Redox control of oxygen sensing in the rabbit ductus arteriosus

Abstract

1. How the ductus arteriosus (DA) closes at birth remains unclear. Inhibition of O₂-sensitive K⁺ channels may initiate the closure but the sensor mechanism is unknown. We hypothesized that changes in endogenous H₂O₂ could act as this sensor.

2. Using chemiluminescence measurements with luminol (50 μm) or lucigenin (5 μm) we showed significantly higher levels of reactive O₂ species in normoxic, compared to hypoxic DA. This increase in chemiluminescence was completely reversed by catalase (1200 U ml⁻¹).

3. Prolonged normoxia caused a significant decrease in K⁺ current density and depolarization of membrane potential in single fetal DA smooth muscle cells. Removal of endogenous H₂O₂ with intracellular catalase (200 U ml⁻¹) increased normoxic whole-cell K⁺ currents (I_K) and hyperpolarized membrane potential while intracellular H₂O₂ (100 nm) and extracellular t-butyl H₂O₂ (100 μm) decreased I_K and depolarized membrane potential. More rapid metabolism of O₂⁻· with superoxide dismutase (100 U ml⁻¹) had no significant effect on normoxic K⁺ currents.

4. N-Mercaptopropionylglycine (NMPG), duroquinone and dithiothreitol all dilated normoxic-constricted DA rings, while the oxidizing agent 5,5′-dithiobis-(2-nitrobenzoic acid) constricted hypoxia-dilated rings. NMPG also increased I_K. We conclude that increased H₂O₂ levels, associated with a cytosolic redox shift at birth, signal K⁺ channel inhibition and DA constriction.

Persistent patency of the DA is the second most common congenital heart defect in infancy and childhood (Coggin et al. 1970). Consequently, an understanding of the normal mechanism of closure is critically important. The DA acts as a right to left shunt in the developing fetus. At birth the rise in alveolar O₂ tension initiates constriction of the DA, thus eliminating the shunt (Heymann & Rudolph, 1975). The incidence of patent DA is greater under hypoxic conditions, as in populations living at high altitude (Alzamora-Castro et al. 1960). While the mechanism by which O₂ initiates constriction is through a direct effect on the DA smooth muscle cells (SMCs; Fay, 1971), the transduction pathway by which this change is signalled is unclear (Smith, 1998). There are substantial data suggesting that the release of endothelin (Coceani et al. 1989, 1992) and decreased levels of dilator prostaglandins (Coceani & Olley, 1973; Clyman et al. 1978) are likely to be important in maintaining DA closure, but the initiating mechanism remains controversial. Roulet & Coburn (1981) showed that the initial closure is, at least partially, dependent on changes in the membrane potential, since O₂-dependent constriction could be significantly reduced by preconstriction with 126 mm KCl. Consistent with this, recent data demonstrate that increases in O₂ tension can inhibit K⁺ channel activity recorded from DA SMCs, depolarize the membrane potential and increase intracellular Ca²⁺ (Nakanashi et al. 1993; Tristani-Firouzi et al. 1996). K⁺ channel inhibition may therefore be involved in normoxic constriction of the DA. We have previously suggested that redox changes provide the signal by which O₂ alters the K⁺ channel gating in pulmonary artery SMCs (Archer et al. 1986, 1993; Reeve et al. 1995). Consequently, if the DA behaves in a similar manner to the lung, in terms of redox changes, an increase in O₂ tension would be expected to increase O₂ radical production (Freeman & Crapo, 1981). We investigated the redox mechanism by which DA SMC K⁺ channels might sense the change in O₂ at birth. We determined that the H₂O₂ level does increase during normoxia in the DA tissue and that the removal of endogenous H₂O₂ by catalase increases I_K. Oxidizing agents constrict isolated hypoxic DA rings, while reducing agents and agents that facilitate electron transfer dilate normoxic DA rings. These results suggest that the shift in the cytosolic redox status of DA SMCs to a more oxidized state, associated with increased H₂O₂, acts as the signalling mechanism for DA K⁺ channel inhibition and constriction at birth.

METHODS

All animal studies were approved by the Institutional Animal Care and Use Committee, and conformed to current National Institutes of Health and American Physiological Society guidelines for the use and care of laboratory animals.

Determination of tissue levels of reactive O₂ species

Rabbit DA rings were isolated from fetal pups as previously described (Tristani-Firouzi et al. 1996) and maintained at 37 ^oC in a Krebs-Hensleit bicarbonate-buffered solution. Rings were bubbled for 20 min with either 2.5 % O₂-2.5 % CO₂-balance N₂ (hypoxia) or 20 % O₂-2.5 % CO₂-balance N₂ (normoxia), and then incubated with either 50 μm luminol or 5 μm lucigenin for a further 15 min (continuous bubbling), before measuring chemiluminescence in a liquid scintillation counter (Packard 1900CA). In some experiments, chemiluminescence was measured in normoxic tissue prior to the 15 min incubation with 1200 U ml⁻¹ catalase, after which chemiluminescence was measured again. At the time of chemiluminescence measurement, pH was 7.37 ± 0.01 (hypoxia) and 7.35 ± 0.01 (normoxia); temperature was 38.8 ± 0.2 ^oC (hypoxia) and 38.6 ± 0.2 ^oC (normoxia) and P_O2 was 30.8 ± 1.8 mmHg (hypoxia) and 140 ± 5.9 mmHg (normoxia). Background counts (blanks) were recorded in luminol- or lucigenin-containing solutions in the absence of tissue and subtracted to obtain final counts. In these blanks, there was no significant difference in counts between normoxia and hypoxia. All counts were adjusted for total weight (mg). Numbers presented (n) indicate the actual number of DAs used which were harvested from fetuses from at least three different mother rabbits to allow for animal variation. The total number of mother rabbits used for the study was 46.

Electrophysiology

The conventional, whole-cell patch-clamp configuration (Hamill et al. 1981), was used to record K⁺ current and membrane potential from freshly isolated fetal DA SMCs. Cells were obtained as previously described by a two-step enzymatic digestion with papain and collagenase (Tristani-Firouzi et al. 1996). For normoxic recordings, cells were allowed to equilibrate with room air throughout the digestion and subsequent refrigeration in low Ca²⁺ Hanks' solution. For hypoxic recordings, all solutions were deoxygenated with N₂. Recordings of both K⁺ current and membrane potential were made using an extracellular perfusate of composition (mm): NaCl, 115; KCl, 5.4; MgCl₂, 1; NaHCO₃, 25; CaCl₂, 1.5; Hepes, 10 and glucose, 10 (pH 7.4 with NaOH). For hypoxic recordings, this solution was bubbled with 2.5 % CO₂-balance N₂ (bath P_O2= 36 ± 3.8 mmHg; n = 12) while for normoxic recordings, the solution was bubbled with 20 % O₂-2.5 % CO₂-balance N₂ (bath P_O2= 125 ± 6.2 mmHg; n = 9). Cells were perfused at a rate of 2 ml min⁻¹. Electrodes were pulled from borosilicate glass and had a resistance of 2.75-4.0 MΩ when filled with a solution of (mm): KCl, 140; MgCl₂, 1; Hepes, 10; ATP, 1 and EGTA, 0.1 (pH 7.2 with KOH). For recording K⁺ current, both current-voltage (I-V) curves and ramp protocols were used. I-V curves were constructed by stepping from a holding potential of -70 mV using 100 ms duration pulses at a frequency of 0.1 Hz. Compensation of whole-cell capacitance allowed an estimation of cell size (10.2 ± 0.4 pF, n = 33) allowing I-V curves to be constructed using both current (pA) and current density (pA pF⁻¹). Ramps of 1 s duration were recorded from the same holding potential. Currents were filtered at 1 kHz and digitized at 2 kHz. Membrane potential recordings were made using the whole-cell configuration with the same electrode solutions at the resting potential of the cell (without current injection). Series resistance and leak were checked at the beginning and end of each membrane potential experiment to eliminate artifactual changes in potential. All data were recorded and analysed using pCLAMP 6.04 software (Axon Instruments, Foster City, CA, USA). For recordings of the effect of catalase, 200 U ml⁻¹ was included in the pipette solution and pH tested to eliminate potential pH-induced effects. To eliminate superoxide (O₂⁻·) effects, 100 U ml⁻¹ superoxide dismutase was included in the pipette. H₂O₂ (100 nm) was included in the pipette while t-butyl H₂O₂ (t-BOOH; 1 and 100 μm) and N-mercaptopropionylglycine (NMPG; 10 mm) were added directly to the bath via a microinjection pipette.

Measurement of changes in tone in isolated DA rings

Ductus arteriosus rings were equilibrated in Earle's bicarbonate buffered medium under hypoxia (0 % O₂-2.5 % CO₂-balance N₂: P_O2≤ 30 mmHg) for 60 min at a tension of 400 mg. This tension has been proven to be optimal for this tissue as determined by previous studies using increasing concentrations of KCl. Three micromolar indomethacin and 100 μml-NAME were present throughout all experiments. These inhibitors of prostaglandin and nitric oxide synthesis, respectively, were used so that a contractile mechanism independent of these mediators could be studied. Drugs were added to the bath in either hypoxia (DA dilated) or normoxia (DA constricted; 20 % O₂-2.5 % CO₂-balance N₂: P_O2= 135 mmHg). Vehicle controls were done for all drugs used. Oxygen tension was recorded continuously using an M1-730 oxygen electrode (Microelectrode Inc., Bedford, NH, USA; tip diameter 2 mm). To determine potential O₂-scavenging effects of NMPG and dithiothreitol (DTT), in vitro experiments were also done. For these experiments Earle's bicarbonate buffer was equilibrated in the organ baths with normoxic gas, as above, but no tissue was added. Increasing concentrations of DTT and NMPG (1, 5 and 10 mm) were then added to the bath and changes in O₂ tension recorded.

Drugs

All drugs and salts were obtained from Sigma-Aldrich Chemicals Inc. (St Louis, MO, USA) except H₂O₂ (General Medical Corp. Richmond, VA, USA).

Statistical analysis

Values are expressed as means ±s.e.m. Student's unpaired and paired t tests were used to compare membrane potential recordings and DA changes in tone, respectively, while repeated measures or factorial ANOVAs were used to compare changes in I_K and chemiluminescence. A value of P < 0.05 was considered significant.

RESULTS

It has previously been shown that the cytosolic redox status shifts to a more oxidized state with an increase in O₂ tension from hypoxia to normoxia (Freeman & Crapo, 1981), such as that which occurs in the DA at birth. To investigate changes in levels of reactive O₂ species (ROS) in the DA, luminol- and lucigenin-enhanced chemiluminescence was recorded in isolated DA rings in either hypoxia or normoxia. Chemiluminescence was consistently higher in normoxic DA tissue than in hypoxic, using either luminol (758 ± 194 counts in hypoxia, 1178 ± 202 counts in normoxia; n = 23 ductuses) or lucigenin (386 ± 34 counts in hypoxia, 986 ± 82 counts in normoxia; n = 11 ductuses), and could be decreased on return to hypoxia (Fig. 1A). This technique, while not selective for specific radicals, has been shown to measure the superoxide dismutase-sensitive production of radicals and peroxides from the isolated lung surface (Archer et al. 1989). Incubation of the tissue with the enzyme catalase (1200 U ml⁻¹), which specifically degrades H₂O₂, completely reversed the normoxic rise in chemiluminescence suggesting that this increase in chemiluminescence may be due to a concomitant rise in H₂O₂ levels (Fig. 1A). The normoxic rise in radical production was also reversed by the electron-shuttling agent duroquinone (100 μm; n = 28; Fig. 1B).

Figure 1. Redox activity.

A, values (means ±s.e.m.) of luminol-enhanced chemiluminescence (CL) recorded from DA rings following equilibration in hypoxia (□) normoxia (▪), on return to hypoxia from normoxia () and normoxia in the presence of 1200 U ml⁻¹ catalase (). Counts are averages of those made over the first minute of recording. *P < 0.05 compared to hypoxia; †P < 0.001 compared to normoxia; ‡P < 0.05 compared to initial hypoxia or P < 0.001 compared to normoxia. B, average values (means ±s.e.m.) of luminol-enhanced CL recorded from DA rings following equilibration in hypoxia (▪) normoxia (□) and normoxia in the presence of 100 μm duroquinone (). §P < 0.05 compared to hypoxia and normoxia. For both A and B, number of DAs studied shown in parentheses.

K⁺ currents recorded from normoxic DA SMCs were significantly decreased and membrane potentials were significantly depolarized as compared to hypoxic DA SMCs (Fig. 2). To determine whether modulation of endogenous H₂O₂ or O₂⁻· levels in DA SMCs could affect K⁺ channel activity, 200 U ml⁻¹ catalase or 100 U ml⁻¹ superoxide dismutase was included in the patch pipette prior to recording K⁺ currents in normoxia. In catalase- treated cells, a time-dependent increase in K⁺ current was observed, with currents increasing at 8 min (46 ± 12 % increase from control at +50 mV; n = 6) and plateauing at 20 min (124 ± 12 % increase from control at +50 mV; n = 6; Fig. 3A and B). Control recordings with either no catalase (n = 6), 200 U ml⁻¹ boiled catalase (n = 8), sodium citrate (catalase buffer; n = 5) or an equimolar concentration of albumin (protein control; n = 3) had no effect on I_K over the same time (-0.1 ± 5 %; 1.2 ± 3 %; -2.5 ± 4 % and 1.6 ± 1 % changes from control at 20 min dialysis at +50 mV, respectively, and Fig. 3D). Following catalase, total normoxic whole-cell current was not significantly different from that recorded in hypoxia without catalase (Fig. 3D). Recordings of membrane potential with 200 U ml⁻¹ catalase in the pipette showed a time-dependent membrane hyperpolarization, which would be consistent with the opening of K⁺ channels (Fig. 4A and C). The hyperpolarization began after approximately 4 min. This is a similar time course to the increase in I_K that was recorded in voltage clamp with 200 U ml⁻¹ catalase in the pipette (Fig. 3). No change in membrane potential was recorded from cells dialysed with boiled catalase (Fig. 4A and B). Bath application of the cell-permeable form of H₂O₂ (t-BOOH) caused partial, reversible inhibition of I_K at 1 μm (Fig. 5A, n = 5; 13.3 ± 9 % inhibition at 2 min at +50 mV; n.s.) as did low concentrations of intracellular H₂O₂ (100 nm) given via the patch-pipette (n = 4, 13.2 ± 4 % decrease at +50 mV; data not shown). Increasing the concentration of t-BOOH to 100 μm caused almost complete suppression of I_K with partial recovery occurring on washout (Fig. 5B; 77 ± 6.2 % decrease at +50 mV; n = 4). Recordings of membrane potential showed that 1 μm t-BOOH had no effect even up to 5 min application, while 100 μm rapidly and reversibly depolarized membrane potential from -29 ± 6.7 mV to -1.4 ± 2.1 mV (an average depolarization of 27.6 ± 4.5 mV; n = 3; P < 0.05; Fig. 5C). Including superoxide dismutase in the pipette had no significant effect on the K⁺ current recorded over the same time period as the catalase experiments (7.4 ± 11 % change in I_K at +50 mV after 16 min dialysis, data not shown; n = 5).

Figure 2. K⁺ channel activity and membrane potential.

A, representative actual current traces recorded from a DA smooth muscle cell in hypoxia (left) and following prolonged (1-3 h) normoxia (right). Cells were held at a membrane potential of -70 mV and stepped to +50 mV in +20 mV increments. B, values (means ±s.e.m.) of whole-cell K⁺ currents recorded from ductus arteriosus (DA) smooth muscle cells at +30 mV following prolonged exposure to hypoxia or normoxia. C, values (means ±s.e.m.) for membrane potential (V_m) recorded from cells under the same conditions as in A and B. Number of cells studied shown in parentheses. *P < 0.05.

Figure 3. Catalase modulation of K⁺ current.

A, values (means ±s.e.m.) I-V relationship of current density (pA pF⁻¹) recorded at 2 min (▪); 8 min (•); 16 min (♦) and 20 min (▴) intracellular dialysis with 200 U ml⁻¹ catalase (n = 6). *P < 0.05; †P < 0.01. B, values (means ±s.e.m.; in pA) of currents taken from data presented in A, showing increasing current amplitudes at -10 mV (approximately the resting potential of the cells in normoxia) with intracellular dialysis of catalase. *P < 0.05. C, actual current traces recorded from a DA SMC by stepping from -70 mV to -10 mV following 2 min (▪) and 16 min (▴) dialysis with catalase. D, I_K recorded at +30 mV from -70 mV, from either normoxic cells dialysed with 200 U ml⁻¹ catalase (), hypoxic cells (▪) or cells dialysed with boiled catalase (). Numbers of cells shown in parentheses.

Figure 4. Catalase modulation of membrane potential.

A, values (means ±s.e.m.) values of membrane potential (V_m) recorded from control normoxic cells (□), normoxic cells dialysed with 200 U ml⁻¹ catalase () and normoxic cells dialysed with boiled catalase (). Number of cells is shown in parentheses. *P < 0.05. B and C, representative traces of membrane potential recorded from either control normoxic DA SMC (B; resting potential -14 mV) or normoxic DA SMC dialysed with 200 U ml⁻¹ catalase (C; resting potential -10 mV).

Figure 5. H₂O₂ modulation of K⁺ current.

A and B, representative voltage ramps recorded over 1 s from -70 mV to +50 mV during hypoxia and following 1 min bath application of 1 μm t-butyl H₂O₂ (t-BOOH; A) or 100 μm t-BOOH (B). C, actual trace of membrane potential recorded during bath application of 100 μm t-BOOH and following wash. Time course of t-BOOH application indicated by bar.

The electron-shuttling agent duroquinone had no effect on tone in hypoxic DA rings but consistently dilated normoxia-constricted rings (Fig. 6A; n = 6). The reducing agents DTT (1 mm; n = 4) and NMPG (10 mm; n = 8) also dilated normoxic rings (Fig. 6B and C, P < 0.01). Neither of these agents had any direct, scavenging effect on P_O2 levels even at 10 mm (Fig. 7A). In patch-clamp experiments, NMPG (10 mm) also significantly increased I_K (35 ± 7.8 % increase at +50 mV after 8 min, n = 4; P < 0.05; data not shown). In contrast to the reducing agents and electron shuttler, the oxidizing agent 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB; 1 mm) partially constricted hypoxic DA rings (n = 5; Fig. 7B). In the presence of DTNB, normoxia only constricted the DA rings to the levels achieved by normoxia in the absence of DTNB, suggesting the two mechanisms of constriction to be similar, not additive.

Figure 6. Effects of reducing agents on tone.

A-C, representative recordings of DA tone after exposure to normoxia and following treatment with the electron-shuttling agent duroquinone (1, 10 and 100 μm; A); the reducing agent dithiothreitol (DTT; 1 mm; B) and the reducing agent and H₂O₂ scavenger N-mercaptopropionylglycine (NMPG; 10 and 30 mm; C). Drugs added as indicated by arrows. Bar indicates period of exposure to normoxia.

Figure 7. Effects of oxidizing agents on tone.

A, in vitro recording of effects of NMPG and DTT on O₂ levels (P_O2) under conditions used for DA ring studies but without tissue present. Drugs added as indicated by arrows. B, values (means ±s.e.m.) of changes in DA tone (Δ constriction) caused by normoxia alone; the oxidizing agent 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB; 1 mm) and normoxia in the presence of DTNB. Number of DAs studied shown in parentheses.

DISCUSSION

The mechanism by which the DA constricts at birth remains controversial (for review see Smith, 1998). While it is widely accepted that dilator prostaglandins (in particular PGE₂) are essential in maintaining patency of the ductus during development (Coceani & Olley, 1988) closure of the DA requires more than just the passive removal of this dilator effect. It has been suggested that the rise in O₂ at birth releases endothelin from endothelial and smooth muscle cells and that this causes constriction of the DA via the ET_A (endothelan A) receptor (Coceani et al. 1992). While it is possible that changes in endothelin levels maintain the closure of the DA in the neonate, there is evidence that this is not the mechanism by which initial DA constriction occurs (Fineman et al. 1998; Smith, 1998). This means that alternative pathways must be investigated. Roulet & Coburn (1981) found that O₂-induced constriction of the DA was associated with smooth muscle depolarization, indicating a role for changes in ion channel activity in closure, in particular K⁺ channel activity which controls resting membrane potential in DA SMCs (Tristani-Firouzi et al. 1996).

While the O₂ sensitivity of K⁺ channels is a generally accepted phenomenon (Lopez-Barneo et al. 1988; Peers, 1990; Post et al. 1992; Youngson et al. 1993; Archer et al. 1996; Buckler, 1997; Ospienko et al. 1997) the signalling mechanism by which the channels sense changes in O₂ remains unknown. We report here a mechanism by which redox changes in the cytosol of SMCs may act as the sensor to trigger changes in K⁺ channel activity. Recent controversy has arisen regarding changes in ROS in hypoxia and normoxia. While it might seem intuitive that levels of ROS would decrease in hypoxia, and indeed this has been shown to be the case in some tissues (Freeman & Crapo, 1981; Archer et al. 1989), recent studies have also reported increased levels of ROS in hypoxia (Chandel et al. 1998; Duranteau et al. 1998; Killilea et al. 2000). The reason for these discrepancies is unknown but may relate to differences in technique for recording redox status (fluorescence assay versus chemiluminescence), differences between fresh tissue and cultured cells and exposures to hypoxia (short-term versus prolonged or ischaemia/reperfusion). Since none of the techniques routinely used for measuring redox status are considered completely specific, it might be wise to always use a combination of techniques for these measurements. For the studies presented, changes in the redox status of the cytosol were measured using chemiluminescence. Chemiluminescence has been criticized for its potential to indirectly produce O₂⁻· in normoxia (Fridovich, 1997). To try to ensure that the change in ROS was not a result of auto-oxidation in normoxia, studies were carried out using both lucigenin and luminol, and similar results recorded. Furthermore, background counts, recorded prior to addition of tissue, did not change significantly between normoxia and hypoxia. Together, these experiments suggest that ROS levels increase in normoxia in the DA. In addition, the change in ROS recorded by switching from hypoxia to normoxia could be completely reversed by incubation of the tissue with the H₂O₂ metabolizing enzyme, catalase. Extracellular catalase has been previously shown to reduce increases in intracellular H₂O₂ as recorded by fluorescence techniques (Maziar-Zafari et al. 1998). This suggests that the H₂O₂ is either freely diffusible or the catalase can enter the cell. The reversal of the increase in ROS by catalase indicates that the majority of the increase in chemiluminescence recorded was due to endogenous increases in levels of H₂O₂, or a subsequent ROS such as the hydroxyl radical (OH·).

There is substantial evidence that K⁺ channels are redox sensitive (Ruppersburg et al. 1991; Vega-Saenz de Miera et al. 1992; Park et al. 1997; Szabo et al. 1997; Filipovic & Reeves, 1997), including elegant studies showing that H₂O₂-sensitive channels have a common cysteine residue in a similar amino acid sequence at the amino (intracellular) terminus of the protein (Vega-Saenz de Miera et al. 1992). However, most previous studies have determined the effects of high concentrations of exogenous redox agents given via the extracellular perfusate. Since physiological changes in the redox status of the cell would primarily affect intracellular sites, this method of application is not optimal. Indeed, previous studies looking at the effects of H₂O₂ often used micromolar concentrations (Szabo et al. 1997; Filipovic & Reeves, 1997) which are a million times greater than concentrations that would be produced by the mitochondria (nm) (Phung et al. 1994), which in our hands could not be tolerated by vascular SMCs if given directly into the cytosol (H. L. Reeve, unpublished observations).

To determine the effect of endogenous H₂O₂ on K⁺ currrents in the DA SMCs, catalase was included in the pipette to enhance the breakdown of cellular H₂O₂ to O₂ and H₂O. As the pipette solution dialysed into the cell, it produced a significant increase in the amplitude of I_K and membrane hyperpolarization. Since we also show that H₂O₂ and/or associated ROS levels increase in normoxia in DA, this observation is consistent with K⁺ channels being open, membrane potential being hyperpolarized and the DA being dilated under hypoxic conditions, when H₂O₂ levels are low. The formation of O₂ from H₂O₂ through the action of catalase cannot be responsible for K⁺ channels being open, as an increase in O₂ results in closure of K⁺ channels and membrane depolarization (Tristani-Firouzi et al. 1996). Since other ROS can be formed during normoxia including O₂⁻· and OH·, studies were also done with superoxide dismutase in the pipette to investigate potential effects of endogenous O₂ on K⁺ channels. There was no significant effect of superoxide dismutase on I_K. Since superoxide dismutase will remove endogenous O₂⁻· by the production of H₂O₂, it might be hypothesized that its presence in the pipette should actually cause a decrease in the amplitude of I_K. In five cells in which this was monitored, there was no consistent decrease in I_K over time. It is likely, therefore, that H₂O₂ produced by the metabolism of superoxide dismutase is removed by endogenous catalase activity and that there is no change in the absolute quantity of H₂O₂ generated. A potential role of OH· in the K⁺ channel effect of catalase should be considered. Theoretically, removal of H₂O₂ with catalase would lead to a decrease in levels of any OH· formed via the Haber-Weiss reaction and so the increase in I_K observed with catalase could be related to reduced levels of OH·. Since the formation of OH· via the Haber-Weiss reaction also requires the presence of O₂⁻·, it might be expected that if the decrease in OH· found with catalase were responsible for the increase in I_K, this increase would also be observed in cells dialysed with superoxide dismutase. This was not the case, suggesting that it is a change in the level of endogenous H₂O₂ which is directly affecting the K⁺ channels.

While levels of endogenous H₂O₂ do appear to change between hypoxia and normoxia in the DA, the source of the changes has not been explored in this study. For instance, the increase in H₂O₂ in normoxia could result from an increase in cytosolic levels as a result of mitochondrial production or the activity of NADPH oxidase.

In addition to modulation of tone, changing K⁺ channel activity has also been shown to directly affect cellular proliferation (Chiu & Wilson, 1989; Nilius & Wohlrab, 1992; Vaur et al. 1998). Indeed a combination of the two effects are often intimately linked in pathophysiological events in the pulmonary circulation. In the DA, normoxic constriction is followed by cellular necrosis and fibrosis to obliterate the shunt pathway (Heymann & Rudolph, 1975; Drayton & Skidmore, 1987). Recent data show that H₂O₂ levels can also directly modulate cellular proliferation, as an over-expression of catalase in smooth muscle (and hence a removal of endogenous H₂O₂ equivalent to hypoxia) inhibits proliferation (Maziar-Zafari et al. 1998; Brown et al. 1999). Whether this also occurs through changes in K⁺ channel activity is unknown but indicates an additional, important correlation between H₂O₂ levels and smooth muscle cellular physiology.

The reducing agent NMPG has been shown to scavenge H₂O₂ in PC12 cells and to reduce levels to those found in hypoxia (Czyzyk-Krzeska et al. 1998). In the DA, NMPG both increased I_K and dilated normoxic-constricted rings. Consistent with these observations, other reducing or electron shuttling agents such as DTT and duroquinone, also dilated normoxia-constricted DA rings. Since duroquinone can act to shuttle electrons in either direction, studies were done to determine its effect on chemiluminescence in isolated rings. It was found to decrease chemiluminescence to levels similar to hypoxia (Fig. 1B) suggesting that it was, in fact, acting like a reducing agent. While the concentrations of these agents used to cause dilatation were higher than those found to affect I_K, they appeared to have no direct O₂-scavenging effects which could have explained their dilator properties (Fig. 7A). It is likely that the higher concentrations used in rings simply reflect the difference between giving a drug directly to the smooth muscle cells (as occurs in the patch-clamp experiments) and requiring it to enter the smooth muscle layer in a ring with intact endothelium and adventitia. Experiments in isolated hypoxic DA with the oxidizing agent DTNB show that there is no significant difference between the nomoxia-induced constriction and the DTNB constriction. In addition, exposure of DTNB-constricted rings to normoxia, in the continued presence of the oxidizing agent, causes no additional change in tone. This suggests that the mechanism by which oxygen constricts the DA is redox dependent.

The DA and pulmonary arteries have exactly opposite responses to O₂. Consequently it is not surprising that in the pulmonary artery and in isolated lungs, reducing agents cause vasoconstriction (Reeve et al. 1995). Similarly, oxidizing agents are known to dilate the pulmonary vasculature (Reeve et al. 1995) while in the DA they cause constriction. Pulmonary arteries constrict in response to hypoxia through inhibition of a K_v channel, (Yuan et al. 1993; Archer et al. 1996) while the rabbit DA may initially constrict in response to normoxia through inhibition of a K_v channel (Tristani-Firouzi et al. 1996). The data presented give rise to the intriguing concept that the same cytosolic redox changes in these two tissues have opposite effects on K⁺ channel gating, accounting for their opposing responses to O₂.