Hsp90 regulates NADPH oxidase activity and is necessary for superoxide but not hydrogen peroxide production

Abstract

The goal of this study was to identify whether heat-shock protein 90 (Hsp90) regulates the production of superoxide and other reactive oxygen species from the NADPH oxidases (Nox). We found that pharmacological and genetic inhibition of Hsp90 directly reduced Nox5-derived superoxide without secondarily modifying signaling events. Coimmunoprecipitation and bioluminescence resonance energy transfer studies suggest that the C-terminus of Nox5 binds to Hsp90. Long-term Hsp90 inhibition reduced Nox5 expression and provides further evidence that Nox5 is an Hsp90 client protein. Inhibitors of Hsp90 also reduced superoxide from Nox1, Nox2 (neutrophils), and Nox3. However, Nox4, which emits only hydrogen peroxide, was unaffected by Hsp90 inhibitors. Hydrogen peroxide production from the other Nox enzymes was not affected by short-term inhibition of Hsp90, but long-term inhibition reduced production of all reactive oxygen species coincident with loss of enzyme expression. Expression of chimeric Nox enzymes consisting of N-terminal Nox1 or Nox3 and C-terminal Nox4 resulted in only hydrogen peroxide formation that was insensitive to Hsp90 inhibitors. We conclude that Hsp90 binds to the C-terminus of Noxes1-3 and 5 and is necessary for enzyme stability and superoxide production. Hsp90 does not bind to the C-terminus of Nox4 and is not required for hydrogen peroxide formation.

FIG. 1.

Structurally distinct inhibitors of heat-shock protein 90 (Hsp90) suppress NADPH oxidase 5 (Nox5) activity. Superoxide production was monitored in COS-7 cells transfected with HA-Nox5. (A) Unstimulated or basal superoxide release in the presence and absence of Hsp90 inhibitors as detected by L-012. Cells were incubated with vehicle (EtOH) or geldanamycin (GA, 3 μM), 17-allylamino-17-demethoxygeldanamycin (17-AAG, 30 μM), and radicicol (RAD, 40 μM) for 30 min. Lower panel: Relative expression of Nox5 and GAPDH in total cell lysates. (B) Superoxide release in the presence and absence of RAD (40 μM) as measured by absorbance of reduced cytochrome c. (C, D) Superoxide release from cells stimulated with ionomycin or phorbol 12-myristate 13-acetate (PMA) as measured by L-012. Arrow indicates injection of ionomycin (1 μM) or PMA (100 nM). Results are presented as mean±SEM (n=4–6); *p<0.05, versus vehicle.

FIG. 2.

Hsp90 inhibitors directly affect Nox5 enzymatic activity and do not influence Nox5 phosphorylation or superoxide dismutation. (A) Superoxide was generated chemically by the addition of NADH (arrow) to phenazine methosulfate (PMS) and was measured using L-012 chemiluminescence in the presence and absence of GA (3 μM). (B) COS-7 cells expressing Nox5 were treated with vehicle or GA for 30 min. in the presence or absence of PMA (100 nM) and phosphorylation of Nox5 at Ser⁴⁹⁰, Thr⁴⁹⁴, and Ser⁴⁹⁸ were determined by western blot using phosphorylation state-specific antibodies. Representative blots from three independent experiments are shown. (C) Superoxide release from an isolated Nox5 activity assay. Partially purified Nox5 was incubated with calcium (1 mM) and FAD (100 μM) and enzyme activity was initiated with NADPH (200 μM) in the presence and absence of varying concentrations of GA. Superoxide was measured using L-012. Arrow indicates the injection of NADPH. Results are presented as mean±SEM (n=6); *p<0.05, versus vehicle.

FIG. 3.

Expression of Hsp90α and Hsp90β, but not Grp94, is necessary for Nox5-derived superoxide. HEK293 cells stably expressing Nox5 were transfected with negative control siRNA or siRNA targeting Hsp90α, Hsp90β, or Grp94 and basal superoxide was measured using L-012. (A) Western blot showing the relative expression of Hsp90α, Hsp90β, and Grp94 versus GAPDH. (B) Relative levels of superoxide from cells treated with Hsp90α, Hsp90β, and Grp94 siRNA. (C) Nox5 activity was determined in an isolated enzyme activity assay in the presence and absence of supplemental Hsp90. Arrow indicates the injection of NADPH. Results are presented as mean±SEM (n=4–6); *p<0.05, versus vehicle.

FIG. 4.

Hsp90 binds to the C-terminal domain of Nox5 and is important for enzyme stability. (A) A schematic diagram depicting the generation of a Myc-tagged C-terminal fragment of Nox5 (Myc-Nox5 C-TERM). (B) COS-7 cells were transfected with the Nox5 C-terminal construct, and 48 h later, cell lysates were immunoprecipitated using either a negative isotype control mouse immunoglobulin (IgG) or a monoclonal antibody against c-Myc. Immune complexes were immunoblotted for Myc, Hsp90α, or Hsp90β. (C) The reverse experiment was performed and lysates from Myc-Nox5 C-TERM–transfected COS cells were immunoprecipitated using control IgG or anti-Hsp90α or -Hsp90β antibodies and immunoblotted with anti-c-Myc. (D) Cells expressing Nox5 were treated with GA at the indicated concentrations for 12 h and cell lysates were immunoblotted for HA-Nox5 versus GAPDH. (E) Schematic of constructs used for bioluminescence resonance energy transfer (BRET) studies. (F, G) COS-7 cells were transfected with different ratios of Rluc-Hsp90:Venus-Nox5 or Rluc-Hsp90:Venus-CDC37 (shown on x-axis) and incubated with 5 μM coelenterazine h, and luminescence light-emission acquisition was performed immediately. Background subtracted BRET ratio is shown on the y-axis. Results are presented as mean±SEM (n=4–6).

FIG. 5.

Superoxide production from the other Nox isoforms, with the exception of Nox4, is sensitive to Hsp90 inhibition. COS-7 cells were singularly transfected with Nox5 or Nox4 or cotransfected with Nox1 or Nox3 together with NOXO1 and NOXA1. (A–D) Cells were treated with either vehicle or GA for 30 min and superoxide release was measured by L-012. (E) Superoxide release from Nox5-transfected cells in the presence of increasing concentrations of RAD. (F, G) Freshly isolated human neutrophils were exposed to GA or RAD for 30 min and superoxide release was measured using L-012. Results are presented as mean±SEM (n=4–6); *p<0.05, versus vehicle.

FIG. 6.

Hydrogen peroxide formation from Nox isoforms, with the exception of Nox4, is sensitive to chronic but not acute inhibition of Hsp90. COS-7 cells were singularly transfected with (A) Nox5 or (D) Nox4 or cotransfected with (B) Nox1 or (C) Nox3 together with NOXO1 and NOXA1. Hydrogen peroxide was measured via background subtracted Amplex red fluorescence in the presence and absence of short-term (3 μM, 1 h, left panel) or long-term (500 nM, 12 h, right panel) exposure to GA. (E) Hydrogen peroxide release from COS-7 cells transfected with Nox5 in the presence of increasing concentrations of RAD. Results are presented as mean±SEM (n=4–6); *p<0.05, versus vehicle.

FIG. 7.

The C-terminal of Nox enzymes is important for Hsp90-dependent production of superoxide and enzyme stability. (A) A schematic diagram outlining the generation of chimeric Nox enzymes consisting of the N-terminus from Nox1 or Nox3 fused to the C-terminus of Nox4. (B) Superoxide release from COS-7 cells expressing Nox1–4, Nox3–4 singularly, and Nox1 together with NOXO1 and NOXA1. (C) Hydrogen peroxide release from COS-7 cells expressing only Nox1–4, Nox3–4, or Nox1 together with NOXO1 and NOXA1 in the presence or absence of long-term (12 h) treatment with GA. Results are presented as mean±SEM (n=4); *p<0.05, versus lacZ control or vehicle.