Role of histidines identified by mutagenesis in the NADPH oxidase-associated H+ channel.

The efflux of protons through a H+ channel acts as the charge compensation pathway for the electrogenic generation of superoxide (O-2) by human neutrophil NADPH oxidase. It has previously been shown that the N-terminal 230 amino acids of the product of the X-linked chronic granulomatous gene gp91(phox) contain all that is required for it to function as the arachidonate-activable, NADPH oxidase-associated H+ channel (Henderson, L. M., Thomas, S., Banting, G., and Chappell, J. B. (1997) Biochem. J. 325, 701-705). To identify functionally important amino acids, Chinese hamster ovary (CHO) cell lines were constructed that expressed point mutations in the N terminus of gp91(phox). No H+ flux was observed in CHO cell lines expressing the N-terminal gp91(phox) mutants H111L, H115L, and H119L, or H115L, or H115K. Partial retention of H+ channel function was, however, observed in the H115D CHO cell line. The addition of arachidonic acid to R91E,R92E CHO cells elicited a full H+ channel response. The buffering capacity and response of 2', 7'-bis(2-carboxyethyl)-5(6)-carboxyfluorescein to H+ were the same in all cell lines. Therefore, it can be concluded that His-115 is important to the ability of gp91(phox) to function as the NADPH oxidase-associated H+ channel and that the mechanism of H+ conduction involves protonation and deprotonation of an amino acid with an appropriate pK value.

An integral membrane component of the NADPH oxidase, gp91 phox (phagocyte oxidase), has previously been demonstrated to function as the oxidase-associated H ϩ channel (1). The NADPH oxidase of phagocytes (neutrophils, monocytes, macrophages, and eosinophils) generates superoxide (O 2 . ) as part of the cellular immune response of the body. Electrons, donated by cytosolic NADPH, are utilized by the oxidase to perform the single electron reduction of O 2 to O 2 . (an oxygen free radical) on the external face of the membrane. The activity of the NADPH oxidase has been shown to be electrogenic (2,3) and is associated with a rapid depolarization of the membrane potential (⌬⌿ ϭ Ϫ60 to Ϫ15 mV) (2) and a slight fall in pH i (4). It has previously been shown that an efflux of H ϩ ions (oxidation of NADPH) through a channel provides the necessary charge compensation for the activity of the oxidase and prevents a large and rapid fall in pH i (2,4). This H ϩ conduction pathway is termed the NADPH oxidase-associated H ϩ channel (5). The pH i of unactivated phagocytes, like that of most eukaryotic cells, is between pH 7.3 and 7.4. Therefore, the mem-brane of phagocytes is relatively impermeable to H ϩ ions, i.e. the channel is normally closed. Henderson and Chappell (5) have previously shown that, like the oxidase itself, the channel is opened following the addition of arachidonic acid (AA). 1 In the presence of AA, the direction of H ϩ flux is dictated by the proton-motive force (⌬p (⌬⌿ Ϫ RT In Patients with chronic granulomatous disease are highly susceptible to infection. This has been demonstrated to be due to an inherited inability of their phagocytes to generate O 2 . (6).
Analysis of the genetic lesions found in chronic granulomatous disease patients has demonstrated that the NADPH oxidase is composed of at least four proteins: a heteromeric, integral membrane cytochrome b 558 (gp91 phox and p22 phox ) and two cytosolic proteins, p47 phox and p67 phox (6). In recent years, additional proteins have been proposed to contribute to the functioning of the oxidase: a small GTP-binding protein, Rac2 (7,8), and an additional cytosolic protein, p40 phox , with sequence similarities to both p47 phox and p67 phox (9,10). Assembly of an active oxidase involves the translocation of the cytosolic proteins to the membrane, where they interact with the cytochrome b, forming a functional oxidase. Through the use of Epstein-Barr virus-transformed B lymphocyte cell lines from chronic granulomatous disease patients and the construction of a stable CHO cell line, Henderson et al. (1) have demonstrated that gp91 phox , the product of the Xlinked chronic granulomatous disease gene (11), functions as the AA-activable, NADPH oxidase-associated H ϩ channel. The hydropathy plot for gp91 phox (11,12) suggests that the protein contains multiple (four to six) transmembrane domains located at the N terminus and a large hydrophilic C-terminal domain that contains the predicted FAD-and NADPH-binding sites and is therefore assumed to be on the cytosolic side of the membrane. Henderson et al. (12) have previously demonstrated that the N-terminal 230 amino acids of gp91 phox (gp91N) contains all that is required for the protein to function as the NADPH oxidase-associated H ϩ channel.
The mutagenesis of amino acids, both man-made and inherited, has been used to investigate and identify functionally important amino acid residues in a range of proteins and multiprotein complexes, including ion transporters, carriers, and channels. For example, the selectivity of the cardiac L-type Ca 2ϩ channel has been reported to be altered from Ca 2ϩ to Na ϩ by the mutation of two Glu residues to either Gln or Lys in the predicted mouth of the channel (13). The flux of H ϩ ions associated with the activity of the NADH:NADPH transhydrogenase is absent when His-91 is changed to Ser, Thr, or Cys (14). In this paper, the measurement of the AA-activated H ϩ flux in CHO cell lines expressing point mutations in gp91 phox identi-fies His-115 as an amino acid important to the functioning of the NADPH oxidase-associated H ϩ channel.

EXPERIMENTAL PROCEDURES
Materials-The following were obtained from Sigma (Poole, Dorset, United Kingdom), and stock solutions were prepared as indicated: arachidonic acid (sodium salt; 10 and 1 mM in 50% EtOH, stored under nitrogen at Ϫ20°C), valinomycin (0.4 mM in EtOH), nigericin (1.5 mM in MeOH), CCCP (8 and 0.8 mM in EtOH, pH adjusted with NaOH), penicillin, streptomycin, trypsin (2.5 mg/ml of phosphate-buffered saline), and hygromycin B (100 mg/ml of sterile H 2 O). Ham's F-12 nutrient mixture with GLUTAMAXI and fetal calf serum were obtained from Life Technologies Inc.. BCECF acetoxymethyl ester was obtained from Molecular Probes, Inc. (Eugene, Oregon), and a stock solution of 1 mM was prepared in dimethyl sulfoxide dried by freeze-thawing. The ECL Western blot detection kit and Hyperpaper (photographic paper) were obtained from Amersham Life Sciences Ltd. (Buckinghamshire, UK). The horseradish peroxidase-conjugated goat anti-mouse antibody, horseradish peroxidase-conjugated avidin, and the biotinylated SDSpolyacrylamide gel electrophoresis molecular mass standards were obtained from Bio-Rad Laboratories Ltd. (Hemel, Hempstead, UK). The compositions of the salt solutions were as follows: 150 mM NaCl, 1 mm KCl, 5 mM Hepes/Tris, and 5.5 mM glucose (pH 7.4) (Na ϩ medium) and 150 mM KCl, 5 mM Hepes/Tris, and 5.5 mM glucose (pH 7.4) (K ϩ medium).
Mutagenesis of the N-terminal Fragment of gp91 phox -Plasmids containing the point mutants of the N terminus (amino acids 1-230) of gp91 phox were constructed by insertion following amplification in two stages, using overlapping PCR products. Synthetic mutant oligonucleotides together with oligonucleotide primers for the 5Ј-and 3Ј-ends of the N-terminal fragment of gp91 phox (12) were used to generate two overlapping PCR products (nucleotides 1-352 and 336 -690). Following purification from 1% agarose gel, the two fragments were annealed and joined by extension from the region of overlap before final PCR amplification using the 5Ј-and 3Ј-end primers (12). The PCR products were inserted as a HindIII/SacII fragment into the plasmid pCMUIV (12,15), generating pCMUIVgp91N(mutant). pCMUIV contains a sequence encoding three tandem copies of the hemagglutinin (HA) epitope (YPYD-VPDYAG). gp91N(mutant)HA was excised as a HindIII/BamHI fragment from pCMUIVgp91N(mutant)HA and inserted downstream of the metallothionein promoter in the multiple cloning site of the plasmid pMEP4 (Invitrogen), generating pMEP4gp91N(mutant)HA. The PCR products for the H115D and H115K mutants were inserted directly into pMEP4 as a HindIII/BamHI fragment using the synthetic 3Ј-oligonucleotide primer AGCTGAGGATCCCTACCCACGTACAATTCG containing a BamHI site (italic) and a stop codon (underlined). The synthetic mutant oligonucleotides used were as follows: T GCG ATT ctc ACC ATT GCA ctt CTA TTT, C AAT GGT gag AAT CGC AGA gag AAG TGC (H111L,H115L,H119L); T GCG ATT ctc ACC ATT GCA CAT CTA TTT, C AAT GGT gag AAT CGC AGA GTG AAG TGC (H115L); T GCG ATT gac ACC ATT GCA CAT CTA TTT, C AAT GGT gtc AAT CGC AGA GTG AAG TGC (H115D); and T GCG ATT aaa ACC ATT GCA CAT CTA TTT, C AAT GGT ttt AAT CGC AGA GTG AAG TGC (H115K).
Construction of CHO Cell Lines-Stable CHO cell lines expressing the mutants of the N-terminal fragment of gp91 phox were established following transfection by electroporation (230 V, 960 microfarads) of the pMEP4 constructs and selection with 100 g/ml hygromycin B, as described previously (1). cDNA for gp91 phox R91E,R92E was constructed by Dr. Karla Biberne-Kinkade in the laboratory of Dr. Mary Dinauer (Indiana University School of Medicine, Indianapolis, IN). The CHORR cell line was established following transfection and selection.
Maintenance of CHO Cell Lines-All manipulations of the cell cultures were performed in a Flow Laboratory flow hood, and the cell lines were grown in a Heraeus incubator at 37°C. The various cell lines were maintained in Ham's F-12 nutrient mixture with GLUTAMAXI, 10% fetal calf serum, 50 units/ml penicillin, and 50 g/ml streptomycin. The medium was replaced every 3-5 days, and the cells were divided (onehalf) once a week following trypsinization. The expression of the mutated proteins was induced by exposure of the cells lines to 10 M Cd 2ϩ for 24 h, as described previously (1,16).
Western Blotting-The expression in the CHO cell lines of the Nterminal fragment of gp91 phox containing the His-to-Leu mutations (3 His residues to 3 Leu residues and His to Leu) was examined by Western blotting using anti-HA epitope monoclonal antibody (16). The cells were treated with and without 10 M Cd 2ϩ for 24 h prior to being collected by trypsinization and washed twice in phosphate-buffered saline (800 g for 10 min). The proteins (10 g/lane) were solubilized in sample buffer (20% (v/v) glycerol, 4% (w/v) SDS, 62.5 mM Tris-HCl (pH 6.8), 10% (v/v) mercaptoethanol, and 4.5% (v/v) saturated bromphenol blue) and separated by electrophoresis on a discontinuous SDS-polyacrylamide gel (4% stacking gel (pH 6.8) and 10% separating gel (pH 8.8)) using 0.3% (w/v) Tris, 1.44 (w/v) glycine, and 0.1% (w/v) SDS as electrophoresis buffer. The separated proteins on the polyacrylamide gel were transfer by electrophoresis to nitrocellulose membrane using cold 0.303% (w/v) Tris, 1.44% (w/v) glycine, and 20% (v/v) methanol as transfer buffer. The membrane was probed with anti-HA monoclonal antibody (primary antibody) for 1 h. The presence of the primary antibody was detected following the addition of horseradish peroxidaseconjugated goat anti-mouse antibody (secondary antibody; 1 h), using the Western blot ECL detection kit and Hyperpaper (17). Biotinylated standards were included to enable calibration of the Western blot for molecular mass.
Immunostaining-Immunocytochemical techniques in combination with confocal optical scanning microscopy were used to examine the expression and cellular localization of the N-terminal fragment of gp91 phox containing the various mutated amino acids as described previously (1,12).
Measurement of Intracellular pH-The activity of the arachidonateactivable, NADPH oxidase-associated H ϩ channel was assessed as a change in pH i in response to the addition of AA as described previously (1,5,12). The response of BCECF was calibrated using additions of HCl to cells in high K ϩ medium in the presence of 10 M nigericin (18,19). The capacity of the different cell lines to buffer H ϩ ions was monitored as the fall in pH i following the addition of NaAc (26 mM).

RESULTS AND DISCUSSION
To understand the mechanism(s) by which the N-terminal 230 amino acids of gp91 phox act as a pathway for the conduction of H ϩ ions, I attempted to identify regions and individual amino acids that are required for function. The amino acid sequence of gp91 phox does not hold a high degree of similarity to any other protein, and the very high degree of homology between different species (20 -23) makes it nearly impossible to identify conserved and, by implication, important residues. However, amino acids implicated in H ϩ conduction by other types of ion transporter have been described. The conduction through F 0 of the F 0 F 1 -ATPase has been described as requiring an acidic residue (Asp or Glu) located within a hydrophobic sequence of amino acids. Similarly, an Asp residue has been implicated in H ϩ transfer in cytochrome bo 3 -ubiquinol oxidase from Escherichia coli (24). The conduction of H ϩ ions into the endosome by the M2 coat protein of the influenza A virus is important to the uncoating of the virus and its subsequent infection of the cell. Within the single transmembrane domain of this protein, Ser-31 and His-37 have been demonstrated to be important in pH-dependent activation and the conduction of H ϩ ions (25)(26)(27)(28). Similarly, ␤His-91 (His-664 in the bovine enzyme) plays an important role in the proton pumping associated with the activity of the nicotinamide-nucleotide transhydrogenase from mitochondria (14). Within the N-terminal fragment of gp91 phox , no candidate acidic residue located in hydrophobic sequence could be identified; however, three histidines (residues 111, 115, and 119) predicted to be contained within a transmembrane domain were identified. I therefore investigated the role of these three histidine residues in the functioning of the NADPH oxidase-associated H ϩ channel.
Cellular Expression of N-terminal Mutants-A CHO cell line was constructed in which His-111, His-115, and His-119 where all mutated to Leu (CHO-N3Leu). The expression of the protein was driven by an inducible metallothionein promoter (1,12,16) with a HA epitope tag included at the C terminus of the protein to assist determination of expression and cellular localization (12). As shown in Fig. 1a, anti-HA antibody detected a prominent broad band running at an apparent meridian molecular mass of 37.5 kDa in CHO-N3Leu cells treated with Cd 2ϩ . That this expression is regulated is demonstrated by the lower intensity of this protein band detected in CHO-N3Leu cells in the absence of Cd 2ϩ (Fig. 1a). The molecular mass of 37.5 kDa is higher than that of 26.9 kDa predicted from the sequence of 230 amino acids plus 39 for the epitope tag. However, gp91 phox is a highly glycosylated protein traveling as a broad band on SDSpolyacrylamide gel electrophoresis. Therefore, the high molecular mass and the broad appearance of the band suggest that this mutated N-terminal gp91 phox may be partially glycosylated in these cells. Immunostaining of Cd 2ϩ -treated CHO-N3Leu cells with anti-HA antibody gave an annular pattern of fluorescence when imaged on a confocal microscope (Fig. 1b,  panel i), which was not observed in CHO-N3Leu cells not treated with Cd 2ϩ (panel ii) and in untransfected CHO cells (panel iii). This pattern of staining is similar to those already observed for CHO cells expressing full-length gp91 phox (1) and just the N-terminal 230 amino acids (12) and implies that the expressed antigen is located at or in the plasma membrane of the cell. From the facts that the protein is expressed, is probably glycosylated, and is located at the plasma membrane of the cell, I suggest that the introduced mutations have not resulted in a major disruption of the folding and three-dimensional structure of the protein. All CHO cell lines constructed demonstrated an annular pattern of immunostaining when observed on the confocal microscope.
AA-activable H ϩ Channel Activity in CHO-N3Leu Cells-The flux of H ϩ ions was monitored as a change in pH i as recorded by the fluorescent pH-sensitive indicator BCECF. In Na ϩ medium, a strong inward ⌬p is established following the addition of valinomycin (⌬⌿-negative) and Hepes (pH o Ͻ pH i ) (1,5). CHO cells expressing non-mutated N-terminal gp91 phox (CHO-N) demonstrated a rapid fall in pH i upon the addition of AA, as described previously (12), indicative of an influx of H ϩ ions (Fig. 2a). The addition of AA to CHO-N3Leu cells treated with Cd 2ϩ failed to elicit the expected fall in pH i (Fig. 2b). However, the predicted fall in pH i was observed following the addition of the protonophore CCCP (Fig. 2b). The level of response observed in CHO-N3Leu cells was comparable to that observed in untransfected control CHO cells (Fig. 2c) and previously reported (1,12). Therefore, the failure to observe an AA-activated H ϩ flux in CHO-N3Leu cells suggests that the mutation of His-111, His-115, and His-119 to Leu severely inhibits the ability of the N terminus of gp91 phox to function as a H ϩ conduction pathway.
It could be argued that the inhibition of conduction through the H ϩ channel observed in CHO-N3Leu cells was the consequence of alterations to the folding of the local region of the protein due to the replacement of three positively charged amino acids with hydrophobic uncharged substitutes. To further investigate the role of histidine residues in the H ϩ channel and in attempt to minimize disruption to the protein folding, I mutated only the central of the three histidines (His-115) to leucine. The resulting CHO cell line (CHO-NLeu), when treated with Cd 2ϩ , failed to demonstrate a rapid fall in pH i following the addition of AA to cells suspended in Na ϩ medium (Fig. 2d). As shown in Fig. 2d, the addition of AA elicited a slightly greater acidification response in CHO-NLeu compared with CHO-N3Leu cells under the same assay conditions (Fig.  2b), suggesting that CHO-NLeu cells express a small residual H ϩ channel activity. However, the sharp contrast in response to that observed in CHO-N cells (non-mutated) indicates that the introduction of H115L severely reduces that ability of the N terminus of gp91 phox to function as a H ϩ conduction pathway.
Mutation to Restore H ϩ Conduction-If His-115 plays a direct role in the ability of gp91 phox to conduct H ϩ ions, it should be possible to introduce a substitute that retains the ability of the protein to function. It could be suggested that the amino acid residue(s) involved must be capable of both accepting and donating H ϩ ions across a membrane at physiological pH. The side chain of histidine, consisting of an imidazole ring with a pK of 6.5 (in solution), is therefore well suited for the role. The side chain of aspartic acid has a pK of 4.5 for its negatively charged carboxyl group (COO Ϫ ), whereas that of lysine has a positively charge (NH 3 ϩ ) amino group that has a pK of 10. To investigate the contribution of pK and/or the charge of the side chain to the ability of gp91 phox to function as a H ϩ channel, I constructed two CHO cell lines: N-terminal gp91 phox H115D (CHO-NAsp) and N-terminal gp91 phox H115K (CHO-NLys). The conduction of H ϩ ions in both CHO-NAsp and CHO-NLys cells was assessed. Following the addition of AA to CHO-NAsp cells, a fall in pH i was observed (Fig. 3a), indicative of an AA-activated influx of H ϩ ions. The extent of the acidification observed in CHO-NAsp cells (Fig. 3a) was greater than that observed in either CHO-N3Leu or in CHO-NLeu cells (Fig. 2, b  and d), but smaller than that observed in cells expressing the non-mutated N-terminal fragment (Fig. 2a). The full pH i response was observed following the addition of CCCP (Fig. 3a). Therefore, the N-terminal fragment in which Asp is substituted for His-115 results in a protein that is partially capable of functioning as a H ϩ conduction pathway. From the failure to observe a similar AA-induced acidification in CHO-NLys cells (His to Lys) (Fig. 3b), it can be concluded that the substitution of histidine with an alternative positively charged amino acid is insufficient to facilitate H ϩ conduction. Therefore, the retention of activity observed in CHO-NAsp cells (Fig. 3a) is probably determined by the pK value and not by the charge or form of the amino acid side chain.
An efflux of H ϩ ions (rise in pH i ) following the addition of AA has previously been reported for both CHO-91 cells (expressing full-length gp91 phox ) and CHO-N cells suspended in K ϩ medium in the presence of valinomycin and Tris (1,12). Under these conditions, no efflux of H ϩ ions was observed in CHO-NLys cells (Fig. 5c), CHO-NLeu cells (data not shown), and CHO-N3Leu cells (data not shown). The expected rise in pH i was observed in all cases following the addition of CCCP. Therefore, point mutations of His-115 restrict both the arachidonate-activated efflux and influx of H ϩ ions, i.e. conduction in both directions.
Calibration of BCECF Response in All Cell Types-The failure of some of the CHO cell lines to show a large fall in pH i despite the presence of AA and a large inward ⌬p could be argued to be due to differences in the properties of BCECF in the different cell lines. The failure to observe a response could also be accounted for by variations in the initial pH i of the different cell lines. This is unlikely, as the addition of CCCP elicited the fully expected pH i response in all cell lines (Figs. 2 and 3). However, to eliminate this possible explanation, the response of BCECF to pH was calibrated in CHO-NLeu, CHO-NAsp, CHO-NLys, and non-mutated control CHO-N cells. Fig. 4a shows the response of BCECF-loaded CHO-NLys cells to changes in pH. The plot of the change in BCECF fluorescence versus pH (Fig. 4b) for all four cell lines is very similar. All cell lines were observed to have an initial pH i similar to that shown for CHO-NLys cells: pH 7.32 (Fig. 4a). Therefore, the failure to observe a response (Figs. 2 and 3) was not due to the acidity of the resting pH i or to alterations in the response of BCECF to pH.
Buffering Capacity-An increased capacity of CHO-N3Leu, CHO-NLeu, and CHO-NLys cells to buffer H ϩ ions could result in failure to observe H ϩ flux, i.e. AA induces an influx of H ϩ ions, but the cellular cytosol buffering obscures the change in pH i . As shown in Fig. 5, CHO-NLeu, CHO-NLys, CHO-NAsp, and CHO-N cells all showed a similar fall in pH i in response to the addition of NaAc. Therefore, they all have a similar buffering capacity.
R91E,R92E Mutation-The mutations introduced above have altered amino acid charge. Therefore, I investigated whether the change of charge alone is sufficient to alter the ability of gp91 phox to conduct H ϩ ions. A CHO cell line with both Arg-91 and Arg-92 mutated to Glu was constructed (CHORR). The addition of AA to CHORR cells suspended in Na ϩ medium resulted in a large acidification of pH i (Fig. 6a), which was comparable to that observed for cells expressing non-mutated gp91 phox (Fig. 2a) (1, 12). As efflux of H ϩ ions was observed in cells suspended in K ϩ medium, the H ϩ channel is functional in both directions (Fig. 6b). Therefore, despite the major alteration in the charge of the amino acid side chains, there is no major change in the ability of gp91 phox to act as a H ϩ conduction pathway.
I have demonstrated that mutation of His-111, His-115, and His-119 to Leu and of His-115 to Leu and to Lys significantly reduced the ability of the N-terminal fragment of gp91 phox to function as the AA-activable H ϩ channel. In contrast, gp91 phox in which His-115 was replaced with Asp retained some H ϩ conduction. That gp91 phox R91E,R92E retained a fully function H ϩ channel is indicative that not all alterations in amino acid charge alter function. Therefore, it can be concluded that His-115 is important to the ability of gp91 phox to function as the AA-activable, NADPH oxidase-associated H ϩ channel. transmembrane domains. In this paper, I have demonstrated that to function as a H ϩ conduction pathway, the N-terminal fragment of gp91 phox requires His-115. The expression of the protein(s) in response to Cd 2ϩ and their observed cellular location at or within the plasma membrane strongly suggest that the mutations introduced in this paper do not drastically alter the folding and three-dimensional conformation of the protein(s), although this possibility cannot be completely eliminated. The identification, in this study, of a histidine residue as being important to the conduction of H ϩ ions across a membrane is in keeping with similar roles for histidine, previously described, in the mitochondrial transhydrogenase (14) and in the M2 coat protein of the influenza virus (25)(26)(27)(28). Aspartic acid, but not lysine or leucine, could partially substitute for histidine and allowed gp91 phox to function as a H ϩ conduction pathway. The pK of the carboxyl group of aspartic acid is closer to that of the imidazole ring of histidine than the amine group of the side chain of lysine, suggesting that function is dependent on an amino acid side chain with an appropriate pK rather than its net charge. This suggests that the mechanism of H ϩ ion flux through gp91 phox may involve a cycle of protonation/ deprotonation with His-115 being exposed alternatively to the interior and exterior faces of the membrane.
The voltage sensor (S4) of the Shaker K ϩ channel contains seven positively charge residues that are believed to initiate opening of the channel as a result of voltage-dependent movement of S4. The mutation of Arg-365 and/or Arg-368 to histidine in a non-conducting, non-inactivating version of the Shaker K ϩ channel has recently been reported to result in a channel that conducts H ϩ ions in response to repeating cycles of opening/closing, gating current (29). The ability to conduct H ϩ ions was suggested to be due to the movement of the voltage sensor, exposing the histidine residues alternatively to opposite sides of the membrane and facilitating the movement of H ϩ ions down the pH gradient.
The assay used in this study for the AA-activable, NADPH oxidase-associated H ϩ channel utilizes the combination of transmembrane K ϩ gradient and the K ϩ ionophore valinomycin to establish and dominate the membrane potential (⌬⌿) and the addition of either Hepes or Tris to impose a pH gradient by rapid alterations to pH o . Under these conditions, a membrane potential with a value more positive than 0 mV cannot be set. However, the flux of K ϩ ions, carried by valinomycin, can balance the charge generated by H ϩ movement, enabling extended observation of both inward and outward fluxes of H ϩ ions, driven by ⌬p. Electrophysiological studies have demonstrated the presence of voltage-activated H ϩ conductance in snail neurons (30), muscle cells (31), oocytes (32), epithelial cells (33), and phagocytes and phagocyte cell lines (34 -36). The currents are elicited from cells exposed to a large pH gradient (Ն1 pH unit) by the imposition of positive membrane potentials (Ն ϩ40 mV). However, in both neutrophils (34) and macrophages (37), it has been reported that in the presence of arachidonate, the H ϩ currents are elicited by more negative membrane potentials, i.e. the voltage-gated H ϩ conductance is activated by AA. The similarities between the previously described phagocyte voltage-activated and the AA-activated, NADPH oxidase-associated H ϩ conductances (i.e. identification in phagocytes and the activation by AA) are intriguing. In neutrophils, the activation of the NADPH oxidase is associated with a depolarization of the membrane potential from -60 mV to -15 to Ϫ10 mV (2, 4) and with a slight fall in pH i (0.1 pH unit). Under physiological conditions, it is unlikely that the membrane potential of active neutrophils obtains a positive value. Therefore, in vivo opening of the H ϩ conductance probably results from a combination of the depolarization of the membrane potential and the action of AA.
It has previously been proposed that gp91 phox functions as the flavocytochrome of the NADPH oxidase (38). As the predicted binding sites for both FAD and NADPH are located in the hydrophilic cytosolic C-terminal domain, there is no conflict with the role proposed in this study for the N-terminal hydrophobic domain as a H ϩ conduction pathway. It has been suggested that the NADPH oxidase may contain more than one heme moiety (39), but currently, their location within the enzyme complex remains unknown. However, like the protontranslocating cytochrome complexes of the mitochondria, cytochrome b 558 (i.e. gp91 phox ϩ p22 phox ) has been reported to have a pH-dependent midpoint potential (40). This suggests a close association between the cytochrome heme moiety and the pathway for the translocation of H ϩ ions. This is therefore in agreement with the proposed roles for gp91 phox as the cytochrome b and as the NADPH oxidase-associated H ϩ channel.
In this paper, I have exploited CHO cell lines to investigate the involvement of amino acids in the ability of gp91 phox to function as a H ϩ conduction pathway and have identified His-115 as being functionally important. The ability of Asp to substitute in part for His and the lack of inhibition in CHORR cells suggest that the mutations exert their effects on the H ϩ conduction pathway directly rather than via changes in the conformation of the protein. A number of different currents have been described that are activated or enhanced by the presence of arachidonate, e.g. N-methyl-D-aspartic acid receptor current (41), outward rectifier K ϩ channel current in smooth muscle (42,43), and K ϩ currents in hippocampus (44). Sequence comparisons with gp91 phox may assist in the identification of region(s) necessary for and important in the AA activation.