Opposing Actions of Heat Shock Protein 90 and 70 Regulate Nicotinamide Adenine Dinucleotide Phosphate Oxidase Stability and Reactive Oxygen Species Production

Abstract

Objective

Excessive reactive oxygen species contribute to vascular dysfunction. We have previously shown that heat shock protein (Hsp90) inhibitors potently suppress Nox 1 to 3 and 5, and the goals of this study were to identify how molecular chaperones regulate Nox function.

Methods and Results

In vitro, protein expression of Nox 1 to 2, 5 was decreased by Hsp90 inhibitors in multiple cell types (human pulmonary artery endothelial cells, neutrophils, macrophages, and human saphenous vein). In mice treated with Hsp90 inhibitors, Nox1 expression was reduced in lung along with reduced reactive oxygen species from leukocytes. Elevated reactive oxygen species production in obese (db/db) aorta was suppressed by Hsp90 inhibition. Hsp90 inhibitors did not alter Nox5 micro RNA levels, and proteasome inhibition prevented Nox2 and 5 protein degradation and increased ubiquitin incorporation. Inhibition of Hsp90 upregulated the expression of Hsp70 and Hsp70-bound Nox2, 5 and promoted degradation. Silencing Hsp70 prevented Hsp90 inhibitor–mediated degradation of Nox5. The Hsp70-regulated ubiquitin ligase, CHIP, also bound Nox5 and promoted increased Nox5 ubiquitination and degradation. The chaperone binding and ubiquitination domains of CHIP were required, and the silencing of CHIP blunted Hsp90 inhibitor–mediated degradation of Nox2 and 5.

Conclusion

We conclude that Hsp90 binds to and regulates Nox protein stability. These actions are opposed by Hsp70 and CHIP, which promote the ubiquitination and degradation of Nox proteins and reduce reactive oxygen species production.

In the vasculature, reactive oxygen species (ROS), including superoxide (O₂⁻), hydrogen peroxide (H₂O₂), and derivative oxygen radicals, are generated by the nicotinamide adenine dinucleotide phosphate oxidase (Nox) family of enzymes.¹ Of the 7 Nox family members, vascular and myeloid cells express Nox1, 2, 4, and 5 (*Nox5 gene is not present in rat and mouse genomes).² Although the importance of ROS to the obligate function of immune cells has been extensively defined,^3,4 the physiological roles of the various Nox isoforms in the blood vessel wall remain largely enigmatic.⁵ What is well known is that cardiovascular diseases promote increased ROS production in blood vessels with the consequence of altered vasomotor tone, compromised endothelial function, and increased vascular cell proliferation and migration.^6,7

Considerable insight has been gained into the posttranslational regulation of Nox enzymes.^1,8 The mechanisms involved vary significantly according to Nox isoform and cell type. The amount of ROS produced can also be regulated through changes in the gene expression of Nox isoforms and subunits. Many vasoactive, prohypertensive, and proproliferative hormones have been shown to increase the expression of Nox proteins via activation of well-characterized transcriptional networks.^1,9,10 In comparison, the mechanisms regulating the protein stability of the Nox isoforms are poorly understood. Previously we identified heat shock protein 90 (Hsp90) as a novel regulator of Nox5-dependent ROS production. Loss of Hsp90 function through genetic or pharmacological approaches potently reduces Nox5-dependent ROS production.¹¹ A chaperone-dependent function of Hsp90 in controlling the folding of Nox5 is further supported by the loss of Nox5 expression in the presence of Hsp90 inhibition.

Hsps are abundant proteins that function as molecular chaperones.¹² Hsp90 regulates the function of ≈200 identified client proteins by influencing their 3-dimensional structure and stability.¹³ The majority of Hsp90 clients are transcription factors, kinases, or other signaling molecules such as the NO synthases. Inhibitors of Hsp90, such as radicicol (RAD), geldanamycin, and analogs (17-AAG, DMAG) occupy an N-terminal ATP-binding site on Hsp90 and impede client protein folding.¹⁴

Recently it has been shown that Hsp90 levels are elevated in both plaques and serum from individuals with atherosclerosis.^15,16 Inhibition of Hsp90 has been shown to suppress inflammation¹⁵ and reduce ROS production in atherosclerosis,¹⁷ but the mechanisms involved are poorly understood. A consequence of reduced Hsp90 activity is increased expression of Hsp70,¹⁸ and the selective upregulation of Hsp70 itself has been shown to provide cardiovascular protection¹⁹ suggesting that Hsp70 might mediate some of the beneficial effects of Hsp90 inhibition.

Thus, the major goals of this study were to determine whether the expression of other Nox proteins is regulated by Hsp90, to identify the role of Hsp70 in this process, and to identify the mechanisms of Nox protein degradation.

Materials and Methods

Cell Culture

COS-7, HEK293, and promyelocytic leukemia cells (HL-60) were grown in DMEM as described.^11,20 Human pulmonary artery endothelial cells and human aortic smooth muscle cells were from Cascade Biologics and grown in EBM-2 or SmBM (Clonetics). Cells were transfected as described.²¹ HA-Nox5 cell line was generated by Flp recombinase-mediated integration (Invitrogen). Cells were exposed to different concentrations of RAD, 17-AAG (Fisher Scientific), geldanamycin (Biosciences), geranylgeranylacetone (GGA), and MG-132 (Sigma) for 0.5 to 24 hours. Bone marrow–derived macrophages from WT and Nox2^−/− mice were obtained from Cell Biologics. Segments of intact human saphenous vein were obtained as discarded tissues. The procurement of these tissues conforms to the principles outlined in the Declaration of Helsinki and was approved by the Human Assurance Committee of the Georgia Health Sciences University.

Mouse Macrophage Isolation and Animal Studies

Male C57Bl/6 mice (7 to 10 weeks of age; Harlan, Indianapolis, IN) were used in all experiments. All animal studies were performed according to protocols approved by the Institutional Animal Care and Use Committee at Georgia Health Sciences University. Mouse macrophages were isolated from the peritoneum after injection of thioglycolate. In brief, 1 mL of thioglycolate was injected into each mouse 3 to 5 days before collection, peritoneal cells were collected by lavage, red blood cells lysed with hypotonic buffer, and isolated macrophages were cultured in complete RPMI medium. Hsp90 inhibitors (17-AAG, dissolved in 10% DMSO, or 17-DMAG, dissolved in saline) were administered immunoprecipitated 16 to 24 hours before euthanization and isolation of lung tissue and blood.

DNA and Adenoviral Constructs

Plasmid DNA-coding Nox5β (AF325189), Nox1, Nox3, and Nox4 were described previously.^11,20,22 The complementary DNAs for myc-CHIP, myc-CHIP H260Q, and myc-CHIP K30A were gifts from Dr Cam Patterson (University of North Carolina). Other cDNAs were obtained from Origene or Addgene. Control (RFP) and HA-Nox5 adenoviruses have been described.²²

Coimmunoprecipitation and Western Blotting Analysis

Cells were lysed on ice in 20-mmol/L Tris-HCl (pH 7.4), 1% Triton X-100, 100-mmol/L NaCl, 1-mmol/L Na₃VO₄, 10-mmol/L NaF, and 1% protease inhibitor cocktail (Sigma). Soluble extracts were incubated for 2 hours at 4°C with relevant antibodies: anti-HA (Roche Applied Science), anti-Nox2 (Sigma), anti-ubiquitin (Cell Signaling Technology), and a negative isotype control mouse immunoglobulin (Santa Cruz Biotechnology), and complexes precipitated with protein A/G agarose (Santa Cruz Biotechnology). Western blotting was performed as described previously¹¹ using anti-HA (Roche), anti-Nox1, 2 (Sigma), anti-Hsp90, myc (BD Biosciences), anti-ubiquitin, anti-GFP, Hsp70 (Cell Signaling Technology), and anti-CHIP.²³

Real-Time Polymerase Chain Reaction

Real-time polymerase chain reaction was performed as described²⁴ using human Nox5β: 5′-ccttctagttgcgcttttgc-3′ (sense) and 5′-aatggagccactgccatc-3′ (antisense), human 18S: 5′-Cagccacccgagattgagca-3′ (sense) and 5′-tagtagcgacgggcggtgtg-3′ (antisense).

Superoxide and Hydrogen Peroxide Measurement

The measurements of superoxide (L-012) and hydrogen peroxide (Amplex Red) have been described.^11,20,22 2-Hydroxyethidium (2-OH-E+), the specific product of the reaction of hydroethidine with superoxide radical anion (O₂⁻), was measured in HL-60 cells as described elsewhere²⁵ with recent modifications.²⁶ Tissue dihydroethidium staining was performed as described.²⁴ In brief, mouse aortas were treated with vehicle or RAD 20 μmol/L and embedded in OCT, sectioned, and incubated with dihydroethidium (5 μmol/L, Invitrogen).

Transient Knockdown of CHIP Gene With Small Interfering RNA

The with small interfering RNA (siRNA)-targeting CHIP (STUB1, siRNA ID: s195025) was obtained from Applied Biosystems. Validated control and targeting siRNA were transfected into HEK293 cells stably expressing Nox5 and HL-60 cells using siPORT Amine (Applied Biosystems).

Statistical Analysis

Data are reported as mean±SE, and statistical analyses were performed using Instat and a 2-tailed Student t test or ANOVA with a post hoc test where appropriate. Differences were considered as significant at P<0.05.

Results

Hsp90 Is Necessary for the Maintenance of Nox Protein Stability

To determine the effect of Hsp90 inhibition on the stability of other Nox family members, we incubated COS-7 cells expressing Nox5, human pulmonary artery endothelial cells, murine macrophages, a human neutrophil cell line (HL-60), and intact segments of human saphenous vein with Hsp90 inhibitors. We found that prolonged inhibition of Hsp90 robustly decreased the expression of Nox5 in COS-7 cells, Nox2 in macrophages and neutrophils, Nox1 and Nox2 in human pulmonary artery endothelial cells, and Nox1, 2, and 5 in human saphenous vein (Figure 1A–1E). Expression of GAPDH was not altered; densitometry for Figure 1D and 1E is shown in Figure I in the online-only Data Supplement. Hsp90 inhibitors reduced superoxide levels in WT macrophages but did not influence the dramatically reduced levels in Nox2^−/− mice (Figure 1B, right). Hsp90 reduced superoxide levels in PMA-stimulated neutrophils as determined by 2-OH-E+ levels (Figure 1C, right). In vivo, exposure of mice to the Hsp90 inhibitor 17-DMAG for 24 hours resulted in reduced Nox1 expression, but not Nox4, in the lung (Figure 1F; densitometry for Figure 1F is shown in Figure I in the online-only Data Supplement). In leukocytes isolated from mice treated with 17-AAG for 16 hours, there was a significantly reduced capacity to generate superoxide (Figure 1G). The upregulation of Hsp70 in leukocytes from mice treated with 17-AAG is indicative of Hsp90 inhibition (Figure I in the online-only Data Supplement). Aortas from obese db/db mice exhibit higher dihydroethidium oxidation levels in the media and adventitia versus that in lean mice (Figure 1H; Figure II in the online-only Data Supplement), and this was reduced in the presence of Hsp90 inhibitors.

Figure 1.

Inhibition of heat shock protein 90 (Hsp90) in vitro and in vivo results in the loss of Nox isoform expression and reduced reactive oxygen species (ROS) production. A, COS-7 cells expressing Nox5 were treated with the Hsp90 inhibitor radicicol (RAD) at the indicated concentration for 24 h, and cell lysates were immunoblotted for HA-nicotinamide adenine dinucleotide phosphate oxidase (Nox)5 and GAPDH (loading control). B, Macrophages were isolated from C57BL/6 mice and were immunoblotted for Nox2 expression. Superoxide production was measured using L-012 chemiluminescence in PMA (100 nmol/L)–stimulated macrophages from C57Bl/6 and Nox2^−/− mice in the presence and absence of RAD (20 μmol/L). C, Neutrophil-differentiated HL-60 cells were treated with or without RAD for 24 h, and cell lysates were immunoblotted for Nox2 expression. PMA (100 nmol/L)–stimulated superoxide was measured via HPLC detection of 2-OH-E+. D, Human pulmonary artery endothelial cells (HPAECs) were treated with the indicated concentration of RAD or vehicle for 24 h, and cell lysates were immunoblotted for Nox1, Nox2, and Nox4. E, Segments of human saphenous vein were incubated with RAD for 24 h and were immunoblotted for Nox1, Nox2, and Nox5 relative to GAPDH. F, C57BL/6 mice were treated with vehicle or with the Hsp90 inhibitor 17-DMAG (40 mg/kg) in saline. Mice were euthanized after 24 h, and lungs were isolated to determine the relative expression of Nox1 and Nox4. G, Blood was isolated from C57BL/6 mice treated with vehicle or with the Hsp90 inhibitor 17-AAG (10 mg/kg for 16 h), and superoxide production was determined in leukocytes using L-012 chemiluminescence in response to PMA (100 nmol/L). H, ROS production was assessed in aortas from obese (db/db) mice using dihydroethidium (DHE) staining in the presence and absence of RAD (20 μmol/L, 24 h) normalized to collagen autofluorescence. Results are representative of at least 3 to 5 separate experiments, presented as mean±SE, *P<0.05 vs vehicle.

Hsp90 Inhibition Promotes the Ubiquitination and Proteasomal Degradation of Nox5 and Nox2

To exclude the possibility that inhibition of Hsp90 decreases Nox5 protein expression via effects on CMV promoter activity or micro RNA levels, we first generated an additional expression vector. We found that RAD effectively decreased Nox5 expression driven by both CMV and EF-1α promoters and did not alter micro RNA levels (Figure II in the online-only Data Supplement; Figure 2A). Inhibition of Hsp90 did not influence the protein expression of p22phox (Figure II in the online-only Data Supplement). Confirmation of the effect of Hsp90 inhibitors on Nox protein stability was derived from the ability of the proteasome inhibitor, MG-132 to block the loss of Nox5 and Nox2 expression in the presence of the Hsp90 inhibitors, RAD (Figure 2B and 2C) and geldanamycin (Figure III in the online-only Data Supplement). Because these data suggest the involvement of the ubiquitin-proteasome pathway, we next examined whether ubiquitin overexpression can promote Nox5 degradation. In cells transfected with ubiquitin and Nox5, we observed significant loss of Nox5 protein compared with cells in control (Figure 2D). To determine whether Nox5 is ubiquitin labeled, cells were treated with or without geldanamycin in the presence of MG-132, and ubiquitin-labeled proteins were isolated by immunoprecipitation. We observed increased ubiquitin-labeled Nox5 in the presence of Hsp90 inhibition (Figure 2E). These results were confirmed by reciprocal experiments immunoisolating Nox5 and blotting for ubiquitin (Figure 2F and 2G).

Figure 2.

Reduced expression of nicotinamide adenine dinucleotide phosphate oxidase (Nox) proteins in the presence of heat shock protein 90 (Hsp90) inhibitors is mediated by posttranslational ubiquitination and proteasomal degradation. A, COS-7 cells expressing Nox5 were treated with or without radicicol (RAD) for 12 h, and Nox5 micro RNA levels were determined by real-time polymerase chain reaction (PCR) relative to 18S. B, COS-7 cells expressing Nox5 were treated with or without RAD (20 μmol/L) and MG-132 (5 μg/mL; MG-132 was added 0.5 h ahead of RAD), and cell lysates were collected after 12 h and analyzed by Western blot. C, Human neutrophils (HL-60) were incubated with or without RAD (5 μmol/L) and MG-132 for 12 h, and lysates were immunoblotted for Nox2 vs GAPDH. D, COS-7 cells were cotransfected with Nox5 and either a control (RFP) or HA-ubiquitin. Forty-eight hours later, cell lysates were immunoblotted for Nox5, HA-ubiquitin, and GAPDH. E, COS-7 cells expressing HA-Nox5 were exposed to vehicle or geldanamycin (GA; 1 μmol/L) and MG-132 (5 μg/mL) for 12 h. Cell lysates were immunoprecipitated (IP) using either a negative isotype control mouse immunoglobulin (IgG) or an anti-ubiquitin(Ub) antibody. Immune complexes were immunoblotted for ubiquitin or HA-Nox5. F and G, COS-7 cells were cotransfected with HA-Nox5 and GFP-ubiquitin and were exposed to GA (1 μmol/L) and MG-132 (5 μg/mL) for 12 h. Cell lysates were immunoprecipitated using anti-HA, and immune complexes were immunoblotted for HA-Nox5 or GFP-ubiquitin. Results are representative of ≤3 separate experiments. Real-time PCR data are expressed as percentage of 18S (mean±SE, n=6).

Hsp70 Physically Associates With Nox and Is Important for Nox Activity and Stability

Inhibition of Hsp90 resulted in the upregulation of Hsp70 in vitro in cells (Figure 3A) and in vivo (Figure I in the online-only Data Supplement). To determine whether Hsp70 is important for Nox5 regulation, we first tested whether Nox5 is a client protein of Hsp70 using coimmunoprecipitation. We found that Hsp70 physically associates with Nox5 (Figure 3B). To investigate the effect of Hsp70 on Nox5 activity and protein stability in the absence of Hsp90 inhibition, we increased Hsp70 levels by cotransfection. Increased Hsp70 decreased Nox5 activity and promoted Nox5 degradation (Figure 3C). GGA, a nontoxic antiulcer drug, has been shown to acutely increase Hsp70 expression. We exposed cells expressing Nox5 to GGA and found that upregulation of Hsp70 decreased Nox5-derived superoxide in a time- and dose-dependent manner (Figure 3D; Figure III in the online-only Data Supplement). Cells treated with GGA exhibited increased binding of Hsp70 to Nox5 (Figure 3E). We also found that Hsp70 upregulation decreased Nox1- and Nox3-derived superoxide, but not Nox4-derived hydrogen peroxide production (Figure IV in the online-only Data Supplement). To determine whether Hsp70 is necessary for Nox5 degradation in response to Hsp90 inhibition, we silenced Hsp70 using a siRNA approach. In cells with diminished Hsp70 levels, there was a reduced ability of Hsp90 inhibitors to decrease Nox5 expression (Figure 3F).

Figure 3.

Heat shock protein 70 (Hsp70) binds and is necessary and sufficient to regulate nicotinamide adenine dinucleotide phosphate oxidase (Nox5) activity and protein stability. A, COS-7 cells expressing Nox5 were treated with or without radicicol (RAD) (20 μmol/L) for 12 h, and lysates were immunoblotted with anti-HA, Hsp70, and GAPDH antibodies. B, COS-7 cells expressing Nox5 were lysed and immunoprecipitated using either control immunoglobulin (IgG) or anti-HA antibody. Immune complexes were immunoblotted for HA-Nox5 or Hsp70. C, Superoxide release from COS-7 cells cotransfected with HA-Nox5 and either control (RFP) or progressively higher amounts of Hsp70-GFP. The relative expression of Nox5, Hsp70 vs GAPDH, was determined by Western blot. D, Human aortic vascular smooth muscle cells (HASMC) expressing Nox5 (MOI of 20) were treated with geranylgeranylacetone (GGA) at the indicated time points, and unstimulated superoxide release and relative expression of Hsp70 and HA-Nox5 were determined. E, HEK cells expressing Nox5 and either a control small interfering RNA or Hsp70 siRNA were exposed to vehicle or RAD (20 μmol/L) for 12 h, and lysates were probed for expressions of HA-Nox5, Hsp70, and GADPH. F, GGA increases HSP70 binding to Nox5. HASMCs were treated with GGA (30 μmol/L) for 4 h, and Nox5 immunoprecipitated and immune complexes blotted for HA-Nox5 and Hsp70. Data are expressed as mean±SE, * P<0.05 vs control. (n=4–6). IP indicates immunoprecipitation.

CHIP Associates with Nox5

In the presence of Hsp90 inhibitors, we observed increased binding to Hsp70 and diminished binding to Hsp90 revealing a reciprocal pattern of chaperone-dependent regulation (Figure 4A). In addition to Nox5, we also examined whether this pathway was operational in cells expressing Nox2 endogenously. Human peripheral blood mononuclear cells were treated with RAD and Nox2 was immunoprecipitated. Inhibition of Hsp90 resulted in reduced amounts of Hsp90 bound to Nox2 and increased Hsp70 binding (Figure 4B). We next assessed whether the Hsp70-interacting protein (CHIP), previously identified as a regulator of protein quality-control of Hsp90-dependent client proteins, binds to Nox proteins. CHIP has a tetratricopeptide repeat (TPR) domain at its N terminus, which interacts with Hsp70, and a U-box domain at its C terminus, which contains an E3 ubiquitin ligase. To examine whether CHIP physically associates with Nox5, cells were cotransfected with HA-Nox5 and myc-tagged CHIP wild type or mutants: CHIP H260Q (a U-box domain mutant) and CHIP K30A (a TPR domain mutant). Using a Co-IP approach, we found a strong association between both the wild-type and the E3 ubiquitin ligase–deficient CHIP mutant (H260Q), whereas CHIP lacking an intact TPR domain did not associate with Nox5 (Figure 4C). These data suggest that CHIP incorporation into Nox complexes is through interaction with the TPR recognition motif of Hsp70.

Figure 4.

Inhibition of heat shock protein 90 (Hsp90) alters the balance of Hsp90 vs Hsp70 bound to nicotinamide adenine dinucleotide phosphate oxidase (Nox2) and Nox5, and the carboxy terminus of Hsp70-interacting protein, CHIP, physically associates with Nox5. A, COS-7 cells expressing Nox5 were treated with or without radicicol (RAD) for 3 h. Cell lysates were immunoprecipitated using anti-HA antibody, and immune complexes were immunoblotted for HA-Nox5 and bound Hsp90 or Hsp70. B, Human peripheral blood mononuclear cells (PBMC) were treated with RAD (20 μmol/L) for 3 h. Cells were lysed, and Nox2 was immunoprecipitated. Immune complexes were immunoblotted for Nox2, Hsp70, and Hsp90. C, COS-7 cells were cotransfected with HA-Nox5 and myc-CHIP WT, myc-CHIP H260Q (U-box mutant), or myc-CHIP K30A (TPR domain mutant). Cell lysates were immunoprecipitated using control immunoglobulin (IgG) or anti-HA antibody and were immunoblotted with anti–HA-Nox5 or anti-myc(CHIP) antibodies. Results are representative of ≤3 separate experiments. IP indicates immunoprecipitation.

CHIP Induces the Degradation of Nox5 and Reduces Superoxide Production

Increased expression of CHIP, but not either of the CHIP mutants, dose-dependently decreased Nox5-dependent ROS production and protein expression (Figure 5A–5F). To broaden the functional significance of CHIP in regulating the activity and stability of other Nox isoforms, we transfected CHIP together with Nox1, Nox3, and Nox4 and found that CHIP decreased the activity of Nox1 and Nox3, whereas ROS production from Nox4 was unaffected (Figure IV in the online-only Data Supplement). The increased expression of wild-type CHIP promoted Nox5 ubiquitination, an effect that was absent in cells expressing either of the CHIP mutants (Figure 5G). To determine for the role of endogenous CHIP in the degradation of Nox proteins, we silenced CHIP expression using siRNA (Figure 6A). Reduced expression of CHIP abrogated the degradation of Nox5 induced by Hsp90 inhibitors (Figure 6B and 6C). Further, the ability of Hsp90 inhibitors to induce Nox5 ubiquitination was reduced in cells with diminished levels of CHIP (Figure 6D and 6E). To determine whether CHIP is necessary to mediate the effects of increased Hsp70 expression on Nox5 activity, we silenced CHIP and induced Hsp70 expression in cells expressing Nox5. In cells treated with a control siRNA, GGA decreased the ability of Nox5 to generate superoxide. In contrast, cells treated with CHIP siRNA reduced the expression levels of CHIP, and superoxide production was unaffected by GGA treatment (Figure 6F). We also investigated whether CHIP can regulate the expression of other Nox enzymes. In a human neutrophil cell line, silencing CHIP reduced the ability of Hsp90 inhibition to decrease endogenous Nox2 expression (Figure 6G). In addition to CHIP, there are multiple chaperone-associated E3 ubiquitin ligases, including Cullin5, valosin-containing protein, Parkin, and MDM2 that are stimulated by the inhibition of Hsp90^27-30. To determine whether additional E3 ubiquitin ligases regulate Nox5 activity and stability, cells expressing Nox5 were transfected with plasmids encoding Cullin5, valosincontaining protein, Parkin, and MDM2 or siRNA to Cullin5 in the presence of Hsp90 inhibition. None of these ligases had a pronounced effect on Nox5 activity or protein expression, suggesting that they are not the ubiquitin ligases that target Nox5 (Figure V in the online-only Data Supplement).

Figure 5.

CHIP promotes nicotinamide adenine dinucleotide phosphate oxidase (Nox5) ubiquitination and subsequent degradation. A–C, COS-7 cells were cotransfected with HA-Nox5 and increasing amounts of CHIP WT, CHIP H260Q, or CHIP K30A, and unstimulated superoxide release was determined by L-012 chemiluminescence. D–F, Cell lysates were immunoblotted for relative levels of Nox5, myc-CHIP, and GAPDH. G, COS-7 cells were cotransfected with HA-Nox5, GFP-ubiquitin and CHIP WT, CHIP H260Q, or CHIP K30A, and were exposed to MG-132 (5 μg/mL) for 12 h. Cell lysates were immunoprecipitated using anti-HA antibody and immunoblotted with anti-GFP or anti-HA antibodies. Results are representative of ≤3 separate experiments. Data are expressed as mean±SE, * P<0.05 vs vehicle (n=4–6).

Figure 6.

Silencing CHIP ameliorates the loss of nicotinamide adenine dinucleotide phosphate oxidase (Nox) protein stability and increased ubiquitination induced by heat shock protein 90 (Hsp90) inhibition or induction of Hsp70. HEK cells stably expressing Nox5 were transfected with negative control small interfering RNA (siRNA) or siRNA-targeting CHIP, for 48 h. Vehicle or progressively higher doses of radicicol (RAD) were added to cells for 12 h in the presence and absence of MG-132 (5 μg/mL). A, Lysates were immunoblotted for CHIP. B, HA-Nox5 and GAPDH. C, Relative densitometry of Nox5 vs GAPDH expression. D, HEK cells stably expressing Nox5 were transfected with control siRNA or CHIP siRNA, and cells were treated with or without RAD for 12 h. Nox5 was immunoprecipitated using an anti-HA antibody and immunoblotted for Nox5 and ubiquitin. E, Relative densitometry of Nox5 ubiquitination. Data are expressed as percentage of control (means±SE, n=3, * P<0.05 vs Vehicle). F, HEK-Nox5 cells were transfected with scrambled or CHIP siRNA and exposed to GGA (0–50 μmol/L) for 4 h, and basal superoxide levels were measured with L-012. G, Neutrophil-differentiated HL-60 cells were transfected with scrambled or CHIP siRNA and were exposed to vehicle or RAD (5–20 μmol/L) for 16 h, and cell lysates were immunoblotted for Nox2, CHIP, Hsp70, and GAPDH. Results are representative of ≤3 separate experiments.

Discussion

Hsp90 and cochaperones regulate the tertiary structure of select client proteins such as nuclear receptors, kinases, and NO synthases to enable specific enzyme functions.¹² We have previously shown that the binding of Hsp90 to Nox5 regulates ROS production,¹¹ and the current study shows that the protein expression of Nox1, Nox2, and Nox5 depends on active Hsp90. Inhibition of Hsp90 both in vitro and in vivo reduced Nox expression and ROS production in isolated leukocytes, lung tissue, human blood vessels, and aorta from db/db mice. These results support the hypothesis that Hsp90 is integral to the function and stability of the Nox family of enzymes.

One exception to this paradigm was the inability of Hsp90 inhibitors to alter Nox4 expression and activity. We have previously found that the C-terminal region of Nox5 mediates the binding to Hsp90, and in domain-swapping experiments we have shown that the equivalent region in Nox4 is sufficient to provide resistance to Hsp90 inhibitors in Nox enzymes that are otherwise sensitive.¹¹ Expression levels of p22phox can influence the stability of Nox proteins, but the inability of p22phox to influence Nox5, which is sensitive to Hsp90 inhibitors and the lack of effect of Hsp90 inhibitors on p22phox levels, suggest that indirect regulation via p22phox is not a major mechanism. The insensitivity of Nox4 to Hsp90 inhibition suggests that functions conferred by Hsp90 binding to other Nox enzymes are not needed for Nox4. Indeed, Nox4 has enzymatic characteristics unique from the other family members. It is regarded as constitutively active^31,32 and emits only detectable hydrogen peroxide instead of superoxide.^33,34 The mechanisms by which Hsp90 supports the stimulus-driven activity of Noxes 1, 2, 3, and 5 and the release of superoxide remain to be determined.

The upregulation of Hsp70 is a well-described outcome of Hsp90 inhibition and is mediated by the loss of Hsp90-dependent repression of HSF1.¹⁸ To delineate a distinct role for Hsp70 in the absence of Hsp90 inhibition, we increased Hsp70 levels via transfection of a GFP-tagged transgene. We found that the elevation of Hsp70 expression alone reduced ROS production from Nox5 and decreased Nox5 expression. These results suggest a direct interaction between Hsp70 and Nox5, and in Co-IP experiments we indeed found that Hsp70 binds to Nox5. This is the first evidence that Hsp70 is an essential component of Nox complexes. In addition to Nox5, we also found that Hsp70 decreases ROS production from Nox1 and Nox3, but not Nox4, and that Hsp70 binds to Nox2. These results further suggest that Hsp70 is a negative regulator of other Nox enzymes, and there is considerable evidence in the literature to support the protective role of Hsp70 against oxidative damage^35,36 and the ability to negatively regulate Nox enzyme stability, and ROS production may contribute, at least in part, to the beneficial effects of Hsp70. These data are supported by previous studies showing that chemical or thermal induction of Hsps in neutrophils reduces ROS production.³⁷ Mitogens and the protein kinase C agonist, PMA, have been shown to activate Noxes 1, 2, 3, and 5, and the ability of these stimuli to upregulate Hsp70³⁸ would also provide a mechanism for feedback inhibition. This feedback inhibition can be taken further because ROS themselves are potent inducers of Hsp70 expression,³⁹ which would promote a reduction in nicotinamide adenine dinucleotide phosphate-dependent ROS production.

Protein quality control is a vital process that determines whether proteins are properly folded and targets misfolded proteins for degradation. Molecular chaperones, including Hsp90, Hsp70, and cochaperones, are important contributors to protein quality control.⁴⁰ Hsp90 regulates the folding and function of >200 known client proteins, and inhibition of Hsp90 promotes the degradation of many of these clients, including Akt, PDK, NO synthases, glucocorticoid receptor, sGC, and CFTR.^23,41-44 Decreased function of Hsp90 results in the upregulation of Hsp70, which recruits specialized chaperone-directed ubiquitin ligases to promote client protein ubiquitination and degradation.⁴⁰ Our current study shows that both Hsp90 and Hsp70 bind to Nox5, suggesting distinct roles for Hsp90 and Hsp70 in the regulation of Nox protein activity and stability. In the presence of Hsp90 inhibitors, we found a decreased association between Nox5 and Hsp90 and increasing binding to Hsp70. These data indicate that the ratio of Hsp90 and Hsp70 bound to Nox proteins is a critical mechanism regulating ROS production. Disruption of the Hsp90:Nox interaction or facilitation of the Hsp70:Nox interaction may be an effective strategy to suppress Nox-derived ROS and might have a clinical benefit in cardiovascular disease and inflammation.

The results of our study also indicate that ubiquitination is a novel and important posttranslational modification of Nox5 and that the ubiquitin-proteasome pathway regulates Nox5 degradation. The ubiquitination of proteins is orchestrated by ≤3 types of enzymes: an E1 ubiquitin-activating enzyme, an E2 ubiquitin-conjugating enzyme, and an E3 ubiquitin ligase. The E3 ubiquitin ligases catalyze the transfer of ubiquitin to substrate proteins which are then recognized by the proteasome and degraded. The E3 ubiquitin ligases are a large and diverse family of enzymes and there are several, including CHIP, Cullin5, valosin-containing protein, Parkin, and MDM2 that have been shown to be regulated by molecular chaperones.^27-30 The best characterized of these, CHIP, possesses a TPR domain at its N terminus, which interacts with molecular chaperones Hsp70/90, and a U-box domain at its C terminus, which contains an E3 ubiquitin ligase domain. Both the TPR and U-box domains of CHIP are necessary for the degradation of Hsp90 client proteins.^23,41-45 However, CHIP does not promote the degradation of all clients of Hsp90. It promotes the partitioning of endothelial NO synthase from a soluble to an insoluble and inactive cellular compartment,⁴⁶ and more surprisingly, in fibroblasts, it can increase the refolding of proteins via its TPR domain.⁴² We found that CHIP associates in a complex with Nox5 and regulates the activity and stability of Nox5 and Nox2. Both the TPR and U-box domains of CHIP are necessary to induce the degradation of Nox5, which suggests that this process is chaperone dependent. Other chaperone-dependent E3 ligases, Cullin5, valosin-containing protein, Parkin, and MDM2 did not appear to play a major role in the regulation of Nox protein stability. Collectively, these data demonstrate that chaperone and CHIP-dependent ubiquitination is a novel mechanism regulating Nox activity and stability.

Despite significant advances in the therapeutic approach, cardiovascular diseases remain the leading cause of morbidity and mortality in the United States.⁴⁷ Elevated Hsp90 levels in individuals with atherosclerosis^15,16 would be supportive of increased Nox activity and ROS production and contrast from the known protective effects of Hsp70 against oxidative damage and ROS levels. We found that altering the balance of Hsp90 to Hsp70 bound to Nox5 could alter protein stability and ROS production. In blood vessels from animal models of obesity and type II diabetes mellitus, elevated ROS levels secondary to increased Nox1 mediate dysfunction of the endothelium and impaired vasomotor control.²⁴ We found that inhibition of Hsp90 reduces ROS production in blood vessels of both lean and obese mice. Although Hsp90 inhibitors have been shown to be powerful anti-inflammatory agents and reduce ROS production in atherosclerosis, clinical efficacy is constrained by side effects. This is not surprising given the multitude of proteins regulated by Hsp90 including endothelial NO synthase⁴⁸ and soluble guanylate cyclase,⁴⁹ which mediate endothelium-dependent relaxation. Therefore, a greater understanding of the mechanisms involved would enable more selective approaches to targeting Hsp90 client proteins and more effective therapeutics for the treatment of inflammation and cardiovascular diseases.

Conclusions

This study reveals a novel mechanism regulating the activity and stability of Nox proteins via the molecular chaperones, Hsp90 and Hsp70. The ratio of Hsp90 and Hsp70 bound to Nox proteins is a critical mechanism regulating the proper folding and ROS production and guides Nox protein triage decisions. Accordingly, disruption of the Hsp90:Nox interaction or facilitation of Hsp70:Nox binding might be an effective strategy to suppress ROS. We have also identified ubiquitination as a novel mechanism and an important posttranslational modification of Nox proteins that is regulated, at least in part, by the molecular chaperone and cochaperone CHIP. Further studies are necessary to elucidate the significance of this pathway in disease conditions.