|
|
|
|
|
|
|
||
ARTICLE |
Correspondence to Clive M. Baumgarten: clive.baumgarten{at}vcu.edu
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
 |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). We have shown that specific stretch of ß1 integrin in ventricular myocytes under isosmotic conditions activates an outwardly rectifying Cl– current (Cl– SAC) (Browe and Baumgarten, 2003, 2004) that resembles the volume-sensitive Cl– current, ICl,swell (Sorota, 1992; Tseng, 1992). Cl– SAC, like ICl,swell, contributes to the background current, activates slowly over several minutes, exhibits outward rectification, partially inactivates at positive potentials, and is blocked by tamoxifen. Regulation of Cl– SAC involves Src and/or focal adhesion kinase (FAK), angiotensin AT1 receptors, and reactive oxygen species (ROS) generated, at least in part, by NADPH oxidase. Furthermore, angiotensin II (AngII) and H2O2 each elicit Cl– SAC in ventricular myocytes in the absence of stretch (Browe and Baumgarten, 2004).
Stretch (Oeckler et al., 2003), integrins (Moro et al., 2002), and AT1 receptors (Shah and Catt, 2003) transactivate epidermal growth factor receptor (EGFR) kinase, a member of the ErbB family of receptor tyrosine kinases (RTKs). Unstimulated EGFR kinase is a monomer consisting of an extracellular ligand binding domain, a single membrane spanning domain, and a cytoplasmic kinase domain (Holbro and Hynes, 2004). With ligand binding, EGFR kinase dimerizes, trans-autophosphorylates six specific tyrosines within its noncatalytic cytoplasmic tail, and translocates from caveolae to costameres (Shah, 2002). EGFR kinase then recruits and phosphorylates various adaptor and signaling molecules that bind phosphotyrosine motifs with their SH2 and PTB domains and serves as a platform for integrating downstream signaling.
Signaling by EGFR kinase and AT1 receptors is mediated by phosphatidylinositol 3-kinase (PI-3K), a family of lipid kinases also activated by integrins, stretch, and H2O2 (Ushio-Fukai et al., 1999; Oudit et al., 2004). The best characterized PI-3K are heterodimers composed of a p110 catalytic subunit tightly coupled to a regulatory subunit. Cardiac myocytes principally express the p110
and 
isoforms that couple with the p85 and p101 regulatory subunits, respectively (Prasad et al., 2003; Oudit et al., 2004). The p110-p85 and p110-p101 dimers are activated by interaction with phosphotyrosines or the ß subunit of heterotrimeric G proteins, respectively, and catalyze inositol ring phosphorylation at the D-3 position to produce phosphatidylinositol 3,4,5-triphosphate (PtdIns(3,4,5)P3) and phosphatidylinositol 3,4-bisphosphate (PtdIns(3,4)P2), potent signaling molecules.
EGFR kinase and PI-3K also participate in the activation of NADPH oxidase (Seshiah et al., 2002; Vignais, 2002). Cardiac myocytes express a transmembrane flavocytochrome b558 complex consisting of gp91phox (Nox2) and p22phox, as well as the cytosolic components p47phox, p67phox, and Rac (Li et al., 2002; Xiao et al., 2002; Heymes et al., 2003). Nox4, a homologue of gp91phox, also is expressed (Byrne et al., 2003). NADPH oxidase activation involves translocation of the cytosolic components to the membrane and their assembly with flavocytochrome b558 (Vignais, 2002). Once assembled, the NADPH oxidase complex uses intracellular NADPH as substrate to catalyze the single electron reduction of molecular O2 to superoxide anion, •O2–, which is subsequently converted to H2O2 by superoxide dismutase (SOD).
Assembly of the active NADPH oxidase requires the binding of gp91phox to p47phox (Vignais, 2002). Recently, a cell permeable, chimeric peptide that inhibits this interaction was developed (Rey et al., 2001). The inhibitor, gp91ds-tat, combines the 9-mer gp91phox docking sequence for p47phox with a 9-mer from the HIV tat sequence that mediates membrane permeation. gp91 ds-tat can be used to selectively evaluate the role of NADPH oxidase in regulating Cl– SAC without affecting ROS production at other sites, such as the flavocytochromes within the mitochondria.
The aim of this study was to examine the role of EGFR kinase, PI-3K, and NADPH oxidase in the activation of Cl– SAC in ventricular myocytes and to determine whether Cl– SAC is volume sensitive. Paramagnetic beads coated with anti–ß1 integrin mAb were used to stretch integrins. Block of EGFR kinase or PI-3K during stretch inhibited Cl– SAC. Moreover, exogenous epidermal growth factor (EGF) fully activated Cl– SAC, and the EGF-induced current was inhibited by PI-3K and NADPH oxidase blockers and by tamoxifen, a blocker of ICl,swell. Finally, Cl– SAC elicited by either integrin stretch or exogenous EGF was volume sensitive and was completely inhibited by osmotic shrinkage. Thus, the Cl– SAC and EGF-induced Cl– current are likely to be carried by the same channel as ICl,swell. A preliminary report appeared previously (Browe and Baumgarten, 2005b).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
3 kg) of either gender. Hearts were excised, attached to a Langendorff perfusion apparatus, and subjected to an enzymatic dissociation procedure described previously (Browe and Baumgarten, 2003). Isolated myocytes were washed and stored in a modified KB medium. All experimental recordings were conducted within 10 h of myocyte isolation. Single myocytes chosen for study were rod-shaped, quiescent, displayed clear striations, and were free of membrane blebs or other morphological irregularities. Tyrode solution for cell isolation contained (in mM) 130 NaCl, 5 KCl, 1.8 CaCl2, 0.4 KH2PO4, 3 MgCl2, 5 HEPES, 15 taurine, 5 creatine, 10 glucose, pH 7.25 (adjusted with NaOH). CaCl2 was replaced with 0.1 mM Na2-EGTA to make Ca2+-free Tyrode solution. Enzyme solution contained 1.5–1.75 mg/ml BSA (Sigma-Aldrich; A7906), 0.5 mg/ml collagenase (type II, Worthington), and 0.05 mg/ml pronase (Sigma-Aldrich; type XIV) in nominally Ca2+-Tyrode solution without EGTA. All isolation solutions were well oxygenated and maintained at 37°C. Modified KB medium contained (in mM) 120 K-glutamate, 10 KCl, 10 KH2PO4, 0.5 K2-EGTA, 10 taurine, 1.8 MgSO4, 10 HEPES, 20 glucose, 10 mannitol, pH 7.2 (adjusted with KOH).
Experimental Solutions and Drugs
Single ventricular myocytes were scattered on a poly-L-lysine–coated, glass-bottomed chamber and placed on the stage of an inverted microscope (Diaphot; Nikon). Hoffman modulation optics (x40; NA = 0.55) and a high resolution TV camera (CCD72; Dage-MTI) were used to visualize the cells. Bath solution designed to isolate anion currents was suprafused at 2–3 ml/min and contained (in mM) 145 N-methyl-D-glucamine (NMDG)-Cl, 4.3 MgCl2, 10 HEPES, 5 glucose, pH 7.4 (adjusted with NMDG). Pipette solution contained (in mM) 110 Cs-aspartate, 20 CsCl, 2.5 Mg-ATP, 8 Cs2-EGTA, 0.1 CaCl2, 10 HEPES, pH 7.1 (adjusted with CsOH; liquid junction potential, –13.2 mV). Pipette free-Ca2+ was estimated to be 35 nM (WinMAXC 2.4; www.stanford.edu/cpatton/maxc.html). All recordings were made at room temperature (22°C–23°C).
The tyrphostin AG1478 (1 mM; Calbiochem), tamoxifen (20 mM; Sigma-Aldrich), wortmannin (1 mM, Calbiochem), and LY294002 (50 mM, Calbiochem) were prepared as stock solutions in DMSO at the indicated concentrations and kept frozen (–20°C) in small aliquots until use. EGF (mouse; Calbiochem) was dissolved directly in bath solution and kept frozen (–20°C) in small aliquots until use at a final concentration of 3.3 nM, which is equivalent to 20 ng/ml. A membrane-permeant inhibitor of NADPH oxidase, gp91ds-tat, an inactive membrane-permeant analogue, scramb-tat, and an active but impermeant gp91ds peptide were synthesized by the Tufts University Core Facility, as previously described by others (Rey et al., 2001). The inhibitor, gp91ds-tat, is a fusion peptide made from a 9-mer, CSTRIRRQL, that inhibits assembly of NADPH oxidase by mimicking the gp91phox docking site for the cytoplasmic p47phox subunit and a 9-mer from the tat HIV coat protein that confers membrane permeability. The gp91phox docking site sequence is identical in a number of experimental species including rabbit (Gauss et al., 2002), but experimental evidence and homology suggest that gp91ds-tat may not distinguish between gp91phox homologues (Rey et al., 2001). The inactive fusion peptide, scramb-tat, was made by scrambling the 9-mer docking peptide to minimize matches in the GenBank/EMBL/DDBJ database, whereas gp91ds is the docking site 9-mer without tat and is active in broken cell systems. Each peptide was prepared as stock solutions (1.2 mg/ml) in 150 mM NaCl acidified with 10 mM acetic acid and kept frozen (–20°C) in small aliquots until use. For certain experiments, bath solution osmolarity was increased from 300 mosmol/kg (1T) to 450 mosmol/kg (1.5T) by addition of mannitol, preserving a constant ionic strength.
Integrin Stretch
Force was applied directly and specifically to ß1 integrins using mAb-coated paramagnetic beads and an electromagnet, as previously described (Browe and Baumgarten, 2003). mAb for the ß1 subunit of integrin (MAB2250, IgG1; Chemicon) was attached by anti–pan IgG mAb to the surface of uniform 4.5 ± 0.2 µm diameter (mean ± SD) superparamagnetic beads containing iron oxides (dynabeads M-450 pan mouse IgG; Dynal Biotech). Anti-ß1 integrin mAb–coated beads were added to myocytes in the bath and were permitted to randomly settle on myocytes from above while the flow of bath solution was turned off. After 5 min, unbound beads were washed away by restoring flow. Myocytes chosen for study typically had three to five coated beads on their surface, and presumably, each bead was bound to multiple ß1 integrins.
A water-cooled electromagnet was placed directly on top of the bath, and patch pipettes were passed through an opening at the base of the coil. Current through the coil was set to generate a magnetic flux density of 35 Gauss (G) and a magnetic flux density gradient of 2,400 G/m that was uniform in the x–y plane occupied by the myocytes on the chamber floor. The resulting force vector imposed on each bead was directed upwards toward the coil, perpendicular to the long axis of the myocyte, and was estimated to have a magnitude of 1.2 pN/bead (Browe and Baumgarten, 2003).
Electrophysiological Recordings
Pipettes were pulled from 7740 thin-walled borosilicate glass capillary tubing (Sutter) and fire polished to give a final tip diameter of 3–4 µm and resistance in bath solution of 2–3 M
. Membrane currents were recorded with an EPC-7 amplifier (List-Medical) using the whole cell configuration. A 150 mM KCl agar bridge served as the ground electrode. Seal resistances of 5–30 G were typically obtained. Membrane potential was corrected for the measured liquid junction potential before forming a seal. After seal formation, the membrane patch was ruptured by application of negative pressure or by a brief zapping pulse. Myocytes were dialyzed for at least 10 min before recordings commenced. Voltage clamp protocols and data acquisition were governed by a Digidata 1200B or 1322A and pClamp 8.0 (Axon Instruments). Successive 500-ms voltage steps were taken from a holding potential of –60 mV to test potentials ranging from –100 to +40 mV in +10 mV increments. Membrane currents were low-pass filtered at 2 kHz (8-pole Bessel 902, Frequency Devices) and digitized at 10 kHz. For presentation, selected records were low-pass filtered at 50 Hz post-digitization with a Gaussian filter in PClamp. Cl– currents exhibited strong voltage-dependent inactivation, and isochronal IV curves were plotted based on the average current recorded 20–35 ms after the onset of the voltage step.
Statistics
Data are reported as mean ± SEM; n denotes the number of cells. Mean currents are expressed as current density (pA/pF) to account for differences in myocyte surface membrane area, and paired comparisons usually are expressed as a percentage. Statistical analyses were performed by SigmaStat 3.1 (Systat). For multiple comparisons, a one-way repeated measures ANOVA was performed and was followed by a Student-Newman-Keuls test. For comparisons of two groups, a one-tailed paired Students t test was conducted. In both cases, P < 0.05 was taken as significant.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
; Oeckler et al., 2003; Shah and Catt, 2003). Because EGFR kinase serves as a nexus for integrating diverse stimuli that regulate Cl– SAC, we assessed its role in the response to ß1 integrin stretch. Fig. 1 illustrates families of anion currents at test potentials between –100 and +40 mV and the corresponding I–V relationships. Under control conditions with anti-ß1 integrin mAb–coated paramagnetic beads attached to the myocyte surface but before the application of integrin stretch, a small outwardly rectifying background current that partially inactivated at positive potentials was present (Fig. 1 A). The background current reversed at –49 mV, which is near the calculated value of –52 mV for ECl (Fig. 1 D). Partial activation of ICl,swell is thought to occur under these conditions, and the background Cl– current is sensitive to the ICl,swell blocker tamoxifen (Hume et al., 2000). Consequently, ICl,swell is likely to comprise much of the background Cl– current observed in 1T. Stretch of ß1 integrins progressively increased the magnitude of the outward Cl– current over 5 min, giving a more than twofold augmentation of outward current at +40 mV in this example (Fig. 1, B and D). After full activation by stretch, the Cl– current partially inactivated at positive potentials, and its I–V relationship showed strong outward rectification and reversed near ECl, as previously shown (Browe and Baumgarten, 2003, 2004).
|
), almost completely abrogated the stretch-induced Cl– current in the continued presence of integrin stretch (Fig. 1, C and E). The I–V relationship was nearly restored to its control level after 10 min, and the reversal potential was unaffected. Overall, integrin stretch increased the Cl– current at +40 mV from 1.33 ± 0.07 to 2.30 ± 0.22 pA/pF, and exposure to AG1478 (10–12 min) with continued integrin stretch reduced the current to 1.46 ± 0.17 pA/pF (n = 4). Thus, block of EGFR kinase inhibited 93 ± 13% (n = 4, P = 0.001) of stretch-induced current at +40 mV, and the current after block was not significantly different from the control current (n = 4, P = 0.385). Neither stretch of ß1 integrin nor subsequent AG1478 application significantly altered inward current at –100 mV, however, as also was shown for stretch (Browe and Baumgarten, 2003, 2004). As a control, we examined the effect of AG1478 (10 µM, 10 min) on the background Cl– current in the absence of integrin stretch. In contrast to its pronounced inhibitory effect on the stretch-induced outward current, AG1478 did not significantly alter the background Cl– current (Fig. 1 E) at either +40 mV (n = 4; P = 0.526) or –100 mV (n = 4; P = 0.306).
AG1478 also prevented activation of Cl– SAC when it was applied intracellularly before integrin stretch. Myocytes were pretreated by including 1 µM AG1478, a 10-fold lower concentration, in the pipette solution. After 15–20 min of dialysis, myocytes were stretched for 10 min, a time sufficient to fully activate Cl– SAC (Browe and Baumgarten, 2003). After pretreatment, stretch elicited a small but statistically insignificant current, 0.23 ± 0.14 pA/pF, at +40 mV (n = 5, P = 0.10). This current was only 23% of that observed without pretreatment, 0.97 ± 0.09 pA/pF (n = 18).
Exogenous EGF Activates Cl– SAC
If transactivation of EGFR kinase is an upstream signaling event in the activation of Cl– SAC, exogenous EGF, which binds EGFR kinase and stimulates FAK and Src activity and ROS production (Brunton et al., 1997; Park et al., 2004), might be expected to mimic the effect of integrin stretch. Fig. 2 illustrates the effect of exogenous EGF on Cl– currents in the absence of integrin stretch. Under control conditions, a typical background current was observed (Fig. 2, A and E). Addition of 3.3 nM EGF to bathing solution for 6 min evoked an outwardly rectifying Cl– current that exhibited partial inactivation at positive potentials (Fig. 2, B and E). The EGF-induced difference current (Fig. 2, D and F) was obtained by digital subtraction. Overall, exogenous EGF (5–6 min) increased the outward Cl– current at +40 mV by 0.81 ± 0.05 pA/pF (n = 24, P < 0.001), from 1.20 ± 0.07 to 2.01 ±0.08 pA/pF. As found with stretch, EGF had a very modest effect on the inward Cl– current. At –100 mV, EGF increased inward current by 0.04 ± 0.01 pA/pF (n = 24; P < 0.001), from –0.24 ± 0.02 to –0.28 ± 0.03 pA/pF. Significant effects of EGF on inward current were not consistently detected in smaller groups, however.
|
; Browe and Baumgarten, 2003). After full activation of Cl– current by 3.3 nM EGF, exposure to 10 µM tamoxifen for 6 min in the continued presence of EGF completely abolished the EGF-induced Cl– current as well as a large fraction of the background Cl– current (Fig. 2, C and E). In these experiments, tamoxifen (6–8 min) blocked 170 ± 8% (n = 4; P < 0.001) and 177 ± 55% (n = 4; P < 0.02) of the EGF-induced current at +40 and –100 mV, respectively. EGF increased the Cl– current at +40 mV from 1.15 ± 0.09 to 1.85 ± 0.06 pA/pF (n = 4; P < 0.001), and tamoxifen reduced the current to 0.67 ± 0.13 pA/pF, a value significantly less than control (n = 4; P < 0.01). Fig. 3 compares the time course of Cl– currents at +40 mV after activation by either EGF (3.3 nM) or integrin stretch, and as a control for time, the background current in the absence of an intervention. The EGF-induced Cl– current was well described by a single exponential with a time constant of 3.2 ± 0.9 min (n = 4), equivalent to a t1/2 of 2.2 ± 0.6 min, and was 0.91 ± 0.07 pA/pF at 20 min. Activation by integrin stretch was slower and sigmoidal with a t1/2 of 4.3 ± 0.6 min (n = 4), and the steady-state current was 1.19 ± 0.07 pA/pF. Once activated, both currents remained at stable levels for the remainder of the recording period. In the absence of an intervention, the background current was stable for 20 min. The more rapid activation of Cl– current by EGF as compared with stretch is consistent with, but does not prove, the idea that EGF is a downstream event in the stretch-induced signaling cascade regulating Cl– SAC. Moreover, these data demonstrate the stability of the current recordings and serve as a time control for studies with blockers.
|
; Franchini et al., 2000). Upon integrin clustering, activation of FAK and Src leads to the recruitment of PI-3K to costameres (Franchini et al., 2000), and PI-3K regulates NADPH oxidase activity (Vignais, 2002). Fig. 4 illustrates a test of the idea that PI-3K also participates in the activation of Cl– SAC in response to ß1 integrin stretch using wortmannin, an irreversible blocker of PI-3K (Stein and Waterfield, 2000). After activating Cl– SAC by ß1 integrin stretch for 6 min (Fig. 4, B and D), exposure to 500 nM wortmannin for 12 min with continued integrin stretch partially inhibited the outward Cl– SAC (Fig. 4, C and D). Wortmannin (12–15 min) blocked 67 ± 6% (n = 7; P < 0.001) of the Cl– SAC at +40 mV. Stretch increased the Cl– current from 1.45 ± 0.14 to 2.47 ± 0.17 pA/pF, and wortmannin reduced the current to 1.81 ± 0.12 pA/pF (n = 7; P < 0.001). Although block of Cl– SAC by wortmannin at +40 mV was substantial, the current remained significantly greater than control (n = 7; P = 0.006). Surprisingly, wortmannin (500 nM, 12 min) augmented the outward background current in the absence of stretch (Fig. 4 E). At +40 mV, the Cl– current was increased by 15 ± 5% (n = 4; P = 0.02), from 1.16 ± 0.14 to 1.33 ± 0.16 pA/pF. The basis for the stimulation of background Cl– current was not investigated further.
|
; Oudit et al., 2004). Fig. 5 confirms the sensitivity of Cl– SAC to block of PI-3K. After activating Cl– SAC by ß1 integrin stretch for 5 min, exposure to LY294002 (20 µM, 15 min) with continued stretch blocked almost all of the Cl– SAC (Fig. 5, A and B). Stretch increased the current at +40 mV from 1.14 ± 0.28 to 1.86 ± 0.36 pA/pF (n = 4; P < 0.001), and LY294002 (20 µM, 15 min) reduced the current by 99 ± 11% (n = 4; P < 0.001). The current remaining in the presence of LY294002, 1.20 ± 0.35 pA/pF, was not significantly different from the control current (n = 4; P = 0.54). In contrast to its pronounced inhibition of the stretch-induced current, LY294002 (20 µM, 15 min) had a negligible effect on the background current in the absence of stretch (Fig. 5 C). At +40 mV, LY294002 reduced background current by only 3 ± 1% (n = 4; P = 0.07), from 0.62 ± 0.07 to 0.60 ± 0.08 pA/pF.
|
). Therefore, we tested whether the Cl– current elicited by exogenous EGF (3.3 nM, 6 min) also depended on PI-3K. LY294002 (20 µM, 12 min) inhibited almost all of the EGF-induced current in the continued presence of EGF. Overall, EGF increased the Cl– current at +40 mV from 0.97 ± 0.10 to 1.66 ± 0.09 pA/pF, and LY294002 (20 µM, 12–14 min) blocked 124 ± 16% (n = 4; P < 0.001) of the current elicited by EGF. The current remaining after LY294002, 0.82 ± 0.14 pA/pF, was not significantly different from the control (n = 4; P = 0.15). The inward currents at –100 mV were small and were not significantly affected in this set of experiments. Thus, block of PI-3K by LY294002 had the same inhibitory effect on the EGF- and integrin stretch–induced Cl– currents but no effect on the outwardly rectifying background current.
Role of NADPH Oxidase in EGF-induced Cl– Current
Previously we provided evidence that activation of Cl– SAC by integrin stretch involves NADPH oxidase and the production of ROS (Browe and Baumgarten, 2004). EGFR kinase and PI-3K are known to participate in the activation of NADPH oxidase (Seshiah et al., 2002). As illustrated in Fig. 6, we tested whether the Cl– current evoked by exogenous EGF also requires NADPH oxidase activity. First, Cl– current was elicited with EGF (3.3 nM, 5 min; Fig. 6, B and D), and then gp91ds-tat, a selective membrane-permeant fusion peptide blocker of NADPH oxidase (Rey et al., 2001), was added. In the continued presence of EGF, gp91ds-tat (500 nM, 15 min) completely inhibited the EGF-induced Cl– current (Fig. 6, C and D). EGF (3.3 nM, 5 min) increased outward Cl– current at +40 mV from 1.16 ± 0.15 to 2.16 ± 0.16 pA/pF (n = 4; P < 0.001), and gp91ds-tat (500 nM, 12 – 15 min) inhibited 133 ± 20% (n = 4; P <0.001) of the EGF-induced Cl– current. The current after gp91ds-tat treatment, 0.94 ± 0.07 pA/pF, was not significantly different from the control current (n = 4; P = 0.19). EGF also significantly increased the inward Cl– current at –100 mV from –0.16 ± 0.02 to –0.23 ± 0.01 pA/pF (n = 4; P < 0.01), and gp91ds-tat reduced the current to –0.18 ± 0.01 pA/pF, blocking 87 ± 27% (n = 4; P = 0.01).
|
). As shown in Fig. 7, Cl– current was elicited with EGF (3.3 nM, 5 min). Then, either scramb-tat (500 nM, 15 min; Fig. 7 A) or gp91ds (500 nM, 15 min; Fig. 7 B) was applied in the continued presence of EGF. Neither negative control peptide suppressed the EGF-induced current. If anything, the currents elicited by EGF continued to increase slightly over most of the voltage range positive to –20 mV. In the experiments with scramb-tat, EGF increased Cl– current at +40 mV from 1.09 ± 0.22 to 2.03 ± 0.18 pA/pF (n = 4; P < 0.001), and it was 2.04 ± 0.21 pA/pF after a 12–15 min exposure to scramb-tat (n = 4; P = 0.94). Thus, the tat sequence was not sufficient to reproduce the effect of gp91ds-tat. Similarly, in experiments with impermeant gp91ds, EGF increased the Cl– current at +40 mV from 0.98 ± 0.07 to 1.65 ± 0.04 pA/pF (n = 3; P = 0.001), and it was 1.61 ± 0.02 pA/pF after a 12–15 min exposure to impermeant gp91ds (n = 3; P = 0.57). The ineffectiveness of active but impermeant gp91ds indicates that the site blocked by gp91ds-tat must be in the cytoplasm or on the internal face of the sarcolemma and that gp91ds cannot block the EGF-activated Cl– channel from the outside.
|
, 2004). Moreover, both Cl– SAC and ICl,swell are regulated by several of the same signaling molecules (Browe and Baumgarten, 2003, 2004; Ren and Baumgarten, 2005). If this idea is correct, current elicited by ß1 integrin stretch should be volume sensitive. A test of this prediction is illustrated in Fig. 8. After recording control currents in isosmotic 1T bath solution (Fig. 8, A and E), Cl– SAC was activated by a 5-min integrin stretch in 1T bath solution (Fig. 8, B and E). Suprafusion of 1.5T hyperosmotic bath solution for 10 min while integrin stretch was continued completely inhibited the stretch-induced Cl– current (Fig. 8, C and E). Moreover, block of the Cl– SAC by osmotic shrinkage was reversed by returning to 1T bath solution for 5 min while maintaining stretch (Fig. 8, D and E). Inhibition of the Cl– SAC by osmotic shrinkage was observed in all myocytes studied. Stretch in 1T bath solution increased Cl– current at +40 mV from 1.10 ± 0.30 to 2.32 ± 0.57 pA/pF (n = 4; P = 0.001), and osmotic shrinkage in 1.5T bath solution (10–12 min) blocked 94 ± 6% (n = 4; P < 0.001) of stretch-activated current, reducing the current at +40 mV to 1.13 ± 0.24 pA/pF, a value not different from control (n = 4; P = 0.91). After reintroduction of 1T bathing solution for 5 min, the current returned to 2.33 ± 0.51 pA/pF, a value not significantly different from that after the initial period of stretch in 1T bathing solution (n = 4; P = 0.95).
|
). To estimate the contribution of background ICl,swell to the current blocked by 1.5T during integrin stretch, unstretched myocytes in 1T were exposed to 1.5T for 10 min. At +40 mV, osmotic shrinkage in1.5T reduced the background current by 0.53 ± 0.09 pA/pF (n = 4; P = 0.005), from 1.69 ± 0.37 to 1.15 ± 0.29 pA/pF (Fig. 8 F). The current remaining after shrinkage was the same in the absence, 1.15 ± 0.29 pA/pF, and presence, 1.13 ± 0.24 pA/pF, of integrin stretch. On the other hand, the initial background current in the set of cells that were stretched, 1.10 ± 0.30 pA/pF, was smaller than that in the set that was not stretched, 1.69 ± 0.37 pA/pF. If the EGF-induced Cl– current is the same as the Cl– SAC and ICl,swell, it also should be volume sensitive, as is demonstrated in Fig. 9. The Cl– current in 1T bathing solution (Fig. 9, A and E) was increased by addition of EGF (3.3 nM, 5 min; Fig. 9, B and E). The EGF-induced current was blocked after 10 min of myocyte shrinkage by hyperosmotic 1.5T solution containing EGF (Fig. 9, C and E), and block was completely reversed after returning to 1T bath solution containing EGF for 5 min (Fig. 9, D and E). Fig. 9 F compares the current blocked by 1.5T bathing solution before and after exposure to EGF. At +40 mV, EGF in 1T bathing solution increased the Cl– current from 1.65 ± 0.19 to 2.32 ± 0.32 pA/pF (n = 4; P = 0.003), and exposure to 1.5T bathing solution with EGF blocked 127 ± 11% (n = 4; P = 0.001) of the EGF-induced current. The current after myocyte shrinkage, 1.46 ± 0.14 pA/pF, was not significantly different than the control current (n = 4; P = 0.22). On returning to 1T bathing solution with EGF, 94 ± 6% (n = 4; P = 0.001) of the initial EGF-induced current recovered; the current recorded under these conditions, 2.25 ± 0.28 pA/pF, was not significantly different from the initial current with EGF (n = 4; P = 0.64).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2004). The present results indicate that EGFR kinase and PI-3K, which couple AT1 receptors to NADPH oxidase (Seshiah et al., 2002), are also components of the signaling cascade that evokes Cl– SAC in response to integrin stretch. Moreover, stretch- and EGF-induced currents were volume sensitive, suggesting that they are carried by the same channels as ICl,swell.
A simple scheme accounting for both our present and previous findings is depicted in Fig. 10. Activation of Cl– SAC was abrogated by blockers of EGFR kinase, and exogenous EGF elicited a tamoxifen-sensitive current that mimicked the Cl– SAC but turned on more rapidly. These data suggest that stretch stimulates Cl– SAC via EGFR kinase. PI-3K must be downstream from EGFR kinase because inhibition of PI-3K suppressed both the stretch- and EGF-induced current. PI-3K has been implicated in the translocation of Rac to the membrane and the activation of NADPH oxidase (Seshiah et al., 2002; Vignais, 2002). Consistent with this idea, the EGF-induced Cl– current was blocked by gp91ds-tat, a selective inhibitor of NADPH oxidase, and we previously demonstrated that stretch-induced Cl– currents are blocked by other NADPH oxidase inhibitors, diphenyleneiodonium (DPI) and AEBSF (Browe and Baumgarten, 2004). Two other signaling molecules in this cascade, AngII and H2O2, also elicit currents that mimic Cl– SAC in the absence of stretch, whereas the AT1 blocker losartan and degradation of H2O2 by catalase inhibit Cl– SAC (Browe and Baumgarten, 2004). The idea that AT1 receptor activation leads to stimulation of NADPH oxidase by a cascade that includes EGFR kinase and PI-3K is well established in other systems (Seshiah et al., 2002). Although the simple linear scheme in Fig. 10 is sufficient to explain our observations on Cl– SAC, we cannot exclude additional interactions between these signaling molecules.
|
, 2004). The present findings support this idea. Both the stretch- and EGF-induced currents were suppressed by osmotic shrinkage, and the EGF-induced current was blocked by tamoxifen, as previously shown for Cl– SAC (Browe and Baumgarten, 2003).
Cl– SAC and ICl,swell also are regulated by some of the same signaling pathways. Block of EGFR kinase inhibits the activation of ICl,swell by osmotic swelling in human atrial and rabbit ventricular myocytes (Du et al., 2004; Ren and Baumgarten, 2005), and recently we found that AT1 receptors, PI-3K, NADPH oxidase, and ROS play the same roles in eliciting ICl,swell upon osmotic swelling in rabbit ventricle (Ren et al., 2005) as in response to integrin stretch. In contrast, block of Src family protein tyrosine kinases with PP2 gives a result that depends on the stimulus. PP2 suppresses Cl– SAC (Browe and Baumgarten, 2003), whereas it stimulates ICl,swell after it has been activated by swelling (Du et al., 2004; Ren and Baumgarten, 2005; Walsh and Zhang, 2005). Walsh and Zhang (2005) also observed stimulation of ICl,swell in neonatal rat myocytes when FAK was inhibited by overexpression of FRNK, an endogenous FAK inhibitor. On the other hand, Sorota (1995) found that nonselective tyrosine kinase inhibitors, which also should have inhibited Src and FAK, prevented activation of ICl,swell in canine atrial myocytes; the reason for this discrepancy is unknown. It is also unclear why inhibition of Src and FAK stimulates ICl,swell in several preparations but inhibited Cl– SAC. Osmotic swelling differentially modulates members of the Src family (Cohen, 2005), whereas PP2 broadly inhibits this protein tyrosine kinase family. Osmotic swelling also may give rise to signaling not elicited by stretch. Other signaling pathways are implicated in the regulation of ICl,swell (Hume et al., 2000; Baumgarten and Clemo, 2003), but their role in regulation of Cl– SAC has not been tested.
Although we termed the stretch- and EGF-induced stimulation of Cl– current an activation, the present studies do not distinguish between de novo opening of channels that are silent in the absence of these stimuli and modulation of ICl,swell channels that are a major contributor to the background outwardly rectifying Cl– current under isosmotic conditions (Hume et al., 2000). Block of EGFR kinase by AG1478 and PI-3K by LY 294002 restored stretch- and EGF-induced current to their control level but did not alter the background current in the absence of these stimuli. In contrast, the ICl,swell/Cl– SAC blocker tamoxifen and osmotic shrinkage inhibited both the stretch- and EGF-induced and the background current. These observations may suggest that background ICl,swell is regulated by signaling processes not studied here, but this requires further verification.
Regulation of Cl– SAC by EGFR Kinase Transactivation and EGF
The ErbB family of receptor tyrosine kinases is represented in mammals by EGFR kinase (ErbB1) and ErbB2–ErbB4, which form homo- and heterodimers (Bogdan and Klämbt, 2001; Holbro and Hynes, 2004). Two lines of evidence implicate EGFR kinase in Cl– SAC activation. First, 10 µM AG1478 completely blocked Cl– SAC during stretch, and intracellular pretreatment with a 10-fold lower concentration via the patch pipette suppressed Cl– SAC activation by subsequent stretch. AG1478 is a highly selective blocker that inhibits ATP binding to EGFR kinase with an IC50 of 3 nM, whereas it is a weak inhibitor of ErbB2 and the platelet-derived growth factor receptor with an IC50 of >100 µM in both cases (Levitzki and Gazit, 1995). Second, exogenous EGF activated an outwardly rectifying, tamoxifen-sensitive Cl– current that resembles Cl– SAC in the absence of stretch. EGF specifically binds to EGFR kinase; ErbB3 and ErbB4 bind neuregulins, and no ligand is known for ErbB2 (Bogdan and Klämbt, 2001; Holbro and Hynes, 2004). While these results clearly implicate EGFR kinase in Cl– SAC activation, we cannot rule out the possibility that other ErbB family members participate in signaling by forming heterodimers with EGFR kinase.
This appears to be the first evidence that EGFR kinase mediates integrin signaling in cardiac myocytes. Several mechanisms for EGFR kinase transactivation have been identified. Integrins and EGFR colocalize and physically interact within macromolecular complexes at the cell membrane in human ECV304 cells and mouse fibroblasts, and integrin engagement transactivates EGFR kinase in the absence of serum or exogenous growth factors (Moro et al., 2002; Cabodi et al., 2004). This depends on the integrin cytoplasmic tail, the adaptor protein p130Cas, and Src, which phosphorylates EGFR kinase at tyr-845 and triggers autophosphorylation and downstream signaling. In addition, mechanical stretch and AT1 receptors transactivate EGFR kinase by a mechanism that depends on Src and H2O2 in a variety of cells (Ushio-Fukai et al., 2001; Oeckler et al., 2003; Kippenberger et al., 2005), and the carboxyl tail of AT1 receptors directly interacts with and transactivates EGFR kinase after phosphorylation of AT1 Y319 (Seta and Sadoshima, 2003). Thus, it is possible that integrin-mediated stretch directly activates EGFR kinase, or a factor directly linked to EGFR kinase activation, in a process that is independent of ligands for this receptor kinase. In this vein, it recently was reported that the AT1 receptor can be activated by stretch by a mechanism independent of AngII binding (Zou et al., 2004).
An autocrine–paracrine mechanism for transactivation of EGFR kinase also is well established in the cardiovascular system (Eguchi et al., 2003; Shah and Catt, 2004). AT1 receptor activation stimulates Zn2+-dependent matrix metalloproteinases (MMPs) that cleave membrane-tethered pro-heparin–binding EGF-like growth factor (proHB-EGF). This causes shedding of soluble heparin-binding EGF (HB-EGF), which is a ligand for EGF kinase. Moreover, integrins bind to and can activate MMP directly (Stanton et al., 1998) and colocalize with proHB-EGF and EGFR kinase at the surface membrane (Nakamura et al., 1995; Moro et al., 2002; Cabodi et al., 2004). Shedding of HB-EGF may explain why the Cl– current elicited by stretch was about the same magnitude as that activated by exogenous EGF when only a small fraction of the integrins were stretched by magnetic beads. A large number of myocytes with attached beads covered the chamber floor and all were stretched simultaneously. Therefore, the cell under study was exposed to locally released HB-EGF and also to HB-EGF delivered by diffusion and solution flow from other myocytes. Nevertheless, the relative importance of the various mechanisms for transactivation of EGFR kinase is unknown.
Identity of the EGF-induced Cl– Current
The primary Cl– currents traditionally recognized in cardiac myocytes include the PKA-dependent Cl– current due to the cystic fibrosis transmembrane conductance regulator (ICFTR,cardiac), the calcium-dependent transient outward Cl– current (ICl,Ca), and ICl,swell (Hume et al., 2000). The biophysical and pharmacological profile and volume sensitivity of the EGF-induced Cl– current are consistent with ICl,swell rather than either ICFTR,cardiac or ICl,Ca. The EGF-induced Cl– current and ICl,swell both activate over several minutes, outwardly rectify, and partially inactivate at positive potentials. In contrast, ICFTR,cardiac is time independent at all voltages under similar conditions (Hume et al., 2000). ICl,Ca is activated by cytoplasmic Ca2+ transients. With normal Ca2+ handling, ICl,Ca displays outward rectification and inactivates at positive potentials (Zygmunt and Gibbons, 1991). If cytoplasmic Ca2+ is clamped at an elevated level, however, ICl,Ca is time independent. Under the present conditions, ICl,Ca should be negligible because the bathing solution was Ca2+ free, and Ca2+ transients were minimized by clamping the cytoplasmic free Ca2+ to 35 nM with 8 mM EGTA. In addition, the EGF-induced Cl– current was blocked by 10 µM tamoxifen as expected for ICl,swell (Vandenberg et al., 1994), but ICFTR,cardiac (Vandenberg et al., 1994) and ICl,Ca (Valverde et al., 1993) are insensitive to tamoxifen. Finally, both EGF-induced Cl– current and Cl– SAC were inhibited by myocyte shrinkage with 1.5T hyperosmotic bathing solution. Volume sensitivity is a fundamental characteristic of ICl,swell that distinguishes it from ICFTR,cardiac and ICl,Ca (Hume et al., 2000). Thus the characteristics of the EGF-induced Cl– current resemble ICl,swell. We note, however, that the inward currents observed here are smaller than typical for ICl,swell. EGF elicits larger inward Cl– currents under different recoding conditions (unpublished data).
EGFR kinase previously was shown to modulate ICl,swell in human atrial (Du et al., 2004), rabbit ventricular (Ren and Baumgarten, 2005), and cultured neonatal rat ventricular (Walsh and Zhang, 2005) myocytes under different experimental conditions. Moreover, exogenous EGF potentiates ICl,swell in liver HTC (Varela et al., 2004) and mouse mammary C127 cells (Abdullaev et al., 2003), as well as hypotonicity-induced 125I– flux in human intestine 407 cells (Tilly et al., 1993). In the latter two studies, EGF did not activate anion current under isosmotic conditions, however. EGF also up-regulated ICl,swell in LNCaP epithelial cells, but this was attributed to enhanced expression of ClC-3 (Lemonnier et al., 2004).
Regulation of Cl– SAC by PI-3K
Block of integrin stretch- and EGF-induced currents by wortmannin and LY 294002 argues for the participation of PI-3K in the response to both stimuli. Because inhibition of PI-3K suppressed the EGF-induced current, these data also require that PI-3K is downstream from EGFR kinase in the signaling cascade (see Fig. 10), as documented by biochemical methods (Bogdan and Klämbt, 2001). Osmotic activation of ICl,swell was shown to involve PI-3K in rabbit ventricular myocytes (Ren et al., 2005), smooth muscle (Wang et al., 2004), and epithelial cells (Shi et al., 2002) Moreover, mechanical stretch (Petroff et al., 2001), osmotic swelling (Bewick et al., 1999), integrin activation (Franchini et al., 2000), AngII (Rabkin et al., 1997), and EGFR kinase (Krieg et al., 2004) all stimulate PI-3K in cardiac myocytes.
Cardiac myocytes predominantly express class I PI-3K heterodimers; p110-p85 is activated by RTK, and p110-p101 is activated by the ß subunits of G proteins (Prasad et al., 2003; Oudit et al., 2004). Wortmannin covalently modifies the p110 catalytic subunit, whereas LY294002 is structurally distinct and competes with ATP for the p110 active site (Stein and Waterfield, 2000; Oudit et al., 2004). The observation that 500 nM wortmannin suppressed only 67% of Cl– SAC may provide a clue as to which PI-3K isoform is involved. Wortmannin inhibits most PI-3K isoforms with an IC50 of 1–10 nM, but monomeric class II PI-3K-C2, which also is expressed in heart, is relatively insensitive to wortmannin with an IC50 of 400 nM (Stein and Waterfield, 2000; Oudit et al., 2004). The relative insensitivity of the PI-3K-C2 isoform may explain why wortmannin only partially inhibited Cl– SAC. Consistent with this idea, engagement of integrins activates PI-3K C2 in both platelets (Zhang et al., 1998) and smooth muscle cells (Paulhe et al., 2002). On the other hand, wortmannin, but not LY 294002, augmented the background current in the absence of stretch. The basis for this is unknown, but a stimulation of a background component of current in the presence of stretch would appear as incomplete block of the stretch-induced current. It also must be recognized that neither wortmannin nor LY 294002 is perfectly selective for PI-3K. Wortmannin inhibits myosin light chain kinase, and LY 294002 blocks casein kinase-2, several K+ channels (Stein and Waterfield, 2000; Oudit et al., 2004), and L-type Ca2+ channels (Welling et al., 2005). Because their nonspecific actions differ, however, it seems likely that both antagonists are working here by suppressing the PI-3K pathway. Although there is no evidence to suggest that both wortmannin and LY 294002 block anion channels, we also cannot definitively rule out this possibility.
Regulation of the EGF-induced Cl– Current by NADPH Oxidase
We previously demonstrated that integrin stretch and AngII elicit Cl– SAC in ventricular myocytes by activating NADPH oxidase and triggering the production of H2O2 (Browe and Baumgarten, 2004). The present results indicate that the EGF-induced current also depends on NADPH oxidase. A membrane-permeant fusion peptide inhibitor of NADPH oxidase assembly, gp91ds-tat, that mimics the docking site on gp91phox for p47phox (Rey et al., 2001) completely blocked Cl– SAC activation by EGF. In contrast, neither the inactive membrane-permeant fusion peptide scramb-tat nor the active but impermeant gp91ds docking sequence 9-mer altered the EGF-induced current. These results are consistent with the expected intracellular site of action of the docking sequence and rule out nonspecific effects of the tat 9-mer and block of Cl– channels by gp91ds from the outside. Two recent reports also conclude that exogenous EGF and/or swelling activate ICl,swell via NADPH oxidase in HeLa and liver HTC cells based on block of current by DPI (Shimizu et al., 2004; Varela et al., 2004) and expression of the dominant negative p47phox mutant p47S379A (Varela et al., 2004).
An advantage of gp91ds-tat relative to more commonly used NADPH oxidase inhibitors is its specificity, which arises from peptide–peptide interactions. This peptide does not affect the activity of other flavocytochrome enzymes or act as an O2– scavenger (Rey et al., 2001). In contrast, DPI, one of the most widely used inhibitors of NADPH oxidase activity, blocks other flavin-containing enzymes, including xanthine oxidase (Doussiere and Vignais, 1992), mitochondrial complex I (Majander et al., 1994), and nitric oxide synthase (Stuehr et al., 1991). Although gp91ds-tat is specific for NADPH oxidase, it apparently does not distinguish between isoforms because the docking sites for p47phox are homologous (Rey et al., 2001). Cardiac myocytes express both the phagocyte-like gp91phox (Nox2) isoform (Li et al., 2002; Xiao et al., 2002; Heymes et al., 2003) and the Nox4 homologue (Byrne et al., 2003). We also cannot rule out the possibility that other sources of ROS are activated by stretch or EGF and participate in the regulation of Cl– SAC. In preliminary experiments, we found that acetylcholine elicits a tamoxifen-sensitive Cl–
= 3.2 ± 0.9 min; n = 4; equivalent t1/2 = 2.2 ± 0.6 min), while activation by integrin stretch followed a slower, sigmoidal time course (t1/2 = 4.3 ± 0.6 min; n = 4). The background current was nearly constant throughout the recording period in unstimulated myocytes. Solid lines are fits to averaged data; dashed line is zero current.
