NADPH oxidase-dependent and -independent mechanisms of reported inhibitors of reactive oxygen generation

Abstract

NADPH oxidase isoform-2 (NOX2) generates reactive oxygen species (ROS) that contribute to neurodegenerative and cardiovascular pathologies. However, validation of NOX2 as a pharmacotherapeutic target has been hampered by a lack of mechanistically-defined inhibitors. Using cellular and biochemical assays, we explored previously reported inhibitors of ROS production (perhexiline, suramin, VAS2870 and two Shionogi patent compounds) as direct NOX2 inhibitors. All but suramin, which presumably lacks cell penetrance, inhibit cellular ROS production. However, only perhexiline and suramin inhibit biochemical NOX2 activity. Indeed, our data suggest that NOX2 inhibition by perhexiline may contribute significantly to its demonstrated cardioprotective effects. Inhibition of protein kinase CβII explains the cellular activity of the Shionogi compounds, whereas VAS2870 inhibits by an as-yet unidentified mechanism unrelated to direct NOX2 function or subunit assembly. These data delineate the mechanisms of action of these compounds and highlight their strengths and limitations for use in future target validation studies.

Keywords::

• Reactive oxygen species
• NADPH oxidase isoform-2
• suramin
• VAS2870
• perhexiline

Introduction

Small-molecule inhibitors are valuable probes for defining the role of a particular enzyme in a given biological process. Application of these compounds to relevant pathophysiological cellular and animal models enables validation of the catalytic activity of that enzyme as a target for pharmacotherapy. However, a thorough interpretation of such studies requires an adequate understanding of the mechanisms of action of these tools. For complex enzymatic systems with multiple modes of regulation and activation, this mechanistic description is even more critical, given the numerous possibilities for small-molecule interaction¹.

The NADPH oxidase (NOX) family of enzymes represents one such complex enzyme system. These are multisubunit, membrane-bound enzymes that generate superoxide anion from NADPH and molecular oxygen². NADPH oxidase isoform-2 (NOX2), the phagocytic isoform of NOX and best-characterized member of this family, initiates production of intraphagosomal reactive oxygen species (ROS) for microbial killing³. However, expression of NOX2 is not limited to phagocytic cell types, as expression in a variety of tissues, including vascular smooth muscle cells, adventitial fibroblasts and cardiomyocytes, has been reported^3,4. NOX2 is activated by phosphorylation of the cytosolic subunit p47^phox, which stimulates translocation and assembly of a cytosolic complex of subunits (p47^phox, p67^phox, p40^phox and Rac2) with the membrane-bound, cytochrome-containing components (gp91^phox and p22^phox)². The interaction of cytosolic and membrane components permits binding and oxidation of NADPH on gp91^phox and subsequent electron transfer through its flavin site and dual heme domains to the oxygen substrate to yield superoxide anion.

Although NOX2 and other ROS-producing enzymes serve beneficial functions, dysregulation of the generation and/or removal of ROS upsets the balance between oxidative and antioxidant processes. Ultimately, cellular damage from persistent oxidative stress leads to impaired tissue function and the onset and progression of a disease state, as has been implicated for a number of diseases, including Alzheimer’s disease, arthritis, hypertension, stroke, heart failure and atherosclerosis^5–8. Indeed, the improved functional endpoints noted in cardiovascular disease models for mice lacking various subunits of the NOX2 enzyme complex^9–12 strongly suggest a pathophysiological role for NOX2. Emerging data also implicate the NOX1 and NOX4 isoforms in various aspects of cardiovascular disease, although these enzymes are not as well-characterized and reagents for detailed analysis are more limited¹³.

Many inhibitors of NOX enzymes have been claimed in the literature, but few have been shown to be specific to NOX or to act directly on the enzyme^14–16. As examples, in vitro pharmacological inhibition of NOX2 by covalent modifiers such as apocynin, diphenyleneiodonium (DPI) and a peptidic inhibitor of NOX2 assembly, gp91ds-tat, has been described, inclusive of demonstrated inhibition of cellular ROS production^17–22. Furthermore, apocynin and gp91ds-tat have been utilized in vivo with reports of cardiovascular benefit^23–26, in support of the hypothesis that NOX2 inhibition would be therapeutic in cardiovascular disease. However, a direct association of cellular and in vivo effects of these compounds must be viewed cautiously, given their potential for multiple off-target cellular effects²⁷.

The complexity of the activation cascade for NOX2 reveals several potential mechanisms for pharmacological disruption: e.g. by inhibiting an initiating kinase, inhibiting oxidase assembly, competing with the substrate NADPH or scavenging the superoxide product. Recently, several new compounds have been described as inhibitors of ROS formation in cellular or tissue preparations (structures shown in Figure 1): perhexiline, an antianginal agent prescribed in New Zealand and Australia^28,29; suramin, a purinergic receptor antagonist used as an antiparasitic agent^30–32; two compounds described by Shionogi, Ltd. as inhibitors of cellular ROS generation (US patent US20060089362A1); and VAS2870, from the company Vasopharm^33,34. Furthermore, several of these have been suggested to act as NOX inhibitors, but of unclear mechanism. The purpose of the current study was to provide an in vitro characterization of these compounds, primarily whether any directly inhibit NOX2 and could serve as better-defined tool molecules for further validating NOX2 as a target for therapeutic intervention.

Figure 1.  Structures of inhibitors of ROS production, proposed to act at NOX.

Methods

General

Structures of perhexiline, suramin, Shionogi compounds 1 and 2 and VAS2870 are shown in Figure 1. Perhexiline was purchased from Sigma Chemical Co. (St Louis, MO). Suramin was purchased from Biomol International (Plymouth Meeting, PA). Apocynin was purchased from Calbiochem (Gibbstown, NJ). Shionogi compound 1 [3-(3-chlorophenyl)-N-{2-[4-(ethylamino)-1-piperidinyl]phenyl}pyrazolo[1,5-a]pyrimidine-5-carboxamide] and Shionogi compound 2 [3-(3-chlorophenyl)-N-[2-(1-piperazinyl)phenyl] pyrazolo[1,5-a]pyrimidine-5-carboxamide] were prepared according to the procedures described in US patent US20060089362A1. VAS2870 [7-(1,3-benzoxazol-2-ylthio)-3-(phenylmethyl)-3H-[1,2,3]triazolo[4,5-d]pyrimidine] was purchased from Butt Park, Ltd. (Okehampton, Devon, UK). All other chemicals were purchased from Sigma Chemical Co (St. Louis, MO), and were of the highest analytical grade.

Isolation of human neutrophils and neutrophil membrane fractions

Human biological samples were sourced ethically and their research use was in accord with the terms of the informed consents. Blood was collected from healthy human donors using 50 mM EDTA as the anticoagulant. Blood was diluted 1:1 in Dulbecco’s phosphate buffered saline without calcium or magnesium (DPBS) and layered over Ficoll-Paque™ PLUS, followed by centrifugation (400× g for 30 min at room temperature). The resultant pellet was resuspended in DPBS to twice the pellet volume and 6% dextran was added to a final concentration of 1%. Following 1 h incubation at room temperature to sediment erythrocytes, supernatants were removed and further treated with BD Pharm Lyse™ buffer (BD Biosciences, San Jose, CA) as per manufacturer’s instructions to lyse remaining erythrocytes. The supernatants were then centrifuged at 1000x g for 10 min and the resultant pellet (neutrophils) was resuspended in either Hank’s balanced salt solution (HBSS, 1.3 mM CaCl₂, 5.4 mM KCl, 0.4 mM KH₂PO₄, 0.5 mM MgCl₂, 0.4 mM MgSO₄, 137 mM NaCl, 0.4 mM NaHCO₃, 0.3 mM Na₂HPO₄, 5.6 mM glucose) for cellular NADPH oxidase activity experiments, or in ice-cold sonication buffer [11% sucrose, 50 mM potassium phosphate pH 7.0, 120 mM NaCl, 5 mM EGTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 µg/mL leupeptin, 1 µM pepstatin, 1% protease inhibitor cocktail and 2 mM diisopropyl fluorophosphate] for isolation of membrane fractions.

To isolate neutrophil membrane fractions, cells were sonicated in sonication buffer and centrifuged at 1000× g for 15 min at 4°C. Supernatants were layered onto a 20/40% (w/v) discontinuous sucrose gradient and centrifuged at 150,000× g for 40 min. Membrane fractions were collected from the 20/40% sucrose interface and protein concentrations were determined using the bicinchoninic acid method (Pierce, Rockford, IL). Membrane fractions were snap-frozen in liquid nitrogen and stored at −80°C.

Recombinant His-tagged p47^phox, p67^phox and Rac2

Recombinant p47^phox, p67^phox and Rac2 were prepared as described by Rotrosen, et al.³⁵. Briefly, DNA encoding the NOX2 cytosolic components (p47^phox, p67^phox and Rac2) were amplified from human neutrophil cDNA using the polymerase chain reaction with the following primer sequences: p47^phox forward = 5′-GGAATT CACCATGC ATCATCACCAT CACCAT GGGGAC ACCTTCAT CCGT-3′, reverse = 5′-GGGGTACCT CAGACGGC AGACGCCAGC-3′; p67^phox forward  = 5′-GGAATT CACCATGCA TCATCACCA TCACCATTC CCTGGTGGAG GCCATC-3′,reverse = 5′-GGGGTACC TAGACTTCTC TCCGAGTGC-3′; Rac2 forward = 5′-GGAATT CACCATG CATCATC ACCATCA CCATCA GGCCATCAAG TGTGTG-3′,reverse  = 5′ -GGGGTAC CTAGAG GAGGCTGC AGGC-3′. For all primer sets, an EcoRI site, a Kozak sequence and a (His)₆ tag were included in the forward primer and a KpnI site was included in the reverse primer. Open reading frames for each subunit were subcloned into pFastBac-1 (Invitrogen, Carlsbad, CA), and transfected into Sf9 insect cells for expression of proteins according to Fornwald, et al.³⁶. Proteins were purified using nickel-nitrilotriacetic acid (nickel-NTA) columns according to manufacturer’s instruction (Qiagen, Valencia, CA).

Immunoblotting

Protein samples were subjected to SDS-PAGE using 4–20% polyacrylamide gels and then transferred to PVDF membranes. Membranes were blocked with 5% nonfat dry milk in tris-buffered saline containing 0.2% Tween-20 (TBST), then probed with primary antibody in 2% nonfat dry milk in TBST, followed by either goat anti-mouse or anti-rabbit IRDye-labeled IgG secondary antibody (LI-COR Biosciences, Lincoln, NE; diluted to 1:10,000, v/v) in 2% nonfat dry milk in TBST. The following primary antibodies were used: mouse anti-p47^phox (BD Biosciences, San Jose, CA; diluted to 1:2,000, v/v), mouse anti-p67^phox (BD Biosciences; diluted to 1:2,000, v/v) and mouse anti-Rac2 (Abnova Corp., Walnut, CA; diluted to 1:2,000, v/v). Blots were imaged using the LI-COR Odyssey infrared imaging system, and, when necessary, bands were quantified with the ImageQuant software package (GE Healthcare, Piscataway, NJ).

Enzyme activities

Cellular NOX2 activity was determined by monitoring the reduction of exogenously-added cytochrome c by stimulated primary neutrophils. Cytochrome c from horse heart (100 µM; Sigma Chemical Co., St. Louis, MO) was added to neutrophils suspended in HBSS at a density of 2 × 10⁶ cells/mL. In a 96 well plate, cells (200,000 cells/well) were incubated with vehicle (0.1% DMSO) or compounds for 20 min at 37°C before stimulation with 1 µM phorbol 12-myristate 13-acetate (PMA). In pilot experiments, this concentration of PMA was shown to stimulate superoxide release to 80% of the maximum level (EC₈₀). The rate of cytochrome c reduction was determined by measuring the change in absorbance at 550 nm for 30 min after PMA stimulation. Superoxide production was determined by subtracting the absorbance of reaction wells containing an excess of superoxide dismutase (SOD, 50 units/100 μL) from those containing no SOD. The change in absorbance at 550 nm was converted to concentration of cytochrome c reduced using an extinction coefficient of 2.1 × 10⁴ M⁻¹ cm⁻¹³⁷. Experiments were performed in triplicate.

For analysis of semi-recombinant NOX2 activity, 1 µg of human neutrophil membrane was combined with purified recombinant cytosolic subunits of NOX2 (2 ng of p47^phox, 2 ng of p67^phox and 1 ng of Rac2) and added to wells containing assay buffer [final concentrations: 50 mM potassium phosphate (pH 7), 1 mM EGTA, 0.01 mM FAD, 4 mM MgCl₂, 1 µM GTPγS, 40 µM arachidonic acid, 100 µM cytochrome c and 1% DMSO]. These concentrations of protein subunits and cofactors were determined in pilot experiments to provide the maximal activity of the enzyme when stimulated with NADPH. Reactions were run at room temperature and initiated by addition of 50 µM NADPH, which was determined in pilot experiments to be the K_m level. The rate of cytochrome c reduction was determined for 10 min after addition of NADPH using the method described above. Experiments were performed in triplicate.

Steady-state inhibition mechanisms, relative to NADPH substrate, were determined for the direct inhibitors of semi-recombinant NOX2 by measuring reaction velocities as a function of varied concentrations of both NADPH and inhibitor. All other reaction components and conditions were as described above. For each inhibitor, entire data sets were globally fit to standard inhibition models for assignment of best-fit mechanism, as described below.

The effects of the test compounds on the neutrophil-abundant isoform of protein kinase C (PKC), PKCβII, were assayed using a commercially available kit (IMAP™ FP kit, Molecular Devices, Sunnyvale, CA) and recombinant human PKCβII (Upstate/Millipore, Billerica, MA), according to manufacturer’s instructions.

To assess xanthine oxidase activity, compounds or vehicle (0.1% DMSO) were incubated with xanthine oxidase (from bovine milk, EMD Biosciences, Carlsbad, CA; 10 milliunits/well in HBSS containing 100 µM cytochrome c) for 20 min prior to stimulation with 50 µM hypoxanthine. Experiments were performed in triplicate wells, with the third well containing an excess of SOD (40 units/100 μL). Cytochrome c reduction was measured by the change in absorbance at 550 nm. Superoxide production was determined by subtracting the absorbance of the SOD-containing wells from the duplicate wells containing no SOD.

NOX2 assembly (p47^phox translocation)

Isolated human neutrophils were incubated with vehicle (0.1% DMSO), 1 µM PMA alone or 1 µM PMA in the presence of compound (10 µM) or apocynin (100 µM), a known inhibitor of NOX2 assembly¹⁸. Neutrophils were preincubated with inhibitors or vehicle for 30 min at 37°C prior to stimulation with PMA. After 10 min stimulation with PMA, cells were pelleted by low-speed centrifugation, membrane fractions were recovered, and protein concentrations were determined as described above. Protein from membrane fractions (20 µg) was used for immunoblot analysis as described above using either mouse anti-p47^phox (BD Biosciences, San Jose, CA; diluted to 1:1,000, v/v) or rabbit anti-gp91^phox (Upstate/Millipore, Billerica, MA, diluted to 1:1,000, v/v) as the primary antibodies.

Data analysis

All data are reported as mean ± standard error of the mean. For kinetic evaluations, nonlinear regression analyses were performed using the GraphPad Prism 5 software package (GraphPad Software, Inc., San Diego, CA). Concentration response curves were fit to the following equation to determine IC₅₀ values:

where [I] represents the concentration of compound and h refers to the Hill coefficient. Given the reduced solubility of the Shionogi compounds (observed as precipitation) at concentrations of 10 µM and above, data obtained at or above this limit were excluded from further analysis. To determine inhibition modality with respect to NADPH substrate, initial velocities obtained at multiple NADPH and inhibitor concentrations were globally fit to competitive, uncompetitive and noncompetitive models, represented, respectively, by the following equations:

The best-fit model was identified using the extra sum-of-squares F-test as implemented by the GraphPad Prism software package (GraphPad Software, Inc., San Diego, CA).

For the Western analysis of p47^phox translocation, statistical comparisons between each treatment group and the PMA-treated control group (n = 3 biological replicates) were made using one-way ANOVA with Dunnett’s multiple comparisons post-tests (**, p < 0.01; ***, p < 0.001). One outlier from the apocynin-treated samples was identified using the Grubb’s test and excluded from data analysis.

Results

Inhibition of cellular and semi-recombinant NOX2 activity by tool compounds

Human neutrophils were used both to assess the specific effects of compounds on NOX2 in an intact cellular assay format and as a source of enzyme to examine the effects of compounds on NOX2 in a biochemical, cell-free format. Namely, these cells selectively express NOX2 over the other NOX isoforms (NOX1, NOX3, NOX4 and NOX5), as reported previously³⁸ and confirmed by us (data not shown). Furthermore, deletion of the gp91^phox subunit of NOX2 leads to loss of the respiratory burst and undetectable superoxide production in neutrophils³⁹.

Extracellular superoxide production by intact, PMA-stimulated human neutrophils was assessed by the spectroscopic detection of cytochrome c reduction (Figure 2A)⁴⁰. Perhexiline, VAS2870, and DPI, a positive control, inhibited cellular superoxide formation with IC₅₀ values of 3.0 ± 0.2 µM, 77 ± 14 nM and 39 ± 5 nM, respectively (Figure 2B). Shionogi 1 and Shionogi 2 inhibited superoxide formation with IC₅₀ values of 56 ± 3 nM and 99 ± 6 nM, respectively. Suramin did not inhibit neutrophil superoxide formation at concentrations ranging from 10 nM to 10 µM.

Figure 2.  Effects of compounds on cellular NOX activity. (A) Representative kinetic traces of cytochrome c reduction by PMA-stimulated human neutrophils alone (black circles) or in the presence of either SOD (white circles) or 100 µM DPI (gray circles). (B) Prior to stimulation with phorbol ester, human neutrophils were incubated in the presence of perhexiline (IC₅₀ = 3.0 ± 0.2 µM, Hill slope = −3.1 ± 0.9), Shionogi 1 (IC₅₀ = 56 ± 3 nM, Hill slope = −1.4 ± 0.1), Shionogi 2 (IC₅₀ = 99 ± 6 nM, Hill slope = −1.8 ± 0.2), VAS2870 (IC₅₀ = 77 ± 14 nM, Hill slope = −1.0 ± 0.2), DPI (IC₅₀ = 39 ± 5 nM, Hill slope = −1.1 ± 0.1), or suramin (inactive). NOX activity was determined using detection of SOD-inhibitable cytochrome c reduction as described in “Methods”.

The activity of NOX2 in a cell-free environment was assessed using a semi-recombinant system in which neutrophil membranes were combined with purified recombinant p47^phox, p67^phox and Rac2. Confirmation of gp91^phox in these membranes and the purity of the recombinant proteins were obtained by electrophoresis and immunoblotting (Figure 3). Enzymatic activity was determined using the same cytochrome c-based assay format as was used for intact neutrophils (Figure 4A). As shown in Figure 4B, perhexiline and suramin inhibited semi-recombinant NOX2 activity (IC₅₀ of 13 ± 2 µM and 1.0 ± 0.1 µM, respectively), when assayed with NADPH at its K_m value (50 µM, as determined in pilot experiments). Neither VAS2870 nor the Shionogi compounds inhibited semi-recombinant NOX2 activity at concentrations ranging from 10 nM to 100 µM (for VAS2870) or 3 µM (for Shionogi 1 or 2). DPI, the positive control, inhibited semi-recombinant NOX2 activity with an IC₅₀ of 90 ± 10 nM.

Figure 3.  Protein components of the semi-recombinant NOX2 assay. (A) Representative anti-gp91^phox immunoblot analysis of neutrophil membranes used in the semi-recombinant assay. A characteristic broad band centered at ~91 kD was observed when the membranes were denatured at 70°C for 5 min (left), consistent with literature reports for glycosylated gp91^phox (see, for example, ref 35. Denaturation at 100°C for 10 min elicited a single band at ~60 kD (right), consistent with deglycosylation of the protein (theroretical MW = 65.3 kD) as a result of sample preparation. (B) Representative gel images of purified, recombinant, His-tagged cytosolic proteins p47^phox (left), p67^phox (middle) and Rac2 (right). For each protein, the Coomassie-stained gel (2 µg of protein) and the corresponding immunoblot (40 ng of protein) are shown.

Figure 4.  Effects of compounds on purified semi-recombinant NOX2 activity. (A) Representative kinetic traces of cytochrome c reduction by NADPH-stimulated semi-recombinant NOX2 alone (black circles) or in the presence of either SOD (white circles) or 2 µM DPI (gray circles). (B) Prior to reaction initiation with NADPH, neutrophil membranes and recombinant cytosolic components were added to perhexiline (IC₅₀ = 13 ± 2 µM, Hill slope = −0.75 ± 0.08), Shionogi 1 (inactive), Shionogi 2 (inactive), suramin (IC₅₀ = 1.0 ± 0.1 µM, Hill slope = −1.0 ± 0.1), VAS2870 (inactive), or DPI (IC₅₀ = 90 ± 10 nM, Hill slope = −0.84 ± 0.07). NOX2 activity was determined using SOD-inhibitable cytochrome c reduction as described in “Methods”.

Determination of inhibition mechanisms of compounds

Given its activity in both NOX2 assay formats, perhexiline was also tested against an alternate superoxide generator, xanthine oxidase (XO), in order to rule out the possibility that this compound acts as a general superoxide scavenger. At concentrations ranging from 100 nM to 100 µM, perhexiline showed no effect on XO activity, whereas the known XO inhibitor luteolin⁴¹ showed characteristic inhibition (Figure 5A).

Figure 5.  Testing for inhibitory effects of compounds by NOX2-independent mechanisms. (A) Inhibition of xanthine oxidase activity by perhexiline (inactive) and luteolin (a positive control) was measured using SOD-inhibitable cytochrome c reduction as described in “Methods”. (B) Inhibition of PKCβII activity by cellular-active inhibitors Shionogi 1 (IC₅₀ = 4.6 ± 0.8 nM, Hill slope = −1.0 ± 0.1), Shionogi 2 (IC₅₀ = 9.4 ± 1.0 nM, Hill slope = −1.0 ± 0.1), perhexiline (inactive) and VAS2870 (inactive) as well as sangivamycin (a positive control, IC₅₀ = 6.8 ± 2.0 µM, Hill slope = −0.7 ± 0.1), was determined using a fluorescence polarization IMAP™ kit as described in “Methods”.

Neutrophils contain an abundance of the βII isoform of protein kinase C (PKCβII), which is critical for the activation of NOX2 by phorbol esters^42,43 such as PMA, the stimulus used in our isolated neutrophil activity assay. Those compounds which inhibited superoxide formation in intact cells, but not semi-recombinant NOX2 − VAS2870, Shionogi 1 and Shionogi 2 − were tested for inhibition of PKCβII. VAS2870 did not inhibit PKCβII activity, whereas both Shionogi compounds displayed potent inhibition (IC₅₀ = 4.6 ± 0.8 nM and 9.4 ± 1.0 nM, respectively, for Shionogi 1 and Shionogi 2) (Figure 5B). Consistent with its action as a direct inhibitor of NOX2, perhexiline did not inhibit PKCβII activity. As a positive control, the PKC inhibitor sangivamycin⁴⁴ displayed characteristic inhibition of PKCβII, with an IC₅₀ of 6.8 ± 2.0 µM (Figure 5B).

The ability of the compounds to affect translocation of p47^phox to the membrane for assembly of NOX2 was examined using Western blotting of membrane fractions from PMA-stimulated neutrophils, as previously described⁴⁵. As shown in Figure 6, apocynin, a known inhibitor of this process^18,20, blocked the appearance of p47^phox in membrane fractions from stimulated cells. Shionogi 1 and 2 also blocked p47^phox translocation, as expected from the role of PKCβII in this process and its inhibition by these compounds (see above). Perhexiline and suramin, which directly inhibited the purified semi-recombinant enzyme (see above), did not affect PMA-stimulated complex assembly. Consistent with a prior report³⁴, VAS2870 also did not inhibit this response, suggesting that this compound does not interfere with NOX2 assembly or activation (e.g. by an activating kinase other than PKCβII).

Figure 6.  Inhibition of p47^phox translocation in PMA-stimulated neutrophils: representative immunoblots (top) and observed intensities of the p47^phox bands (normalized to the intensity of the corresponding band for the integral membrane component of NOX2, gp91^phox), relative to the PMA-treated samples from corresponding gels (bottom). Neutrophils were treated with vehicle, PMA (1 µM), or PMA following pretreatment with apocynin, VAS2870, perhexiline, suramin, Shionogi 1, or Shionogi 2, then membrane fractions were isolated and analyzed by Western analysis as described in “Methods”. With the exception of apocynin (100 µM), all compounds were tested at 10 µM. The order of sample lanes (top) correspond to the order of intensity bars (bottom).

The inhibition modalities of perhexiline and suramin with respect to NADPH were determined in the semi-recombinant NOX2 assay. As shown in Figure 7A, suramin exhibited NADPH-competitive inhibition of NOX2 with a K_i of 0.7 ± 0.1 µM. In contrast, perhexiline (Figure 7B) yielded a noncompetitive pattern with respect to NADPH (K_i = 13 ± 2 µM).

Figure 7.  Mode of NOX2 inhibition by suramin (A) and perhexiline (B). Inhibition modalities were determined by global nonlinear fitting and model comparison as described in “Methods”. Lineweaver-Burk plots (foreground) were generated by reciprocal transformation of the best nonlinear fits (inset). Solid lines represent the best-fit global trends. (A) Suramin exhibits a competitive pattern of inhibition with respect to NADPH (best-fit parameters: K_i = 0.7 ± 0.1 µM, K_m = 52 ± 8 µM, V_max = 0.013 ± 0.001 µM/s). (B) Perhexiline exhibits a noncompetitive pattern (best-fit parameters: K_i = 13 ± 2 µM, K_m = 44 ± 8 µM, V_max = 0.46 ± 0.03 µM/s). Data shown are representative; differences in V_max values between the two data sets reflect the typical variability observed with neutrophil membrane preparations.

Discussion

The data presented in this paper address mechanisms of action of reported inhibitors of cellular ROS production to determine whether any represent tools which could be used to validate NOX2 inhibition as a pharmacological approach for therapeutic intervention. We envisioned several distinct, readily-testable mechanisms by which these compounds might inhibit ROS production through the NOX2 pathway: direct inhibition of NOX2 enzymatic activity (perhaps by competition with NADPH substrate), general antioxidant activity, inhibition of PKC and inhibition of NOX2 assembly. As described below, our approach of evaluating inhibition both within the intact cellular milieu and directly against purified, semi-recombinant NOX2 enabled discrimination between compounds with respect to mechanisms of action and targets of effect. Furthermore, this approach underscored the limitations of these molecules as potential tool molecules to report on direct effects of NOX2 inhibition in cellular studies.

Of the compounds tested in the current study, only suramin and perhexiline directly inhibit NOX2 activity in vitro. Suramin inhibits NOX2 via competitive inhibition of NADPH. This is the first report of an NADPH-competitive inhibitor of NOX, and is consistent with its demonstration as a purinergic receptor antagonist³⁰ and competitive inhibitor of NAD⁺-dependent sirtuins such as SIRT5⁴⁶. The lack of cellular activity of suramin is consistent with its lack of membrane permeability, as previously suggested⁴⁷, and indicates that further work to improve its membrane permeability would be necessary to exploit suramin as a tool for validating NOX2 as a therapeutic target. By contrast, we have yet to determine the precise mechanism of the observed noncompetitive inhibition of NOX2 by perhexiline. However, we interpret the similar potencies between biochemical and cellular activities of this compound as evidence that direct enzymatic inhibition is a major cause of cellular functionality.

Of all the compounds examined, only perhexiline inhibits NOX2 activity of both intact neutrophils and the purified enzyme. The use of perhexiline clinically and its efficacy in heart failure patients have been attributed to its inhibition of carnitine palmitoyl transferase-1 (CPT-1⁴⁸). Inhibition of CPT-1 shifts myocardial energy usage from fatty acid to glucose metabolism, which lowers oxygen demand⁴⁹. Interestingly, perhexiline has also been reported to inhibit superoxide formation in cardiovascular tissues and neutrophils²⁸. Because our results also rule out antioxidant or kinase-dependent mechanisms, the inhibition of myocardial oxidative stress via direct NOX2 inhibition could be a major contributor to the beneficial effects of perhexiline, above and beyond its effects on cardiac energetics.

VAS2870 and both Shionogi compounds are extremely potent inhibitors of cellular superoxide formation, but not semi-recombinant NOX2 activity. Consistent with other studies of VAS2870 and a close analog, VAS3497⁵⁰, we have shown that the mechanism of VAS2870, as well as the Shionogi compounds, cannot include generalized antioxidant activity, as this would present as inhibition in both cellular and purified enzyme assays. Therefore, these compounds must act upstream of the activated NOX2 enzyme complex. Such a conclusion appears inconsistent with prior findings from ten Freyhaus, et al.³⁴, who reported inhibition of NOX2 by VAS2870 in whole cell lysates. However, this inhibition appeared to be dependent on the dynamics of complex assembly and order of stimulus addition (unpublished observation in reference 50) that could reflect events upstream of complex activation not recapitulated by our semi-purified enzyme preparation. As such, their observations with this compound likely reflect events more complex than direct inhibition of NOX2 catalysis.

Because intact neutrophils were stimulated with phorbol ester to activate PKCβII and trigger p47^phox phosphorylation and subsequent translocation to membrane-associated gp91^phox, a possible mechanism for cell-active compounds is inhibition of PKCβII. Both Shionogi compounds inhibit PKCβII with IC₅₀ values similar to those seen in intact cells. Thus, PKCβII inhibition is responsible for the observed inhibition of cellular superoxide production by these compounds.

On the other hand, VAS2870 is not an inhibitor of PKCβII. The lack of effect of VAS2870 on PKCβII or as an antioxidant left open the possibility that VAS2870 might inhibit cellular NOX activity by blocking some other aspect of oxidase assembly. For example, apocynin, following peroxidase-dependent dimerization, inhibits NOX via covalent modification of p47^phox and thereby blocking translocation of the subunit to the membrane¹⁸. Indeed, apocynin inhibited the appearance of p47^phox in PMA-stimulated neutrophil membrane fractions in our experiments, but VAS2870 had no effect at concentrations two orders of magnitude above its IC₅₀ in intact cells, supportive of previous results with this compound³⁴. Thus, the precise mode of action of VAS2870 against NOX is undetermined, but distinct from the mechanisms examined in the current study.

Conclusions

In this study, we profiled several emerging tool compounds as possible inhibitors of NOX2 and sought to probe their mechanisms by monitoring effects on direct activity of the purified enzyme and in intact cells. VAS2870 appears to be the most potent cellular inhibitor of NOX2 reported, but its precise mechanism of action remains to be determined. Our definition of whether VAS2870 is a direct inhibitor of NOX2 contrasts with that given by Wind et al.⁵⁰, which probably reflects differences in our assay formats. However, we reach a similar conclusion that the cellular inhibition of ROS generation by this class of compounds is not yet defined and may furthermore involve unrecognized elements of NOX2 regulation. The demonstration of NADPH-competitive inhibition of NOX2 by suramin is intriguing, and perhaps represents a route to identify other inhibitors of the enzyme. However, its activity at purinergic receptors and its absent or poor membrane penetrance render suramin, per se, an unsuitable tool for cellular studies of NOX2 inhibition. The identification of perhexiline as a direct inhibitor of NOX2 may represent a novel “add-on” for this compound, which is already used as a therapeutic for heart failure²⁹. For example, the β-adrenergic receptor antagonist carvedilol has shown added cardiovascular benefit over drugs in the same class due to added beneficial mechanisms of action, including antioxidant activity⁵¹.

Additional work is required to identify a direct NOX2 inhibitor of sufficiently-defined mechanism for use as a tool for in vivo pharmacological target validation. However, our analysis has highlighted the various aspects of the NOX2 activation pathway that can be used to improve our understanding of the role of NOX2 in disease.

Acknowledgments

This work is dedicated to the memory of our friend and colleague Stephen A. Douglas. We are grateful to David J. Behm for discussions and critical review of this paper.

Declaration of interest

All authors are either currently employees of GlaxoSmithKline or were employed by GlaxoSmithKline at the time that this work was performed.