Abstract

Soluble guanylate cyclase (sGC) is an important downstream intracellular target of nitric oxide (NO) that is produced by endothelial NO synthase (eNOS) and inducible NO synthase (iNOS). In this study, we demonstrate that sGC exists in a complex with eNOS and heat shock protein 90 (HSP90) in aortic endothelial cells. In addition, we show that in aortic smooth muscle cells, sGC forms a complex with HSP90. Formation of the sGC/eNOS/HSP90 complex is increased in response to eNOS-activating agonists in a manner that depends on HSP90 activity. In vitro binding assays with glutathione S-transferase fusion proteins that contain the α- or β-subunit of sGC show that the sGC β-subunit interacts directly with HSP90 and indirectly with eNOS. Confocal immunofluorescent studies confirm the subcellular colocalization of sGC and HSP90 in both endothelial and smooth muscle cells. Complex formation of sGC with HSP90 facilitates responses to NO donors in cultured cells (cGMP accumulation) as well as in anesthetized rats (hypotension). These complexes likely function to stabilize sGC as well as to provide directed intracellular transfer of NO from NOS to sGC, thus preventing inactivation of NO by superoxide anion and formation of peroxynitrite, which is a toxic molecule that has been implicated in the pathology of several vascular diseases.

  • smooth muscle cells
  • endothelium
  • vascular endothelial growth factor
  • bradykinin
  • cGMP accumulation

soluble guanylate cyclase (sGC), an α/β-heterodimeric heme protein, catalyzes the conversion of GTP to cGMP in many cells including vascular endothelial cells (ECs) and vascular smooth muscle cells (SMCs). Activation of sGC is by direct binding of nitric oxide (NO) to the sGC heme prosthetic group. Formation of the nitrosyl heme adduct induces a conformational change in sGC that results in an increase in its enzymatic activity (21). The NO that activates sGC in various cells is the product of a reaction that is catalyzed by one of three distinct NO synthase (NOS) molecules, which catalyze the oxidation of l-arginine to produce l-citrulline and NO (1). In ECs, NO production is mediated by the constitutively expressed endothelial NOS (eNOS). Activation of eNOS is by Ca2+-calmodulin (CaM) after agonist-stimulated elevations in intracellular Ca2+ concentrations. Two signaling pathways exist that involve eNOS and sGC. The first is an intercellular pathway whereby NO, which is produced by eNOS in ECs, diffuses to the underlying SMCs and promotes blood vessel relaxation (16). The second is an intracellular eNOS-sGC pathway that is essential for vascular endothelial growth factor (VEGF)-induced increases in EC permeability and proliferation (24, 25, 30).

Initially, eNOS was thought to function as an isolated homodimer. It is now known, however, that eNOS exists in multiprotein complexes in which it interacts with other proteins. These proteins include Akt (10), caveolin-1 (13, 19, 23), heat shock protein 90 (HSP90; Refs. 12, 26), certain G protein-coupled receptors (18, 22), eNOS-interacting protein (NOSIP; Ref. 7), and dynamin-2 (6). Interactions of eNOS with Akt, caveolin-1, HSP90, and G protein-coupled receptors are regulated in an agonist-dependent manner (9, 12, 18, 22, 26). Currently, no protein partners for sGC are known, and the enzyme is thus believed to exist in the cytoplasm as a free protein that does not interact with other proteins. However, because sGC is a downstream intracellular effector of NO, in the present study we have examined whether sGC forms a complex with eNOS in bovine aortic ECs (BAECs). We have also defined the role of HSP90 in mediating the sGC-eNOS interaction and identified the subunit of sGC that is responsible for complex formation. In addition, we investigated whether complex formation is regulated in an agonist-dependent manner and whether the capacity of sGC to be activated by NO is dependent on HSP90 activity.

METHODS

Materials. Anti-sGC antibodies were purchased from Cayman Chemical (catalog no. 160897). Anti-eNOS and anti-HSP90 antibodies were obtained from Transduction Laboratories (catalog nos. 61097 and 610419, respectively). VEGF was purchased from R&D Systems. Geldanamycin was purchased from Life Technologies. Purified recombinant human HSP90 was purchased from Stressgen Biotechnology. BioRad protein assay was purchased from Bio-Rad. Bradykinin (BK) and sodium nitroprusside (SNP) were obtained from Sigma. The cGMP enzyme immunoassay kits, the glutathione S-transferase (GST) cloning vector pGEX-4T-l, and glutathione Sepharose 4B were purchased from Amersham Pharmacia Biotech. Protein A/G agarose came from Santa Cruz Biotechnology, and SuperSignal chemiluminescence substrate came from Pierce. Plasmids containing the cDNAs for the human sGC α1- and β1-subunits were a generous gift from Dr. Andreas Papapetropoulos, University of Athens, Athens, Greece.

Cell culture. In-house harvested BAECs were passaged from primary cultures and used in experiments during passages 2–5. Cultures were maintained in M199 supplemented with 10% fetal bovine serum, 5% iron-supplemented calf serum, 20 μg/ml l-glutamine, 1× MEM amino acid and vitamin solutions, 0.6 μg/ml thymidine, 500 IU/ml penicillin, and 500 μg/ml streptomycin. In-house harvested rat aortic SMCs (RASMCs) were also passaged from primary cultures and used in experiments during passages 2–5. Cultures were maintained in DMEM that contained 10% fetal bovine serum, 500 IU/ml penicillin, and 500 μg/ml streptomycin.

Preparation of soluble and particulate fractions of BAECs and RASMCs. Cells were lysed in hypotonic buffer that contained 50 mM Tris · HCl, pH 7.4, 5 mM EGTA, 2 mM EDTA, 10 μg/ml leupeptin, 10 μg/ml pepstatin A, and 1% PMSF. Samples were homogenized by 30 up-and-down strokes in a tight-fitting Dounce homogenizer followed by sonication. Lysates were then centrifuged at 100,000 g for 30 min to obtain particulate (pellet) and soluble (supernatant) fractions. After resuspension of the pellet to the original volume of the homogenate, equal volumes of supernatant and pellet fractions were analyzed via immunoblot with anti-sGC antibody and densitometry of the immunoblots.

Immunoprecipitation and immunoblotting. BAECs and RASMCs were lysed in ice-cold buffer that contained 20 mM Tris · HCl, pH 7.4, 2.5 mM EDTA, 50 mM NaF, 10 mM Na4P2O7, 1% Triton X-100, 1% PMSF, 10 μg/ml leupeptin, 10 μg/ml pepstatin A, and 5 μg/ml aprotinin. Lysates were centrifuged at 10,000 g for 25 min to remove insoluble material. Anti-sGC, -eNOS, and -HSP90 antibodies were then added for 3 h at 4°C. Protein A/G agarose was added, and samples were incubated for an additional 3 h at 4°C. Immunoprecipitated proteins bound to the agarose beads were then washed twice, and proteins were eluted from the beads by boiling the samples in SDS sample buffer. Agarose beads were pelleted by centrifugation, and protein supernatants were analyzed on SDS polyacrylamide gels. Proteins were transferred to nitrocellulose membranes by electroblot. Membranes were then probed with anti-eNOS, -sGC, or -HSP90 antibodies or nonimmuno-Ig isotypes to confirm specificity of interaction. Bound proteins were visualized using the Pierce SuperSignal chemiluminescence substrate.

Expression and purification of eNOS in a baculovirus system. Bovine eNOS was expressed in a baculovirus-Sf9 insect cell system and purified to >95% homogeneity as described previously (29).

In vitro binding assays. Plasmid constructs that encode GST-sGC α1- and β1-subunits (GST-sGC-α1 and -β1, respectively) were created by subcloning the full-length cDNAs for human sGC-α1 and sGC-β1 into the GST fusion protein-cloning vector pGEX-4T-1. Fusion proteins and a GST nonfusion protein were expressed in Escherichia coli and purified by affinity binding to glutathione-Sepharose as described by Frangioni and Neel (11). GST or GST-sGC fusion proteins (100 pmol each, quantified by Bio-Rad protein assay) prebound to glutathione-Sepharose beads were incubated overnight (with shaking at 4°C) with 100 pmol purified HSP90, purified eNOS, or a combination of both purified proteins in a buffer that contained 50 mM Tris · HCl, 20% glycerol, and 1% PMSF. Beads were washed six times in buffer that contained 50 mM HEPES, pH 7.5, 1.0 M NaCl, 1 mM EDTA, 0.5% 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), and 1% PMSF when binding assays were performed with purified HSP90 only. The same wash buffer was used for assays with purified eNOS or the combination of eNOS and HSP90 except for a change in salt concentration to 0.2 M. Bound proteins were eluted with 100 mM reduced glutathione, 150 mM NaCl, 3 mM EDTA, and 7.5 mM Tris·HCl, pH 8.0. Eluted proteins were then immunoblotted with anti-HSP90 or anti-eNOS antibodies either separately or in combination.

Colocalization studies. BAECs and RASMCs grown on glass coverslips were fixed in 100% methanol at –20°C for 20 min, washed in PBS, blocked in 5% goat serum, and incubated with rabbit polyclonal anti-sGC and mouse monoclonal anti-HSP90 antibodies. After being washed and blocked as described, specimens were exposed to secondary rhodamine-tagged anti-rabbit and fluorescein-tagged anti-mouse antibodies and to Hoechst 33258. Preparations were examined using an Olympus IMT-2 microscope equipped for epifluorescence digital imaging. With the use of separate FITC, rhodamine, and 4′,6-diamidino-2-phenylindole filter sets, image stacks were collected with 0.2-μm spacing between images at each wavelength and were processed for digital deconvolution optical sectioning using Microtome software (Vaytek; Fairfield, IA). Corresponding optical sections at the three separate wavelengths were pseudocolored and overlaid using IPLab Spectrum software (Scanalytics; Fairfax, VA).

cGMP accumulation. Estimation of cGMP accumulation was carried out by a modification of the procedure of Ishii et al. (17) that was originally designed to measure NO release by determining sGC activity in RFL-6 rat lung fibroblast reporter cells. BAEC or RASMC culture medium was removed and replaced by Locke's solution that contained 154 mM NaCl, 5.6 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 3.6 mM NaHCO3, 5.6 mM glucose, 10 mM HEPES, pH 7.4, 20 U/ml superoxide dismutase (SOD), and 0.3 mM isobutyl methyl xanthine. Cells were treated for 15 min with various concentrations of SNP following either no pretreatment or pretreatment with geldanamycin (1 μg/ml for 1 h) and then were lysed in ice-cold 20 mM sodium acetate, pH 4.0. Lysates were frozen at –20°C until assay for cGMP concentration using an enzyme immunoassay kit.

Determination of extracellular release of superoxide anion. BAECs or RASMCs grown in 24-well multiwell plates were washed with Earle's balanced salt solution (EBSS) and were incubated for 60 min with EBSS that contained 1 or 10 μg/ml geldanamycin and 40 μM ferricytochrome c with or without 350 U/ml Cu/Zn SOD. SOD-inhibitable reduction of ferricytochrome c to ferrocytochrome c was assessed spectrophotometrically at 550 nm, and superoxide production was estimated based on the extinction coefficient of 21.1 mM1 · cm1.

Whole animal studies. Experiments were approved by the Institutional Committee for Animal Use in Research and Education and adhered to the American Physiological Society standards of humane animal experimentation. Sprague-Dawley rats (250 g) were anesthetized, tracheostomized, and artificially ventilated, and catheters were introduced into the left carotid artery and right jugular vein. Systemic arterial pressure was continuously recorded on a Gould recorder. Bolus injections of SNP in saline (0.4–0.7 ml iv, 1–3 μg/ml) were given before and 60 min after administration of 3 mg/ml iv geldanamycin. Each animal received two injections of equal volume and concentration before and after geldanamycin administration, and the maximal fall in mean arterial pressure (MAP) values was averaged. The volume and concentration of SNP before geldanamycin administration were adjusted to produce an ∼30-mmHg maximal decrease in MAP.

RESULTS

sGC is so named because it is thought to exist predominately or exclusively in the cytosolic fraction of most if not all cell types. In contrast, particulate guanylate cyclase (which is not activated by NO) is thought to exist exclusively in the plasma membrane as a transmembrane-spanning protein (21). Possible membrane association of sGC, however, has not been extensively investigated. To determine whether a portion of sGC might be membrane associated in BAECs or RASMCs, we prepared soluble and particulate fractions from cells after homogenization and lysis in nondetergent buffer and quantified the relative amounts of sGC in each fraction by immunoblotting with an anti-sGC antibody and performing densitometry of the immunoblots. This analysis showed that 82 ± 4.2% (means ± SE, n = 6) of sGC-β1 in BAECs and 16 ± 1.9% (n = 6) of sGC-β1 in RASMCs exist in the particulate fraction of these cells (Fig. 1). These numbers remained essentially the same following treatment with geldanamycin (1 μg/ml for 1 h, n = 3 for each cell type). These data suggest that at least in the case of BAECs, the term “soluble guanylate cyclase” may be somewhat of a misnomer.

Fig. 1.

Subcellular distribution of soluble guanylate cyclase (sGC) β1-subunit in bovine aortic endothelial cells (BAECs, A) and rat aortic smooth muscle cells (RASMCs, B). BAECs and RASMCs were lysed, homogenized, and sonicated in hypotonic nondetergent lysis buffer, and particulate (pellet) and soluble (supernatant) fractions were separated by centrifugation at 100,000 g. Relative amounts of sGC β1-subunit were quantified by densitometry of immunoblots. Shown are representative blots and means ± SE from six different experiments.

Successive enzymes in intracellular signaling pathways are often physically associated in multiprotein complexes. Moreover, it was shown previously (12) that eNOS forms a complex with HSP90. Therefore, we examined whether sGC, eNOS, and HSP90 exist in a trimeric complex in BAECs. BAECs (passages 2–5) were grown in serum-containing medium and lysed, and lysates were immunoprecipitated with anti-eNOS, -HSP90, or -sGC antibodies. Immunoprecipitated proteins were then immunoblotted with an anti-eNOS antibody. As shown in Fig. 2A, anti-HSP90 and -sGC antibodies coimmunoprecipitated eNOS, which suggests that the three proteins do indeed form a complex in BAECs. To further confirm this conclusion, lysates were immunoprecipitated using anti-eNOS, -HSP90, or -sGC antibodies, and immunoprecipitates were immunoblotted with an anti-HSP90 antibody. Anti-eNOS and -sGC antibodies coimmunoprecipitated HSP90 in these experiments (Fig. 2B), thereby providing further evidence for the existence of an sGC/eNOS/HSP90 complex in BAECs. In an additional set of experiments, BAEC lysates were immunoprecipitated with anti-eNOS, -HSP90 or -sGC antibodies, and immunoprecipitated proteins were immunoblotted with an anti-sGC antibody. In these experiments, the anti-eNOS and -HSP90 antibodies coimmunoprecipitated sGC (Fig. 2C), which again confirms the existence of an sGC/eNOS/HSP90 complex in these cells.

Fig. 2.

Coimmumoprecipitation of sGC, endothelial nitric oxide (NO) synthase (eNOS), and heat shock protein 90 (HSP90) from BAECs. Unstimulated BAECs were lysed, and lysates were immunoprecipated (IP) with either anti-eNOS, -HSP90, or -sGC antibody and then immunoblotted (IB) with either an anti-eNOS antibody (A), an anti-HSP90 antibody (B), or an anti-sGC antibody (C). Amounts of protein loaded in each lane were not equivalent but were adjusted to amounts that gave a clear signal without overexposure. Similar results were obtained in three different experiments.

We also examined the potential interactions of sGC and HSP90 in cultured RASMCs. Coimmunoprecipitation experiments showed that anti-sGC antibodies coprecipitated HSP90 (Fig. 3A) and that anti-HSP90 coprecipitated sGC (Fig. 3B). These data suggest that an sGC/HSP90 complex exists in RASMCs that is similar to the sGC/eNOS/HSP90 complex found in BAECs.

Fig. 3.

Coimmunoprecipitation of sGC and HSP90 from RASMCs. Unstimulated RASMCs were lysed, and lysates were immunoprecipitated with either anti-sGC or anti-HSP90 antibody and then immunoblotted with either anti-HSP90 antibody (A) or anti-sGC antibody (B). Amounts of protein loaded in each lane were not equivalent but were adjusted to amounts that gave a clear signal without overexposure. Results shown are representative of three different experiments.

The amount of eNOS associated with HSP90 is increased following stimulation of ECs with the eNOS-activating agonists VEGF, histamine, and estrogen (12, 26). We therefore examined whether sGC interactions with eNOS and HSP90 in BAECs might increase in response to BK or VEGF stimulation. BAECs were serum starved overnight and then treated with BK (1 μM) or VEGF (20 ng/ml) for 0, 1, 5, or 10 min. Cells were lysed, and lysates were immunoprecipitated with an anti-sGC antibody. Immunoprecipitated proteins were then immunoblotted with either anti-eNOS or anti-HSP90 antibody. As shown in Fig. 4A, BK stimulated an increased association of sGC with both eNOS and HSP90 that was maximal at 1 min and declined thereafter. Likewise, VEGF stimulated an increased association of sGC with eNOS and HSP90 that was maximal at 5 min (Fig. 4B). These results suggest that sGC interactions with eNOS and HSP90 are regulated in an agonist-dependent manner.

Fig. 4.

Agonist stimulation of sGC/eNOS/HSP90 complex formation in BAECs. BAECs were treated with bradykinin (BK; 1 μmol/l, A) or vascular endothelial growth factor (VEGF, 20 ng/ml, B) for 0, 1, 5, or 10 min, cells were lysed, and equal quantities of lysate were immunoprecipitated with an anti-sGC antibody. Immunoprecipitated proteins were then immunoblotted with either an anti-eNOS antibody or an anti-HSP90 antibody. Similar results were obtained in three different experiments.

Complex formation between HSP90 and HSP90-interacting proteins generally requires the ATP-dependent conversion of HSP90 from an inactive to an active conformation. For example, formation of a trimeric complex of HSP90, the aryl hydrocarbon receptor (AhR), and AhR-interacting protein (AIP) is dependent on HSP90 activity (4). To determine whether the interactions of sGC with eNOS or HSP90 are dependent on HSP90 activity, we used a specific inhibitor of HSP90 activity, geldanamycin (14). BAECs were serum starved overnight and then treated with BK (1 μM) for 0, 1, and 5 min following either no pretreatment or pretreatment with geldanamycin (1 μg/ml for 1 h). Cells were lysed, and lysates were immunoprecipitated with an anti-sGC antibody. Immunoprecipitates were immunoblotted with anti-eNOS and -HSP90 antibodies. As shown in Fig. 5, A and B, geldanamycin inhibited BK-stimulated complex formation of sGC with both eNOS and HSP90, which suggests that complex formation requires that HSP90 be in an active conformation. Interestingly, geldanamycin inhibited sGC-eNOS association by 20–40%, whereas it inhibited sGC-HSP90 association by 45–91%. The significance of these differences is not apparent.

Fig. 5.

Effects of geldanamycin (Gld) on BK-stimulated sGC/eNOS/HSP90 complex formation in BAECs. BAECs were treated with BK (1 μmol/l) for 0, 1, or 5 min following either no pretreatment or pretreatment with geldanamycin (1 μg/ml for 1 h). Cells were lysed, and equal quantities of lysate were immunoprecipitated with an anti-sGC antibody. Immunoprecipitated proteins were then immunoblotted with either an anti-eNOS (A), an anti-HSP90 (B), or an anti-sGC-β1 antibody (C). Equivalent results were obtained in three different experiments.

To determine whether HSP90 or eNOS interacts directly with sGC and, if so, with which subunit, we prepared GST fusion proteins that contained the full-length human sGC-α1 and -β1 subunits. The fusion proteins and a GST nonfusion protein were expressed in E. coli and purified by affinity chromatography on glutathione-Sepharose. The GST fusion proteins or GST only were then used in in vitro binding assays with purified human HSP90, purified bovine eNOS, or a combination of the two purified proteins. GST or GST fusion proteins prebound to glutathione-Sepharose beads were incubated with HSP90, eNOS, or both proteins overnight at 4°C, washed six times in buffer that contained either 0.2 M or 1.0 M NaCl, and eluted with reduced glutathione. The wash buffers used with purified eNOS or purified eNOS plus purified HSP90 contained only 0.2 M NaCl to maximize our chances of detecting binding, because it was reported previously that the eNOS-HSP90 complex disassembles in a highionic strength buffer that contains 0.5 M NaCl (12). Eluted proteins were immunoblotted with anti-HSP90, -eNOS antibody, or a combination of both antibodies. As shown in Fig. 6, left, HSP90 bound to GST-sGC-β1 but not to GST only or GST-sGC-α1 subunit. Furthermore, it appears that HSP90 binds to sGC-β1 with very high affinity, because the binding interaction is not disrupted by very high-stringency washes in buffer that contains 1.0 M NaCl. However, eNOS did not interact with either of the sGC subunits in the same experiment (Fig. 6, middle), which suggests that eNOS does not interact directly with either subunit of sGC. However, when eNOS was incubated with the GST fusion proteins in the presence of HSP90, eNOS binding to GST-sGC-β1 was detected (Fig. 6, right). In addition, when eNOS was incubated together with HSP90 and GST-sGC-β1 in the presence of geldanamycin (1 μg/ml), binding of both HSP90 and eNOS to GST-sGC-β1 was completely blocked. These data suggest that the eNOS-sGC interaction is indirect, and that it occurs due to binding of both eNOS and sGC to HSP90 with HSP90 functioning as an adaptor protein for the eNOS-sGC interaction.

Fig. 6.

In vitro binding of HSP90 and eNOS to GST-sGC fusion proteins. GST-sGC α1- and β1-subunit fusion proteins and GST alone were expressed in Escherichia coli and purified by affinity binding to glutathione-Sepharose beads. Proteins, prebound to beads, were incubated overnight at 4°C with purified HSP90, purified eNOS, or a combination of both purified proteins in the absence or presence of geldanamycin. Beads were washed six times, and bound proteins were eluted with reduced glutathione and immunoblotted with either anti-HSP90 (left) or anti-eNOS (middle) antibody or a combination of the two antibodies (right).

To confirm the in vivo subcellular colocalization of sGC and HSP90, BAECs and RASMCs were fixed, stained with either anti-HSP90 (fluorescein green) antibody, anti-HSP90 (rhodamine red) antibody, or both, and examined using fluorescence deconvolution optical sectioning microscopy. As shown in Fig. 7, wide extranuclear cytoplasmic distribution of sGC and HSP90 was observed in both cell types. Colocalization of the two proteins was confirmed by the overlay of pseudo-colored red and green images, which resulted in a yellow signal at sites of colocalization throughout most of the extranuclear domain in BAECs and RASMCs stained with both anti-sGC and -HSP90 antibodies. These data make it appear highly likely that sGC and HSP90 coexist in common subcellular compartments.

Fig. 7.

Fluorescent colocalization of HSP90 and sGC in BAECs and RASMCs. BAECs (left) and RASMCs (right), which were grown on glass coverslips, were fixed and stained with either rabbit polyclonal anti-HSP90 antibodies (top), mouse monoclonal anti-sGC antibodies (middle), or both (bottom) and were then exposed to secondary fluorescein-tagged anti-mouse antibodies, rhodamine-tagged anti-rabbit antibodies, or both, respectively, and were examined with fluorescence deconvolution optical sectioning microscopy. Yellow hue (bottom) indicates colocalization of sGC and HSP90.

One potential role for HSP90 in the sGC/eNOS/HSP90 complex is to facilitate sGC-eNOS interactions in a manner similar to that described recently for the HSP90/AhR/AIP complex (4). Another possible role is that HSP90 affects sGC either directly or indirectly to make the enzyme more susceptible to activation by NO. To test this possibility, BAECs were either not pretreated or were pretreated with geldanamycin (1 μg/ml for 1 h) and then exposed for 15 min to 0, 1, 10, or 100 μM of the NO donor SNP. The amount of cGMP produced in the absence and presence of geldanamycin, basally, or in response to increasing concentrations of SNP was then quantified by enzyme immunoassay. Geldanamycin had little or no effect on basal cGMP production in BAECs. However, geldanamycin significantly attenuated SNP stimulation of cGMP synthesis. Exposure to 100 μM SNP for 15 min produced a 4.3 ± 0.2-fold increase (mean ± SE, n = 3 wells, each in duplicate) in cGMP accumulation in BAECs that was reduced to a 1.9 ± 0.3-fold increase (n = 3 wells, each in duplicate) following pretreatment with geldanamycin (Fig. 8A). The effect of geldanamycin (1 μg/ml for 1 h) on SNP-induced cGMP accumulation in RASMCs was also investigated. At 1 μM concentration, SNP produced a 17.1 ± 1.4-fold (n = 3 wells, each in duplicate) increase in cGMP content of RASMCs in 15 min, which is consistent with the much higher levels of sGC activity that are known to exist in RASMCs compared with ECs. Pretreatment with geldanamycin reduced the SNP-induced increase in cGMP concentration to 8.0 ± 0.7-fold (Fig. 8B). Recent reports suggest that geldanamycin stimulates the production of superoxide anion (8). Superoxide could interact with NO to form peroxynitrite and reduce the amounts of NO that are available to interact with sGC independent of the actions of geldanamycin on HSP90. To prevent this possibility, all cGMP-accumulation experiments were performed in the presence of SOD. In addition, the ability of geldanamycin to produce superoxide was investigated in both BAECs and RASMCs following a 60-min incubation in the absence of SOD. As shown in Fig. 9, at 10 μg/ml, geldanamycin elicited significant increases in superoxide levels, but at 1 μg/ml (the concentration used in all our studies), there was no significant increase from basal superoxide levels.

Fig. 8.

Effects of geldanamycin on the ability of the NO donor sodium nitroprusside (SNP) to stimulate cGMP accumulation in BAECs and RASMCs. BAECs (A) or RASMCs (B) were pretreated with either vehicle or geldanamycin (1 μg/ml) for 1 h and were then exposed for 15 min to various concentrations of SNP. Cells were lysed, and lysates were assayed for cGMP by enzyme immunoassay (means ± SE, n = 3; *P < 0.05, paired t-test).

Fig. 9.

Effects of geldanamycin on superoxide generation in BAECs and RASMCs. BAECs (A) or RASMCs (B) grown in 24-well multiwell plates were incubated for 60 min with 1 or 10 μg/ml geldanamycin and 40 μM cytochrome c. Superoxide dismutase (SOD)-inhibitable reduction of cytochrome c was assessed spectrophotometrically at 550 nm.

The effects of in vivo delivery of geldanamycin to anesthetized rats (n = 5) on SNP-induced decreases in arterial blood pressure were also examined. Bolus injections of SNP in saline (0.4–0.7 ml iv, 1–3 μg/ml) were given before and 60 min after administration of 3 mg/ml geldanamycin. Baseline MAP was 114 ± 7 mmHg. SNP produced an immediate and transient decrease in systolic pressure (SAP) values, diastolic pressure (DAP) values, and MAP values that averaged –27 ± 7, –30 ± 5, and –35 ± 3 mmHg, for SAP, DAP, and MAP, respectively. After geldanamycin treatment, the same doses and volumes of SNP elicited reduced responses (–11 ± 3, –6 ± 1, and –8 ± 1 mmHg, respectively; Fig. 10). When expressed as a percent change from baseline, before geldanamycin administration, SNP produced 25 ± 2, 25 ± 6, and 25 ± 4% decreases in SAP, DAP, and MAP, respectively; these values were reduced to 10 ± 2, 18 ± 5, and 13 ± 2% decreases from baseline 1 h after geldanamycin administration. After geldanamycin administration, MAP increased maximally by 10 ± 2 mmHg or 10 ± 3% from baseline.

Fig. 10.

Effects of geldanamycin on the ability of the NO donor SNP to reduce arterial pressure in anesthetized rats. Systolic (SAP), diastolic (DAP), and mean arterial pressures (MAP) were continuously recorded from a carotid arterial catheter in anesthetized rats. Bolus injections of SNP (0.4–0.7 ml iv, 1–3 μg/ml) were given before and 1 h after administration of 3 mg/ml iv geldanamycin. Data are expressed as absolute changes in pressure (A) and percent changes from control (B). Maximal changes in resting pressures following geldanamycin administration are shown (C). Values are means ± SE, n = 5; P < 0.05, paired t-test.

DISCUSSION

The results of the present study demonstrate that at least a portion of total cellular sGC exists in a complex with eNOS and HSP90 in BAECs. Furthermore, formation of the sGC/eNOS/HSP90 complex in BAECs is stimulated by the eNOS-activating agonists BK and VEGF in a manner that depends on HSP90 being in an active conformation. Complex formation of sGC with HSP90 appears to facilitate cGMP production in cultured cells and in the whole animal by increasing the capacity of sGC to be activated by NO (see Fig. 11). This could be important when iNOS is induced by cytokines in SMCs and for intracellular signaling through the eNOS-sGC pathway in ECs. Despite the fact that sGC activity is low in ECs compared with SMCs, the enzyme is known to play a critical role in VEGF signal transduction in ECs. Both of the major consequences of VEGF stimulation of ECs, namely, increased cell proliferation and permeability, appear to be completely dependent on signaling through the eNOS-sGC pathway (24, 25, 30).

Fig. 11.

Schematic representation of sGC/HSP90 complexes. At least part of sGC is in a multiprotein complex with HSP90 and eNOS. Additional proteins may participate. We propose that this arrangement maximizes intracellular cGMP production in endothelial cells (affecting, e.g., permeability and proliferation) and smooth muscle cells (affecting, e.g., tone and proliferation) in response to endogenous or drug-released NO, minimizes NO scavenging by superoxide and formation of peroxynitrite, and possibly prolongs sGC stability by retarding its degradation. GPCR; G protein-coupled receptor; CAV1, caveolin 1; CBD, calmodulin (CAM) binding domain.

Garcia-Cardeña et al. (12) reported previously that geldanamycin (10 μg/ml) does not influence the vasorelaxant properties of the NO donor nitroglycerin on rat aortic rings, nor does it (at a 1 μg/ml concentration) influence the cGMP accumulation in response to 100 μM SNP in human umbilical vein ECs. Shah et al. (27) also reported that there is no effect of geldanamycin (3 μg/ml) on SNP-incluced vasodilation of perfused rat mesenteric vessels. In contrast, in the present study, we found that geldanamycin (1 μg/ml) inhibits SNP-induced cGMP accumulation in early-passage BAECs and RASMCs as well as SNP-induced vasodilation in anesthetized rats. The reason for this discrepancy is not clear but may relate to the differences in the model systems [BAECs and RASMCs vs. human umbilical vein ECs (HUVECs); in vivo vs. perfused mesenteric bed and aortic rings]. In experiments most comparable to the ones reported here, Garcia-Cardeña et al. (12) measured cGMP accumulation in response to SNP in HUVECs; unfortunately, they did not mention the degree of control SNP-induced sGC activation (i.e., fold increase in cGMP levels by SNP compared to vehicle); they only cite SNP-stimulated cGMP levels in the presence or absence of geldanamycin. Obviously, the degree of inhibition by geldanamycin depends on the degree of stimulation of sGC by SNP; thus if SNP only poorly stimulates sGC (as for example in late-passage cells that may express significantly less sGC), the effect of geldanamycin may be very small or even masked. We harvest and grow our own cells and use them in passage 2 or 5; they express robust amounts of sGC.

An additional point that should be made is that whereas in most cases the effects of geldanamycin on HSP90 are to reduce its activity by blocking interaction of the protein with ATP, in other cases, the compound exerts its effects by targeting HSP90-interacting proteins for degradation (28). Thus it is possible that the effects of geldanamycin on sGC in the present study may be due at least in part to sGC degradation, which suggests that additionally or alternatively, HSP90 may serve to stabilize sGC as is the case for another protein of the sGC/eNOS/HSP90 complex, Akt (2).

It appears likely that the sGC/eNOS/HSP90 complex involves several additional proteins. Already, Akt, caveolin-1, select G protein-coupled receptors, NOSIP, and dynamin-2 have been shown to associate with HSP90 directly or indirectly (6, 7, 10, 13, 22); others are currently under investigation (Fig. 11). It is therefore reasonable to speculate that for optimum cellular function, several multimeric dynamic complexes exist with components that vary according to the subcellular localization of the complex and the magnitude of physiological or pathological stimulation of the cell.

The functional significance of sGC-HSP90 interaction remains to be fully determined. Endothelial sGC plays a critical role in VEGF-induced increased cell proliferation and permeability (24, 25, 30), angiogenic processes implicated in solid tumor growth, myocardial revascularization, and diabetic retinopathies. Smooth-muscle sGC is a major regulator of vascular tone and remodeling, which are processes implicated in heart disease, hypertension, stroke, shock, and erectile dys-function. One likely reason for the existence of these complexes is that they serve to target intracellular NO produced by eNOS and iNOS directly to sGC and thus prevent inactivation of NO by reaction with superoxide anion (15) and prevent production of the highly toxic molecule peroxynitrite (see Fig. 11). Peroxynitrite formation is injurious to cells because it modifies and inactivates tyrosine residues in proteins, a process that is implicated in the pathology of human atherosclerosis, myocardial ischemia, septic and distressed lung, inflammatory bowel disease, and amyotrophic lateral sclerosis (3). Additionally, the sGC/eNOS/HSP90 interaction may be important in facilitating the actions of NO on leukocyte or platelet vessel-wall interactions including expression of adhesion proteins on the endothelium and platelet aggregation. Our in vivo data also raise the possibility of a previously unrecognized role of the HSP90-sGC complex in the optimum mechanism of action of NO donors such as SNP and nitroglycerin. More studies are needed to discern whether the observed functional effects of HSP90-sGC association are due to optimization of sGC activity, sGC stabilization, or a yet-unknown mechanism.

DISCLOSURES

This work was supported by National Institutes of Health (NIH) Grants HL-57201, HL-62152 (to R. C. Venema), and HL-52958 (to J. D. Catravas) and by an American Heart Association Grant-in-Aid (to J. D. Catravas). R. C. Venema is an Established Investigator of the American Heart Association. M. B. Harris is supported by a National Research Service Award from NIH. M. E. Maragoudakis is supported by an Institutional Research Service Award from NIH.

Footnotes

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

References

View Abstract