Inhibition of NADPH Oxidase Activation by 4-(2-Aminoethyl)-benzenesulfonyl Fluoride and Related Compounds*

The elicitation of an oxidative burst in phagocytes rests on the assembly of a multicomponental complex (NADPH oxidase) consisting of a membrane-associated flavocytochrome (cytochrome b 559), representing the redox element responsible for the NADPH-dependent reduction of oxygen to superoxide (O·̄2), two cytosolic components (p47 phox , p67 phox ), and the small GTPase Rac (1 or 2). We found that 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF), an irreversible serine protease inhibitor, prevented the elicitation of O·̄2 production in intact macrophages and the amphiphile-dependent activation of NADPH oxidase in a cell-free system, consisting of solubilized membrane or purified cytochrome b 559 combined with total cytosol or a mixture of recombinant p47 phox , p67 phox , and Rac1. AEBSF acted at the activation step and did not interfere with the ensuing electron flow. It did not scavenge oxygen radicals and did not affect assay reagents. Five other serine protease inhibitors (three irreversible and two reversible) were found to lack an inhibitory effect on cell-free activation of NADPH oxidase. A structure-function study of AEBSF analogues demonstrated that the presence of a sulfonyl fluoride group was essential for inhibitory activity and that compounds containing an aminoalkylbenzene moiety were more active than amidinobenzene derivatives. Exposure of the membrane fraction or of purified cytochrome b 559, but not of cytosol or recombinant cytosolic components, to AEBSF, in the presence of a critical concentration of the activating amphiphile lithium dodecyl sulfate, resulted in a marked impairment of their ability to support cell-free NADPH oxidase activation upon complementation with untreated cytosol or cytosolic components. Kinetic analysis of the effect of varying the concentration of each of the three cytosolic components on the inhibitory potency of AEBSF indicated that this was inversely related to the concentrations of p47 phox and, to a lesser degree, p67 phox . AEBSF also prevented the amphiphile-elicited translocation of p47 phox and p67 phox to the membrane. These results are interpreted as indicating that AEBSF interferes with the binding of p47 phox and/or p67 phox to cytochromeb 559, probably by a direct effect on cytochromeb 559.

The production of reactive oxygen radicals represents the major microbicidal mechanism of phagocytes (1). Oxygen radicals are also generated, in lesser amounts, by some nonphagocytic cells, sharing the enzymatic machinery characteristic of phagocytes (2), and, under certain conditions, by plant cells (3). Interest in reactive oxygen species has also been stimulated by accumulating evidence for their involvement in the pathogenesis of diseases, ranging from respiratory distress syndrome to ischemia-reperfusion injury in several organs (4).
The primordial oxygen radical produced by phagocytes is superoxide (O 2 . ). 1 It is generated, in response to appropriate stimuli, by NADPH-derived one-electron reduction of molecular oxygen, a reaction catalyzed by a membrane-bound heterodimeric flavocytochrome (cytochrome b 559 ) (reviewed in Refs. [5][6][7]. Cytochrome b 559 contains two redox centers, FAD and heme, and electron flow from NADPH to oxygen in initiated by the interaction of cytochrome with two cytosolic proteins, p47 phox and p67 phox , and the small GTPase, Rac1 or Rac2 (reviewed in Ref. 8). In the intact cell, this interaction is made possible by the stimulus-dependent translocation of the cytosolic components to the plasma membrane, leading to the assembly of what is known as the NADPH oxidase complex. Activation of NADPH oxidase, resulting in O 2 . generation, can be reproduced in vitro by a cell-free system consisting of either phagocyte membranes and cytosol (9,10) or of a mixture of purified or recombinant components, exposed to a critical amount of an anionic amphiphile (arachidonate or SDS) (11,12). A number of inhibitors of the O 2 . generating NADPH oxidase have been described (reviewed in Ref. 13). The search for such inhibitors is propelled by two incentives: (a) to serve as tools for understanding the structure and mechanism of activation of the NADPH oxidase, and (b) their potential use as therapeutic agents in diseases associated with production of oxygen radicals at an inappropriate site or time. In most studies, inhibitors were tested for an effect on intact phagocytes. The main drawback of such an approach is that it does not permit a distinction between an effect on membrane receptors and signal transduction and a direct effect on the components of the NADPH oxidase complex. Alternative strategies rest on examining the effect of inhibitors on subcellular fractions originating from stimulated cells or on NADPH oxidase activation in the cell-free system. These latter approaches should allow the selection of inhibitors with a direct effect on the assembled complex or on individual components of the NADPH oxidase complex. So far, the only compound having gained wide acceptance as a relatively specific direct inhibitor of NADPH oxidase is diphenylene iodonium (14).
In the present paper we report that 4-(2-aminoethyl)-benzenesulfonyl fluoride (AEBSF; also known as Pefabloc SC), originally developed as an irreversible serine protease inhibitor, prevents the activation of the O 2 . generating NADPH oxidase in both intact stimulated macrophages and in cell-free systems. The effect is shared by some structurally related compounds, indicating that we are dealing with a new class of NADPH oxidase inhibitors. Evidence is presented in favor of the proposal that AEBSF interferes with the interaction of p47 phox and/or p67 phox with cytochrome b 559 , probably by chemical modification of cytochrome b 559 .
Protease inhibitors were dissolved in water at a concentration at least 50-fold higher than the highest concentration used in NADPH oxidase inhibition assays, witih the exception of PMSF, TPCK, and DCIC, which were dissolved in ethanol, methanol, and dimethyl sulfoxide, respectively.

Preparation of Macrophages and Subcellular Fractions
Macrophages were obtained from the peritoneal cavity of guinea pigs injected with mineral oil 5-6 days before cell harvest. The cells were washed and made erythrocyte-free, as described previously (10), and suspended in Earle's balanced salt solution (2 ϫ 10 8 cells/ml), for assays involving intact cells. The membrane and cytosol fractions were obtained from cells disrupted by sonication, as described before (15), with some modifications (16).

Solubilization of Membranes and Purification of Cytochrome b 559
Membranes were washed in 1 M KCl in cell homogenization buffer (15,16) and solubilized in 40 mM octyl glucoside, as described (17). Before use in the cell-free assay, solubilized membrane was diluted in detergent-free solubilization buffer, to bring the concentration of octyl glucoside to 4 -5 mM. Cytochrome b 559 was purified from the original solubilized membrane and relipidated by a modification (18) of our original method (19). Spectral analysis of cytochrome b 559 preparations was performed as described in Ref. 19.

Preparation of Recombinant Cytosolic NADPH Oxidase Components
Baculoviruses carrying cDNAs for human p47 phox and p67 phox were kind gifts of Dr. Thomas L. Leto (National Institutes of Health, Bethesda, MD). Suspension cultures of Sf9 cells (2 ϫ 10 6 cells/ml), grown in serum-free medium (SF-900 II SFM, Life Technologies) were infected with the recombinant baculoviruses and harvested 72 h after infection. The recombinant proteins were prepared as described by Leto et al. (20). Briefly, the cells were disrupted by sonication using three 30-s pulses, in a 250 watt apparatus (Sonics and Materials), centrifuged at 12,000 ϫ g for 10 min, and the supernatant fractions recentrifuged at 100,000 ϫ g for 1 h. The recombinant proteins were purified by ion exchange chromatography of the high speed supernatants, referred to as "crude extracts." Thus, crude extract, containing p47 phox , was applied to a SP Sepharose column (HiLoad 16/10, Pharmacia Biotech), equilibrated with 50 mM sodium phosphate buffer, pH 7.0, supplemented with 1 mM dithioerythritol, and was eluted with a 0 -0.3 M NaCl gradient in the same buffer (210 ml, at a flow rate of 1.5 ml/min). Crude extract, containing p67 phox , was applied to a Q-Sepharose column (HiLoad 16/ 10, Pharmacia Biotech Inc.) equilibrated with 20 mM Tris-HCl, pH 7.5, supplemented with 1 mM dithioerythritol, and eluted with a 0 -0.5 M NaCl gradient in the same buffer (210 ml, at a flow rate of 1.5 ml/min). Purified recombinant proteins were analyzed by SDS-polyacrylamide gel electrophoresis; p47 phox was 99% pure, whereas p67 phox was 90% pure. Recombinant Rac1 was isolated from Escherichia coli transformed with Rac1 cDNA subcloned into the bacterial expression vector pGEX2T (a kind gift of Dr. Thomas L. Leto, National Institutes of Health, Bethesda, MD), as described by Kwong et al. (21). Briefly, the pellet of bacterial culture overexpressing the glutathione S-transferase-Rac1 fusion protein was disrupted by sonication, by three pulses of 30 s, and the homogenate was clarified by centrifugation at 12,000 ϫ g for 5 min. The fusion protein was absorbed from the supernatant on glutathione-agarose (Sigma) and Rac1 was cleaved from the resin-bound fusion protein by thrombin. Thrombin was removed by absorption on benzamidine-Sepharose 6B (Pharmacia Biotech Inc.). Purified Rac1 was loaded with GTP␥S, as described before (22). . production was assayed by the continuous recording of superoxide dismutase inhibitable ferricytochrome c reduction, measured at 550 nm, and the maximal rate was calculated from the linear sector of the change in absorbance curve (23).
Assays for Cell-free Activation of NADPH Oxidase O 2 . Production-Anionic amphiphile (LiDS)-activated O 2 . production in vitro was assayed by ferricytochrome c reduction in mixtures consisting of solubilized membrane and either total cytosol or a combination of p47 phox , p67 phox , and Rac1-GTP␥S, as described (10,16). In some experiments, the membrane was replaced by purified relipidated cytochrome b 559 . Cell-free assays were performed either by the "one-step" method, in which activation by LiDS and NADPH-dependent O 2 . production occur simultaneously, or by the "two-step" method (24) permitting, at least partial, separation of the activation and catalytic stages of the reaction. NADPH Oxidation-LiDS-dependent NADPH oxidation by cell-free mixtures, of a composition identical to those utilized in the O 2 . production assays, was measured as described in Ref. 25.
Oxygen Consumption-LiDS-and NADPH-dependent oxygen consumption by cell-free mixtures, consisting of membrane and recombinant cytosolic components, was assayed with the aid of a Clark oxygen electrode (Yellow Springs Instruments), as described (26).

Translocation of p47 phox and p67 phox to the Membrane
The translocation of p47 phox and p67 phox to the membrane fraction of macrophages was studied in the LiDS-activated cell free system. Macrophage membranes (equivalent to 1.6 ϫ 10 7 cells) were mixed with total macrophage cytosol (equivalent to 1.5 ϫ 10 7 cells) in a total volume of 1 ml of NADPH oxidase assay buffer (10), not containing ferricytochrome c, and incubated for 5 min at 24°C, in the absence or presence of 185 M LiDS (a concentration found to induce maximal O 2 . production under these conditions). The mixture was centrifuged at 15,800 ϫ g for 30 min at 4°C and, after careful removal of the supernatant, the membrane pellet was resuspended in 1 ml of NADPH oxidase assay buffer and recentrifuged at 15,800 ϫ g for 15 min. After removal of the supernatant, the sedimented membranes were resuspended in 30 l of electrophoresis sample buffer containing 2% SDS, heated at 95°C for 5 min, and subjected to SDS-polyacrylamide gel electrophoresis and immunoblotting, as described (19). The blots were probed with a mixture of goat polyclonal anti-recombinant p47 phox and anti-recombinant p67 phox antibodies (20) (kind gifts of Dr. Thomas L. Leto, National Institutes of Health, Bethesda, MD), followed by peroxidase-conjugated rabbit anti-goat IgG (Jackson Immu-noResearch). The reactive bands were detected by enhanced chemiluminescence, using reagents manufactured by Amersham International, and exposure to Rx medical x-ray film (Fuji). NADPH oxidase can be activated in a cell-free system by certain anionic amphiphiles, a mechanism which circumvents the transductional pathways active in the intact phagocyte (9, 10). We, therefore, examined the effect of AEBSF on O 2 . production in several forms of cell-free activation systems. The simplest one consisted of solubilized macrophage membranes and unfractionated cytosol, both derived from unstimulated cells. AEBSF was found to exhibit a concentration-related, inhibitory influence on LiDS-induced O 2 . production, with an IC 50 of 0.206 Ϯ 0.014 mM (n ϭ 8) (Fig. 1B). We next investigated the effect of AEBSF on a semi-recombinant cell-free system (11,12), consisting of either solubilized macrophage membrane or purified cytochrome b 559 , incorporated in phospholipid liposomes, combined with recombinant p47 phox , p67 phox , and Rac1-GTP␥S. AEBSF was found to act as an inhibitor of O 2 . production in this system, too, composed of four purified NADPH oxidase components in the virtual absence of any other protein (Fig. 1C). In the course of the latter experiments we became aware of an unexpected variability in the inhibitory potency of AEBSF which was not found in the cell-free system consisting of solubilized membrane and unfractionated cytosol. Analysis of this phenomenon revealed that, in the presence of a constant amount of membrane or purified cytochrome b 559 , the inhibitory effect of AEBSF was inversely related to the concentration of p47 phox . A more detailed exploration of the mechanism of this correlation is provided in the penultimate section.

AEBSF
To further ascertain that inhibition by AEBSF of O 2 . production in the cell-free system is due to a direct effect on NADPH oxidase, we examined the influence of AEBSF on two additional indicators of NADPH oxidase activation. These were LiDS-elicited NADPH oxidation (25) and LiDS-and NADPHdependent oxygen consumption (26), by mixtures of membrane and cytosol or membrane and recombinant cytosolic components. We found that AEBSF suppressed NADPH oxidation in a concentration-related manner; when cell-free mixtures were analyzed in parallel by ferricytochrome c reduction, the IC 50 of AEBSF values were identical in the two assays. Also, at 5 mM, AEBSF inhibited oxygen consumption by 95%. We concluded that AEBSF acts by interacting with one or more of the components of the NADPH oxidase complex. same kinetic characteristics (9,10). The effect of AEBSF (2 mM) on NADPH oxidase activation in the cell-free system was examined at concentrations of NADPH varying from 10 to 400 M and analyzed by Michaelis-Menten plotting. This demonstrated that inhibition of AEBSF was noncompetitive with NADPH. Membrane solubilization and purification of cytochrome b 559 results in a partial loss of FAD from the cytochrome (11,19,28), which explains the dependence of NADPH oxidase activation, in the cell-free system, on exogenous FAD (9). We, therefore, investigated whether AEBSF interfered with the reflavination of cytochrome b 559 , by assessing the effect of varying the concentration of exogenous FAD on the inhibitory effect of AEBSF. We found that the degree of inhibition by AEBSF (IC 50 ) was independent of the concentration of exogenous FAD (from 10 to 100 M). Finally, we explored the possibility that AEBSF competes with the activating amphiphile for interaction with a component of the NADPH oxidase complex. We found that increasing the concentration of LiDS, from the optimal activating range of 120 -140 M to up to 180 M, did not reverse or reduce the inhibitory effect of AEBSF (see also Fig. 4). AEBSF Interferes with Activation of NADPH Oxidase but Not with Electron Flow-Activation of NADPH oxidase, in intact cells and in broken cell preparations, is considered to be the result of the assembly of the individual components of the enzyme into a membrane-localized complex (reviewed in Ref. 8), this being followed by the induction of electron flow within the flavocytochrome b 559 dimer. The "activation" and "electron flow" stages can be separated in a two-step cell-free activation assay (24). This consists of exposing a mixture of NADPH oxidase components to an optimal concentration of activating amphiphile in a small volume (100 l) for a fixed time interval, in the absence of substrate NADPH, followed by 10-fold dilution and the initiation of electron flow by addition of NADPH. The 10-fold dilution results in a reduction in the concentration of amphiphile to a subactivating level and causes the virtual interruption of further assembly of the enzyme complex. We utilized this approach to identify the event affected by the inhibitor, by adding AEBSF either to the activation mixture at time 0 or after the completion of activation. As apparent in Fig.  2, AEBSF was a much more effective inhibitor when present during the assembly of the NADPH oxidase complex than when added at the completion of assembly. The finding that AEBSF was not totally inactive when added at the end of step one is probably due to the fact that some enzyme assembly continues taking place in the course of the second step. The ability of AEBSF to suppress O 2 . production when added to the components of the NADPH oxidase complex, in the absence of electron flow, also indicates that its mechanism of action is distinct from that of diphenylene iodonium, which acted as an inhibitor only under reducing conditions, such as those generated by NADPH oxidase turnover (29). Reversibility of Inhibition by AEBSF-Use was made of the two-step cell-free assay for determining whether inhibition of NADPH oxidase activation by AEBSF was reversible. For this purpose, activation mixtures were prepared in the absence and presence of a single concentration of AEBSF (2 mM) and incubated with an optimal concentration of LiDS for 5 min. Following this, the reaction mixtures were diluted in assay buffer to reduce the amphiphile concentration, and re-exposed or not to an activating concentration of LiDS for 3 min. The results illustrated in Fig. 3  Specificity of AEBSF-mediated Inhibition-AEBSF was developed originally as an irreversible serine protease inhibitor (30,31). It was, therefore, essential to establish whether its inhibitory effect on NADPH oxidase activation was related to its anti-protease activity. Hence, we examined the ability of five additional serine protease inhibitors to affect NADPH oxidase activation in the cell-free system. These were the irreversible inhibitors PMSF (0.1-2 mM) and TPCK (10 -200 M), the "mechanism based" irreversible inhibitor, DCIC (0.25 mM), and the reversible inhibitors aprotinin (0.3 M) and leupeptin (20 M). None of these exhibited an inhibitory effect on NADPH oxidase activation. We next explored the possibility that inhibition might be related to the concentration of activating amphiphile. As apparent in Fig. 4, PMSF, TPCK, and DCIC did not possess an inhibitory effect at any LiDS concentration whereas AEBSF was a potent inhibitor over the whole range of amphiphile concentrations. On the contrary, PMSF, TPCK, and DCIC had an enhancing effect on NADPH oxidase activation, at concentrations of LiDS below those required for maximal activation.
Structural Requirements for Inhibitory Action-We investigated the capacity of several analogues of AEBSF to interfere with the activation of NADPH oxidase. One group consisted of aminoethylbenzene derivatives and included: MAEBSF, identical to AEBSF except for the presence of an amino-linked methyl group; AEBSAc, a product of alkaline hydrolysis of AEBSF, found to be inactive as a protease inhibitor (31); and AEBSNH 2 , in which an amide group replaces the fluoride found in AEBSF, also reported to lack protease inhibitory activity (32). As apparent in Fig. 5A, AEBSAc and AEBSNH 2 were totally inactive, whereas MAEBSF exhibited an inhibitory potency identical to that of AEBSF (IC 50 ϭ 0.285 Ϯ 0.005 mM; n ϭ 3). The second group consisted of amidinobenzene derivatives. Among these, pAPMSF was reported to be a potent irreversible serine protease inhibitor (33), and benzamidine acts as a peptidase inhibitor (34). As seen in Fig. 5B, only pABSF was a moderate inhibitor of NADPH oxidase activation; its IC 50 (1.49 Ϯ 0.05; n ϭ 3) was 7.5-fold higher than that of AEBSF. We also examined the influence of an excess of the two inactive analogues, AEBSAc and AEBSNH 2 , sharing the aminoethylbenzene structure with AEBSF, on inhibition of NADPH oxidase activation by AEBSF. Neither of the two analogues, at 5 mM, was capable of reducing the inhibitory effect of AEBSF, assayed at concentrations ranging from 0.1 to 5 mM. These results demonstrate that a sulfonyl fluoride group, adjacent to the benzene ring, is essential for inhibitory activity and that aminoalkylbenzene derivatives are more potent inhib- itors than amidinobenzene compounds. Identification of the Molecular Target of AEBSF-We next investigated whether interaction between AEBSF and a specific NADPH oxidase component can be demonstrated. The basic design of these experiments was to expose subcellular fractions or individual NADPH oxidase components to AEBSF for a fixed time interval in a small volume, followed by 10-fold dilution into assay buffer containing the untreated complementary fraction or a mixture of the other, untreated NADPH oxidase components, activating LiDS and NADPH. Dilution resulted in a reduction in AEBSF concentration to subinhibitory levels, limiting its effect to the component initially exposed to the compound. We examined the effect of AEBSF on total membrane, total cytosol, relipidated cytochrome b 559 , and recombinant p47 phox , p67 phox , and Rac1-GTP␥S. Each of these was exposed or not to AEBSF in the absence or presence of LiDS, the maximal concentration of amphiphile being chosen not to exceed that permitting the recovery of NADPH oxidase activity, upon complementation with untreated components.
As apparent in Table I, exposure of total solubilized membrane or purified and relipidated cytochrome b 559 to AEBSF, in the presence but not in the absence of LiDS, resulted in marked inhibition of their capacity to support O 2 . production upon complementation with cytosol and activation by LiDS. Total cytosol as well as p47 phox and Rac1 were insensitive to treatment by AEBSF; a minor effect of AEBSF on p67 phox was evident but this result was difficult to interpret because of considerable interexperimental variability and because, among all NADPH oxidase components, p67 phox was the most sensitive to denaturation by LiDS. The ability of AEBSF to inactivate cytochrome b 559 was investigated in more detail by assessing the dose-dependence of the inactivation of solubilized membrane by AEBSF, in the presence and absence of LiDS. As seen in Fig. 6, AEBSF inhibited the capacity of the membrane to support O 2 . production only in the presence of LiDS, with an IC 50 of 2.18 mM. AEBSF Interferes with the Interaction between Cytochrome b 559 and p47 phox and/or p67 phox -This series of experiments was initiated as an extension of the preliminary finding that, in the presence of a constant amount of cytochrome b 559 , the inhibitory effect of AEBSF could be overcome by increasing the concentration of cytosolic component p47 phox . The consensus opinion is that an essential event in the assembly of the NADPH oxidase complex is the binding of p47 phox to cytochrome b 559 (reviewed in Ref. 8). In light of the results described in the preceding section, it seemed likely that AEBSF competes with p47 phox for interaction with cytochrome b 559 . We, therefore, investigated the influence of varying the concentration of p47 phox on the inhibition of NADPH oxidase activation by AEBSF, in the presence of a constant amount of membrane and saturating concentrations of p67 phox and Rac1-GTP␥S (Fig. 7A). In additional experiments, we varied the concentrations p67 phox (Fig. 7B) and Rac1-GTP␥S (Fig. 7C), in the presence of saturating concentrations of the other two cytosolic components. In all experiments, three concentrations of cytosolic component were chosen, two of which were on the ascending slope of previously established dose-response curves, whereas the third one represented a saturating concentration. The data were analyzed by plotting the "concentration of AEBSF" (on a logarithmic scale) against "inhibition of NADPH oxidase activation" and IC 50 of AEBSF values, for each concentration of cytosolic component, were calculated. As seen in Fig.  7A, when p47 phox was present in subsaturating concentrations, sigmoid curves were generated and IC 50 values for AEBSF decreased with decreasing concentrations of p47 phox . IC 50 of AEBSF values were moderately reduced by lowering the concentration of p67 phox (Fig. 7B) and not significantly affected by varying the concentration of Rac1-GTP␥S (Fig. 7C). These results are compatible with the hypothesis that AEBSF interferes with the binding of p47 phox to cytochrome b 559 but we production in the cell-free system Various components, at the concentrations shown in the footnotes, were incubated with 2 mM AEBSF for 5 min, at room temperature, in a volume of 100 l, in the absence or presence of LiDS, at concentrations shown in the footnotes. This was followed by the addition of the missing untreated NADPH oxidase components at saturating concentration in 0.9 ml of assay buffer containing a concentration of LiDS, to result in a final concentration of 120 M. Preparations, not exposed to AEBSF, were supplemented, at this stage, with 0.2 mM AEBSF, as a control for an eventual effect of AEBSF on electron flow. O 2 . production was elicited 90 s later by the addition of 0. . In 100 l, was exposed to AEBSF in the absence or presence of 300 M LiDS. It was complemented with cytosol (equivalent to 4.5 ϫ 10 6 cells) in 0.9 ml.
b Cytosol (equivalent to 4.5 ϫ 10 6 cells), in 100 l, was exposed to AEBSF in the absence or presence of 120 M LiDS. It was complemented with solubilized membrane (4.5 pmol of cytochrome b 599 ), in 0.9 ml.
c Purified cytochrome b 559 (4.92 pmol), in 100 l, was exposed to AEBSF in the absence or presence of 150 M LiDS. It was complemented with cytosol (equivalent to 6 ϫ 10 6 cells), in 0.9 ml. d p47 phox (12.2 pmol), in 100 l, was exposed to AEBSF in the absence or presence of 140 M LiDS. It was complemented with the missing components, in 0.9 ml, at the following final concentrations: solubilized membrane (2.4 nM cytochrome b 599 ), p67 phox (6 nM), and Rac1-GTP␥S (100 nM). e p67 phox (6 pmol), in 100 l, was exposed to AEBSF in the absence or presence of 140 M LiDS. It was complemented with the missing components in 0.9 ml, at the following final concentrations: solubilized membrane (2.4 nM cytochrome b 599 ), p47 phox (12.2 nM), and Rac1-GTP␥S (100 nM).
f Rac1-GTP␥S (100 pmol), in 100 l, was exposed to AEBSF in the absence or presence of 300 M LiDS. It was complemented with the missing components in 0.9 ml, at the following final concentrations: solubilized membrane (2.4 nM cytochrome b 559 ), p47 phox (12.2 nM), and p67 phox (6 nM). cannot rule out the possibility that AEBSF also inhibits a direct interaction between p67 phox and cytochrome b 559 (see "Discussion").
The hypothesis that AEBSF acts as a competitive inhibitor of p47 phox was further tested by performing a Michaelis-Menten analysis of the effect of AEBSF on the activation of NADPH oxidase at four concentrations of p47 phox (14,21,28, and 35 nM), all within the ascending slope of the concentration-response curve. As seen in the Lineweaver-Burk plot in Fig. 8, AEBSF, in the concentration range of 0.5 to 2 mM, behaves as a competitive inhibitor with respect to p47 phox . At concentrations of AEBSF exceeding 2 mM, a noncompetitive or mixed type of inhibition pattern was obtained. Similar kinetic analysis of the effect of AEBSF on NADPH oxidase activation at varying concentrations of either p67 phox or Rac1-GTP␥S revealed noncompetitive or mixed type inhibition patterns.
AEBSF Prevents Amphiphile-dependent Translocation of p47 phox and p67 phox to the Membrane-Activation of NADPH oxidase in intact cells is accompanied by translocation of fractions of p47 phox and p67 phox to the plasma membrane (35). The anchoring of both cytosolic components to the membrane is dependent on the presence of cytochrome b 559 (36). Also, translocation of p67 phox to the membrane occurs only in the presence of p47 phox (36), suggesting that at least one of the functions of p47 phox is to serve as an escort protein for p67 phox . It was, therefore, of interest to investigate the effect of AEBSF on the translocation of cytosolic components to the membrane in the cell-free system (37). As seen in Fig. 9, the addition of AEBSF to mixtures of macrophage membranes and cytosol, activated by a concentration of LiDS to result in maximal NADPH-dependent O 2 . production, markedly inhibited the translocation of both p47 phox and p67 phox to the membrane. Parallel testing of the sedimented membranes, derived from cell-free mixtures activated in the absence and presence of 4.5 mM AEBSF, for O 2 . production, showed that lack of translocation was accompanied by an 82% inhibition of NADPH oxidase activation. As apparent in Fig. 9B, translocation of p47 phox and p67 phox (and its inhibition by AEBSF) was strictly dependent on the presence of membranes and was not the result of nonspecific precipitation of cytosolic components by LiDS (35). 6. Treatment of the membrane fraction with AEBSF results in impairment of its ability to cooperate with recombinant cytosolic components in NADPH oxidase assembly. Solubilized membrane (3.45 pmol of cytochrome b 559 ) was exposed or not to various concentrations of AEBSF (0.5-10 mM), in the presence or absence of 300 M LiDS, for 5 min at room temperature, in a total volume of 100 l. This was followed by the addition of p47 phox (24 nM), p67 phox (19.8 nM), and Rac1-GTP␥S (108 nM) in 0.9 ml of assay buffer containing 120 M LiDS, bringing the total volume of the assay to 1 ml, and the mixture was kept for a further 90 s at room temperature, at which time O 2 .
production was elicited by the addition of 0.2 mM NADPH. Reaction mixtures which contained solubilized membrane not pre-exposed to AEBSF were supplemented with AEBSF, at one-tenth of the concentration added to the membranes, at the stage of adding the cytosolic components, as a control for an effect of AEBSF on cytosolic components or electron flow. The figure illustrates the results of a typical experiment, with each measurement being performed in duplicate. For each measurement, the results are expressed as % inhibition of NADPH oxidase activation by relating NADPH oxidase activity of mixtures, containing membrane pretreated with AEBSF, to that of control preparations containing untreated membrane interacting with cytosolic components in the presence of AEBSF at 10-fold lower concentration.

DISCUSSION
We found that AEBSF, originally developed as a watersoluble irreversible serine protease inhibitor of relatively low toxicity, interferes with the activation of the O 2 . generating NADPH oxidase both in intact macrophages and in cell-free systems in a concentration-dependent manner. A remarkable feature of this inhibition is that it affects all forms of cell-free activation, from broken cell systems, consisting of total membrane and unfractionated cytosol, to mixtures of purified cytochrome b 559 and three recombinant cytosolic components. While this paper was in preparation, we became aware of the report by Remold-O'Donnell and Parent (38), describing the inhibitory effect of 2 mM AEBSF on O 2 . production by human neutrophils in response to PMA and opsonized zymosan. AEBSF (0.1-0.2 mM) was recently found to block priming of human monocytes for enhanced O 2 . production in response to PMA and fMLP but was reported not to affect the response of unprimed cells, in the above concentration range (32). There is a large body of literature linking the elicitation of an oxidative burst to the activation of cellular proteases (summarized in Ref. 39). Some of the evidence for such a connection is based on the effect of synthetic or natural protease inhibitors, particularly serine protease inhibitors, on oxygen radical production by intact phagocytes. Results of such experiments have to be interpreted with caution since it became evident that the effects of protease inhibitors on O 2 . production were frequently unrelated to their antiproteolytic activity (39 -42). Our results indicate, indeed, that AEBSF affects NADPH oxidase activation, at least in cell-free preparations, by a mechanism unrelated to its protease inhibitory activity. This claim is supported by the following arguments: (a) five other serine protease inhibitors (PMSF, TPCK, DCIC, leupeptin, and aprotinin) were found to be inactive when tested under identical conditions (it is of interest that DCIC which, like AEBSF, was found to block O 2 . production by intact neutrophils (38), did not inhibit NADPH oxidase activation in the cell-free system); (b) among protease inhibitors, structurally related to AEBSF (PMSF and pAPMSF), there was no relation between NADPH oxidase activation inhibitory and antiproteolytic activities, and (c) AEBSF was inhibitory in a cell-free system composed exclusively of purified native and recombinant proteins, and, there-fore, free of contaminating proteases. The inhibitory effect of AEBSF on NADPH oxidase activation exhibited a pronounced structural specificity. A number of derivatives of benzylamine and benzamidine, to which categories AEBSF is related, were shown to function as competitive serine protease inhibitors (43). The presence of a reactive sulfonyl fluoride group on the aromatic ring generates compounds capable of forming a covalent bond with their target enzymes (30,44). We found that, among four compounds sharing an aminoethylbenzene moiety (AEBSF, AEBSAc, AEBSNH 2 , and MAEBSF), only AEBSF and MAEBSF, which possess a reactive sulfonyl fluoride group, were inhibitory. Among three amidinobenzene derivatives (benzamidine, pAPMSF, and pABSF) only pABSF, possessing a sulfonyl fluoride group adjacent to the benzene ring exhibited moderate NADPH oxidase inhibitory activity. PMSF and TPCK, two serine protease inhibitors bearing limited structural similarity to AEBSF, but lacking the basic amino group present in AEBSF, were found to be inactive in the cell-free system. We conclude that two structural elements are the minimal prerequisites for the possession of NADPH oxidase inhibitory property: a positively charged aminoalkyl moiety and a reactive sulfonyl fluoride group.
The search for the molecular target of AEBSF led to the conclusion that this is, most likely, cytochrome b 559 . This pro- FIG. 9. AEBSF interferes with the amphiphile-dependent translocation of p47 phox and p67 phox to the membrane in a cellfree system. A, macrophage membranes were incubated with total cytosol for 5 min at 24°C, in the absence or presence of 185 M LiDS, and in the absence or presence of 4.5 mM AEBSF. Translocation of p47 phox and p67 phox from cytosol to membranes was assessed as described under "Experimental Procedures." B, in this experiment, identical in design to that described in panel A, a mock control was included to demonstrate that LiDS-dependent translocation of p47 phox and p67 phox and its inhibition by AEBSF occur only in the presence of membranes. The band cathodal to p67 phox is unrelated to p67 phox and was detected independently of the presence of membrane or activation by LiDS.
posal is based principally on the finding that treatment of either total membrane fraction or purified and relipidated cytochrome b 559 with AEBSF, in the presence of a critical concentration of LiDS, resulted in a dose-dependent reduction in their ability to cooperate with untreated components in the assembly of NADPH oxidase. Neither total cytosol nor any of the three cytosolic components were sensitive to AEBSF to a comparable degree. A remarkable feature of this effect is its absolute dependence on the simultaneous presence of an anionic amphiphile. The amphiphile may make the liposome-enclosed cytochrome b 559 more accessible to AEBSF or may induce a conformational change in cytochrome b 559 itself, normally associated with the process of activation, but also leading to exposure of an AEBSF-binding domain. Evidence for a direct effect of anionic amphiphiles, at NADPH oxidase activating concentrations, on cytochrome b 559 was recently presented by the demonstration that arachidonate and SDS enhance interaction of cytochrome b 559 with the NADPH oxidase inhibitory heme ligand, butyl isocyanide (45).
The results of systematic screening of the effect of varying the concentration of each of the three cytosolic components on the IC 50 of AEBSF support the proposal that AEBSF prevents the binding of p47 phox to cytochrome b 559 and, consequently, interferes with the assembly of a functionally active NADPH oxidase complex. Such a mechanism is also in agreement with the Michaelis-Menten analysis of the inhibition of NADPH oxidase activation by AEBSF at various concentrations of p47 phox . There is extensive evidence demonstrating interaction between p47 phox and the small (46,47) and large (48 -50) subunits of cytochrome b 559 . Several groups also reported that, in p47 phox , a region extending from residues 315 to 347, or part of it, represents a site of interaction with cytochrome b 559 (51)(52)(53). This region is rich in basic residues, especially arginines, and it is possible that it forms electrostatic bonds with acidic residues on one of the two subunits of cytochrome b 559 .
However, the IC 50 of AEBSF is also affected, albeit to a lesser degree, by the concentration of p67 phox . This effect can be explained by assuming that p67 phox binds to cytochrome b 559 and that AEBSF interferes with this binding. There is, indeed, recent functional evidence for a direct interaction between p67 phox and cytochrome b 559 (54,55). An alternative explanation could be provided by an indirect effect of varying the concentration of p67 phox on the binding of p47 phox to cytochrome b 559 . Thus, based on kinetic evidence, an enhancement of p47 phox binding to cytochrome b 559 by p67 phox was demonstrated (56), and a marked reduction in the translocation of p47 phox to the plasma membrane in neutrophils of p67 phoxdeficient patients, was recently reported (cited in Ref. 57). Our finding that AEBSF inhibits the activation-related translocation of both p47 phox and p67 phox to the membrane can be explained by either distinct effects on the binding of each of the two cytosolic components to cytochrome b 559 or by an effect on the binding of one component affecting the binding of the other component.
When working as a protease inhibitor, AEBSF acts as a bifunctional reagent (30). The charged aminoethyl moiety acts as a substrate analogue, capable of forming ionic complexes with trypsin and related proteases because of its similarity to basic amino acids, whereas the chemically reactive sulfonyl fluoride group endows the inhibitor with the capacity of forming a stable covalent bond with the enzyme (30,43,44). We propose that AEBSF also acts as a bifunctional inhibitor of NADPH oxidase assembly, binding by the intermediary of its aminoethyl end, to an acidic domain on one of the subunits of cytochrome b 559 . This, essentially reversible step is followed by the formation of an irreversible bond, mediated by the sulfonyl fluoride group. The finding that, in the 0.5-2 mM range, AEBSF behaved as a competitive inhibitor, with respect to p47 phox , but that this behavior was replaced by a noncompetitive pattern, at higher concentrations, suggests that the establishment of stable bonds becomes increasingly dominant with increasing concentrations of AEBSF. While this manuscript was under review, we became aware of a recent communication 2 that AEBSF, under certain conditions (such as, at high concentrations), forms covalent adducts with proteins, other than serine proteases. These interactions appear to involve tyrosine and lysine residues and the free amino terminus.
Further exploration of the mechanism of inhibition of NADPH oxidase by AEBSF in particular and by aminoalkylbenzenesulfonyl fluorides in general is of both theoretical and practical relevance. Thus, it should lead to a better understanding of the molecular basis of the assembly of the NADPH oxidase complex and to the design of drugs to be used in circumstances associated with the undesired production of oxygen radicals. This latter aim, however, is dependent on the clarification of the relevance of effects found in the cell-free system to the situation in intact cells. Nevertheless, the fact that AEBSF is a stable, water-soluble and relatively nontoxic compound, possessing both NADPH oxidase suppressory and anti-proteolytic properties, makes it an interesting model compound for the design of antiinflammatory drugs.