The gp91 phox Component of NADPH Oxidase Is Not the Voltage-gated Proton Channel in Phagocytes, but It Helps*

During the “respiratory burst,” the NADPH oxidase complex of phagocytes produces reactive oxygen spe-cies that kill bacteria and other invaders Electron efflux through NADPH oxidase is electrogenic and is compensated by H (cid:1) efflux through proton channels that reportedly are contained within the gp91 phox subunit of NADPH oxidase. To test whether gp91 phox functions as a proton channel, we studied H (cid:1) currents in granulocytes from X-linked chronic granulomatous disease patients lacking gp91 phox (X-CGD), the human myelocytic PLB-985 cell line, PLB-985 cells in which gp91 phox was knocked out by gene targeting (PL-B KO ), and PLB-985 knockout cells re-transfected with gp91 phox (PLB 91 ). H (cid:1) currents in unstimulated PLB

A voltage-gated proton conductance is activated during the respiratory burst in human neutrophils (1)(2)(3)(4)(5)(6). The resulting H ϩ efflux compensates for the electrogenic action of NADPH oxidase (1). Several lines of evidence have suggested that the gp91 phox component of the NADPH oxidase complex might be the proton channel that is activated during the respiratory burst (2,3,7,8). The presence of H ϩ currents in monocytes from gp91 phox -deficient CGD 1 patients appeared to refute this idea (9), but Henderson and Chappell (10) argued that these data were inconclusive. Furthermore, it has been reported that heterologous expression of gp91 phox results in the appearance of proton fluxes or proton currents resembling those activated during the respiratory burst (8,(11)(12)(13)(14)(15). The expression systems employed to date provide ambiguous results, because CHO and HEK-293 cells express endogenous voltage-gated proton channels (14 -17) and mRNA for gp91 phox , and four gp91 phox homologs have been detected by reverse transcriptase polymerase chain reaction in HEK-293 cells (18). An increase in H ϩ currents after transfection might reflect expression of channels formed by the transfected gene product but could simply reflect up-regulation of constitutively expressed H ϩ channels. It is also possible that expression of gp91 phox in a background lacking p22 phox might induce non-physiological behavior that is not exhibited in phagocytes. The stability of gp91 phox and p22 phox expression in phagocytes is enhanced by the formation of heterodimers of these two components of flavocytochrome b 558 (19,20). We therefore studied stable PLB-985 cell lines with gp91 phox genetically knocked-out and with gp91 phox re-expressed in the same background (21).

EXPERIMENTAL PROCEDURES
Cells-The PLB-985-derived cell lines were developed by Dinauer and colleagues (21). Wild-type PLB-985 cells (PLB WT ), PLB KO (PLB-985 cells targeted with a construct that prevents gp91 phox expression), and PLB 91 (gp91 phox knockout cells after rescue by stable transfection with gp91 phox cDNA) were all induced by incubation with 0.5% N,N-dimethylformamide (DMF; Sigma) for 4 -7 days. Some whole-cell studies were done on PLB KO cells before DMF induction, designated PLB KO* . The absence of gp91 phox expression in the PLB KO granulocytes is well documented (20 -23). X-CGD granulocytes (mainly neutrophils) were isolated by density gradient centrifugation as described (24) from three patients with CGD, all of whom had documented absent neutrophil superoxide production and mutations that would prevent stable expression of gp91 phox (25). The specific mutations were (a) Cys 1347 3 Ala in Exon 11, changing the codon for Cys 445 to a premature STOP codon; (b) deletion of Cys 1028 in Exon 9, leading to a frameshift after Pro 339 and a premature STOP three codons downstream; and (c) insertion of Cys after Gly 169 in Exon 3, leading to a frameshift after Leu 52 and a premature STOP codon in Exon 5. In patient c, the absence of cytochrome b 558 was demonstrated spectrophotometrically in neutrophil extracts. Blood from patient c was refrigerated overnight before use, and most surviving granulocytes were identified as eosinophils in a Wrightstained cytospin preparation.
Electrophysiology-Whole-cell and permeabilized patch voltageclamp recordings were done as described (26,27) with micropipettes pulled from 7052 glass (Garner Glass). Whole-cell solutions (pipette and bath) included 100 mM buffer near its pK a with tetramethylammonium ϩ and methanesulfonate Ϫ as the main ions, 1 mM EGTA, and 1-2 mM * This work was supported in part by the NHLBI, National Institutes of Health (to T. E. D. and M. C. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (27). Currents are shown without correction for leak or liquid junction potentials. Data were collected at 20 -21°C or at room temperature. (21,22), Express Large Voltage-gated Proton Currents-PLB-985 cells induced by DMF to granulocytic differentiation express all NADPH oxidase components and are capable of a respiratory burst (21). PLB WT cells had large voltage-gated proton currents ( Fig. 1, A and B) that resemble those in other phagocytes and related cells (17). Proton currents in DMF-induced PLB 91 cells ( Fig. 1, C and D) were similar to those in DMF-induced PLB WT cells, as expected. PLB KO cells, which do not express gp91 phox protein (21,22), also had large H ϩ currents both before ( Fig. 1, E and F) and after induction with DMF. These results demonstrate unequivocally that gp91 phox is not the voltage-gated proton channel in unstimulated phagocytes.

PLB KO Cells, Which Lack the gp91 phox Protein
The Selectivity and Gating Kinetics of Voltage-gated Proton Channels Are Identical Regardless of Whether gp91 phox Is Present-To explore whether expression of gp91 phox might alter the properties of H ϩ channels, we characterized the H ϩ currents thoroughly. Tail currents reversed near the Nernst potential for H ϩ in the three PLB lines ( Fig. 2A), confirming that protons carry these currents. The slope of the data is 51.8 mV/unit pH, which is close to the 58.2 mV given by the Nernst equation. The largest deviation from the Nernst prediction indicates that H ϩ is Ͼ10 6 more permeant than tetramethylammonium ϩ , the main cation present. Like other H ϩ channels (17), those in PLB cells are essentially perfectly H ϩ -selective.
The voltage dependence of H ϩ current activation was very similar in PLB WT , PLB 91 , and PLB KO cells, as evident in average H ϩ chord-conductance voltage (g H -V) data (Fig. 2B). H ϩ currents in PLB knockout cells studied before (PLB KO ) and after induction with DMF (PLB KO* ) were identical. The effects of changing pH o from 7.0 ( Fig. 1, A, C, and E) to 5.5 (Fig. 1, B, D, and F) were similar in all cell types and to effects reported previously (17,28).
The behavior of the g H in cells studied at pH i 6.5 (not shown) was also similar in PLB 91 and PLB KO cells and to that described previously (17,28). Fig. 2, C and D shows that the kinetics of H ϩ channel opening ( act ) and closing ( tail ), respectively, were indistinguishable in PLB-985 cells expressing (PLB WT and PLB 91 ) or lacking gp91 phox (PLB KO and PLB KO* ). Thus, the physiological properties of H ϩ channels in unstimulated phagocytes are not altered by gp91 phox expression.

Activation of NADPH Oxidase by PMA Can Be Detected as an Electron Current in PLB 91 Cells Studied in Permeabilized
Patch Configuration-The response of individual PLB 91 cells to PMA was variable, possibly reflecting variable levels of induction by DMF. We observed DPI-sensitive electron currents, which reflect NADPH oxidase activity (24, 27, 29 -31), at the holding potential in about half (9 of 17) of PLB 91 cells stimulated with PMA. Electron currents usually appeared after a delay (up to 10 min) and in conjunction with a slowing of tail current decay. The average peak electron current was Ϫ2.4 Ϯ 1.8 pA (mean Ϯ S.D.; n ϭ 9), similar to Ϫ2.3 pA in human neutrophils (27). This similarity is consistent with the similar levels of superoxide anion production in PMA-stimulated PLB-985 cells and human neutrophils (21).
H ϩ Currents in PLB 91 Cells Studied in Permeabilized Patch Configuration Are Enhanced by PMA-The demonstration that gp91 phox is not the voltage-gated proton channel in unstimulated PLB-985 cells is compatible with a recent suggestion that two types of H ϩ channels exist in phagocytes and that gp91 phox functions as a proton channel only when NADPH oxidase is active (29). In human neutrophils or eosinophils studied in permeabilized patch configuration, both NADPH oxidase and H ϩ channels can be activated by PMA or arachidonic acid (24, gp91 phox Is Not a Proton Channel in Phagocytes 36064 27,31). The H ϩ currents in these activated phagocytes closely resemble the NADPH oxidase-related variety described by Bá nfi et al. (29). The H ϩ current response of PLB 91 cells to PMA was qualitatively like that of human neutrophils and eosinophils (24,27). Fig. 3A illustrates H ϩ currents during test pulses to ϩ60 mV in a PLB 91 cell. PMA stimulation produced four changes in H ϩ currents in PLB 91 cells that displayed electron currents as follows (Fig. 3C): (a) the H ϩ current amplitude (I H ) increased; (b) activation of H ϩ current during depolarizing pulses ( act ) became faster; (c) deactivation of H ϩ currents ( tail ) became slower; and (d) the threshold for activating H ϩ currents (V threshold ) was shifted 32 mV toward more negative voltages. Each change increases the likelihood of H ϩ channel opening in intact cells.

PMA Increases H ϩ Currents in PLB-985 Cells and Human Neutrophils to the Same Extent Regardless of Whether gp91 phox
Is Present-Because the increase in I H after PMA stimulation might reflect the appearance of a distinct type of proton channel related to gp91 phox (29), evaluating the PMA response of PLB KO cells was of great interest (Fig. 3B). No electron current was detected, consistent with the absence of a complete NADPH oxidase complex. PMA stimulation increased I H to a similar extent in PLB KO , X-CGD, PLB 91 cells, and neutrophils (Fig. 3C). Because I H during a test pulse is an arbitrary measure, we also compared the maximum g H , which increased after PMA stimulation by a factor of 2.06 Ϯ 0.39 (mean Ϯ S.D.; n ϭ 7) in PLB KO cells and 2.43 Ϯ 1.37 (n ϭ 9) in PLB 91 cells (p Ͼ 0.5). In all cell types, H ϩ current activation became faster. In parallel, we studied granulocytes from three CGD patients with mutations that prevent expression of gp91 phox (X-CGD).
The X-CGD cells had normal or larger than normal H ϩ currents, and their response to PMA was similar to that of PLB KO cells. Although the mean change in act was larger in PLB 91 than PLB KO or X-CGD cells, our exclusion from analysis of PLB 91 cells without electron currents may account for this difference, because this criterion could not be used to exclude non-responding PLB KO or X-CGD cells. The PMA-induced changes in H ϩ currents in the eight PLB 91 cells with no detectable electron currents (not shown) were identical to those in PLB KO cells. Because I H increased to a similar extent in PLB KO , X-CGD, PLB 91 cells, and human neutrophils, the increased g H during the respiratory burst (1,2,4,5) is not because of the appearance of proton currents conducted through the gp91 phox molecule.
PMA Elicits Fewer Changes in Gating Kinetics of Proton Channels in gp91 phox -deficient Cells-Although I H increased to the same extent after PMA stimulation, the response of H ϩ currents to PMA was different in cells expressing or lacking gp91 phox . The slowing of tail and large hyperpolarizing shift of V threshold were not observed in PLB KO or X-CGD cells (Fig. 3C). The slowing of tail was less pronounced in PLB 91 cells than in neutrophils and eosinophils. In most cells there was a distinct but relatively subtle slowing. The hyperpolarizing voltage shift was almost as large in PLB 91 cells (Ϫ32 mV) as in neutrophils (Ϫ39 mV) and in eosinophils (Ϫ43 mV) stimulated with PMA under similar conditions (24,27). This voltage shift was sufficient to result in the appearance of inward H ϩ currents in some cells, a hallmark property of the NADPH oxidase-related H ϩ channel (29). DISCUSSION The presence of robust H ϩ currents in PLB KO cells demonstrates unequivocally that the voltage-gated proton channel in unstimulated phagocytes is not gp91 phox nor does it require gp91 phox expression. Similarly, granulocytes (this study) or monocytes (9) from CGD patients lacking gp91 phox exhibit normal levels of H ϩ currents. Furthermore, genetic knockout of gp91 phox did not detectably alter the amplitude or behavior of whole-cell H ϩ currents. Voltage-gated proton channels in whole-cell studies of unstimulated phagocytes function independently of gp91 phox .
Bá nfi et al. (29) proposed that there were two types of H ϩ channels in eosinophils, one in resting cells and a novel variety that is observed only under conditions that permit NADPH oxidase function. This novel channel reportedly differs from that in resting cells in (a) activating at more negative voltages, (b) activating more rapidly, (c) deactivating more slowly, and (d) being more sensitive to inhibition by Zn 2ϩ . We observed novel H ϩ channel gating behavior during NADPH oxidase function in human neutrophils and eosinophils stimulated with PMA or arachidonic acid in permeabilized patch studies (24,27,31). However, we saw no evidence of multiple kinetic components in stimulated phagocytes, no correlation between the amplitude of the NADPH oxidase-generated electron currents and the amplitude of PMA-activated H ϩ currents (27), and identical Zn 2ϩ sensitivity of H ϩ currents in resting and activated cells displaying both types of channel behavior (24). We conclude that there is one type of H ϩ channel in phagocytes, whose properties are greatly altered during the respiratory burst.
Here we examined whether the increased H ϩ conductance in stimulated cells is because of the appearance of additional channels formed by gp91 phox . PMA stimulation clearly increased I H in cells that lack gp91 phox (PLB KO and X-CGD). This increase was not statistically different from that in cells expressing gp91 phox (PLB 91 and neutrophils). If a small gp91 phoxmediated H ϩ conductance were also activated in neutrophils . Test pulses to ϩ60 mV were applied before and after application of 60 nM PMA (currents labeled with the time after treatment). Note the slowing of tail current decay in A but not in B. The response of the cell in A was larger than typical but was selected to illustrate the changes in gating kinetics observed. C, average changes in H ϩ current kinetics ( act and tail ), amplitude (I H ), and threshold voltage (V thr ; right axis) after PMA stimulation, compared with previously published data from human neutrophils (27) (PMN) and human eosinophils (24) (EOS). The mean Ϯ S.E. ratio of the peak response, usually measured 5-10 min after PMA addition, to the control measurement is plotted. Numbers of cells are as follows: X-CGD, 6 (two from each patient); PLB KO , 7; PLB 91 , 9; PMN, 11-14; EOS, 12-14. Only PLB 91 cells exhibiting electron currents upon stimulation with PMA are included; those without electron currents (not shown) responded identically to PLB KO cells.

gp91 phox Is Not a Proton Channel in Phagocytes 36065
and PLB 91 cells, it could be only a small fraction of the total g H . It is conceivable that under some conditions, such as heterologous expression in non-phagocytes, gp91 phox might function as a proton channel, but the evidence presented here indicates that it does not contribute significantly to the total proton conductance of phagocytes.
In CHO cells transfected with gp91 phox , arachidonic acid stimulated larger proton fluxes than in control cells (8,11,12). Although suggestive of enhanced H ϩ channel activity, these measurements are indirect. It is difficult to determine which part of these H ϩ fluxes was mediated by H ϩ channels, because suppression of flux by the H ϩ channel inhibitor Zn 2ϩ was not demonstrated. Patch-clamp studies of CHO cells transfected with gp91 phox (13) reveal a large conductance with properties fundamentally different from H ϩ channels in native cells. A quintessential feature of H ϩ channels is potent inhibition by Zn 2ϩ , which slows act (17,26). The conductance in CHO cells was weakly inhibited by Zn 2ϩ , and no slowing of activation was evident at 200 M Zn 2ϩ (13), whereas even 1 M Zn 2ϩ slows act 3-10-fold in cells expressing voltage-gated proton channels (24,26). In all cells with H ϩ channels, increasing pH i shifts the voltage-activation curve by ϳ40 mV/unit pH (16,17,28). In contrast, the conductance in CHO cells was activated at Ϫ20 mV at pH i 6.9, but no H ϩ current was seen at pH i 7.5 at voltages up to ϩ140 mV (13). The failure to see H ϩ current at pH i 7.5 is especially surprising, because the currents at pH i 6.9 are an order of magnitude greater than in any mammalian cell. Finally, the outward currents in CHO cells activate anomalously rapidly, within Ͻ100 ms, whereas act for phagocyte H ϩ channels is typically seconds (9,17,24,27,29,31). It was reported recently that transient gp91 phox expression in COS-7 cells results in voltage-gated proton currents (15). However, the currents shown appear to reverse roughly near 0 mV at pH o 7.5 and pH i 5.7, where the Nernst potential for H ϩ is Ϫ105 mV; thus, this conductance is not H ϩ -selective. Evidently, expression of gp91 phox in alien cell lines can induce novel conductances that differ markedly from H ϩ currents in resting or activated phagocytes or any cell studied to date.
The gating kinetics of H ϩ channels responded differently to PMA in cells lacking gp91 phox . Although it is possible that gp91 phox itself modulates H ϩ channels, we propose that these modulations of H ϩ channel function occur only in the presence of a functioning NADPH oxidase complex. The properties that are influenced by NADPH oxidase function, slower tail and hyperpolarization of the g H -V relationship, promote activation of the g H at membrane potentials that might occur in intact phagocytes. The alterations in H ϩ channel gating during NADPH oxidase activity probably contribute more to activating H ϩ flux during the respiratory burst than does the increase in g H, max .