SDS-polyacrylamide gels. Proteins were then transferred to Immobilon-P membranes and probed with 0.1% anti-nNOS and 1 g/ml AC88 for hsp90. The immunoblots were then incubated a second time with 125 I-conjugated goat anti-mouse and anti-rabbit IgGs to visualize the immunoreactive bands.
Purification of nNOS-Lysates from infected Sf9 cells (2 ϫ 10 9 ) were centrifuged at 100,000 ϫ g for 1 h. The supernatant fraction was loaded onto a 2Ј,5Ј-ADP-Sepharose column (2 ml), and the nNOS was affinity purified as described (28), except that 10 mM 2Ј-AMP in high salt buffer was used to elute the protein. The nNOS-containing fraction was concentrated and dialyzed against 50 mM Tris-HCl, pH 7.4, containing 100 mM NaCl, 10% glycerol, 0.1 mM EDTA, 0.1 mM dithiothreitol, and 100 M tetrahydrobiopterin with the use of a ProDicon and a 100-kDa cut-off membrane. The concentrated sample was stored at Ϫ80°C.
Assay of nNOS Activity-nNOS activity was assayed by the following three methods: 1) conversion of oxyhemoglobin to methemoglobin; 2) assay of nitrate and nitrite released into the cell growth medium; and 3) conversion of [ 14 C]arginine to citrulline. Conversion of oxyhemoglobin was assayed by adding 20 l of insect cell lysate to an assay solution containing 100 M CaCl 2 , 100 M NADPH, 100 M arginine, 100 M tetrahydrobiopterin, 100 unit/ml catalase, 10 g/ml calmodulin, and 25 M oxyhemoglobin in a total volume of 200 l of 50 mM potassium phosphate, pH 7.4. The mixture was incubated at 25°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 (29). The amount of nitrate and nitrite in the cell medium was assayed by the use of nitrate reductase and quantitation by the Griess method as described (30). To assay conversion of arginine to citrulline, an aliquot (5 l) of incubation mixture was added to an assay solution containing 1 mM CaCl 2 , 1 mM NADPH, 50 M [ 14 C]arginine (32 mCi/mmol), 100 M tetrahydrobiopterin, 10 g/ml calmodulin, in a total volume of 100 l of 40 mM potassium phosphate, pH 7.4. The assay mixture was incubated at 37°C for 5 min, and the amount of [ 14 C]citrulline was determined as described previously (31).
DE52 Chromatography of Reticulocyte Lysate-Rabbit reticulocyte lysate (25 ml) was adsorbed to a 2.5 ϫ 20-cm column of DE52 equilibrated with HE buffer (10 mM Hepes, 1 mM EDTA, pH 7.35); the column was washed with 150 ml of HE buffer with 20 mM sodium molybdate followed by 150 ml of HE buffer alone, and proteins were eluted with a 400-ml gradient of 0 -0.5 M KCl. Proteins in the drop-through and wash or in DE52-retained material eluted with salt were concentrated and dialyzed against HKD buffer (10 mM Hepes, 100 mM KCl, 5 mM dithiothreitol, pH 7.35). hsp90, hsp70, Hop, and p23 were detected in the DE52-retained material by resolving an aliquot of each fraction by SDS-PAGE and Western blotting with appropriate antibodies, and fractions were combined in three pools designated A-C as described by Dittmar et al. (32). Pooled fractions were dialyzed against HKD buffer, concentrated to 1 ml (1/25 the original volume of lysate), and flashfrozen in small aliquots.
Purification of hsp70 and hsp90 -hsp70 was purified from the DE52 fraction pool A of rabbit reticulocyte lysate by chromatography on ATP-agarose and elution with ATP followed by ammonium sulfate precipitation exactly as described previously (33). hsp90 was purified from the DE52 fraction B of rabbit reticulocyte lysate by chromatography on hydroxylapatite followed by chromatography over ATP-agarose as described previously (33).
Glucocorticoid Receptor Heterocomplex Reconstitution-Glucocorticoid receptors were immunoadsorbed from 250 to 300-l aliquots of L cell cytosol by rotation for 2 h at 4°C with 8 l of protein A-Sepharose prebound with BuGR antibody. Prior to incubation with reticulocyte lysate, immunoadsorbed receptors were stripped of associated hsp90 by incubating the immunopellet an additional 2 h at 4°C with 0.5 M NaCl followed by one wash with 1 ml of TEG and a second wash with 1 ml of Hepes buffer (10 mM Hepes, pH 7.4). BuGR immune pellets (8 l of protein A-Sepharose) containing GR stripped of hsp90 were incubated with 50 l of rabbit reticulocyte lysate, with the DE52 drop-through and wash material, with the DE52-retained proteins of reticulocyte lysate, or with fractions A-C of the retained proteins (each at 10 l) and adjusted to 50 l with HKD buffer. Dithiothreitol (1 l) was added to each incubation to a final concentration of 5 mM, and 5 l of an ATPregenerating system (50 mM ATP, 250 mM creatine phosphate, 20 mM MgOAc, and 100 units/ml creatine phosphokinase) were added to all assays to yield a final assay volume of 56 l. The assay mixtures were incubated for 20 min at 30°C with resuspension of the pellets by shaking the tubes every 5 min. At the end of the incubation, one-fourth of the suspension was removed for assay of steroid binding, and the remainder was used for Western blotting of receptor and associated hsp90 as described previously (32). Steroid binding is expressed as the counts/min of [ 3 H]triamcinolone acetonide bound per GR immune pellet prepared from 100 l of L cell cytosol.
Cell-free Activation of Apo-nNOS-Apo-nNOS (5.6 g) purified from Sf9 cells incubated in heme-free medium was incubated for 1.5 h at 20°C with 2.5 l of the ATP-regenerating system, 3.5 l of a cofactor mixture (100 M FAD, 100 M FMN, 1 mM L-arginine, 100 M tetrahydrobiopterin, 200 units/ml superoxide dismutase, 2000 units/ml catalase), and other additions as noted in the figure legends adjusted to a total incubation volume of 35 l with HKD buffer. At the end of the incubation, nNOS activity was assayed by conversion of [ 14 C]arginine to citrulline. It should be noted that the apo-nNOS contained minor amounts of holoprotein, which is produced during recombinant expression of nNOS in insect cells due to endogenous heme synthesis.
Heme Assay-The amount of heme was determined by high pressure liquid chromatography as described previously (34). Samples were injected onto a high pressure liquid chromatography 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 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 400 nm was monitored. Myoglobin (horse heart) was used as a standard.
The ferrous-carbonyl complex was determined from the CO difference spectrum of the reduced protein with the use of an extinction coefficient of 91 mM Ϫ1 (25,35). Apo-nNOS was treated with the DE52retained proteins of reticulocyte lysate as described above. A portion of this mixture (200 l) was added to a solution containing 20% glycerol and 0.1 mM EDTA in 1.8 ml of 0.1 M potassium phosphate, pH 7.4. The difference spectrum was measured with the use of an Aminco-DW2 spectrophotometer modified with the OLIS-DW2 operating system.

Neuronal NOS Exists in Native
Heterocomplexes with hsp90 -To detect native nNOS⅐hsp90 heterocomplexes, cytosol from HEK 293 cells expressing rat nNOS was immunoadsorbed with nNOS-specific rabbit antiserum, and both the nNOS and the co-immunoadsorbed hsp90 in the washed immune pellet were detected by Western blotting. As shown in Effect of Geldanamycin Treatment on HEK Cell nNOS-Geldanamycin is a benzoquinonone ansamycin antibiotic that binds to the nucleotide-binding site of hsp90 and specifically blocks hsp90 function (36 -39). In geldanamycin-treated cells, proteins that are regulated by hsp90 are degraded more rapidly by a ubiquitin-dependent proteosomal mechanism (for review, see Ref. 2). In Fig. 2A, HEK 293 cells were treated with 10 M geldanamycin, and the amount of nNOS protein decreased ϳ50% in 24 h. Consistent with a geldanamycin-dependent increase in the rate of nNOS turnover, a similar decrease in nNOS occurred when translation of new enzyme was blocked with cycloheximide ( Fig. 2A). The effect of geldanamycin is concentration-dependent, with a substantial decrease in nNOS protein being observed at 1 M antibiotic (Fig. 2B). Fig. 3 shows the effect of a 6-h treatment with 10 M geldanamycin on HEK 293 cell nNOS activity. In that 24 h of treatment with geldanamycin are required to decrease nNOS protein levels by 50% (Fig. 2), the 80% decrease in nNOS activity observed after 6 h of treatment with geldanamycin ( Fig. 3) cannot be accounted for by increased nNOS degradation. As also shown in Fig. 3, treatment with the calcium ionophore A23187 increases nNOS activity and co-treatment with geldanamycin inhibits the enzyme activity. Both the geldanamycin-mediated reduction in nNOS activity and the increase in rate of nNOS turnover support the proposal that hsp90 regulates nNOS function in intact HEK 293 cells.
Effect of Geldanamycin on Heme Stimulation of nNOS Activity in Sf9 Cells-Sf9 cells have very low levels of endogenous heme, and addition of exogenous heme to the culture medium causes an increase in nNOS activity in infected cells. In the experiments shown in Fig. 4, Sf9 cells expressing rat nNOS were incubated in the presence of heme, and cells were harvested at various times for assay of nNOS activity by conversion of oxyhemoglobin to methemoglobin. Addition of heme (closed circles) increases nNOS activity 7-fold with respect to the activity in the absence of exogenous heme (open squares). This heme-dependent increase in nNOS activity is inhibited ϳ30% by geldanamycin (open circles and bars in inset).
Fractionation of the nNOS Activating Activity in Reticulocyte Lysate-Reticulocyte lysate contains all of the components required to assemble heterocomplexes between hsp90 and steroid  2, 4 and 6), and the immune pellets were washed three times under the following conditions: lanes 1 and 2, TEGM buffer with 0.1% Triton X-100; lanes 3 and 4, TEGM buffer with 0.4% Triton X-100; lanes 5 and 6, RIPA buffer containing 20 mM molybdate. nNOS and hsp90 were resolved by SDS-PAGE and Western blotting. B, molybdate stabilization of nNOS⅐hsp90 heterocomplexes. Immune pellets prepared as above and washed three times with TEGM plus 0.4% Triton X-100 were suspended in HE buffer and incubated for 25 min at 25°C. At the end of the incubation, all samples were washed three times with TEGM buffer plus 0.4% Triton X-100. Conditions are as follows: lanes 1 and 2, zero time (no incubation); lanes 3 and 4, incubation at 25°C with 20 mM molybdate; lanes 5 and 6, incubation at 25°C without molybdate. receptors or protein kinases (for review, see Refs. 1 and 2). As shown in Fig. 5A, incubation of immunoadsorbed, hsp90-free, mouse GR with rabbit reticulocyte lysate results in assembly of a GR⅐hsp90 heterocomplex (Western blot) and generation of steroid binding activity (bar graph). When the components of reticulocyte lysate are separated into drop-through and retained fractions by passage through DE52, the GR⅐hsp90 assembly activity is present in the DE52-retained fraction (Fig.  5A).
In the cell-free GR⅐hsp90 heterocomplex assembly system, the receptor is immobilized by immunoadsorption to protein A-Sepharose beads prior to incubation with reticulocyte lysate. We attempted to mimic this system by immobilizing nNOS by adsorption to ADP-agarose beads and then incubating with reticulocyte lysate, but this was not successful, and we found it necessary to incubate purified apo-nNOS in solution with the activating system of reticulocyte lysate. A second major difference in cell-free conditions of steroid receptor and nNOS activation is that the apo-nNOS is somewhat unstable at higher temperatures (40), and we incubate it with reticulocyte-activating system at room temperature rather than at 30°C.
Cell-free reactivation of apo-nNOS is shown in Fig. 5B. nNOS was purified from Sf9 cells grown in the presence or absence of heme to yield the heme-containing holo-nNOS and heme-depleted apo-nNOS standards shown at the left of Fig.  5B. When the apo-nNOS is incubated at 20°C in reactivation mix but without lysate, the level of enzyme activity decreases by about half, as indicated by the apo-nNOS control (Apo Ctl). Incubation with 1.2 M heme (HM) yields a somewhat higher activity as does incubation in the presence of reticulocyte lysate (RL). But unfractionated reticulocyte lysate does not activate nNOS activity like it activates the steroid binding activity of the GR shown in Fig. 5A. Reticulocyte lysate has an extremely high concentration of heme, and high concentrations of heme are toxic to nNOS. The vast majority of the heme is present in the DE52 drop-through fraction (DT) of lysate, which does not activate apo-nNOS. However, the DE52-retained fraction (RET) of lysate activates the apo-nNOS to the level of holo-nNOS (cf. RET with Holo Std in Fig. 5B). This DE52-retained fraction does not have nNOS activity or contain nNOS by immunoblotting, but it contains 33 M total (bound and free) heme. It is important to note that geldanamycin did not inhibit reactivation by the DE52 retentate, suggesting that hsp90 is not required for activation of nNOS by this cell-free system.
Fractionation of the Activity in the DE52 Retentate-The DE52-retained portion of lysate eluted with salt was divided into three subfractions (A-C) containing different components of the hsp90-based chaperone machinery as described by Dittmar et al. (32). The fractions eluting in the first half of the salt gradient were combined to form subfraction A, which contains most of the hsp70, the majority of hsp40, and all of the Hop, Hip, and BAG-1, fractions eluting at higher salt were combined to form subfraction B, which contains hsp90, and fractions eluting at highest salt form subfraction C which contains p23 (3,32). After contraction and dialysis, these three pooled subfractions were used to activate GR steroid binding activity and nNOS enzymatic activity. As we have reported previously (32) and as shown in Fig. 6A, all three subfractions of reticulocyte lysate are required for activation of GR steroid binding activity.
In the in vitro activation experiment of Fig. 5B, nNOS was reactivated under conditions where heme was present. In contrast, in the experiments of Figs. 5A and 6A, the GR was first assembled into a stable GR⅐hsp90 heterocomplex and subsequently incubated with steroid. In Fig. 6B, the GR was reactivated in the presence of ligand in the same manner as for apo-nNOS activation. When the GR is incubated with subfraction A in the presence of steroid, which binds to steroid-binding sites immediately as they are opened up, no steroid binding is generated unless purified hsp90 is also present (Fig. 6B). It should be noted that GR⅐hsp90 heterocomplexes are highly unstable in the absence of the stabilizing protein p23 (in subfraction C) or molybdate; thus, steroid must be present during heterocomplex assembly to detect conversion of the receptor to the steroid binding state (41). The activation of GR steroid binding activity by the combination of hsp90 and subfraction A is inhibited by geldanamycin (Fig. 6B). It is important to note that hsp90 alone is inactive in the absence of the co-chaperones in subfraction A (Fig. 6B).
In contrast to the inability of the hsp90-free subfraction A to activate GR steroid binding activity, this fraction is fully active at activating nNOS, and subfractions B and C produce no activation beyond that obtained with heme alone (Fig. 7A). Addition of purified hsp90 to subfraction A does not increase nNOS activation beyond the effect achieved with A alone, nor does either purified hsp90 or hsp70 alone activate nNOS, nor does hsp90 or hsp70 increase the activation produced by addition of free heme (data not shown). Thus, in vitro activation of GR steroid binding activity is strictly hsp90-dependent (Fig.  6B), whereas activation of nNOS does not require hsp90 (Fig.  7A), and neither protein is activated by hsp90 alone.
Subfraction A contains about 100 M heme, whereas B and C contain ϳ2 M heme. However, the presence of a high concentration of heme in subfraction A cannot account for the nNOS activating activity of subfraction A. The activation of 1 M purified apo-nNOS by free heme is maximum at 1-1.5 M heme, and at higher concentrations free heme becomes inhibitory (Fig. 7B). Essentially all of the heme in the DE52-retained fraction of reticulocyte lysate and subfractions A-C is bound to DE52-retained proteins, and we do not know the concentrations of free heme when these fractions are incubated with purified apo-nNOS. The major hemoprotein found in the retained fraction is hemoglobin, which is known to be able to release its heme prosthetic group and reconstitute apomyoglo-bin (42). Thus, it is likely that hemoglobin-bound heme and not free heme is the source of the heme during reconstitution. It is clear that cell-free activation of apo-nNOS is accompanied by an increase in enzyme-associated heme. We have assayed the nNOS-bound heme by its absorbance in the ferrous-carbonyl complex. The purified apo-nNOS preparation contained approximately one-fifth the amount of heme per mol of monomer as the purified holo-nNOS preparation, and after activation by the DE52-retained fraction of lysate, the apo-nNOS contained the same amount of heme as the holo-nNOS. The 5-fold increase in nNOS-associated heme is consistent with the ϳ5-fold increase in enzyme activity obtained under our cell-free activation conditions (Figs. 5B and 7A).
Despite the fact that nNOS activation in vitro does not require hsp90, the DE52-retained fraction of lysate, which was shown to activate nNOS in Figs. 5B and 7A, forms nNOS⅐hsp90 heterocomplexes (Fig. 7C, lane 4). In Fig. 7C, hsp90 was immunoblotted with the AC88 monoclonal IgG, which recognizes the rabbit hsp90 assembled into a heterocomplex with the rat nNOS by the chaperone machinery of reticulocyte lysate but does not recognize insect hsp90 from the Sf9 cells that produced the nNOS (43). As shown in Fig. 7D, the apo-nNOS preparation contains a significant amount of insect (Sf9-derived) hsp70 and a trace amount of insect hsp90, both of which were detected by immunoblotting with an antiserum that recognizes both of the insect hsps (43). Immunoadsorption of the purified apo-nNOS preparation is accompanied by co-immunoadsorption of the Sf9 hsp70 (data not shown); thus, a portion of the purified apo-enzyme is bound by this component of the chaperone system.

DISCUSSION
It is clear from Fig. 1 that nNOS in HEK 293 cells is associated with hsp90, as was reported recently for eNOS (22). In contrast to hormone-free steroid receptors, which are present in molybdate-stabilized cytosols entirely as stable 9 S receptor⅐hsp90 heterocomplexes, we estimate that only 5-10% of the nNOS in HEK 293 cytosol is bound to hsp90. This suggests that the association of nNOS with hsp90 may be much more dynamic than that of steroid receptors. The binding of steroid to the receptor⅐hsp90 heterocomplex causes the complex to dissociate (1), and a similar dissociation might occur when heme binds to nNOS in HEK 293 cells. This could also explain the relatively low level of nNOS⅐hsp90 heterocomplexes obtained in the heme-containing cell-free system from reticulocyte lysate (Fig. 7C). Despite the low fraction of nNOS that is bound to hsp90 at any particular time, treatment of HEK 293 cells with the hsp90 inhibitor geldanamycin over a period of several hours substantially increases the rate of nNOS turnover (Fig. 2) and decreases nNOS activity (Fig. 3), implying that hsp90 is important for both nNOS stability and function in the intact cell.
Geldanamycin was first reported by Joly et al. (44) to inhibit inducible NOS activity in rat smooth muscle and murine brain endothelial cells. These authors were unaware that geldanamycin, which was thought for many years to be a protein kinase inhibitor, had been shown by Whitesell et al. (36) to interact with hsp90 rather than with the hsp90-bound protein kinases themselves. Thus, Joly et al. (44) suggested that geldanamycin inhibited a protein kinase required for inducible NOS induction. Garcia-Cardena et al. (22) then showed that geldanamycin inhibited eNOS activity in cultured human endothelial cells. Because they found that purified hsp90 bound directly to purified eNOS and facilitated eNOS-mediated catalysis in vitro, they speculated that hsp90 may act in cells as an allosteric modulator of eNOS by inducing a conformational change in the enzyme or by stabilizing the dimeric form. Although we find that geldanamycin treatment of intact HEK 293 cells results in profound inhibition of nNOS activity, we find no effect of purified hsp90 on the activity of purified nNOS, and we suggest an alternative mechanism may explain the hsp90 effect on nNOS activity in HEK 293 cells.
Because a major action of the hsp90-based chaperone ma-  2 and 4) antibody. The nNOS immune pellets were incubated for 20 min at 30°C with reticulocyte lysate (lanes 1 and 2) or the DE52-retained fraction of lysate (lanes 3 and 4). The immune pellets were then washed, and nNOS and hsp90 were detected by Western blotting. The AC88 antibody used to probe for hsp90 recognizes rabbit hsp90 but not the hsp90 of the Sf9 cells that produced the nNOS (43). D, purified rat apo-nNOS contains insect hsp70. Samples (8 g) of the purified nNOS preparation were resolved by SDS-PAGE and stained with Coomassie Blue (CB) or immunoblotted (IB) with anti-nNOS or rabbit antiserum that reacts with both hsp70 and hsp90 of Sf9 cells. The photos are of the bands developed with the peroxidase-conjugated counterantibody. chinery on the GR is to open the steroid binding cleft sufficiently to create a high affinity steroid-binding site, we speculated that it might act in a similar manner to open the heme binding cleft in nNOS to facilitate heme entry. Thus, we asked if geldanamycin would block the heme-induced increase in nNOS activity in Sf9 cells. Although geldanamycin inhibited about 30% of the heme-induced increase (Fig. 4), which occurs within an hour in Sf9 cells, the effect of geldanamycin over several hours in HEK 293 cells was much greater (Fig. 3). When we then utilized the DE52-retained fraction of reticulocyte lysate to activate purified apo-nNOS in vitro, there was no inhibition by geldanamycin (Fig. 5B). A reasonable explanation of the progressive loss in geldanamycin sensitivity as we proceed from the HEK cell to apo-nNOS activation in vitro may relate to the levels of free heme that are available to the enzyme in the different environments. For example, apo-nNOS activation in vitro is occurring in the presence of heme that can enter a heme binding cleft as soon as it is opened. Thus, even transient opening of the heme binding cleft by components of a reticulocyte protein folding/unfolding machinery that segregate with subfraction A (e.g. hsp70, hsp40, Hop, Hip, and BAG-1) may be sufficient when a significant concentration of free heme is present. In intact HEK 293 cells, however, very little free heme is available to the enzyme, and a longer term of cleft opening that is produced by the entire hsp90-based chaperone system is required to facilitate heme entry. Heme activation of apo-nNOS in Sf9 cells may reflect an intermediate situation.
Here, there is a gradient of free heme from the outside to the inside of the cell, with sufficient free heme in the nNOS environment to permit activation of a significant portion of the enzyme via transient opening of the heme binding cleft. Inasmuch as the hsp90 component of the machinery is not required to ensure a longer term opening of the heme-binding site, the inhibition of apo-nNOS activation produced by geldanamycin in the heme-treated Sf9 cells is less than the geldanamycin inhibition of nNOS activity in HEK 293 cells.
The reactivation of apo-nNOS achieved under cell-free conditions in this work is the most extensive activation achieved to date. Hemmens et al. (45) have reported activation of rat apo-nNOS to a specific activity of 344 nmol of citrulline/min/mg when the purified enzyme reconstituted with hemin was treated with 100 M tetrahydrobiopterin and 1 mM L-arginine in the presence of 30 mM dithiothreitol over a period of 4 h. Here, we activate apo-nNOS to the level of purified holo-nNOS (500 -600 nmol of citrulline/min/mg) with the DE52-retained components of reticulocyte lysate. When we assay the heme in our purified holo-nNOS preparation, we find there is approximately 1 mol of heme per 2 mol of nNOS. However, the amount of nNOS is based on the protein concentration of the purified enzyme preparation which, as shown by staining with Coomassie Blue (Fig. 7D), contains considerable (ϳ30 -40%) NOS cleavage products and hsp70. The holo-nNOS was purified from Sf9 cells that were incubated for 3 h with oxyhemoglobin, which produces maximal nNOS activation, and when our holo-nNOS preparation is further purified over a Sephacryl S300HR column, the specific activity increases to approximately 1000 nmol of citrulline/min/mg. For comparison, the activity of native nNOS is in the range of 1000 to 1200 nmol of citrulline/ min/mg. Thus, the reactivation we achieve may be nearly complete. Our findings on the reactivation of apo-nNOS under cell-free conditions are limited to the recombinantly expressed apo-nNOS produced in Sf9 cells grown in heme-deficient medium. In this respect, the chemical preparation of an apo-nNOS that could be reactivated would be a major technical advance.