Regulation of swelling-activated Cl⁻ current by angiotensin II signalling and NADPH oxidase in rabbit ventricle

Abstract

Aims

We assessed whether hypoosmotic swelling of cardiac myocytes activates volume-sensitive Cl⁻ current (I_Cl,swell) via the angiotensin II (AngII)-reactive oxygen species (ROS) signalling cascade. The AngII-ROS pathway previously was shown to elicit I_Cl,swell upon mechanical stretch of β_1D integrin. Integrin stretch and osmotic swelling are, however, distinct stimuli. For example, blocking Src kinases stimulates swelling-induced but inhibits stretch-induced I_Cl,swell.

Methods and results

I_Cl,swell was measured in rabbit ventricular myocytes by whole-cell voltage clamp. Swelling-induced I_Cl,swell was completely blocked by losartan and eprosartan, AngII type I receptor (AT₁) antagonists. AT₁ stimulation transactivates epidermal growth factor receptor (EGFR) kinase. Blockade of EGFR kinase with AG1478 abolished both I_Cl,swell and AngII-induced Cl⁻ current, whereas exogenous EGF evoked a Cl⁻ current that was suppressed by osmotic shrinkage. Phosphatidylinositol 3-kinase (PI-3K) is downstream of EGFR kinase, and PI-3K inhibitors LY294002 and wortmannin blocked I_Cl,swell. Ultimately, AngII signals via NADPH oxidase (NOX) and superoxide anion, $•{\text{O}}_2^{-}$. NOX inhibitors, diphenyleneiodonium, apocynin and gp91ds-tat, eliminated I_Cl,swell, whereas scramb-tat, an inactive gp91ds-tat analogue, was ineffective. $•{\text{O}}_2^{-}$ rapidly dismutates to H₂O₂. Consistent with H₂O₂ being a downstream effector, catalase inhibited I_Cl,swell, and exogenous H₂O₂ overcame suppression of I_Cl,swell by AT₁ receptor, EGFR kinase, and PI-3K blockers. H₂O₂-induced current was not blocked by osmotic shrinkage, however.

Conclusion

Activation of I_Cl,swell by osmotic swelling is controlled by the AngII-ROS cascade, the same pathway previously implicated in I_Cl,swell activation by integrin stretch. This in part explains why I_Cl,swell is persistently activated in several models of cardiac disease.

1. Introduction

Volume-sensitive Cl⁻ current, I_Cl,swell, is elicited by osmotic swelling and isosmotic hydrostatic pressure-induced inflation of cardiac myocytes, and the current is persistently activated in myocytes isolated from ischaemic and non-ischaemic models of dilated cardiomyopathy in which angiotensin II (AngII) levels are elevated (for review,^1–3). I_Cl,swell influences cardiac electrical activity^4,5 and cell volume regulation⁶ and may have a role in ischaemic preconditioning⁷ and apoptosis.⁸

I_Cl,swell also is activated by stretching β_1D integrins with paramagnetic beads coated with anti-β_1D monoclonal antibodies; integrin-bound beads are pulled towards an electromagnet placed above the cells.^9–11 Integrin stretch elicits I_Cl,swell via Src, focal adhesion kinase (FAK), and the AngII AT₁ receptor signalling cascade. AT₁ signalling involves transactivation of epidermal growth factor receptor (EGFR) kinase, phosphatidylinositol 3-kinase (PI-3K), sarcolemmal NADPH oxidase (NOX), superoxide anion ( $•{\text{O}}_2^{-}$), and ultimately H₂O₂.^12–14 AngII is released by cardiac stretch, stretch causes AT₁-dependent hypertrophy,¹⁵ and AngII activates NOX in cardiomyocytes^16–19 and the vasculature.^14,20 AngII and stretch also were recently found to destabilize Kv4.3 mRNA in neonatal rat myocytes via NOX-dependent $•{\text{O}}_2^{-}$ production.¹⁹

Although I_Cl,swell is evoked both by stretching β_1D integrins and by osmotic swelling, these are different stimuli and may signal by different pathways. Osmotic swelling dilutes the intracellular milieu and reduces its ionic strength, while integrin stretch is localized and does not alter the contents of the cytoplasm. Furthermore, PP2, an inhibitor of Src family tyrosine kinases, blocks I_Cl,swell activation upon integrin stretch⁹ consistent with its role as an upstream mediator of NOX activity,¹² whereas PP2 augments I_Cl,swell in osmotically swollen myocytes.^21–23

The goal of the present study was to determine whether osmotic control of I_Cl,swell utilizes the same AngII signalling cascade engaged by β_1D integrin stretch, despite differences between the stimuli and observations that Src kinase inhibition has opposite effects on swelling- and stretch-induced I_Cl,swell. We found that activation of I_Cl,swell by osmotic swelling was abrogated by inhibition of AT₁, EGFR kinase, PI-3K, or NOX and by scavenging H₂O₂. Moreover, exogenous epidermal growth factor (EGF) elicited I_Cl,swell and exogenous H₂O₂ overcame block of AT₁ receptors, EGFR kinase, and PI-3K. In contrast, osmotic shrinkage failed to suppress H₂O₂-induced I_Cl,swell. These data argue that the AngII-ROS signaling cascade participates in the response of cardiomyocytes to osmotic swelling. AngII-dependent I_Cl,swell activation may modulate electrical activity and cell volume in cardiac disease.

2. Methods

2.1. Ventricular myocytes

Studies conform to Guide for the Care and Use of Laboratory Animals (NIH Publication 85-23, revised 1996). Left ventricular myocytes were isolated from anesthetized New Zealand rabbits (~3–4 kg) using collagenase (type II) and pronase (type XIV).^11,22 Cardiomyocytes were washed twice and stored in modified Kraft–Brühe solution (pH 7.2; 295 mosmol/kg).¹¹ Rod-shaped quiescent cells with clear striations and no membrane blebs were studied within 8 h of isolation.

2.2. Solutions and drugs

Bath solutions designed to isolate anion currents were isosmotic (1T; 300 mosmol/kg; T, times isosmotic), hypoosmotic (0.7T), or hyperosmotic (1.5T) and contained (mM): 90 N-methyl-D-glucamine-Cl, 3.0 MgCl₂, 10 HEPES, 10 glucose, 5 CsCl, 1 BaCl₂, 0.2 CdCl₂, 0–250 mannitol (pH 7.4, CsOH). These solutions suppressed the time-dependence of I_Cl,swell at positive potentials²² and allowed osmotic challenges at constant ionic strength. Osmolarity was verified by freezing-point depression.

Stock solutions of apocynin, AG1478, diphenyleneiodonium-Cl (DPI), LY294002, and wortmannin in DMSO and EGF, losartan-K, and eprosartan mesylate in H₂O were frozen (−20°C) in aliquots until use. A membrane-permeant fusion peptide inhibitor of NOX, gp91ds-tat, and a permeant but inactive scrambled analogue, scramb-tat, were synthesized.²⁴ The inhibitor is a 9-mer that blocks NOX assembly by mimicking the gp91^phox (NOX2) docking site for p47^phox joined to a tat 9-mer that drives transmembrane uptake. Peptide stocks (1.2 mg/ml) were made in 150 mM NaCl plus 10 mM acetic acid and frozen (−20°C) in aliquots until use. Final diluent concentrations, 0.1–0.5%, did not alter I_Cl,swell.

2.3. Electrophysiology

Myocytes were placed in a poly-L-lysine-coated chamber and super-fused at ~2 ml/min (21–22°C). Pipettes (2–3 MΩ) were filled with (mM): 110 Cs-aspartate, 20 CsCl or 20 TEA-Cl, 2.5 Mg-ATP, 8 Cs₂-EGTA, 0.15 CaCl₂, 10 HEPES (pH 7.1, CsOH; liquid junction potential, −11.5 ±0.7 mV, n = 9).²² This gave a free-[Ca²⁺]_i of ~60 nM (WinMAXC 2.40; http://www.stanford.edu/~cpatton). Junction potentials were corrected, and ground was a 3-M KCl agar bridge. Seal resistances of 5–30 GΩ were achieved.

Myocytes were dialyzed for 10 min before data were taken. Whole-cell currents were recorded with an Axoclamp 200B and Digidata 1322A under pClamp 8. Currents were low-pass filtered (Bessel, 2 kHz) and digitized (5 kHz). Membrane capacitance was calculated from 5-mV steps. Successive 500-ms steps were made from −60 mV to test potentials between −100 and +60 mV in +10 mV increments. I–V relationships were obtained at 1-min intervals to track responses to interventions and were plotted from quasi steady-state currents. In preliminary studies, I_Cl,swell fully activated in <5 min and was stable for at least 45 min. All agents were applied for sufficient time for currents to reach steady state.

Because myocytes were studied under whole-cell conditions, interventions that alter I_Cl,swell are not expected to significantly alter cell volume.^21,25 This was confirmed in preliminary studies using video microscopy methods as previously described.⁶

2.4. Statistics

Percent block or activation (±SEM) were calculated using each myocyte as its own control and assessed with paired t-tests or, in one case, with a Wilcox rank sign test because the data were not normally distributed; n represents the number of cells, and P < 0.05 was taken as significant. Analysis of mean current densities (pA/pF) for each treatment group by repeated-measures ANOVA and Student–Newman–Keuls tests gave identical statistical conclusions.

3. Results

3.1. AT₁ signalling

Blocking AT₁ receptors with losartan inhibits I_Cl,swell evoked by integrin stretch, and exogenous AngII elicits a Cl⁻ current attributed to I_Cl,swell.¹⁰ Figure 1A–C shows that AT₁ receptors also are involved in I_Cl,swell activation upon osmotic swelling in 0.7T. As expected, swelling turned on an outwardly rectifying current that reversed at −44 ± 1 mV, near the Cl⁻ equilibrium potential (E_Cl), −42 mV. Adding losartan (30 μM, 10–13 min) to 0.7T strongly suppressed I_Cl,swell. The losartan-sensitive difference current at +60 mV was 101 ±6% (P < 0.001, n = 10) of the current activated by swelling. Although currents were smaller at −100 than +60 mV, similar results were obtained at both voltages in this and subsequent protocols. Eprosartan (30 μM, 10 min), a carboxybenzylimidazol AT₁ antagonist with structural features distinct from losartan, a biphenylte-trazole,²⁶ also fully inhibited swelling-induced currents (98 ±2%; P < 0.001, n = 9) (Figure 1D and E). In contrast, block of I_Cl,swell by a lower concentration of losartan (10 μM, 10 min) was incomplete (68 ±4%; P < 0.001, n = 6).

Figure 1.

AT₁ blockers inhibit I_Cl,swell during osmotic swelling, and block is reversed by H₂O₂. (A) Currents and (B) I–V relationships in 1T, 0.7T (10 min), after losartan (30 μM, 10 min) in 0.7T (LOS), and after H₂O₂ (100 μM, 10 min) in presence of losartan (H₂O₂). (C ) Time course of I_Cl,swell at +60 mV; dash line, control current in 1T. (D and E) Analogous experiments with eprosartan (EPRO; 30 μM, 10 min). I_Cl,swell is time-independent because bath contained Cd²⁺ and Ba^2+.22

Integrin stretch elicits I_Cl,swell in cardiomyocytes via AT₁ signalling and its downstream effector H₂O₂.^9,10 If the same pathway is stimulated by osmotic swelling, exogenous H₂O₂ (100 μM) should overcome block of I_Cl,swell by losartan (Figure 1A–C). H₂O₂-induced current in the presence of losartan was 154 ±6% (P < 0.001, n = 6) of I_Cl,swell, confirming that ROS are downstream from AngII in the regulatory pathway.

3.2. EGFR kinase

EGFR kinase is a receptor protein tyrosine kinase that undergoes transactivation upon AT₁ occupancy and is downstream from AT₁ in the AngII-ROS cascade.^12,27 Figure 2A–C demonstrates that AG1478, a specific EGFR kinase blocker (IC₅₀ = 0.003–0.9 μM), fully inhibited I_Cl,swell. Addition of AG1478 (1 μM, 10–13 min) to 0.7T suppressed 98 ±3% (P < 0.001, n = 11) of the swelling-induced current at +60 mV. Similar results were obtained in human atrial²¹ and rabbit ventricular²² myocytes using two other EGFR kinase blockers, AG556 and PD153035. As shown for losartan, block of I_Cl,swell by AG1478 was overcome by exogenous H₂O₂. The H₂O₂-induced current in the presence of AG1478 was 150 ±15% (P < 0.001, n = 6) of I_Cl,swell, indicating that ROS are down-stream from EGFR kinase.

Figure 2.

Block of EGFR kinase suppresses I_Cl,swell and is reversed by exogenous H₂O₂; exogenous EGF activates I_Cl,swell. (A) Currents and (B) I–V relationships in 1T, 0.7T (10 min), after AG1478 in 0.7T (1 μM, 10 min), and after H₂O₂ (100 μM, 10 min) in presence of AG1478. (C) Time course of I_Cl,swell at +60 mV. (D) Currents and (E) I–V relationships in 1T, after EGF in 1T (3.3 nM, 10 min; 1T+EGF), and after osmotic shrinkage in 1.5T (10 min) in presence of EGF (1.5T+EGF).

The putative role of EGFR kinase suggests that exogenous EGF should elicit a volume-sensitive Cl⁻ current under isosmotic conditions. Consistent with this prediction, Figure 2D and E shows that an outwardly rectifying Cl⁻current was activated by 3.3 nM EGF in 1T (also see,¹¹). EGF (10–15 min) increased the current at +60 mV by 137 ±19% (P < 0.001, n = 6) in the absence of osmotic swelling. Moreover, osmotically shrinking myocytes in hyperosmotic 1.5T bath solution inhibited 91±7% (P < 0.001, n = 6) of the EGF-induced current in the continued presence of EGF. Thus, the EGF-induced Cl⁻ current was volume-sensitive, and osmotic shrinkage must act at site(s) downstream to EGF.

Previously we reported that exogenous AngII elicits a tamoxifen-sensitive Cl⁻ current attributed to I_Cl,swell.¹⁰ Figure 3 shows that AngII, like swelling, signals to Cl⁻ channels via EGFR kinase. Adding AngII (5 nM, 10 min) to 1T bathing media evoked an outwardly rectifying current that reversed near E_Cl, and AG1478 (1 μM, 10 min) in the continued presence of AngII inhibited 99 ±3% (P < 0.001, n = 6; +60 mV) of the AngII-induced current.

Figure 3.

EGFR kinase inhibition blocks AngII-induced I_Cl,swell. (A) Currents and (B) I–V relationships in 1T, after AngII (5 nM, 10 min), and after AG1478 (1 μM, 10 min) in presence of AngII (AG1478). (C) Time course of I_Cl,swell at +60 mV.

3.3. Phosphatidylinositol 3-kinase

PI-3K is downstream of EGFR kinase in the AngII-ROS signalling cascade and participates in NOX activation.^12,28 We tested whether PI-3K also regulates I_Cl,swell using selective inhibitors, LY294002 (IC₅₀ = 10 μM) and wortmannin (IC₅₀ = 1−10 nM). Figure 4A–C illustrates that LY294002 (100 μM, 10–15 min) fully blocked I_Cl,swell. Addition of LY294002 to 0.7T inhibited the swelling-activated current by 95 ±3% (P < 0.001, n = 8) at +60 mV. As expected if H₂O₂ is downstream of PI-3K, the suppression of I_Cl,swell was abrogated by exogenous H₂O₂. The H₂O₂-induced current in the presence of LY294002 was 147 ±13% (P < 0.001, n = 5) of the swelling-induced current.

Figure 4.

PI-3K blockade inhibits I_Cl,swell and is reversed by exogenous H₂O₂. (A) Currents and (B) I–V relationships in 1T, 0.7T (10 min), after LY294002 in 0.7T (100 μM, 10 min; LY), and after H₂O₂ (100 μM, 10 min) in presence of LY294002 (H₂O₂). (C ) Time course of I_Cl,swell at +60 mV. (D and E) Analogous experiments with wortmannin (500 nM, 13 min; WORT).

Wortmannin possesses a distinct chemical structure and mechanism of action from those of LY294002.²⁹ After activating I_Cl,swell in 0.7T, adding wortmannin (500 nM, 10–15 min) reduced the current by 60 ±2% (P < 0.001, n = 6) at +60 mV (Figure 4D and E).

3.4. NADPH oxidase

AT₁ activation ultimately generates ROS by initiating assembly of sarcolemmal NOX from cytoplasmic and membrane subunits,²⁸ and NOX and ROS participate in the activation of I_Cl,swell in response to integrin stretch.¹⁰ Therefore, we evaluated the role of NOX in the response to osmotic swelling. Apocynin (IC₅₀ = 80 μM) prevents NOX assembly by conjugating thiol residues.³⁰ As shown in Figure 5A–C, apocynin (500 μM, 10 min) inhibited 84 ±12% (P = 0.002, n = 5) of I_Cl,swell at +60 mV. Figure 5D and E demonstrates the effect of a structural distinct NOX inhibitor, DPI (IC₅₀ = 0.9 μM), which binds to flavins and the haem b redox centres of gp91^phox and, thereby, suppresses $•{\text{O}}_2^{-}$ production.³¹ Addition of DPI (60 μM, 10–13 min) to 0.7T completely blocked I_Cl,swell (116 ±16%; P = 0.002, n = 5).

Figure 5.

NOX inhibition blocks I_Cl,swell. (A) Currents and (B) I–V relationships in 1T, 0.7T (10 min), after apocynin in 0.7T (500 μM, 10 min; APO). (C) Time course I_Cl,swell at +60 mV. (D and E) Analogous experiments with DPI (60 μM, 13 min).

Although block by DPI or apocynin is widely taken as sufficient evidence to infer participation of NOX, questions can be raised regarding specificity. To address this issue, we utilized gp91ds-tat, a membrane permeant fusion peptide that is regarded as a highly selective blocker of NOX assembly.²⁴ Figure 6A and B shows that exposure to gp91ds-tat also fully suppressed I_Cl,swell. Addition of gp91ds-tat (500 nM, 10–12 min) to 0.7T reduced I_Cl,swell by 106 ±5% (P < 0.001, n = 5) at +60 mV. To rule out non-specific actions of the inhibitor peptide, the inactive peptide scramb-tat²⁴ was studied (Figure 6C). Exposure to scramb-tat (500 nM, 10–15 min) did not alter I_Cl,swell (P = ns, n = 4).

Figure 6.

Block of NOX with gp91ds-tat inhibits I_Cl,swell. (A) Currents and (B) I–V relationships in 1T, 0.7T (10 min), and after gp91ds-tat in 0.7T (500 nM, 10 min). (C) Analogous experiment with scramb-tat (500 nM, 10 min), a membrane permeant but inactive peptide.

3.5. Role of H₂O₂

NOX generates $•{\text{O}}_2^{-}$ by single e⁻ reduction of molecular O₂, and enzymatic and spontaneous dismutation forms H₂O₂, a membrane-permeant ROS, that is, a stronger oxidant.²⁸ H₂O₂ is a potent agonist of cardiac I_Cl,swell, participates in stretch-induced I_Cl,swell activation,¹⁰ and as shown in Figures 1, 2 and 4, overcomes block of I_Cl,swell by inhibitors of AT₁ receptors, EGFR kinase, and PI-3K. Therefore, we tested whether catalase, a H₂O₂ scavenger, would inhibit I_Cl,swell after osmotic swelling (Figure 7). Addition of catalase (1000 U/ml, 10–15 min) to 0.7T blocked 90 ±11% (P = 0.001, n = 5) of I_Cl,swell at +60 mV. This is consistent with the idea that H₂O₂ is a downstream mediator of I_Cl,swell activation.

Figure 7.

Scavenging of H₂O₂ by catalase inhibits I_Cl,swell. (A) Currents and (B) I–V relationships in 1T, 0.7 T (10 min), and after 1000 U/ml catalase to 0.7 T (10 min, Catalase). (C) Time course of I_Cl,swell at +60 mV.

To further characterize the site of action of H₂O₂, we tested whether H₂O₂-induced current was blocked by osmotic shrinkage. Exposure to 100 μM H₂O₂ in 1T increased outwardly rectifying current at +60 mV by 1.6 ±0.6 pA/pF (n = 5), but osmotic shrinkage in 1.5T in the continued presence of H₂O₂ failed to significantly alter the current (1.9 ±0.7 pA/pF, P = 0.34; data not shown). This suggests that H₂O₂ acts at sites distal to those regulated by osmotic shrinkage.

4. Discussion

We found that osmotic swelling activated I_Cl,swell by means of the AngII AT₁ receptor signalling cascade that involves EGFR kinase, PI-3K, production of $•{\text{O}}_2^{-}$ by NOX, and its dismutation to H₂O₂.^12,14 Previously, we showed that the same pathway stimulates I_Cl,swell when β_1D-integrins are stretched with paramagnetic beads.^10,11 A scheme illustrating the results is presented in Figure 8. As expected from this scheme, osmotic activation of I_Cl,swell was blocked by inhibitors of AT₁ receptors, EGFR kinase, PI-3K, and NOX and by scavenging H₂O₂ with extracellular catalase, _10,11 whereas exogenous AngII, EGF, and H₂O₂ elicited I_Cl,swell. H₂O₂ must act late in the cascade because it overcame the suppression of I_Cl,swell by AT₁ receptor, EGFR kinase, and PI-3K blockade. AT₁ receptor activation must be upstream from EGFR kinase because AG1478 blocked the AngII-induced current. The EGF-induced current was blocked by osmotic shrinkage, and we previously showed that it also is blocked by inhibitors of PI-3K and gp91ds-tat under different conditions.¹¹ These data imply that EGFR kinase is upstream from both PI-3K and NOX. While osmotic shrinkage blocked the EGF-induced current, it failed to affect H₂O₂-induced currents. Thus, osmotic shrinkage must regulate I_Cl,swell at one or more sites between EGFR kinase and H₂O₂. Taken together, these data imply that osmotic shrinkage does not simply oppose the effect of osmotic swelling at its initial transduction site. Rather, shrinkage and swelling must be sensed by different signalling molecules.

Figure 8.

Proposed scheme for regulation of I_Cl,swell by AngII-ROS signalling. Osmotic swelling releases AngII and/or directly activates AT₁ receptors. AT₁ activation evokes downstream signalling via EGFR kinase and PI-3K, causing assembly of NOX from membrane-bound (gp91^phox, p22^phox) and cytoplasmic (p47^phox, p67^phox, Rac) components. NOX produces $•{\text{O}}_2^{-}$, which rapidly undergoes dismutation by ecSOD to membrane permeant H₂O₂. H₂O₂ may modulate I_Cl,swell directly or via a variety of redox sensitive kinases and phosphatases.

The scheme presented here does not include other molecules that participate in AngII signalling in vasculature³² and heart.³³ For example, a PKC-induced activation of NOX by AngII that is independent of EGFR kinase is found in vascular tissue.³² Such a mechanism apparently does not significantly regulate I_Cl,swell in ventricular myocytes because EGFR kinase inhibitors fully blocked I_Cl,swell. Nevertheless, a more complex interplay between components of the proposed cascade cannot be excluded. Interactions with other PKC- and PKA-dependent mechanisms that regulate I_Cl,swell (for review,^1–3) are unexplored.

Osmotic swelling also does not appear to stimulate I_Cl,swell by activating integrin-dependent signalling mediated by FAK and Src. Block of the Src family by PP2 or FAK by over expression of the endogenous peptide inhibitor FRNK augments swelling-induced current in rabbit²² and neonatal rat ventricular²³ and human atrial²¹ myocytes, whereas PP2 suppresses stretch-induced I_Cl,swell.⁹ Notably, PP2 prevents Src- and integrin-dependent FAK autophosphorylation at Tyr397 and Tyr577, whereas G protein-dependent phosphorylation at Tyr397 can occur by a PP2-insensitive, Src-independent pathway.³⁴ Src family members are differentially modulated by osmotic swelling,³⁵ and unique Src family members may dominate the responses to swelling and stretch.

4.1. AT₁ receptors

Autocrine-paracrine regulation of ventricular myocytes by stretch-induced AngII release has been described.¹⁵ Although AngII secretion was not measured, it seems likely that AngII released from numerous myocytes in the chamber upon swelling initiated autocrine-paracrine AT₁ signalling, and I_Cl,swell blockade by losartan and eprosartan is consistent with this idea. In contrast, Zou et al.³⁶ recently made a strong case that cardiomyocyte AT₁ receptors are mechanosensors and respond to stretch by an AngII-independent mechanism that, nonetheless, is blocked by the AT₁ antagonist candesartan. In the present study, complete block of I_Cl,swell occurred with 30 μM but not 10 μM losartan. This implies an IC₅₀ higher than the 31 nM reported for [¹²⁵I]-AngII binding to AT₁ receptors in rabbit ventricular membranes at low ionic strength,³⁷ but lower than that for AT₂ receptors, 81 μM.³⁸ Losartan is a non-competitive/surmountable antagonist,²⁶ and its IC₅₀ for putative block of direct AT₁ activation by stretch is unknown. The involvement of AT₁ receptors is strongly supported by data implicating downstream components of its signalling cascade¹² in I_Cl,swell regulation. Nevertheless, we cannot rigorously exclude the possibility that losartan and eprosartan act at another unidentified site.

4.2. EGFR kinase

AT₁ activation triggers transactivation of EGFR kinase^12,27 by stimulating membrane-bound metalloproteases that clip and release heparin-binding EGF, as well as by association of AT₁ receptors with EGFR kinase.¹³ We found that the EGFR kinase inhibitor AG1478 blocked both I_Cl,swell and the AngII induced Cl⁻ current, and Cl⁻ current elicited by exogenous EGF was inhibited by osmotic shrinkage. These data implicate EGFR kinase in the signalling pathway and further suggest that osmotic shrinkage must act at a distal site in the signalling cascade. Block of volume-sensitive, EGF-induced Cl⁻ current by tamoxifen, an I_Cl,swell blocker, also was reported in studies under different experimental conditions.¹¹ Previously, we showed that block of EGFR kinase with PD153035 and AG556 inhibit I_Cl,swell in rabbit ventricular²² and human atrial²¹ myocytes, respectively. In contrast, tyrphostin A51 (AG183), another EGFR kinase blocker, failed to suppress I_Cl,swell in canine atrial myocytes.²⁵ The basis for this discrepancy is unknown. Consistent with the present results, activation of EGFR kinase by exogenous EGF or transfection with bovine papilloma virus recently were shown to potentiate I_Cl,swell in C127 mammary cells,³⁹ and EGF also stimulates I_Cl,swell in liver-derived HTC cells,⁴⁰ suggesting this is a common mode of I_Cl,swell regulation in several tissues.

4.3. Phosphatidylinositol 3-kinase

PI-3K is downstream of EGFR kinase^12,28 and is stimulated by osmotic swelling.^41,42 In turn, PI-3K produce phosphoinositol-3,4,5-triphosphate and phosphoinositol-3,4-bisphosphate,²⁹ which activate NOX both by binding to the PX domain of p47^{phox 28} and by augmenting Rac-mediated nucleotide exchange.^12,29 Block of I_Cl,swell by LY294002 and wortmannin argue that PI-3K participates in the regulation of cardiomyocyte I_Cl,swell. We did not investigate why block by 500 nM wortmannin was incomplete. One possibility is involvement of monomeric class II PI-3K-C2α, a cardiac isoform that is more resistant to wortmannin (IC₅₀ = 400 nM) than dimeric PI-3K (IC₅₀ = 1−10 nM).²⁹ Inhibition of PI-3K also suppresses I_Cl,swell in hepatocytes⁴³ and pulmonary artery smooth muscle⁴² and swelling-induced Na⁺-K⁺ pump stimulation in rabbit ventricular myocytes.⁴⁴

4.4. Reactive oxygen species

A major component of AT₁ signalling depends on generation of $•{\text{O}}_2^{-}$ by NOX.^14,33 The involvement of NOX and ROS in hypoosmotic activation of I_Cl,swell is strongly supported by findings that three distinct NOX inhibitors all blocked I_Cl,swell. DPI is promiscuous in binding to FAD-containing enzymes^45,46 and does not distinguish between mitochondrial, microsomal, and sarcolemmal sources of ROS. Apocynin does not alter $•{\text{O}}_2^{-}$ production by cardiac mitochondria,⁴⁷ but it inhibits P450 monooxygenase in endothelial cells.⁴⁸ In contrast, gp91ds-tat is thought to be highly NOX selective because of specific interactions between the 9-mer docking sequence and p47^phox or its homologues (e.g. NOXO1, assembles with NOX1).²⁴ NOX2 (gp91^phox) and NOX4 are expressed in cardiomyocytes.^17,49 The present experiments do not establish which isoform is responsible, but two factors favour involvement of NOX2. First, knockout of NOX2 fully abrogates AngII-induced, NOX-dependent, $•{\text{O}}_2^{-}$ production in cardiomyocytes, indicating that NOX2 rather than NOX4 is the AngII-responsive isoform.^17,50 Second, NOX4 activity in native cells and heterologous expression systems does not depend on cytoplasmic subunits including p47^phox and its homologues.⁴⁹ Thus, gp91ds-tat binding to p47^phox-like proteins should not have suppressed NOX4 activity.

NOX transfers e⁻ from intracellular NADPH to extracellular O₂ via FAD bound to its cytoplasmic C-terminal domain and two haem complexes in its membrane-spanning domains.²⁸ Consequently, NOX produces $•{\text{O}}_2^{-}$ at the extracellular membrane face, where it rapidly undergoes dismutation to H₂O₂ by extracellular SOD (SOD-3) anchored to proteoglycans.⁵¹ H₂O₂ is a longer-lived species and stronger oxidant than $•{\text{O}}_2^{-}$ and is membrane permeant.²⁸ Consistent with a role for H₂O₂, scavenging H₂O₂ by adding catalase to the bath suppressed I_Cl,swell in osmotically swollen myocytes. Previously, we showed that tamoxifen-sensitive I_Cl,swell is activated by exogenous H₂O₂ (EC₅₀ = 8 μM) in rabbit ventricular myocytes.¹⁰ Recent reports indicate that exogenous H₂O₂ also elicits I_Cl,swell in HeLa (200–500 μM) and HTC (EC₅₀ = 100 μM) cell lines.^40,52 Both of these groups detected ROS production attributed to NOX upon swelling. Expression of the dominant negative p47^S379A NOX subunit suppresses activation of I_Cl,swell in HeLa cells,⁴⁰ as does the non-selective flavin-inhibitor DPI in both HeLa and HTC cells.^40,52 It remains to be seen whether the ultimate target of H₂O₂ is the Cl⁻ channel itself or redox-sensitive signalling pathways.^33,53 Cysteines with an acidic pK_a are particularly sensitive to oxidation by H₂O₂. Such cysteines are found in the active site of all protein tyrosine phosphatases, including PTEN, which opposes PI-3K by dephosphorylating inositols.⁵³ Nevertheless, we cannot rule out the possibility that other reactive species contribute to the regulation of I_Cl,swell.

4.5. Implications

I_Cl,swell is persistently activated in models of dilated cardiomyopathy,² but the underlying mechanism is obscure. The present results suggest that upregulation of the AngII cascade, which occurs in ventricular hypertrophy, heart failure, hypertension, atherosclerotic coronary artery disease, hypercholesterolemia, and diabetes, may in part explain persistent activation of I_Cl,swell in dilated cardiomyopathy and predicts similar findings in other settings with elevated AngII or NOX activity. For example, NOX is upregulated by subpressor doses of AngII and aortic banding,^16,17 after myocardial infarction,⁵⁴ during the progression from hypertrophy to failure,⁵⁵ and in atrial fibrillation.⁵⁶ Because I_Cl,swell outwardly rectifies, its upregulation in cardiac disease promotes reduction of action potential duration, which favours reentrant tachyarrhythmias by decreasing the minimum wavelength for a reentrant circuit.