Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2016 Feb 1;35(3):319-34.
doi: 10.15252/embj.201592394. Epub 2016 Jan 7.

Targeted redox inhibition of protein phosphatase 1 by Nox4 regulates eIF2α-mediated stress signaling

Celio X C Santos  1 , Anne D Hafstad  2 , Matteo Beretta  1 , Min Zhang  1 , Chris Molenaar  1 , Jola Kopec  3 , Dina Fotinou  3 , Thomas V Murray  1 , Andrew M Cobb  1 , Daniel Martin  1 , Maira Zeh Silva  3 , Narayana Anilkumar  1 , Katrin Schröder  4 , Catherine M Shanahan  1 , Alison C Brewer  1 , Ralf P Brandes  4 , Eric Blanc  5 , Maddy Parsons  6 , Vsevelod Belousov  7 , Richard Cammack  8 , Robert C Hider  8 , Roberto A Steiner  6 , Ajay M Shah  9
Affiliations

Targeted redox inhibition of protein phosphatase 1 by Nox4 regulates eIF2α-mediated stress signaling

Celio X C Santos et al. EMBO J. .

Abstract

Phosphorylation of translation initiation factor 2α (eIF2α) attenuates global protein synthesis but enhances translation of activating transcription factor 4 (ATF4) and is a crucial evolutionarily conserved adaptive pathway during cellular stresses. The serine-threonine protein phosphatase 1 (PP1) deactivates this pathway whereas prolonging eIF2α phosphorylation enhances cell survival. Here, we show that the reactive oxygen species-generating NADPH oxidase-4 (Nox4) is induced downstream of ATF4, binds to a PP1-targeting subunit GADD34 at the endoplasmic reticulum, and inhibits PP1 activity to increase eIF2α phosphorylation and ATF4 levels. Other PP1 targets distant from the endoplasmic reticulum are unaffected, indicating a spatially confined inhibition of the phosphatase. PP1 inhibition involves metal center oxidation rather than the thiol oxidation that underlies redox inhibition of protein tyrosine phosphatases. We show that this Nox4-regulated pathway robustly enhances cell survival and has a physiologic role in heart ischemia-reperfusion and acute kidney injury. This work uncovers a novel redox signaling pathway, involving Nox4-GADD34 interaction and a targeted oxidative inactivation of the PP1 metal center, that sustains eIF2α phosphorylation to protect tissues under stress.

Keywords: Nox4; eIF2α; metal center; protein phosphatase; redox signaling.

PubMed Disclaimer

Figures

Figure 1
Figure 1. Nox4 selectively regulates ATF4 during ER stress
  1. A

    Tunicamycin (Tn, 2 μg/ml) increased Nox4 protein levels in H9c2 cells. Tubulin was used as a loading control. n = 4–6/group. *, significant compared to baseline.

  2. B

    Thapsigargin (Tp, 1 μg/ml) increased Nox4 protein levels in H9c2 cells. n = 4–6/group. *, significant compared to baseline.

  3. C

    Effect of Nox4 on the unfolded protein response. Nox4 was depleted in H9c2 cells by shRNA‐mediated knockdown (Ad.shNox4), or cells were treated with a control adenovirus (Ad.Ctl). In cells with Nox4 knockdown, tunicamycin treatment resulted in lower increases in protein levels of the ER chaperones Grp94, Grp78, and calreticulin than in control cells. Nuclear protein levels of ATF4 were substantially lower in Nox4‐depleted cells than in control cells, but the levels of cleaved ATF6 (ATF6c) were similar. Histone was used as a loading control. The relative mRNA levels of Xbp1‐s (a readout of IRE1 signaling) were unaltered after Nox4 knockdown. Mean data are shown in Appendix Fig S1E. Similar results were obtained with an independent siRNA approach (Appendix Fig S2A).

  4. D

    Effect of adenoviral‐mediated overexpression of Nox4 (Ad.Nox4) or a control β‐galactosidase protein (Ad.β‐Gal) on tunicamycin responses of H9c2 cells. Nox4 enhanced the increase in cellular ER chaperones and nuclear ATF4 levels but did not affect tunicamycin‐induced changes in nuclear ATF6c levels and caused minor reduction in Xbp1s mRNA levels. Mean data are shown in Appendix Fig S2B.

  5. E, F

    Effect of Nox4 knockdown or overexpression, respectively, on the tunicamycin‐induced changes in mRNA levels of ATF4 target genes. n = 4/group. Psat1, phosphoserine aminotransferase; Phgdh, 3‐phosphoglycerate dehydrogenase; Asns, asparagine synthetase; Slc6a9, glycine transporter 1.

  6. G

    Effect of ATF4 silencing with two different siRNAs on Nox4 protein levels in tunicamycin‐treated H9c2 cells. Scrambled siRNAs were used as a control (Ctl). Representative immunoblots shown to the top (captions at bottom of bar graphs refer also to the immunoblots); tubulin was used as a loading control. n = 4/group. *, significant compared to baseline; #, significant comparing siATF4 versus corresponding siCtl.

  7. H

    Effect of ATF4 overexpression on Nox4 mRNA and protein levels. n = 3/group.

Data information: All blots are representative of at least three independent experiments. Data are presented as mean ± SEM. Comparisons in (A, B) were made by one‐way ANOVA and in other panels by Student's t‐test. P < 0.05 was considered significant. Values above bar graphs denote the level of significance.
Figure 2
Figure 2. Nox4 selectively inhibits PP1 activity and prolongs eIF2α phosphorylation
  1. A

    The knockdown of endogenous Nox4 resulted in a substantial inhibition of tunicamycin‐induced eIF2α phosphorylation in H9c2 cells, with no change in phospho‐Thr980‐PERK (PERK‐P) levels. GADD34 levels were significantly decreased after Nox4 knockdown, while there was no change in PP1 protein levels.

  2. B

    Overexpression of Nox4 in H9c2 cells caused prolongation of tunicamycin‐induced eIF2α phosphorylation, with minimal change in phospho‐PERK levels.

  3. C, D

    Mean levels of phosphorylated eIF2α relative to total eIF2α protein in tunicamycin‐treated cells after Nox4 knockdown or overexpression, respectively. n = 3/group. *, significant compared to baseline; #, significant comparing Nox4 knockdown (Ad.shNox4) or overexpression (Ad.Nox4) versus corresponding controls (Ad.Ctl or Ad.β‐Gal, respectively).

  4. E, F

    Effect of Nox4 knockdown or overexpression, respectively, on okadaic acid‐resistant Ser/Thr phosphatase activity in membrane fractions of tunicamycin‐treated H9c2 cells. n = 4/group. *P < 0.05, **P < 0.01 cf. baseline; # P < 0.05, ## P < 0.01 comparing Nox4 knockdown (Ad.shNox4) or overexpression (Ad.Nox4) versus corresponding controls (Ad.Ctl or Ad.β‐Gal, respectively).

  5. G, H

    Nox4 knockdown or overexpression, respectively, had no effect on the phosphorylation of glycogen synthase at Ser641 (GS‐P) or histone H3 at Ser57 (H3‐P) in H9c2 cells.

  6. I

    Nox4−/− MEF cells (KO) showed blunted tunicamycin‐induced increases in levels of phospho‐eIF2α, ATF4, and ER chaperones as compared to wild‐type (WT) MEFs, a response that was rescued by reintroduction of Nox4 (KO + Nox4). The latter had no effect on GS‐P or H3‐P levels.

Data information: All blots are representative of at least 3 independent experiments. Data are presented as mean ± SEM. Comparisons were made by ANOVA and P < 0.05 was considered significant. Values above bar graphs denote the level of significance. Mean data from quantification of immunoblots are shown in Supplementary Appendix Fig S3.
Figure 3
Figure 3. Nox4 binds to GADD34 to mediate spatially localized PP1 inhibition and enhance eIF2α phosphorylation

  1. Subcellular localization of Nox4. Tunicamycin (Tn 2 μg/ml, 6 h) increased Nox4 levels in H9c2 cells as assessed by spinning disk confocal microscopy (scale bar, 10 μm). 3D SIM images (scale bars, 2 μm) showed localization of Nox4 (green) to the ER, which was labeled with an anti‐KDEL antibody (red). At higher magnification (right), yellow dots denote co‐localization of Nox4 and KDEL signals. Cell nuclei were stained with DAPI (blue). 1 Z slice from 3D stack is shown.

  2. Progressive enrichment of GADD34, PP1, eIF2α, and Nox4 in membrane fractions of tunicamycin‐treated H9c2 cells.

  3. After sucrose gradient fractionation of lysates of tunicamycin‐treated H9c2 cells, GADD34, PP1, eIF2α and Nox4 co‐eluted in fractions 12 and 13 (F12, F13).

  4. Immunoprecipitation (IP) of pooled fractions 12/13 with an anti‐GADD34 antibody revealed the presence of both PP1 and Nox4.

  5. The association of Nox4 with GADD34 was validated in HEK293 cells co‐transfected with Nox4 and either Flag‐tagged or non‐tagged GADD34, followed by IP with an anti‐Flag antibody.

  6. Co‐transfection of HEK293 cells with GADD34‐Flag and different myc‐tagged Nox4 constructs, followed by IP with an anti‐myc antibody. GADD34 binds to full‐length Nox4 (FL) and the Nox4 transmembrane domain (TD), but not the C‐terminal domain (CD).

  7. Representative pseudocolor images of simultaneous ER and cytosolic ROS measurement with HyPer‐ER and HyPer‐Red Cyto, respectively, in tunicamycin‐treated MEF cells. Redox‐insensitive mutant probes were used as negative controls and to exclude pH changes. Extracellular H2O2 (200 nM) was added as a positive control. The pseudocolor scale is shown along the left vertical edge of each image. KO = Nox4−/−. Scale bars, 2 μm.

  8. Transfection of HEK293 cells with PP1 and GADD34 increased PP1 activity (bar graph) and decreased phospho‐eIF2α levels (immunoblots). Co‐transfection of full‐length Nox4 reduced PP1 activity and increased phospho‐eIF2α levels (captions at bottom refer both to the bar graphs and immunoblots). These effects were abrogated when either Nox4 P437H or the Nox4 transmembrane domain (TD) was transfected. Nox4 did not affect phosphorylation of glycogen synthase (GS‐P) or histone H3 (H3‐P). Data are presented as mean ± SEM.

Data information: All experiments were performed with n = 3/group. Values below the immunoblots are mean ± SEM levels for phospho‐eIF2α/total‐eIF2α. *P < 0.05 comparing third and fourth lanes; # P < 0.05 compared to lane 4. All comparisons were made by Student's t‐test, with P < 0.05 considered significant; levels of significance for comparisons of PP1 activities are shown above the bar columns. See also Fig EV1 and Appendix Fig S4.
Figure EV1
Figure EV1. Association of Nox4 with GADD34 at the ER
  1. A

    Confocal microscopic images of U2OS cells co‐transfected with Myc‐tagged Nox4 and Flag‐tagged GADD34 showed a co‐localization of Nox4 (red) and GADD34 (green). Yellow dots in the top panel denote co‐localization of fluorescence signals. Scale bars, 1 μm.

  2. B

    Confocal microscopic images of U2OS cells co‐transfected with Myc‐tagged Nox4 and Flag‐tagged GADD34 as in (A), then stained for GADD34 (red) and KDEL (green) as an ER marker. GADD34 co‐localized with KDEL. Scale bars, 5 μm.

  3. C

    Confocal microscopic images of H9c2 cells treated with tunicamycin (Tn, 2 μg/ml, 6 h), showing increased endogenous levels and co‐localization between Nox4 (green) and GADD34 (red). Scale bars, 10 μm.

  4. D, E

    Mean ± SEM data for changes in HyPer‐ER and HyPer‐Red Cyto fluorescence from 3 independent cell preparations/group and at least 12 cells imaged/preparation. *, significant comparing Tn versus basal; #, significant comparing KO versus other groups. Extracellular H2O2 (200 nM) was added as a positive control, and the changes in HyPer‐ER and HyPer‐Red Cyto fluorescence were quantified.

  5. F

    ROS levels were increased in HEK293 cells transfected with full‐length Nox4 (Nox4FL), but not with Nox4 transmembrane domain (Nox4 TD) or a Nox4(P437H) mutant. ROS levels were measured by HPLC‐based quantification of the dihydroethidium (DHE) oxidation products, 2‐hydroxyethidium (EOH) and ethidium (E). n = 4/group. *, significant comparing full‐length Nox4 versus control (Ctl); #, significant comparing full‐length Nox4 versus Nox4 TD or Nox4 P437H.

Data information: Data are presented as mean ± SEM. Comparisons were made by Student's t‐test or one‐way ANOVA, with P < 0.05 considered significant. Values above bar graphs denote the level of significance.
Figure EV2
Figure EV2. X‐ray crystallographic features of the PP1 metal coordination center and its redox regulation
  1. A

    Purified recombinant PP1 was dose‐dependently inhibited by H2O2.

  2. B

    Overall representation of the PP1 catalytic domain. The catalytic Mn‐Mn or Mn‐Fe metal ions shown as magenta spheres are located in a shallow groove at the molecular surface. PP1 side chains coordinating the metal ions and forming the base of the groove are highlighted in black. A phosphate ion, shown in stick representation with phosphate and oxygen atoms, colored cyan and red, respectively, is bound to the dinuclear center.

  3. C

    The X‐ray fluorescence emission spectrum of PP1 crystals reveals a mixture of Mn and Fe ions with the latter ion typically present in a lower amount. The ratio for the sample shown here is Mn:Fe 1:0.175. No other metals are present at significant concentration. The emission spectrum in the 4–8 keV energy region is shown by the gray thin line. Fitting for the individual Mn (purple dotted line, expected emission energies 5,899 eV and 6,491 eV) and Fe (green dotted line, expected emission energies 6,404 eV and 7,058 eV) metals as well as the total fit (black continuous line) are also shown. Fitting was carried out with the package PyMca (Solé et al, 2007). Excitation energy was 18 keV.

  4. D

    Cartoon representation of the PP1 active site with the dinuclear (M1, M2) metal center represented by purple spheres. The metal ions are pseudo‐octahedrally coordinated. Residues at coordinating distance (D64, H66, D92, N124, H173, H248) as well as the catalytically important H125 hydrogen‐bonded to the phosphate moiety are shown as sticks. P, N, C, O atoms are in cyan, blue, white, and red, respectively. The bridging μ‐OH − and the terminal water (W) are represented by small spheres. In crystallo metal analysis shows the presence of both Mn and Fe metals with the former being the most abundant. Anomalous difference maps (shown in purple at the +8σ level) calculated from data collected at the 6,876.6 eV energy where the Fe anomalous contribution is negligible indicate that Mn is present at both M1 and M2 centers. An attempt to specifically locate Fe ions using a double difference anomalous map approach (Than et al, 2005) was unsuccessful likely owing to the low Fe content.

  5. E

    Correlation between change in metal–ligand (M–L) coordination distances determined by X‐ray experimental and theoretical methods. There is a good correlation between theory and experiment with 75% of the Δ values (black circles) lying on the diagonal within error while three values (red circles) can be considered outliers. Most points lie on the lower‐left quadrant, implying a contraction of the average (M–L) distance upon metal oxidation. See Appendix Table S1 and related text for further details.

  6. F

    PP1 cysteine residues Cys127 (A) and Cys273 (B) were often seen oxidized to their sulfenic acid (CSO) derivative. Occasionally, Cys291 is also oxidized. 2mFo‐DFc electron density maps are shown in blue at the 1.1σ level. S, C, O atoms are in yellow, white, and red, respectively. Cys oxidation appears independently of H2O2 treatment as we have observed sulfenic acid derivatization also in ascorbate‐treated crystals. Cysteine oxidation does not affect PP1 catalysis as the PP1 C127, 273S double variant displays the same activity as WT PP1 (see Fig 4A).

  7. G, H

    Effect of H2O2 on the activity of WT PP1 and PP1 variants bearing mutations in amino acids involved in metal coordination, N124D and D64N. The pH optimum for each PP1 variant was determined by assessing PP1 activity over a range of pH values (buffers: 100 mM Tris–HCl pH 6–8 and 100 mM glycine–NaOH pH 9–10) (G). The pKa for WT PP1 and N124D PP1 was 7.2, and for D64N PP1, it was 9.8. The PP1 variants were incubated with 0.1 mM H2O2 for 20 min at 37°C, and PP1 activity was assessed at optimum pH (H). The red bars show PP1 activity under reduced conditions and the blue bars after H2O2 treatment. The relative inhibition by H2O2 was substantially increased in the mutant proteins.

  8. I

    Low‐temperature EPR spectra of purified PP1 at baseline and after treatment with H2O2 (1 mM). The central six‐line signal from 2.5 to 4.5 kG is consistent with Mn2+. After H2O2 treatment, there is a consistent signal in the region of 1.5 to 2 kG, which is typical for Fe(III).

  9. J

    Effect of ascorbate (Asc, 0.5 mM) on eIF2α phosphorylation in tunicamycin (Tn)‐treated H9c2 cells with overexpression of Nox4 (Ad.Nox4). The Nox4‐induced increase in eIF2α phosphorylation after 4 h of Tn was inhibited by ascorbate (see also Fig 4F).

Data information: Data are means ± SEM or representative of at least three independent experiments.
Figure 4
Figure 4. Redox inhibition of PP1 involves metal center oxidation

  1. Recombinant PP1 was inhibited by H2O2 (0.2 mM) and activity was not restored by glutathione (GSH), cysteine (Cys), or dithiothreitol (DTT). A Cys127Ser/Cys273Ser PP1 mutant was inhibited by H2O2 similarly to wild‐type PP1. Values above bars denote level of significance for the inhibitory effect of H2O2.

  2. Ascorbate (Asc) dose‐dependently restored PP1 activity. #, significant effect of Asc compared to H2O2 alone.

  3. EPR spectra of PP1 incubated with ascorbate (1 mM) alone (a) or PP1 exposed to H2O2 followed by catalase treatment, then incubation with ascorbate (b). (b) shows a typical spectrum for the ascorbyl radical (hyperfine splitting constant, aH = 1.8 G), similar to the positive control obtained by exposing ascorbate to H2O2 (c). (d) shows that no ascorbyl radical is detected if H2O2 is degraded by catalase in the absence of PP1, prior to ascorbate addition.

  4. Cartoon representation of the active site of H2O2‐treated PP1 as in Fig EV2D. 2mF o‐DF c electron density map at the 2.2‐Å resolution is shown in yellow at the 1.1σ level. H2O2 treatment causes an overall shrinkage of the PP1 coordination sphere by 0.12 Å compared to ascorbate‐treated crystals consistent with the oxidation of the dinuclear center. This increases the energy barrier for the catalytic steps involving μ‐OH − attack on the phosphorous center of the bridging phosphate and rupture of the P‐O scissile bond with the assistance of H125 (black arrows). Reported coordination distances in Å are averaged over the two PP1 molecules in the crystallographic asymmetric unit. See also Appendix Table S1.

  5. Effect of Nox4 on GADD34/PP1‐mediated eIF2α dephosphorylation in transfected HEK293 cells. Nox4 increased eIF2α phosphorylation in cells transfected with GADD34 and WT PP1, and resulted in even higher phospho‐eIF2α levels in cells transfected with N124D or D64N PP1 variants.

  6. Effect of ascorbate (Asc, 0.5 mM) on phosphatase inhibition in tunicamycin‐treated H9c2 cells with overexpression or knockdown of Nox4 (Ad.Nox4 and Ad.shNox4, respectively). In control cells, ascorbate enhanced tunicamycin‐stimulated increases in phosphatase activity. Phosphatase activity was lower in Nox4‐overexpressing than control cells but was normalized by ascorbate to the same level as in control cells. In Nox4 knockdown cells, tunicamycin‐induced increases in phosphatase activity were enhanced and ascorbate had minimal additional effect.

Data information: All experiments were performed with n = 3/group. Data are presented as mean ± SEM. Comparisons were made by Student's t‐test or one‐way ANOVA, with P < 0.05 considered significant. Values above bar graphs denote the level of significance. See also Fig EV2..
Figure 5
Figure 5. Nox4 enhances cell survival and protects hearts against I/R injury through increase in eIF2α phosphorylation

  1. H9c2 cells treated with tunicamycin (2 μg/ml, 12 h) showed significantly lower survival when endogenous Nox4 was silenced (siNox4) as compared to cells treated with a scrambled siRNA (siCtl). Cell survival was restored by treatment with either guanabenz (Gbz, 5 μM) or salubrinal (Sal, 50 μM) but was unaffected by clonidine (Cld, 5 μM). n = 3/group.

  2. Nox4‐depleted cells had lower levels of phospho‐eIF2α and ATF4 than control cells, but these were restored in the presence of guanabenz (Gbz).

  3. Schematic representation of the effect of the small molecule inhibitors, guanabenz and salubrinal, on the GADD34/PP1/eIF2α interaction.

  4. Hearts from Nox4 knockout (KO) mice and WT controls were subjected to global ischemia followed by aerobic reperfusion (I/R). Infarct size assessed by triphenyltetrazolium chloride (TTC) staining was greater in Nox4 KO hearts compared to WT and was significantly reduced by guanabenz (Gbz). In the representative heart sections shown at the top, white denotes infarct area and red the viable area. Scale bars, 1 mm. Numbers of hearts are indicated within the bars.

  5. Immunoblotting of heart homogenates after I/R showed lower levels of phospho‐eIF2α, ATF4, and ER chaperones, and higher levels of cleaved caspase‐12, in Nox4 KO compared to WT. Tubulin was used as a loading control. Treatment with guanabenz (Gbz) reversed these changes (blots shown to the right).

Data information: Data are presented as mean ± SEM. Comparisons were made by one‐way ANOVA, with P < 0.05 considered significant. Values above bar graphs denote the level of significance. See also Appendix Fig S5.
Figure 6
Figure 6. Nox4 protects against acute kidney injury in vivo

  1. Plasma urea levels were elevated to a greater extent in tunicamycin‐treated Nox4 KO mice than WT. Co‐treatment with guanabenz (Gbz) reduced urea levels in both groups. Numbers of animals are indicated within bars.

  2. Forty‐eight hours after systemic tunicamycin treatment, kidneys of Nox4 KO mice showed a marked surface pallor (bottom right).

  3. TUNEL staining revealed a significantly higher number of apoptotic cells in tunicamycin‐treated Nox4 KO mice. n = 4/group.

  4. Immunoblotting of kidney homogenates showed significantly elevated cleaved caspase‐12 and cleaved PARP levels in tunicamycin‐treated Nox4 KO mice compared to WT.

  5. Survival curves showed that a very high proportion of Nox4 KO mice died after AKI. Guanabenz (Gbz) treatment dramatically improved survival in tunicamycin‐treated KO mice. Number of animals as indicated. Levels of significance by Kaplan–Meier analysis are reported to the right.

  6. Schematic depicting the effect of Nox4‐generated ROS on PP1 activity and the balance between eIF2α phosphorylation and dephosphorylation. Nox4 is upregulated by ATF4 and binds to GADD34. It inhibits GADD34‐bound PP1 through the local generation of H2O2 and oxidation of the metal (M) center of the serine–threonine phosphatase. The consequent prolongation of eIF2α phosphorylation promotes cell survival in the face of acute protein unfolding stress. M = iron or manganese, which are oxidized from the M (II) to the M (III) species.

Data information: Data are presented as mean ± SEM. Comparisons in (A, C) were made by one‐way ANOVA, with P < 0.05 considered significant. Values above bar graphs denote the level of significance.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Alessi DR, Street AJ, Cohen P, Cohen PT (1993) Inhibitor‐2 functions like a chaperone to fold three expressed isoforms of mammalian protein phosphatase‐1 into a conformation with the specificity and regulatory properties of the native enzyme. Eur J Biochem 213: 1055–1066 - PubMed
    1. Anilkumar N, Weber R, Zhang M, Brewer A, Shah AM (2008) Nox4 and Nox2 NADPH oxidases mediate distinct cellular redox signaling responses to agonist stimulation. Arterioscler Thromb Vasc Biol 28: 1347–1354 - PubMed
    1. Baird TD, Wek RC (2012) Eukaryotic initiation factor 2 phosphorylation and translational control in metabolism. Adv Nutr 3: 307–321 - PMC - PubMed
    1. Barford D, Keller JC (1994) Co‐crystallization of the catalytic subunit of the serine/threonine specific protein phosphatase 1 from human in complex with microcystin LR. J Mol Biol 235: 763–766 - PubMed
    1. Bedard K, Krause KH (2007) The NOX family of ROS‐generating NADPH oxidases: physiology and pathophysiology. Physiol Rev 87: 245–313 - PubMed

Publication types