O(2) sensing by airway chemoreceptor-derived cells. Protein kinase c activation reveals functional evidence for involvement of NADPH oxidase.

Accumulating evidence suggests that neuroepithelial bodies are airway O(2) sensors. Recently, we have established the H-146 small cell lung carcinoma line as a suitable model to study the biochemical basis of neuroepithelial body cell chemotransduction. Here we explore the possibility that hypoxic modulation of K(+) channels is intimately linked to activity of NADPH oxidase. Graded hypoxia caused graded inhibition of whole cell K(+) currents, which correlated well with membrane depolarization. Pretreatment with the phorbol ester, 12-O-tetradecanoyl (TPA), inhibited K(+) currents at all potentials. Although 4alpha-phorbol 12,13-didecanoate and TPA in the presence of bisindolylmaleimide were also able to depress K(+) currents, only TPA could significantly ameliorate hypoxic depression of these currents. Thus, protein kinase C (PKC) activation modulates the sensitivity of these cells to changes in pO(2). Furthermore, because the addition of H(2)O(2), a downstream product of NADPH oxidase, could only activate K(+) currents during hypoxia (when endogenous H(2)O(2) production is suppressed), it appears likely that PKC modulates the affinity of NADPH oxidase for O(2) potentially via phosphorylation of the p47(phox) subunit, which is present in these cells. These data show that PKC is an important regulator of the O(2)-transduction pathway and suggests that NADPH oxidase represents a significant component of the airway O(2) sensor.

Ventilation-perfusion matching is essential for efficient oxygenation of pulmonary blood in the face of varying demands of the body's tissues. Numerous physiological processes have long been known to ensure optimization of ventilation-perfusion matching, but the cellular mechanisms underlying such processes are only now beginning to be understood. One such important mechanism is that of O 2 chemoreception, i.e. an active cellular response to an acute general and/or regional reduction of pO 2 . Most studies to date have focused on the carotid body arterial chemoreceptors. Type I cells of these sensors express O 2 -sensitive K ϩ channels, and exposure to hypoxia causes rapid channel inhibition. This leads to cell depolarization, Ca 2ϩ entry via voltage-gated Ca 2ϩ channels, and subsequent triggering of neurosecretion, an essential step in the initiation of afferent information, which leads to reflex increases of ventilation (1,2). In vascular smooth muscle, O 2 -sensitive K ϩ channels (3) and Ca 2ϩ channels (4) have more recently been reported, and their rapid modulation by hypoxia evokes appropriate vasodilatation or vasoconstriction (depending on the location of the blood vessel (5)).
Compelling evidence is emerging that neuroepithelial bodies (NEBs), 1 located in clusters at bronchial bifurcations throughout the lung, serve as airway chemoreceptors (6,7). Like their vascular counterparts in the carotid body, NEB cells possess numerous transmitters, which appear to be released in hypoxia (8). These transmitters (particularly serotonin) may act to initiate afferent nerve activity to medullary respiratory centers and may also act to control local vasomotor tone, hence directly influencing local ventilation-perfusion matching (9). Compared with type I cells of the carotid body, our understanding of O 2 chemoreception by NEB cells is in its infancy, due largely to the technically demanding nature of their isolation/primary culture or to the difficulty in recording currents in situ using the lung slice. However, it has been established that they are electrically excitable and, most importantly, possess O 2 -sensitive K ϩ channels (7,10), suggesting that they may respond, at least superficially, like carotid body type I cells. In recent studies, we have built on the suggestions of others to establish that small cell lung carcinoma (SCLC) cells, which are believed to be derived from NEB cells (11) and with which they share numerous similarities (12), represent immortal airway chemoreceptors. Thus, we have recently identified, in the SCLC cell line H-146, a specific component of the whole cell K ϩ current, which is sensitive to hypoxia and influences membrane potential (13,14). This current is 4-aminopyridine-(4-AP) and Ca 2ϩinsensitive, quinidine-sensitive, and may be a novel member of the tandem P-domain family of K ϩ channels. The ease of use of cell lines as compared with isolated NEB cells or lung slices suggests that they may be exploited to understand the molecular and biochemical mechanisms underlying rapid effects of hypoxia on chemoreceptor cells.
The mechanism(s) which couple a fall of pO 2 to K ϩ channel inhibition remain elusive, but recent work has implicated the involvement of NADPH oxidase as an O 2 sensor in both NEB cells and SCLC cells (15). The NADPH oxidase model for O 2 chemoreception suggests that, under normoxic conditions, the oxidase tonically generates superoxide from O 2 , which is rapidly converted to H 2 O 2 by a number of enzymes including superoxide dismutase and catalase. This H 2 O 2 is believed to promote channel activity. It follows, therefore, that H 2 O 2 levels will decline in hypoxia, causing a concomitant reduction in channel activity. Several independent lines of evidence support this idea: (i) NEB and SCLC cells express the mRNA encoding K ϩ channels known to be modulated by H 2 O 2 ; (ii) in situ hybridization and immunohistochemistry have indicated that specific subunits of NADPH oxidase (gp91 phox and p22 phox ) are present in both cell types; (iii) H 2 O 2 generation in NEB cell clusters is significantly greater than in surrounding tissue, can be enhanced by protein kinase C (PKC) activation using phorbol ester (which specifically phosphorylates another subunit, p47 phox , of the oxidase complex), and can be suppressed by diphenyleneiodonium, a known NADPH oxidase inhibitor and; iv), K ϩ currents can be enhanced (albeit transiently) by application of H 2 O 2 (10,15). Collectively, these observations support the idea that hypoxic inhibition of K ϩ channels in airway chemoreceptor cells and SCLC cells involves reduced levels of H 2 O 2 derived from NADPH oxidase activity, but a direct link between the observations is still lacking. Indeed, it has, to date, never been demonstrated directly that hypoxia causes a fall of H 2 O 2 levels in these cells.
In the present study, we demonstrate for the first time that H 2 O 2 production declines during hypoxia and have exploited the fact that NADPH oxidase activity can be regulated by PKC-dependent phosphorylation (16) to test further this model for O 2 chemoreception in SCLC cells. Our results provide compelling functional evidence to support a central role for NADPH oxidase in this important process, and we propose that PKC may modulate chemoreception by altering the affinity of the oxidase for O 2 .

Cell Culture
The small cell lung carcinoma cell line, H-146, was purchased from American Tissue Type Cell Collection (Rockville, MD) and was of unknown passage number. Cells were grown in suspension culture in RPMI 1640 medium (containing L-glutamine) supplemented with 10% fetal calf serum, 2% sodium pyruvate, and 2% penicillin/streptomycin (all from Life Technologies, Inc.) in a humidified atmosphere of 5% CO 2 /95% air at 37°C. Medium was changed every 2 days and cells were passaged every 6 -7 days by splitting in the ratio 1:5. Cells were used between nominal passage numbers 1 and 6.
All tubing was gas-impermeant (Tygon tubing BDH, Atherstone, Berkshire, UK). Normoxic solutions were equilibrated with room air. Solutions were made hypoxic, where appropriate, by bubbling with N 2(g) for at least 30 min prior to perfusion of cells. This procedure produced no shift in pH. Solution flow rate was ϳ5 ml⅐min Ϫ1 . Graded hypoxia was achieved by splicing variable lengths of gas-permeant tubing into the perfusion lines. pO 2 was measured (at the cell) using a polarized (Ϫ800 mV), calibrated carbon fiber electrode (16); for the experiments reported herein, the pO 2 values were 150 (normoxia), 85, 45, 25, and 15 (hypoxia) mm Hg.
Whole Cell Recording-Following trituration (10 passes through a 1-ml automatic pipette tip), cells were allowed to adhere at 37°C for at least 1 h to poly-L-lysine-coated glass coverslips before being placed in a temperature-controlled perfusion chamber (Brook Industries, Lake Villa, IL) mounted on the stage of a Nikon TMS inverted microscope. All experiments were carried out at 21 Ϯ 1°C. Patch pipettes were manufactured from standard walled borosilicate glass capillary tubing on a two-stage Narishige PP-83 pipette puller (Narishige Scientific Instrument Lab, Kasuya, Tokyo, Japan), were heat-polished on a Narishige microforge, and had measured tip resistances of 3-8 M⍀ (when filled with K ϩ -rich pipette solution).
Resistive feedback voltage-clamp was achieved using an Axopatch 200A amplifier (Axon Instruments, Forster City, CA). Voltage protocols were generated, and currents were recorded using pClamp 6.0.3 software employing a Digidata 1200 A/D converter (Axon Instruments). Data were filtered (4-pole Bessel) at 2 kHz and digitized at 5 kHz. Following successful transition to the whole-cell recording mode (12), capacitance transients were compensated for and measured. Where necessary, series resistance compensation was used at 100%.
To evoke ionic currents in H-146 cells, two voltage protocols were used: (a) ramp protocol: holding potential ϭ Ϫ70 mV, Ϫ100 to ϩ60 mV, ramp duration ϭ 1 s, frequency ϭ 0.1 Hz; (b) time series: holding potential ϭ Ϫ70 mV, single increment to 0 mV, step duration ϭ 50 ms, frequency ϭ 0.1 Hz. Fast current clamp was achieved using the same amplifier, and the solutions were the same as those used in the voltage-clamp experiments. Cells were clamped at I ϭ 0 pA and the recorded voltage was filtered at 1 kHz and digitized at 2 kHz.

Single Cell Fluorescence
H-146 cells were loaded for 30 min at 37°C with 10 M (in 0.01% dimethyl sulfoxide) H 2 O 2 fluorescent indicator 2Ј,7Ј-dichlorodihydrofluorescein diacetate (H 2 DCFDA, Molecular Probes, Leiden, Netherlands) during adherence to poly-L-lysine-coated coverslips (17). Coverslips were mounted on the stage of Nikon Diaphot 300 and perfused with the standard normoxic (150 mm Hg) bath solution (see above for composition). H 2 DCFDA was excited with light of wavelength 488 nm using a monochromator and emitted fluorescence (Ͼ510 nm) collected by a CCD camera (Hamamatsu, C4880 -80). Cells were imaged using Openlab software at 2-s intervals, and pixel intensity was calculated off-line. Following a control period of 20 s, cells were perfused with standard bath solution for 2 min, which was either air-equilibrated (150 mm Hg) or hypoxic (15 mm Hg). Recovery was followed for a further 3 min.

Data Handling and Calculations
Electrophysiology-The magnitude of the steady-state outward currents (time-series protocol) was measured as the mean current between 46 and 49 ms of the voltage-pulse. In the time-series plots, the currents from each separate experiment were normalized by dividing the current at each point by the mean of the first five currents in that series; in the figures, the numbers represent the mean (Ϯ S.E.); n refers to the number of cells. Current amplitudes were not corrected for cell size, because this did not vary between the four groups of cells studied, as determined by membrane capacitance measurements; control cells, 5.0 Ϯ 0.13 picoFarad (n ϭ 93); TPA-treated cells, 4.8 Ϯ 0.23 picoFarad (n ϭ 57); 4␣-PDD treated cells, 4.7 Ϯ 0.21 picoFarad (n ϭ 23); TPA-and BIM-treated cells, 5.7 Ϯ 0.50 picoFarad (n ϭ 16). Statistical comparisons were made using the paired or unpaired Student's t test, as appropriate, with p Ͻ 0.05 being considered significant.
Fluorescence-H 2 DCFDA is effectively trapped within cells and, in common with most other flourescence indicators, undergoes photobleaching. Because it is not a dual excitation or emission dye, this artifact cannot be removed by signal ratioing. Thus, during the experimental period, fluorescence intensity declines exponentially with time. Nevertheless, a change in the rate of decay is indicative of a change in cellular H 2 O 2 content/production. Mean pixel intensity was calculated from each cell area at intervals of 2 s, background fluorescence was subtracted, and the subtracted intensity was plotted against time. The rate of change of fluorescence was calculated as a rolling average of slope of five consecutive frames for the entire duration of the experiment. To account for any differential dye loading, slope data from each cell were normalized such that the first average from each protocol was unity. In Fig. 2, the mean data plot shows this rolling average versus time. A downward deflection (indicative of an increase in normalized slope value) represents a decrease in H 2 O 2 production/content. Exemplar, nonadjusted data are shown in the insets of Fig. 2.
RT-PCR-Total RNA was extracted from pelleted H-146 cells using the RNeasy Micro Kit (Qiagen, Crawley, W. Sussex, UK). The extracted RNA was then divided; 50% was treated (cleaned) with RQ-1 RNasefree DNase (1 unit⅐g Ϫ1 RNA; Promega, Southampton, Hampshire, UK) to remove genomic DNA contamination, before re-extraction using the RNeasy Micro Kit. The remaining 50% was kept in an untreated (uncleaned) state. The yield, purity, and integrity of the RNA was verified by spectrophotometry at 260/280 nm, followed by electrophoresis on 1% agarose, and was then stored in aqueous solution at Ϫ80°C. RT was performed on 1-g aliquots of both cleaned and uncleaned RNA using the reverse transcription system A3500 (Promega), comprising Avian Myeloblastosis Virus reverse transcriptase and oligo(dT) (15) primers (42°C, 15 min). The resulting cDNA was amplified by PCR and screened for p47 phox RNA expression using oligonucleotide primers (22-mers) designed against the published sequence of human neutrophil cytosolic factor 1 (GenBank TM accession number M55067 (18)). These primers (upstream: 5Ј-atgcaggtgagccatacgtcgc-3Ј, downstream: 5Ј-tctttcctgatgacccaccagc-3Ј) were chosen because they were predicted to amplify a region within a single exon. Therefore, products from both reverse transcribed mRNA and genomic DNA (the positive control used in this study) would be of identical length. PCR reactions using the cleaned and uncleaned (as positive control) cDNA were run in parallel. Amplification of 1-l cDNA (equivalent to 160 ng of reverse-transcribed RNA) was performed using a Hybaid (Ashford, Middlesex, UK) express thermal cycler, in a volume of 50 l, containing 2.5 units of Taq DNA polymerase (Promega), under optimized conditions with 2.0 mM MgCl 2 : hot start ϭ 94°C/1 min; denaturing ϭ 94°C/1 min, annealing ϭ 64°C/1 min, extension ϭ 72°C/1 min for 30 cycles followed by a final cycle, which had a 5-min extension. Products were separated on 2% agarose gels and visualized with ethidium bromide/UV transillumination. Because signal was relatively low from cleaned cDNA, a "turbo" PCR was performed on 5 l of the primary PCR products. Turbo PCR conditions were identical to primary PCR. Sequencing was carried out by dye terminator PCR with an ABI PRISM automated sequencer (School of Biology, University of Leeds, UK).

RESULTS
Response to Graded Hypoxia-In H-146 cells, we have recently shown that acute hypoxia (pO 2 ϳ15 mm Hg) causes rapid and reversible inhibition of whole cell K ϩ currents and membrane depolarization (13). Fig. 1 extends our earlier observations to show that reducing perfusate pO 2 in a graded manner (between 150 and 15 mm Hg) results in correspondingly graded inhibition of these K ϩ currents and also causes graded membrane depolarization (Fig. 1A). In this series of experiments, a pO 2 of 15 mm Hg significantly reduced mean outward K ϩ current amplitudes (measured at 0 mV) from 223.2 Ϯ 28.9 pA to 149.6 Ϯ 41.2 pA (p Ͻ 0.005, n ϭ 7), a reduction of ϳ33%. The same degree of hypoxia caused a significant depolarization from -47.9 Ϯ 1.3 mV to Ϫ38.4 Ϯ 3.8 mV (p Ͻ 0.05, n ϭ 7), a reduction of 9.5 mV. These values are similar to our previously published data (13). Fig. 1B shows a plot of the mean current amplitude versus membrane potential at the five grades of hypoxia examined. Hypoxic reduction in whole cell K ϩ current correlates well with membrane potential (r ϭ 0.96) and strongly suggests, therefore, that the O 2 -sensitive current contributes to the resting membrane potential in this airway chemoreceptor cell model.

Effects of Hypoxia on H 2 O 2 Production in H-146
Cells-Central to the hypothesis that NADPH oxidase acts as the O 2 sensor in airway chemoreceptors is the notion that H 2 O 2 levels are decreased during hypoxia. Fig. 2 demonstrates that this is indeed the case in H-146 cells. During normoxia, fluorescence intensity gradually declined because of photobleaching ( Fig. 2A  and inset). However, perfusion with the hypoxic solution (15 mm Hg) caused a far greater decline in fluorescence intensity, approaching twice the base-line bleaching rate (observed under normoxic conditions (Fig. 2B and inset)). This effect was maximal around 30 s after exchange to the hypoxic solution and was reversible on reperfusion with normoxic solution. The time-course of this effect was similar to that of hypoxic inhibition of K ϩ currents (see Ref. 13 and also Figs. 5 and 6). These data show clearly that H 2 O 2 levels in H-146 cells are regulated by environmental pO 2 and suggest that K ϩ channels are under tonic control by a H 2 O 2 -generating system.
Effects of Phorbol Esters on H-146 K ϩ Currents-Preincubation of H-146 cells for 10 min with the membrane-permeable PKC activator, TPA (100 nM), caused a significant (p Ͻ 0.05, n ϭ 11 control cells and 8 TPA-treated cells; unpaired Student's t test) reduction in ramp currents at all potentials (Figs. 3, A  and B). To determine the contribution which PKC activation made to this reduction in current, we employed: (a) a biologically inactive phorbol ester, 4␣-PDD (100 nM); and (b) TPA together with the PKC inhibitor, BIM (1 M bath and pipette). These maneuvers resulted in smaller, nonsignificant (p Ͼ 0.29, n ϭ 5; p Ͼ 0.52, n ϭ 10, respectively) reductions in currents (Fig. 3, C-E), than were seen with TPA. This suggests that the inhibition caused by TPA was attributable to both nonspecific actions of phorbol esters and activation of PKC. Fig. 4A shows currents evoked by ramp depolarizations from a typical H-146 cell before, during, and after exposure to two levels of acute hypoxia. These currents exemplify the reversible, graded inhibition caused by hypoxia, which is summarized in Fig. 1A. In contrast to these findings, currents recorded in cells pretreated with 100 nM TPA, although reduced in amplitude (see above), were virtually unaffected by hypoxia (e.g. Fig. 4B). Average current amplitudes (measured at 0 mV) at all levels of hypoxia examined are plotted in Fig. 4C. Thus, TPA markedly and significantly (p Ͻ 0.01, n ϭ 6 control cells; n ϭ 4 TPA-treated cells) suppressed hypoxic inhibition of K ϩ currents in H-146 cells. The results shown in Fig. 4D (determined from currents recorded at 0 mV during normoxia and hypoxia at 15 mm Hg) indicate that this was attributable to PKC activation, because neither 4␣-PDD (p ϭ 0.18, n ϭ 4) nor TPA in the presence of BIM (p ϭ 0.87, n ϭ 5) could significantly suppress the level of hypoxic inhibition of currents.

Effects of PKC Activation on Hypoxic Inhibition of K ϩ Currents in H-146 Cells-
We have previously shown that the O 2 -sensitive K ϩ current is resistant to 4-AP (14). Following treatment with TPA, the proportion of the whole cell current, which was 4-AP-insensitive, remained essentially unchanged compared with control (Fig. 5, A and D). However, in the presence of a maximally effective concentration of 4-AP (10 mM), hypoxia (15 mm Hg) caused over 50% inhibition of the residual current (e.g. Fig. 5, A, C, and D). In contrast, hypoxic inhibition of the 4-AP-resistant current was significantly (p Ͻ 0.005, n ϭ 8) reduced to around 25% following TPA treatment (e.g. Fig. 5, B, C, and E). Fig. 5C shows the mean time course of these effects. This observation shows that although TPA caused a generalized reduction in current (see also Fig. 3), the 4-AP-resistant component was less O 2 -sensitive in these cells when PKC was activated.
Although there are many PKC-dependent mechanisms that could account for these data, one potential site of action for PKC might be the NADPH-oxidase complex, which has been recently implicated in NEB cell O 2 sensing (15). If this is the case in our NEB cell model, we would predict that products downstream of NADPH oxidase activity would regulate the O 2 -sensitive K ϩ channel. We addressed this possibility by applying one such product, H 2 O 2 , in the presence of 4-AP. Fig. 6A shows that under normoxic conditions, H 2 O 2 was without effect. However, during hypoxic inhibition of the K ϩ current, H 2 O 2 caused a dramatic and transient reactivation of the current. The time course of this effect was similar to that previously reported in NEB cells (15).
A probable site of PKC-dependent phosphorylation of the FIG. 6. Effect of H 2 O 2 on 4-AP-insensitive K ؉ current. A, typical currents evoked from an example cell before (Normoxia), during (Normoxia ϩ H 2 O 2 ), and after (Normoxia) bath application of 2 mM H 2 O 2 . B, typical currents evoked from a representative cell before (Normoxia) and during hypoxia (Hypoxia) and during continued exposure to hypoxia but in the additional presence of 2 mM H 2 O 2 (Hypoxia ϩ H 2 O 2 ). Currents were evoked by step depolarizations from Ϫ70 to 0 mV, and scale bars apply to both A and B. 10 mM 4-AP was present throughout in both cases. C, mean time series plot of currents evoked from five cells as in A but normalized to the first five currents evoked after the effects of 10 mM 4-AP had stabilized. H 2 O 2 was applied for the period indicated by the horizontal bar. D, as C, except cells (n ϭ 7) were exposed to hypoxia before and during H 2 O 2 application, as indicated by horizontal bars.
heteromeric enzyme, NADPH oxidase, is p47 phox (16). However, although there is good evidence for the presence in both native NEB cells and H-146 cells of some of the components of the enzyme (15), evidence for the presence of p47 phox in H-146 cells is still lacking. Fig. 7 shows the use of RT-PCR to address this question directly. Amplification of a product compatible with the presence of p47 phox (130 base pairs) in genomic DNA (uncleaned lanes) shows that the PCR protocol that we employed can detect p47 phox (Fig. 7A). Using this protocol, we could detect only a small signal from cDNA reverse transcribed from DNase-treated H-146 RNA (cleaned lanes). However, a turbo PCR reaction produced a strong signal (130 base pairs) from the cleaned sample (Fig. 7A), indicating that p47 phox was indeed being transcribed by these cells (confirmed by sequencing, Fig. 7B) but that its expression was low because of either rapid degradation or low copy number of p47 phox mRNA; PCR cannot distinguish between these two possibilities. Regardless, p47 phox is present and is, therefore, a possible site of PKC-dependent regulation of the oxidase in our model airway chemoreceptor cells.

DISCUSSION
Electrophysiological studies have demonstrated that acute reductions in pO 2 suppress K ϩ currents in airway chemoreceptor NEBs (7) in a manner comparable to their vascular counterparts, the type I cells of the carotid body (1,19). Such an effect underlies membrane depolarization, which presumably leads to activation of voltage-gated Ca 2ϩ channels, thereby permitting Ca 2ϩ influx to trigger neurotransmitter release in both cell types (8,20). Previous studies, including our own, have demonstrated that SCLC cell lines (such as H-146 used here), which are derived from the same precursor pool as NEBs (11), provide an excellent model for studying mechanisms underlying hypoxic inhibition of NEB cell K ϩ channel inhibition (13,14). This is an important advance given the present limitations of acutely isolated, primary cultured NEB cells.
An important aim of the present study was to use PKC activation as a method by which to investigate the transduction pathway responsible for hypoxic suppression of K ϩ currents in our recently established model of NEB O 2 sensing. In NEB cells, it has been proposed, but not directly substantiated, that NADPH oxidase is an important upstream component of the chemoreception pathway (6,15). NADPH oxidase is a heteromultimeric enzyme that contains, importantly, a PKC-activable subunit, p47 phox (16). Fig. 7 shows, for the first time, that mRNA for this subunit is present in H-146 cells. This observation has allowed us to exploit both PKC-regulation of NADPH oxidase and the fact that the candidate K ϩ channel is not selectively regulated by PKC per se (21) to test the hypothesis that O 2 transduction relies, at least in part, on the sensing of pO 2 by NADPH oxidase. Clearly, there are many targets for PKC-dependent phosphorylation. However, there is accumulating evidence in NEB and NEB-derived cells that, in terms of O 2 signal transduction, NADPH oxidase activity, and its regulation by PKC, are of paramount importance. First, H 2 O 2 , a product of the sequential actions of NADPH oxidase and catalase/superoxide dismutase, appears to provide the required oxidized environment needed for tonic K ϩ channel activity in normoxia (15) ; Fig. 2 shows clearly that production of H 2 O 2 by H-146 cells is suppressed during hypoxia with a time course similar to that of hypoxic K ϩ current inhibition (13). Second, phorbol 12-myristate 13-acetate treatment (another PKC-stimulating phorbol ester) causes a dramatic increase in intracellular H 2 O 2 concentration; presumably via activation of p47 phox (15). Based upon these observations, we would predict that any agent that increases the affinity of NADPH oxidase for substrate will ameliorate the inhibitory action of hypoxia.
To investigate the potential involvement of PKC in O 2 sensing by H-146 cells, it was important first to characterize more completely the O 2 sensitivity of hypoxic inhibition of K ϩ currents in these cells. Fig. 1A shows that the inhibition of K ϩ currents (measured at a test potential of 0 mV) by hypoxia was clearly graded, with significant inhibition only observed at a pO 2 of 45 mm Hg or lower. This is in accordance with studies of the [Ca 2ϩ ] i response of arterial chemoreceptor cells (20) and further validates our own model system of airway chemoreceptor cells. Most importantly, we found an excellent correlation between hypoxia-induced K ϩ current suppression and hypoxiaevoked membrane depolarization (Fig. 1B). This finding justifies our use of measuring current amplitude at 0 mV (which provides an acceptable compromise between signal to noise ratio and resolution of the physiological action of hypoxia) as a reliable indicator of events occurring at resting membrane potential, as we have earlier suggested (14).
Phorbol ester activation of PKC often requires several minutes to become maximal, and so to investigate the effects of PKC activation on K ϩ currents in H-146 cells, we pretreated cells for 10 min with such agents. Nonspecific inhibitory effects of phorbol esters on ion channels have been documented previously (22), and the data shown in Fig. 3 agree with this finding; phorbol ester treatment suppressed mean current amplitude but this can be divided into both PKC-dependent and -independent components. Because the effects of TPA were partially reversed by inhibiting PKC with BIM, it appears that the PKC-dependent component is significant. However, 4␣-PDD, a phorbol ester that is inactive with respect to PKC activation, also caused some current depression. Although these data contrast with recent studies in type I carotid body cells, which indicate that TPA selectively inhibits O 2 -sensitive Ca 2ϩ -dependent K ϩ channels solely via PKC activation (23), the O 2 -sensitive K ϩ current in H-146 cells is clearly not Ca 2ϩdependent (13). Indeed, our most recent study (14) has shown that the O 2 -sensitive K ϩ current in H-146 cells is likely to be a member of the recently characterized tandem-P domain family of K ϩ channels (TWIK (24), TREK (25) TASK-1 (21), TASK-2 (26), and TRAAK (27)) and shows structural similarities to TASK. Furthermore, our data in Fig. 5D show very clearly that proportionally the 4-AP-insensitive current (of which the O 2sensitive current is a large component) is unaffected by phorbol ester treatment. Therefore, although TPA causes a generalized depression of K ϩ currents, it does not affect the O 2 -sensitive component per se, but, instead, it specifically ameliorates the hypoxia sensitivity of that component.
Indeed, the most important observation of the present study is that TPA modulates the O 2 sensitivity of K ϩ currents in H-146 cells (Figs. 4 and 5), and such a finding has important implications for the frequently proposed involvement of NADPH oxidase in O 2 chemoreception in various tissues, including airway chemoreceptors (15), arterial chemoreceptors (28), and pulmonary vascular smooth muscle (29). This model proposes that O 2 sensing is dependent on the generation of H 2 O 2 by NADPH oxidase under normoxic conditions. This enzyme has been shown to produce H 2 O 2 tonically, and both native and recombinant subtypes of K ϩ channels (including those found in NEB cells and SCLC cells (15)) have been shown to be enhanced by H 2 O 2 , raising the attractive concept that hypoxia inhibits K ϩ currents by reducing H 2 O 2 production by NADPH oxidase simply by limiting the enzyme's substrate, O 2 . We have now provided direct evidence that acute hypoxia reduces H 2 O 2 production (Fig. 2). The ability of NADPH oxidase to generate H 2 O 2 can be enhanced by the phorbol ester phorbol 12-myristate 13-acetate, which is another known activator of PKC. Our findings indicate that the PKC-activating phorbol ester, TPA, suppressed the ability of hypoxia to inhibit K ϩ currents in H-146 cells. This specific action of TPA can be attributed to PKC activation, because it was not mimicked by another phorbol ester, 4␣-PDD (which does not activate PKC) and was prevented by the PKC inhibitor, BIM. Because the nonspecific suppressing effects of TPA on K ϩ currents measured under normoxic conditions were largely because of a PKCindependent action of TPA (see above), the PKC-dependent suppression of hypoxic inhibition is likely to occur at a site upstream of the channel itself. Furthermore, the channel which is likely to underlie the hypoxia-sensitive K ϩ current in H-146 cells is possibly TASK-related, and TASK is insensitive to the actions of phorbol esters (21). The cytosolic protein p47 phox is one of several subunits comprising the NADPH oxidase complex and is a known substrate for PKC (16). Its stimulation by phorbol 12-myristate 13-acetate (presumed to be mediated by PKC phosphorylation) has been suggested to account for the enhanced H 2 O 2 production under normoxic conditions in NEB cells, as determined by rhodamine 123 fluorescence (15). In the context of this finding, our results provide compelling evidence that supports the idea that NADPH oxidase is the O 2 sensor in the hypoxic signal transduction pathway. In this model, the tonic generation of H 2 O 2 in H-146 cells provides the appropriate environment for normoxic channel activity. That this activity is maximal in normoxia is evidenced by the ability of exogenously applied H 2 O 2 to enhance the O 2 -sensitive current amplitude only during hypoxia (Fig. 6B) and not during normoxia (Fig. 6A). This is in contrast to results reported for native NEB cells where H 2 O 2 transiently activates K ϩ currents in normoxia (15). This may be because of differences between NEB and H-146 cell NADPH oxidase turnover or specific class of K ϩ channel targeted. Although this may suggest limitations of our model, it does not change the picture of the basic transduction pathway per se. Thus, following PKC activation, H 2 O 2 levels are likely to be enhanced in H-146 cells as they are in NEB cells (15) due possibly to increased affinity of NADPH oxidase for O 2 . Evidence for a PKC-dependent increase in substrate affinity is shown by the kinetic analyses of depression of K ϩ currents during graded hypoxia. These analyses, which are summarized in Fig. 8, show that PKC increases O 2 affinity ϳ6-fold and provide an explanation for how H 2 O 2 , and therefore, channel open state probability is maintained in the face of reduced pO 2 .
In summary, PKC activation reduces the sensitivity of H-146 cells to changes in pO 2 . Furthermore, because the addition of H 2 O 2 , a downstream product of NADPH oxidase, only activates K ϩ currents during hypoxia (when H 2 O 2 production is sup- pressed), it appears likely that PKC-dependent phosphorylation modulates the affinity of NADPH oxidase for O 2 , and the PKC substrate p47 phox (a component of the functional NADPH oxidase) is clearly present in these cells. Taken together, these data show that PKC is an important regulator of the O 2transduction pathway and provide functional evidence that suggests that NADPH oxidase is a major airway O 2 sensor.