NADPH Oxidase-dependent Acid Production in Airway Epithelial Cells*

The purpose of this study was to determine the role of NADPH oxidase in H (cid:1) secretion by airway epithelia. In whole cell patch clamp recordings primary human tracheal epithelial cells (hTE) and the human serous gland cell line Calu-3 expressed a functionally similar zinc-blockable plasma membrane H (cid:1) conductance. However, the rate of H (cid:1) secretion of confluent epithelial monolayers measured in Ussing chambers was 9-fold larger in hTE compared with Calu-3. In hTE H (cid:1) secretion was blocked by mucosal ZnCl 2 and the NADPH oxidase blockers acetovanillone and 4-(2-aminoethyl)benzene-sulfonyl fluoride (AEBSF), whereas these same blockers had no effect in Calu-3. We determined levels of transcripts for the NADPH oxidase transmembrane isoforms (Nox1 through -5, Duox1 and -2, and p22 phox ) and found Duox1, -2, and p22 phox to be highly expressed in hTE, as well as the intracellular subunits p40 phox , p47 phox , and p67 phox . In contrast, Calu-3 lacked transcripts for Duox1, p40 phox , and p47 phox . Anti-Duox antibody staining resulted in prominent apical staining in hTE but no significant staining in Calu-3. of Duox1 of

trachea was found to be 6.85 using microelectrode measurements (8). Using a pH-sensitive fluorophore, pH ASL was found to be 6.8 in bovine cultures and 7.1 in vivo in mice (9). In the inflammatory airway diseases asthma, cystic fibrosis, and chronic obstructive pulmonary disease an acidic pH ASL or acidic exhaled breath has been demonstrated (10 -12). These observations suggested a regulated mechanism of acid secretion by the airways that is up-regulated during inflammation. Recently we found that human airway cells express a voltageactivated, zinc-sensitive H ϩ conductance and secrete acidequivalents into the mucosal bath (13).
H ϩ -selective, voltage-activated currents were first recorded from snail neurons (14) and since then have been found in many other cell types (15). The transmembrane protein complex NADPH oxidase has been suggested to mediate the H ϩ conductance in phagocytic cells (16), although this notion has been challenged recently (17). In phagocytic cells NADPH oxidase plays a crucial role in killing invading organisms by producing superoxide radicals (O 2 . ) by transferring an electron from intracellular NADPH across the membrane to extracellular O 2 . During the process of NADPH oxidation in neutrophils, intracellular H ϩ is released from NADPH and thereby contributes to intracellular acid production (18). The phagocytic NADPH oxidase is a multisubunit enzyme consisting of two transmembrane subunits (the cytochrome gp91 phox and p22 phox ), three cytosolic subunits (p40 phox , p47 phox , and p67 phox ), and a small ubiquitously expressed GTP-binding protein p21 rac (19). In phagocytes, all subunits were shown to be necessary for a regulated release of O 2 . during the respiratory burst (20). gp91 phox is a member of the Nox gene family, where it is referred to as Nox2. In addition to the small Nox homologs Nox1 through -5, two large homologs have been described recently, called dual oxidases Duox1 and -2 (21)(22)(23). Expression of Duox1 and -2 has been found in the epithelium of the thyroid gland by immunofluorescence (22), and expression of Duox1 transcript by in situ hybridization and Duox1 antisense-sensitive H 2 O 2 production by airway epithelial cultures has been demonstrated (24).
Recent studies in our laboratory identified a zinc-sensitive H ϩ conductance and H ϩ secretion across the apical membrane of well differentiated human airway epithelia (13). We were intrigued by the notion that this physiological process was provided by the NADPH oxidase complex. This was addressed by studying H ϩ conductance, H ϩ secretion, NADPH oxidase expression, and intracellular H ϩ production in two functionally different types of human airway epithelia. Ciliated cells from the airway surface epithelium function to absorb fluid and acidify the airway lumen slightly (8,9,13), whereas cells from the submucosal serous glands secrete a bicarbonate-rich, alkaline fluid (25,26). In this report we compare the properties of airway surface versus submucosal gland cells by using primary * This work was supported in part by National Institutes of Health Grants HL071829, DK51799, and 1P50HL60288 and a Pilot and Feasibility Grant by the Cystic Fibrosis Foundation (Grant FISCHE02I0). 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 U.S.C. Section 1734 solely to indicate this fact.
¶ cultures of human tracheal surface epithelia (hTE) and the Calu-3 submucosal gland cell line. These cells were chosen to relate high (hTE) and low rates (Calu-3) of H ϩ secretion to the expression of NADPH oxidase and its function. The experiments in this report demonstrate the localization of the dual oxidase Duox in the apical membrane of hTE surface cells but not in Calu-3 gland cells. Our data suggest that a Duox-based NADPH oxidase complex supports epithelial H ϩ secretion by generating intracellular H ϩ , which provides a H ϩ gradient across the apical cell membrane. The plasma membrane H ϩ conductance did not correlate with Duox expression suggesting that intracellular H ϩ exited through a parallel pathway.
Whole Cell Patch Clamp Recordings-Cells were whole cell patchclamped as previously described (13,33). On an inverted microscope cells were bathed in (in millimolar): 120 Hepes, 70 gluconic acid, 100 trimethylammonium hydroxide, 2 calcium gluconate, 1 magnesium gluconate (pH 7.3). Pipettes were filled with (in millimolar): 100 MES, 60 gluconic acid, 90 trimethylammonium hydroxide, 10 EGTA, 1 glucose, 1 magnesium gluconate, 3.3 magnesium ATP, 0.07 lithium GTP (pH 5.3). Junction potentials at the electrodes were measured, zeroed, and carefully observed for stability. Noted instabilities were smaller than Ϯ2 mV. With these solutions pipette resistance was ϳ8 M⍀. After establishing the whole cell configuration the membrane potential was continuously clamped to Ϫ50 mV and pulsed every 14 s for 6 s to 20 mV. The access resistance (R a ) and the cell membrane capacitance (C m ) were measured by fitting the current transients caused by a 10-mV pulse with a single exponential. Current-voltage step protocols from Ϫ80 mV to ϩ40 mV were applied, and resulting step-currents were recorded. Time constants of current activation () during voltage pulses were fitted with a single exponential of the form I(t) ϭ A ⅐ e (Ϫt/) ϩ K, where t is time, I(t) is the time-dependent current, A is the amplitude of the current transient, and K is a constant current offset. Whole cell conductance was calculated from steady-state currents as the chord conductance measured between Ϫ50 mV and ϩ20 mV. For the calculation of the specific membrane conductance (G m in picosiemens (pS)/picofarad (pF)) the whole cell conductance was corrected for R a (on average, 22 Ϯ 5 M⍀ (hTE) and 30 Ϯ 3 M⍀ (Calu-3)) and normalized to C m with G m ϭ (G c Ϫ1 Ϫ R a ) Ϫ1 ⅐C m Ϫ1 . Patch clamp recordings and fits were done using pClamp 8 (Axon Instruments, Fremont, CA). Currents during current-voltage step protocols were sampled at 500 Hz and filtered at 200 Hz. For the plots in the figures, currents are shown sampled at 50 Hz by averaging adjacent samples. In the figures original currents are shown without leak subtraction or corrections. Measurements were done at 26°C.
Quantification of Proton Secretion-Proton secretion was measured using the pH-stat titration technique in an Ussing chamber as described previously (13). Briefly, cultures were bathed serosally with Hepes-buffered solution and mucosally with buffer-free solution (3 ml each). Solutions were constantly gassed with oxygen and were nominally free of HCO 3 Ϫ /CO 2 . Standard NaCl Ringer solutions contained (in millimolar): mucosal, 140 NaCl, 2 KCl, 15 glucose, 2 CaCl 2 , 1 MgCl 2 ; serosal, 140 NaCl, 2 KCl, 5 glucose, 10 Hepes, 2 CaCl 2 , 1 MgCl 2 (pH 7.3) with HCl/NaOH. During the experiment the pH of the mucosal solution was continuously measured and titrated to a target value of pH 7.3 (range, 7.25-7.35) with 10 mM NaOH to determine H ϩ secretion (J H ) by the cultures. Rates are expressed in nmol⅐h Ϫ1 ⅐cm Ϫ2 , and positive rates refer to acidification of the mucosal medium. Measurements were done at 37°C.
Molecular Expression of NADPH Oxidase Subunits-Total RNA was prepared from 2 ϫ 10 5 hTE and Calu-3 cells, respectively, grown on permeable filter supports, and poly(A) mRNA was isolated using the RNeasy and Oligotex mRNA kits (Qiagen, Valencia, CA). Reverse transcriptase-PCR was performed using 100 ng of mRNA, 40 units of avian myeloblastosis virus reverse transcriptase, 20 units of RNase inhibitor, 4 l of 5ϫ buffer (Roche Applied Science), 2 l of 5 mM dNTP, and 2.5 M random hexamer oligonucleotides (Applied Biosystems, Foster City, CA) for 60 min at 42°C. One microliter of first strand cDNA was used as template in polymerase chain reactions with NADPH oxidase-specific oligonucleotides (see Table I). PCR was performed in a DNA thermal cycler using AmpliTaq Gold® DNA polymerase (Applied Biosystems). Following a 10-min 94°C incubation, reactions proceeded for 30 s at 94°C then 60 s at 60°C for 43 cycles followed by 7 min at 72°C. One-tenth of the amplification products were analyzed on a 1% agarose gel. For quantitative PCR total RNA was isolated from hTE and analyzed using an ABI Prizm 7700 sequence detection system (PE Applied Biosystems, Inc.). Samples were normalized to glyceraldehyde-3-phosphate dehydrogenase.
Immunodetection of Duox-For immunoblotting, hTE cultures were washed with phosphate-buffered saline and cells were transferred directly into SDS-PAGE sample buffer. Proteins were separated by 6% SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membrane was blocked with 1% casein and probed with a rabbit anti-Duox antibody (kindly provided by F. Miot (22)), and horseradish peroxidase-conjugated secondary anti-rabbit antibody labeling was visualized using the ECL Western blotting analysis system (Amersham Biosciences). For immunocytochemistry, hTE cells were dissected from confluent cultures, fixed with 2% paraformaldehyde, and incubated with 0.3% (v/v) Triton X-100 followed by 1% (w/v) bovine serum albumin in phosphate-buffered saline. Cells were immunostained for Duox, and the tight junction protein ZO-1 was used as a marker of the apical region using a monoclonal anti-ZO-1 antibody (BD Biosciences). Secondary antibodies were Alexa Fluor 488 anti-rabbit (Molecular Probes) for Duox and Alexa Fluor 546 anti-mouse for ZO-1. Cells were embedded in Crystal Mount (Biomedia) and observed with a 63ϫ/1.4 numerical aperture oil-immersion objective on a Solamere spinning disc confocal microscope (excitation, 457 nm; emission, 535 Ϯ 20 nm for Duox, and excitation, 514 nm; emission, 605 Ϯ 30 nm for ZO-1).
Measurement of Intracellular pH (pH i )-Cells on glass coverslips were incubated for 30 min with 20 M BCECF-AM (2Ј,7Ј-bis-(2-carboxyethyl)-5-(and-6)-carboxyfluorescein, acetoxymethyl ester, Molecular Probes Inc.), washed, and allowed to cleave the dye for 60 min. Cellular fluorescence was investigated in an open perfusion chamber on an inverted microscope. Excitation wavelengths were 440 and 495 nm, and emission was collected between 525 and 550 nm with a cooled chargecoupled device camera (Photometrics CoolSnap HQ, Roper Scientific) controlled by a computerized imaging system (Metafluor, Universal Imaging Corp.). Emission intensities of individual cells were background-subtracted, and ratios of 495 to 440 nm were calculated. pH i was calculated from ratios using a four-point pH calibration in presence of 20 M nigericin in KCl solution. Calibration points were pH 7.5, 7, 6.5, and 6.0. Resulting calibration ratios for individual cells were fitted to second order polynomials, and fitted parameters were used to calculate pH i from measured ratios. Individual rates of acidification of cells during the first 15 min of treatment with amiloride were determined by linear regression. Dye loading and experiments were done at 26°C. NaCl solution composition was (in millimolar) 145 NaCl, 5 KCl, 1 CaCl 2 , 1 MgCl 2 , 12.5 Hepes (pH 7.4). pH calibration was performed in (in millimolar) 5 NaCl, 145 KCl, 1 CaCl 2 , 1 MgCl 2 , and Hepes and/or MES (total buffer 12.5 mM).
Statistics-Data are presented as original values or as mean Ϯ S.E.; n refers to the number of cells or tissues investigated. Effects were tested with t tests or with ANOVAs followed by Bonferroni-corrected t tests. Calculations were done with StatView (Abacus Concepts, Berkeley, CA). Calculated p values are given, and p Ͻ 0.05 was considered significant.

Proton Conductances in Airway Cells-Single hTE and
Calu-3 cells were whole cell patch-clamped to characterize and compare their H ϩ conductance under conditions selective for H ϩ currents with a pipette to bath pH gradient of 5.3 to 7.3. Calu-3 are large round cells with an average C m of 29.9 Ϯ 5.9 pF (n ϭ 7). hTE were visibly smaller with a C m of 10.4 Ϯ 1.4 pF (n ϭ 6, significantly lower than Calu-3, p ϭ 0.012, t test). hTE had visible cilia on one pole of the elongated cells. Reported conductances were normalized to C m to account for the difference in cell size.
Measured currents were identified as H ϩ currents by the established characteristics: (i) H ϩ -selective reversal potentials, (ii) zinc sensitivity, and (iii) activation by depolarization with time constants in the order of seconds (34). Both airway cell types expressed whole cell H ϩ currents, H ϩ -selective reversal potentials, and activation by depolarizing pulses. Continuous  Fig. 2 (E and F). Current-voltage relations recorded under control conditions (triangles) showed outward rectification and negative reversal potentials indicative for H ϩ currents. Control steadystate currents reversed at Ϫ21.6 Ϯ 4.7 mV (hTE) and Ϫ25.9 Ϯ 6.1 mV (Calu-3). Addition of Zn 2ϩ shifted the reversal potential to positive values in both cell types (Fig. 1, E and F, circles, by on average 37.8 Ϯ 3.5 mV, n ϭ 11; not different between cell types). On average, hTE and Calu-3 expressed similar specific H ϩ conductances of 19.6 Ϯ 3.1 pS/pF and 11.6 Ϯ 2.8 pS/pF, respectively (p ϭ 0.085, Fig. 1G).
As a biophysical identifier of the H ϩ conductance we used the time constants of current activation when stepping the mem-brane potential from Ϫ50 mV to 20 mV ( 20 ) and from Ϫ50 mV to 40 mV ( 40 ). Fig. 2 shows current activations evoked by voltage pulses in hTE ( Fig. 2A) and Calu-3 (Fig. 2B). In both cell types the depolarization-activated currents were fitted well with a single exponential (lines in Fig. 2, A and B (36)) showed values similar to our data under comparable recording conditions. These data showed that hTE and Calu-3 cells expressed functionally similar H ϩ conductances, as judged by the quantitatively similar average conductances, rates of voltage-activated currents, and zinc sensitivity.
Acid Secretion by hTE and Calu-3 Cultures-Acid secretion of intact epithelia was measured using the pH-stat titration technique in Ussing chambers. Block of H ϩ channels (by ZnCl 2 ) and block of NADPH oxidase (by acetovanillone or AEBSF) was used to 1) determine the role of H ϩ channels in epithelial acid secretion, 2) verify an apical location of H ϩ channels in polarized epithelial monolayers, and 3) determine the role of NADPH oxidase during acid secretion. Both cell types showed basal acid secretion into the mucosal compartment. In hTE, J H ϭ 939 Ϯ 49 nmol⅐h Ϫ1 ⅐cm Ϫ2 , n ϭ 12. Calu-3 showed a 9-fold smaller J H of 106 Ϯ 9 nmol⅐h Ϫ1 ⅐cm Ϫ2 , n ϭ 18 (p Ͻ 0.0001, Fig.  3). When cultures were pre-treated for 1 h with ZnCl 2 (100 M mucosally), acid secretion by hTE was significantly reduced (to 480 Ϯ 29 nmol⅐h Ϫ1 ⅐cm Ϫ2 , n ϭ 6, p Ͻ 0.0001, unpaired t test). In contrast, no significant effect of ZnCl 2 was found on acid secretion by Calu-3 (block to 84 Ϯ 4 nmol⅐h Ϫ1 ⅐cm Ϫ2 , n ϭ 5, p ϭ 0.2, Fig. 3).
In addition we tested the NADPH oxidase blockers AEBSF and acetovanillone (500 M each) on H ϩ secretion by hTE and Calu-3. After incubation of cultures for 14 -16 h with blockers, H ϩ secretion was inhibited significantly in hTE (p Ͻ 0.0001, ANOVA) but not in Calu-3 (Fig. 2). In hTE, AEBSF reduced J H to 515 Ϯ 49 nmol⅐h Ϫ1 ⅐cm Ϫ2 , and acetovanillone reduced J H to 606 Ϯ 58 nmol⅐h Ϫ1 ⅐cm Ϫ2 . These values were not different from inhibition of acid secretion achieved with ZnCl 2 . The remaining H ϩ secretion in hTE was partially blocked by apical ouabain (50 M, by 15 Ϯ 5.8%, n ϭ 7, different from zero, p ϭ 0.04) and apical bafilomycin A1 (100 nM, by 11 Ϯ 3.9%, n ϭ 7, different from zero, p ϭ 0.04) indicating that hTE expressed additional mechanisms for acid secretion (12,37) operating at a low rate under our recording conditions. Fig. 3 shows a quantitative comparison of the effects of ZnCl 2 and NADPH oxidase blockers. These experiments showed that about half of the epithelial acid secretion in hTE was blocked by either blocking the apical H ϩ conductance or by blocking NADPH oxidase, indicating that both are involved in epithelial acid secretion in hTE. In contrast, treatment of Calu-3 with ZnCl 2 , acetovanillone, or AEBSF did not block acid secretion (Fig. 3) indicating that acid secretion by Calu-3 epithelia involves neither a H ϩ conductance nor NADPH oxidase activity. Thus, despite a similar H ϩ conductance in hTE and Calu-3 in patch clamp recordings (Fig.  1), H ϩ secretion by intact epithelia was mediated by a H ϩ conductance and by NADPH oxidase only in hTE. This difference was further investigated by determining the expression of NADPH oxidase isoforms in these cells.
Expression of NADPH Oxidase Genes in Airway Cells-We investigated the expression of described NADPH oxidase subunits, including the transmembrane subunits Nox1, Nox2, Nox3, Nox4, Nox5, Duox1, Duox2, and p22 phox , and the intracellular subunits p40 phox , p47 phox , and p67 phox . Actin and glyceraldehyde-3-phosphate dehydrogenase were included as control genes. The sequence-specific primers used for these genes are shown in Table I. Fig. 4A shows the expression pattern of mRNA of NADPH oxidase subunits in Calu-3 (top panel) and hTE (bottom panel) using high amplification (43 cycles). Transcripts for Nox1, Nox2, and Nox5 were detected in both cell types. Nox4 was expressed at low levels in hTE but not detected in Calu-3 cells. The large Nox genes, Duox1 and Duox2, were the major isoforms in hTE; in contrast, Duox1 was not detected and Duox2 was expressed at comparably low levels in Calu-3 cells. Nox3 was not detected in either cell type (not shown). The small transmembrane subunit of NADPH oxidase, p22 phox , was expressed highly in both cell types. Quantification of the relative expression of the transmembrane subunit in hTE using real-time PCR showed that in hTE transcripts for Duox1, Duox2 and p22 phox were expressed at Ͼ1000-fold higher levels than the small transmembrane Nox isoforms (Fig. 4B). The cytosolic subunits p40 phox , p47 phox , and p67 phox (Fig. 4A, right panels) were expressed in hTE, whereas Calu-3 expressed p67 phox but lacked transcripts for the two smaller subunits p40 phox and p47 phox . Thus the expression of NADPH oxidase subunits showed significant differences between hTE and Calu-3 cells. The lack of Duox1 and the comparably low expression of Duox2 in Calu-3, both of which were highly expressed in hTE, suggested a role for Duox during epithelial acid secretion.
Localization of Duox-We investigated the localization of Duox in hTE using immunofluorescence and confocal microscopy. A polyclonal antibody raised against Duox1 (22) was used to immunostain airway cells. Note that this antibody does not distinguish well between the Duox1 and -2 isoforms 2 and is thus referred to here as Duox antibody. Fig. 5A shows an immunoblot for Duox of cell lysates from hTE and Calu-3. A band of ϳ180 kDa was detected in hTE (right lane), and no specific staining was found in Calu-3 (left lane). Fig. 5 (B and C) shows an hTE cell group in a confocal side view. Fig. 5B shows immunostaining of Duox, and Fig. 5C shows for comparison immunostaining for the tight junction protein zonula occludens 1 (ZO1) as a marker for the luminal side of the epithelium. The staining pattern indicates luminal localization of Duox. Fig. 5D shows another group of hTE cells in a confocal side view with cilia clearly visible on one pole of the cells. Prominent anti-Duox immunostaining of the ciliated pole indicates apical localization of Duox. For comparison and as a negative control, Calu-3, which expressed low levels of Duox2 mRNA but no detectable transcripts for Duox1 (Fig. 4A), were immunostained with anti-Duox (Fig. 5E) and anti-ZO1 (Fig. 5F). A confocal image slice of the apical region of Calu-3 is 2 F. Miot, personal communication. shown (identified by ZO1 expression). No specific staining of Duox was detected indicating that Calu-3 do not express Duox at their luminal membranes. Intracellular Acid Production by NADPH Oxidase-Although hTE and Calu-3 both expressed a similar Zn 2ϩ -sensitive plasma membrane H ϩ conductance, only hTE secreted acid in a Zn 2ϩ -blockable manner. These results suggested that the electrochemical gradient for H ϩ exit across an apical H ϩ conductance was larger in hTE than in Calu-3. In the following experiments we tested the notion that the activity of NADPH oxidase generates an intra-to extracellular H ϩ gradient that drives H ϩ through a plasma membrane H ϩ conductance. Under control conditions cytosolic pH in hTE, Calu-3, or in hTE treated with AEBSF was not significantly different (Fig. 6C, black bars). Treatment of cells with amiloride (1 mM) acidified all cells (Fig. 6, A and C), but hTE acidified at a significantly higher rate than Calu-3 (Fig. 6B), and after 90 min of amiloride treatment pH i was significantly more acidic in hTE than in Calu-3 (Fig. 6C, gray bars). The amiloride-induced acidification was significantly reduced by treating hTE with AEBSF (Fig.  6A), which reduced the rate of acidification (Fig. 6B) and increased pH i after 90 min of amiloride treatment over untreated hTE (Fig. 6C, gray bars), suggesting that the activity of NADPH oxidase significantly contributed to the intracellular acid production of hTE. These measurements are consistent with an NADPH oxidase-mediated acidification of pH i in hTE. This process likely provides the driving force for H ϩ exit through apical membrane H ϩ channels in intact epithelia. For comparison, Calu-3 showed small rates of intracellular acid production and thus may not generate an appreciable H ϩ gradient across an apical membrane H ϩ conductance, consistent with the low rates of H ϩ secretion (Fig. 3) by these cells. DISCUSSION The data presented in this report suggest a close correlation between intracellular acid production, acid secretion, NADPH oxidase function, and Duox expression in airway epithelial

FIG. 4. Expression of NADPH oxidase subunits in airway cells. A, PCR gels of transmembrane subunits (left panels) and intracellular subunits (right panels) of Calu-3 (top) and hTE (bottom).
Arrows show the 100-bp band in the marker lanes. Glyceraldehyde-3-phosphate dehydrogenase and actin are shown as controls. Primers for the reactions are given in Table I. Reaction was run for 43 cycles to visualize also low expressing small Nox isoforms. B, quantification of expression of transmembrane subunits in hTE using real-time PCR. Note log scaling of the yaxis. Each bar represents determination in triplicate. Signal was normalized to glyceraldehyde-3-phosphate dehydrogenase. cells: hTE showed high expression of Duox1 and -2, high acid secretion, and high rates of intracellular acid production, which were reduced by blocking NADPH oxidase function, whereas Calu-3 showed no expression of Duox1 and low expression of Duox2, little acid secretion and intracellular acid production, and no effects of NADPH oxidase blockers on these measurements. At the same time both cell types expressed a functionally similar plasma membrane H ϩ conductance. These observations suggest a model for acid secretion by airway surface epithelia in which a Duox-based NADPH oxidase generates intracellular H ϩ adjacent to the apical membrane, and this H ϩ both activates an apical membrane H ϩ conductance and provides the driving force for H ϩ secretion across this H ϩ conductance. This model and its implications are discussed below.
Acid Secretion Is Driven by a H ϩ Gradient Generated by the Activity of NADPH Oxidase-Activation by intracellular acidity and membrane depolarization are typical characteristics of H ϩ conductances in various cell types (15). In our study, whole cell patch clamp measurements were performed with an insideto-outside H ϩ gradient that was used both to activate the H ϩ conductance and to provide a driving force for H ϩ movement, and the membrane voltage was controlled. Under these conditions both airway cell types showed similar specific H ϩ conductances with common biophysical characteristics and zinc sensitivity, suggesting a similar type of H ϩ channel in the plasma membrane of hTE and Calu-3 (Fig. 1). However, in intact epithelial preparations, hTE showed zinc-sensitive H ϩ secretion, whereas Calu-3 did not (Fig. 3). These observations suggested that in Calu-3 epithelia either the H ϩ conductance was inactive or there was no proton-motive driving force across the apical membrane to support epithelial H ϩ secretion.
In intact cells several factors affect the electrochemical driving forces for H ϩ exit, such as the apical H ϩ gradient and the apical membrane potential (V a ), the latter of which in turn is determined by the transepithelial resistance (R t ), the relative resistances of the tight junctions and the cells, and the relative conductance of the apical membrane to Na ϩ and Cl Ϫ . Compared with Calu-3 cells, hTE have larger R t (569 versus 257 ⍀⅐cm 2 ), more negative V t (Ϫ26 versus Ϫ5.6 mV), and a larger apical Na ϩ conductance (30,38). Thus, compared with Calu-3, the hTE cultures can be expected to possess relatively depolarized V a values. Consistent with the observed transepithelial differences in our report, the reported V a for hTE was V a ϭ Ϫ26 mV (38) or V a ϭ Ϫ19 mV (39), and in Calu-3 cells V a ϭ Ϫ52 mV (40). Thus, one explanation for the lack of zinc-sensitive H ϩ secretion in Calu-3 is the hyperpolarized V a , which is expected both to inactivate the H ϩ conductance (15) and to reduce the electrochemical driving force for H ϩ exit. The presence of conductive H ϩ secretion by hTE in the face of a negative V a indicates an intra-to extracellular H ϩ gradient across the apical membrane (13). In alveolar type II cells it was shown that intracellular acidification and a H ϩ gradient were necessary both to activate the H ϩ conductance and to drive H ϩ currents (41). In neutrophils, activation of NADPH oxidase caused the cytosol to acidify from pH 7 to 6.2 when the Na ϩ /H ϩ exchanger was blocked by Na ϩ free solutions (18). In airway epithelia the Na ϩ /H ϩ exchanger was shown to be a major regulator of intracellular pH, however, it is present only in the basolateral membrane (42). Therefore, any H ϩ produced apically by NADPH oxidase may not be readily transported by the Na ϩ /H ϩ exchanger but likely, instead, to cause a local acidification of the intracellular face of the apical membrane.
In analogy to acid production by NADPH oxidase in neutrophils, addition of amiloride (to block the Na ϩ /H ϩ exchanger) caused an AEBSF-blockable acidification of hTE cells (Fig. 6). Basal pH i was not affected by treatment with AEBSF ( Fig. 6), indicating a tight control of pH i by the Na ϩ /H ϩ exchanger under control conditions. Thus, treatment with amiloride uncovered intracellular acid production by NADPH oxidase activity in hTE cells.
The effect of AEBSF on both the rates of intracellular acidification (Fig. 6C) and epithelial acid secretion by hTE (Fig. 3) is consistent with the notion that NADPH oxidase-generated intracellular H ϩ both activates the plasma membrane H ϩ conductance and provides a driving force for H ϩ exit across the apical membrane. When assuming that the pH i measured in the presence of amiloride corresponds to the local pH at the cytosolic face of the apical membrane in untreated epithelia, then, with the measured pH gradient (6.89 to 7.4) and an average V a of Ϫ22.5 mV (43,44), an outwardly directed protonmotive force of 11.5 mV is predicted using the Nernst equation for conductive H ϩ secretion across the apical membrane. The critical role of intracellular acidification for H ϩ secretion is further supported by the data found in Calu-3 cells, which show little intracellular acid production and no evidence for an active, zinc-sensitive H ϩ secretion into the mucosal solution despite a significant plasma membrane H ϩ conductance.
Expression of NADPH Oxidase in Airway Epithelia-Duox1 and -2 were initially identified in thyroid epithelia where they were found to be expressed in the apical membrane (22,45). Investigation of a panel of tissues showed a high expression of Duox1 and -2 transcripts in thyroid and of Duox1 in whole lung (23). In this report we show that in human tracheal surface epithelia both Duox1 and -2 are expressed at levels Ͼ1000-fold higher than other Nox isoforms (Fig. 4) and thus represent the major Nox isoforms in airways. In contrast, in Calu-3 cells Nox isoforms were generally expressed at low levels, and no transcript for Duox1 was found. The high expression of Duox1 in hTE, its apical localization, and the high rates of acid secretion, compared with the lack of expression of Duox1 in Calu-3, which also showed no zinc-sensitive acid secretion, suggest Duox1 as the isoform responsible for acid production and secretion in airways. The effects of block of NADPH oxidase by AEBSF on both H ϩ production and H ϩ secretion underline the correlation between these two functional parameters and Duox-expression.
Regulation and function of the NADPH oxidase of phagocytes greatly depends on the proper assembly of Nox2 with its subunits (46). Currently there is no known subunit that interacts with Duox. Interestingly, hTE express at high levels p22 phox , p40 phox , p47 phox , and p67 phox (Fig. 4). In contrast Calu-3 lacked the two subunits p40 phox and p47 phox . Although we currently have no evidence for a functional interaction between Duox and additional subunits of NADPH oxidase, our expression data support the notion of a fully functional NADPH oxidase complex in hTE that allows for acid production, whereas Calu-3 may not assemble a functional complex and do not generate H ϩ by this mechanism.
H ϩ Channels in Airways-Several small Nox isoforms have previously been suggested to function as H ϩ channels, including Nox2 (16), Nox5 (47), and a truncated form of Nox1 (48). However, DeCoursey and colleagues (34) challenged this conclusion owing to (i) the expression of the Nox constructs in cell systems that already expressed native H ϩ currents, (ii) a lack of H ϩ selectivity of the resulting currents, and (iii) the much shorter time constants of current activation recorded after Nox expression in these studies when compared with time constants of native H ϩ currents. Several additional studies then presented evidence that expression of Nox2 did not correlate with H ϩ channel function, but a parallel non-Nox H ϩ channel structure was suggested to be the H ϩ conductive site (summarized in Ref. 34). Thus the identity of the H ϩ channel and the H ϩ conductance of the NADPH oxidase is controversial. Because the putative H ϩ permeation pathway across Nox2 involves strictly conserved histidine residues (16,49), the argu-ment made for Nox2 can be extended to the other Nox family members, including Duox.
We found in airway cells a H ϩ conductance with characteristics very similar to the native H ϩ conductances found in white blood cells or alveolar type II cells: H ϩ currents activated slowly with time constants in the order of seconds, steady-state H ϩ currents were outwardly rectifying and showed H ϩ selectivity, and ZnSO 4 readily blocked the depolarization-activated currents ( Figs. 1 and 2). Both hTE and Calu-3 expressed qualitatively and quantitatively similar H ϩ conductances suggesting the same type of H ϩ channel was active in both cell types. However, the high expression of Duox1 and -2 in hTE, and the total lack of Duox1 and low expression of Duox2 in Calu-3, suggest that the Duox isoforms do not function as H ϩ channels in airways, but a parallel non-Duox entity serves as a H ϩ channel. In analogy this argument may be extended to the other Nox family members based on the high homology between Duox and Nox in the histidine-containing transmembrane domains. However, our data cannot exclude the possibility that the lowly expressed isoforms Nox1, -2, and -5 may function as H ϩ channels, because they were found in both cell types. Thus a general conclusion about the channel characteristics of the Nox family from our data is precluded by the expression of multiple Nox isoforms in our cell system. Our data indicate that in airways a Duox-based NADPH oxidase generates intracellular H ϩ and a proton-motive force and H ϩ is conducted across the membrane through a parallel non-Duoxmediated pathway.
Two previous reports showed acid secretion of airway epithelia governed by different mechanisms. Coakley et al. (12) reported a ouabain-sensitive non-gastric K ϩ /H ϩ -ATPase in human primary cultures, and Inglis et al. (37) found a bafilomycin-sensitive vacuolar-type H ϩ -ATPase in distal pig bronchi. Using the effects of ouabain and bafilomycin as indicators for these mechanisms, in this report we found their quantitative contributions to be 15 and 11%, respectively, of total acid secretion by hTE airway epithelia in pH-stat experiments. Although the quantitative contribution of these mechanisms is likely dependent on a number of factors, such as tissue origin, species, and experimental pH values, these observations suggest that the airway epithelium expresses additional mechanisms to acidify the airway surface liquid. Fig. 7 shows a model that displays the relation of acid transporting mechanisms and NADPH oxidase in the apical membrane of airway cells as derived from our study. See Fig. 7 legend for details.   FIG. 7. Model of acid release in the apical membrane of airway surface epithelial cells. In this model a Duox-based NADPH oxidase releases intracellular H ϩ from NADPH and acidifies pH i . Intracellular H ϩ exits across a Zn 2ϩ -sensitive H ϩ channel whose molecular identity is currently unclear. This mechanism is similar to the model proposed for NADPH oxidase function in phagocytes (34). In addition, airway cells release acid by way of a bafilomycin-sensitive H ϩ ATPase and an ouabain-sensitive K ϩ /H ϩ ATPase. Percentages give relative contribution to total H ϩ secretion as determined from blocker experiments. Residual, non-blockable H ϩ secretion may be through a separate pathway, or may be due to incomplete inhibition of transporters by used blockers. Quantitative contribution of individual H ϩ secretory mechanisms may be different under experimental circumstances. Note that the source of H ϩ for the H ϩ channel is NADPH (which results in an acidification of pH i as observed in this study), whereas the ATPases release H ϩ from water resulting in an alkalinization of pH i in this model. For simplicity all transporters are shown located in one cell. Electron transport of NADPH oxidase is not shown in this model.
Function of Duox and NADPH Oxidase in the Airway Epithelium-In the thyroid gland the function of Duox has been identified as a source of oxidant for the iodination and crosslinking of tyrosine to generate thyroid hormone (21). In a similar reaction Duox of C. elegans was shown to be involved in the oxidative cross-linking of tyrosine residues for the stabilization of the cuticle of the worm (23).
In contrast, for the airway epithelium, it appears prudent to assume a functional role of Duox during bacterial killing, which is the major function of the airways. The airway epithelium is the first line of defense against inhaled bacterial pathogens, and the epithelium secretes a number of antimicrobial factors, which act in combination with the mechanical movement of the ciliary escalator to keep the distal lungs sterile. Several antimicrobial factors have been isolated from airway secretions indicating the role of the airways in bacterial defense. Production of O 2 . by airway cultures by NADPH oxidase has been shown (24). Acidification of the airway surface liquid during acid secretion by the epithelium is expected to support conversion of O 2 . into H 2 O 2 and HOCl, which show bactericidal activity only in an acidic environment (50). High activity of extracellular superoxide dismutase in airways (51) and the presence of significant lactoperoxidase activity in airway gland secretions suggest a rapid turnover of O 2 . in support of bactericidal reaction products. Thus we suggest that the expression of NADPH oxidase in the airway epithelium and acid secretion represents an innate defense function of the airway epithelium.