TASK-1 current is inhibited by phosphorylation during human and canine chronic atrial fibrillation

Abstract

Atrial fibrillation (AF) is a common arrhythmia with significant morbidities and only partially adequate therapeutic options. AF is associated with atrial remodeling processes, including changes in the expression and function of ion channels and signaling pathways. TWIK protein-related acid-sensitive K⁺ channel (TASK)-1, a two-pore domain K⁺ channel, has been shown to contribute to action potential repolarization as well as to the maintenance of resting membrane potential in isolated myocytes, and TASK-1 inhibition has been associated with the induction of perioperative AF. However, the role of TASK-1 in chronic AF is unknown. The present study investigated the function, expression, and phosphorylation of TASK-1 in chronic AF in atrial tissue from chronically paced canines and in human subjects. TASK-1 current was present in atrial myocytes isolated from human and canine hearts in normal sinus rhythm but was absent in myocytes from humans with AF and in canines after the induction of AF by chronic tachypacing. The addition of phosphatase to the patch pipette rescued TASK-1 current from myocytes isolated from AF hearts, indicating that the change in current is phosphorylation dependent. Western blot analysis showed that total TASK-1 protein levels either did not change or increased slightly in AF, despite the absence of current. In studies of perioperative AF, we have shown that phosphorylation of TASK-1 at Thr³⁸³ inhibits the channel. However, phosphorylation at this site was unchanged in atrial tissue from humans with AF or in canines with chronic pacing-induced AF. We conclude that phosphorylation-dependent inhibition of TASK-1 is associated with AF, but the phosphorylation site responsible for this inhibition remains to be identified.

over 2 million americans suffer from chronic atrial fibrillation (AF), and since this condition is most common among the elderly, this number is expected to more than double by 2050 as the population ages (30). Our current understanding of the pathophysiology of AF has failed to provide a clear explanation of the underlying events that lead to the arrhythmia, although it is certain that the atria undergo both electrical and structural remodeling as AF persists. These changes tend to make the atrium more susceptible to AF and thus perpetuate the arrhythmia (34). AF is associated with an increased risk of stroke and other thrombotic events leading to increased morbidity and mortality, yet consistently effective treatments are still lacking. The acute onset of AF can be associated with cardiothoracic surgery, and perioperative AF specifically is associated with inflammation (15, 17, 26). More often, however, AF is a chronic condition that is associated with widespread structural and metabolic changes in the heart. These changes include altered cell signaling pathways as well as changes in the expression and activity of numerous ion channels (25). For example, Voigt et al. (33) have shown that chronic AF is associated with an increase in the expression of a single PKC isoform, PKC-ε. Interestingly, there is also evidence that the activities of nonspecific protein phosphatases, protein phosphatase (PP)1 and PP2A, are increased. But in the face of these changes in enzymatic activities, the levels of several phosphorylated targets may be either increased or decreased, suggesting a complex regulation of levels of phosphoproteins in the heart (13, 16).

Our previous studies have focused on the activity and regulation of TWIK protein-related acid-sensitive K⁺ channel (TASK)-1, a two-pore domain K⁺ (K_2P) channel that plays a critical role in the repolarization of cardiomyocytes (2, 3). Inflammatory mediators, such as platelet-activating factor, initiate PCK-ε-mediated phosphorylation of TASK-1 on Thr³⁸³, an event that inhibits channel activity and disrupts cardiomyocyte repolarization (15). This inflammatory pathway has been clearly associated with a model of acute perioperative AF, but its role in chronic AF has yet to be explored. Since recent studies of mRNA levels of TASK-1 and TWIK-related K⁺ channel (TREK)-1 (a related K_2P channel) in a porcine rapid pacing model of chronic AF have suggested a role for these background channels in the condition (28, 29), we evaluated the functional status of TASK-1 in atrial tissue from a canine chronic pacing model of AF as well as from human subjects undergoing cardiac surgery. TASK-1 current was absent in all myocytes tested from AF tissue, suggesting that this small background current may be a significant factor in the changes leading to and/or associated with AF.

EXPERIMENTAL PROCEDURES

Human atrial myocyte isolation.

Samples of the left atrial (LA) free wall were obtained as surgical specimens from patients undergoing cardiac surgery, along with corresponding deidentified patient data, according to a protocol approved by the Institutional Review Board of Columbia University. Most patients in normal sinus rhythm were undergoing mitral valve repair, whereas those in chronic AF were also undergoing the Maze procedure. The control population included five men and three women (mean age: 58 ± 4 yr); the AF population included three men and two women [mean age: 60 ± 10 yr, not significant (NS)]. The general characteristics of the patients are shown in Table 1.

Table 1.

Patient characteristics: electrophysiological recordings

Characteristic	NSR Group	AF Group	P Value
n	8	5
Age, yr	57.6 ± 4.4	60.2 ± 10.2	0.7949
Male, n (%)	5 (62.5)	2 (40.0)	0.5921
Female, n (%)	3 (37.5)	3 (60.0)
Ejection fraction, %	49.8 ± 4.7	53.0 ± 6.4	0.4563
LA diameter, cm	4.40 ± 0.28	5.08 ± 0.24	0.1123
Surgery performed, n (%)
Maze	0 (0.0)	3 (60.0)	0.0008*
MVR	7 (87.5)	5 (100)	1.0000
AVR	2 (25.0)	0 (0.0)	0.4872
TVR	1 (12.5)	0 (0.0)	1.0000
CABG	1 (12.5)	1 (20.0)	1.0000
PFO closure	1 (12.5)	0 (0.0)	1.0000
Ventricular aneurism plication	1 (12.5)	0 (0.0)	1.0000
Thrombectomy	0 (0.0)	0 (0.0)
LA appendage resection	0 (0.0)	0 (0.0)
ASD repair	0 (0.0)	0 (0.0)
Medications, n (%)
β-Blockers	2 (25.0)	2 (40.0)	1.0000
Coumadin	1 (12.5)	4 (80.0)	0.0319*
Diuretics	2 (25.0)	1 (20.0)	1.0000
Statins and other cholesterol meds	3 (37.5)	1 (20.0)	1.0000
Digoxin	0 (0.0)	2 (40.0)	0.1282
ASA	4 (50.0)	0 (0.0)	0.1049
ACE inhibitor	1 (12.5)	1 (20.0)	1.0000
AT2 receptor antagonists	4 (50.0)	1 (20.0)	0.5649
Amiodarone	0 (0.0)	1 (20.0)	0.3846
Ca2+ channel blockers	1 (12.5)	1 (20.0)	1.0000
Na+ channel blockers	0 (0.0)	1 (20.0)	0.3846
Diabetes medications	2 (25.0)	0 (0.0)	0.4872
Antihistamine	0 (0.0)	0 (0.0)
Antacid/proton pump inhibitor	3 (37.5)	1 (20.0)	1.0000
Antidepressant/hypnotic	1 (12.5)	1 (20.0)	1.0000
Gout medications	2 (25.0)	0 (0.0)	0.4872
Corticosteroids	0 (0.0)	0 (0.0)
Plavix	0 (0.0)	0 (0.0)
α-Blockers/BPH medications	1 (12.5)	0 (0.0)	1.0000

Values are means ± SE or number of subjects (n) with percentages. NSR, normal sinus rhythm; AF, atrial fibrillation; LA, left atrial; MVR, mitral valve repair; AVR, aortic valve repair; TVR, tricuspid valve repair; CABG, coronary artery bypass graft; PFO, patent foramen ovale; ASD, atrial septal defect; ASA, aspirin; ACE, angiotensin-converting enzyme; AT₂, ANG II type 2; BPH, benign prostatic hyperplasia.

^*Significant difference.

Atrial tissue was first minced in physiological solution [composed of (in mM) 140 NaCl, 5.4 KCl, 5 HEPES, 2.3 NaOH, 10 glucose, 1 CaCl₂, and 1 MgCl₂; pH 7.4] and then rinsed in Ca²⁺-free solution [composed of (in mM) 140 NaCl, 5.4 KCl, 5 HEPES, 0.5 MgCl₂, 1.2 KH₂PO₄, 5.5 glucose, and 50 taurine; pH 6.9]. Next, the pieces were immersed in an enzyme solution and bubbled with O₂ at 37°C. The enzyme solution was made by adding collagenase (59.4 U/ml, Worthington CLS2, 198 U/mg) and protease (0.52 U/ml, Sigma type XIV, 5.2 U/mg) to the Ca²⁺-free solution. Forty-five to sixty minutes after the initial exposure of the tissue to the enzymes and every 10–15 min thereafter, the supernatant was checked for the appearance of myocytes. If myocytes were present, the supernatant was centrifuged, and the pellet was resuspended in Ca²⁺-free physiological solution and then slowly adapted to a physiological Ca²⁺ concentration (1 mM). Only rod-shaped and striated myocytes were used for electrophysiological experiments.

Canine rapid pacing model of AF.

This study was performed under a protocol approved by the Institutional Animal Care and Use Committee of Columbia University. Pacemakers were implanted in female adult mongrel dogs weighing 24–26 kg following the procedure previously described by Yagi et al. (35). Animals were anesthetized with thiopental sodium (17 mg/kg iv) and ventilated with isoflurane (1.5–2%) and O₂ (2 l/min). Active fixation leads were implanted in the right atrial (RA) appendage and right ventricular free wall and connected to an Itrel pulse generator and Thera 8962 pacemaker, respectively (Medtronics). Complete atrioventricular block was generated by injection of 40% formaldehyde (0.1–0.3 ml) into the His bundle. Immediately after surgery, the ventricular pacemaker was programmed at a rate of 60 beats/min to ensure a physiological ventricular rate during recovery. After 2 wk of recovery, atrial pacing was initiated at a rate of 600–900 beats/min and maintained for 4–6 wk. During this period, dogs were studied weekly with measurements of the effective refractory period and recording of the ECG. Following this protocol, the majority of the paced animals developed chronic AF (defined as >5 days of AF in the absence of continued pacing) and were in this rhythm at the time of euthanasia. Some animals underwent sham operations with placement of a nonfunctioning RA lead. The ventricles of these sham animals were paced at 60 beats/min for 50 days. At the time of euthanasia, animals were anesthetized with pentobarbital (30 mg/kg), and hearts were removed. Atrial myocytes were prepared as described above for human tissue except that cells were derived from the RA free wall.

Electrophysiological recordings and solutions.

Atrial myocytes were placed into a perfusion chamber mounted on the stage of an inverted microscope. Since TASK-1 is an open rectifier, it is possible to linearize the current-voltage relation by recording from cells that are superfused at room temperature (22–25°C) with a high-K⁺ external solution. Thus, for these experiments, cells were superfused with the following solution (in mM): 100 NaCl, 50 KCl, 1 CaCl₂, 1 MgCl₂, 5 HEPES, 10 glucose, 5 CsCl, and 1 tetraethylammonium with 5 μM nifedipine and adjusted to pH 7.4 with an osmolarity of ∼298 mosmol/kgH₂O. TASK-1 current was defined as the methanandamide-sensitive current (methanandamide is a TASK-1 blocker, 10 μM) (24), which was recorded in the whole cell configuration from rod-shaped, striated myocytes using a ramp protocol from −50 to +30 mV in 6 s (see Fig. 1). Borosilicate glass pipettes with a tip resistance between 3 and 5 MΩ were used. Pipettes were filled with a solution containing (in mM) 130 aspartic acid, 146 KOH, 10 NaCl, 2 CaCl₂, 5 EGTA, 10 HEPES, and 2 MgATP adjusted to pH 7.2 with an osmolarity of ∼295 mosmol/kgH₂O. In recordings with PP2A (1 U/ml), the tip of the pipette was first filled with the phosphatase and the remainder with regular pipette solution. Recordings started 10–12 min after rupture to allow the enzyme to dialyze into the cell.

Fig. 1.

TWIK protein-related acid-sensitive K⁺ channel (TASK)-1 current is missing in myocytes from human atrial fibrillation (AF) hearts and can be rescued by intracellular application of protein phosphatase (PP)2A. A: net current-voltage (I–V) relation recorded from an isolated human atrial myocyte generated by a ramp protocol from −50 to +30 mV in 6 s under control conditions and in the presence of the TASK-1 blocker methanandamide (10 μM). The methanandamide curve was subtracted from the control result to generate a difference current (B), which was defined under these conditions as the TASK-1 current. These data were normalized to cell capacitance and thus are expressed as current density and have been corrected for junction potential (−9.8 mV). This recording is typical of at least 10 cells. C and D: atrial myocytes were isolated from human hearts that were either in normal sinus rhythm (NSR) or AF. TASK-1 current was recorded, and the average current density ± SE is shown for each group of cells. TASK-1 currents differed significantly under control conditions between myocytes from human NSR (C, ○, n = 10 cells from 6 subjects) and AF hearts (D, ○, n = 9 cells from 5 subjects, P < 0.01). In some cases, the patch pipette contained the nonspecific serine/threonine phosphatase PP2A (1 U/ml). PP2A rescued TASK-1 currents from AF cells (D, ●, n = 13 cells from 4 subjects) but had no effect on TASK-1 current in cells from hearts in NSR (C, ●, n = 8 cells from 5 subjects).

Voltage-clamp protocols were generated using Clampex 8.0 software applied by means of an Axopatch 200-B and a Digidata 1200 interface (Axon Instruments). Current signals were filtered at 2 kHz and acquired at 500 Hz. Currents are expressed as current density after normalization for cell capacitance.

Drugs.

Methanandamide (BioMol) was prepared as a stock solution in DMSO and then diluted to its final concentration in the external recording solution (final DMSO < 0.1%). A stock solution of PP2A (Upstate) was prepared in a buffer solution (provided by the manufacturer) and then diluted to the final concentration of 1 U/ml in the pipette solution.

Western blot analysis.

Frozen tissue from the canine RA or human LA was thawed and homogenized in detergent-free homogenization buffer [50 mM Tris·HCl (pH 7.4), 150 mM NaCl, 1 mM EGTA, 250 mM sucrose, 0.1 mM PMSF, 1 ng/ml aprotinin, 1 ng/ml leupeptin, 0.1 μM pepstatin, 1 μg/ml benzamidine, 20 μM activated sodium orthovanadate, and 20 μM NaF]. Homogenates were centrifuged at 160,000 g, and crude membrane pellets were suspended in homogenization buffer supplemented with 1% Triton X-100. Total protein concentration was determined by a modification to the method of Lowry (27). Twenty micrograms of total protein from each sample were loaded and run onto duplicate 8% or 10% polyacrylamide Tris-glycine gels and transferred to nitrocellulose membranes according to standard protocols. Proteins were visualized by incubation with an antibody that recognizes total TASK-1 (1.4 μg/ml, BioMol) or phosphorylated (p)Thr³⁸³ TASK-1 (10 μg/ml) (15) followed by an incubation with horseradish peroxidase-conjugated anti-rabbit antibody and chemiluminescence substrate (Pierce West Pico) before the blot was exposed to X-ray film. Films were scanned, and band pixilation was calculated using ImageJ software (1). All total TASK-1 and pThr³⁸³ blots were run and exposed in parallel. To approximate the fraction of TASK-1 channels that were phosphorylated, the average pThr³⁸³ signal was expressed as a ratio of the average TASK-1 signal. Western blots were repeated two to three times as sample volumes allowed.

The human tissue used in these biochemical experiments was distinct from the tissue used in the electrophysiological experiments. These data were derived from 19 subjects in normal sinus rhythm (NSR; 15 men and 4 women, mean age: 59 ± 2 yr) and from 21 AF subjects (9 men and 12 women, mean age: 69 ± 3 yr, P < 0.05). The general characteristics are shown in Table 2. Because of the significant difference in age between the two populations, a reanalysis of the data was performed on an age-matched subset of 13 subjects in NSR (10 men and 3 women, mean age: 61 ± 2 yr) and 10 AF subjects (7 men and 3 women, mean age: 60 ± 2 yr, NS). The characteristics of this age-matched group are shown in Table 3.

Table 2.

Patient characteristics: Western blot analysis

Characteristic	NSR Group	AF Group	P Value
n	19	21
Age, yr	58.9 ± 2.4	68.8 ± 2.6	0.0089*
Male, n (%)	15 (78.9)	9 (42.9)	0.0270*
Female, n (%)	4 (21.1)	12 (57.1)
Ejection fraction, %	51.9 ± 3.0	47.5 ± 2.6	0.2767
LA diameter, cm	4.63 ± 0.19	5.95 ± 0.27	0.0009*
Surgery performed, n (%)
Maze	0 (0.0)	15 (71.4)	<0.0001*
MVR	19 (100)	19 (90.5)	0.4885
AVR	0 (0.0)	4 (19.0)	0.1079
TVR	0 (0.0)	4 (19.0)	0.1079
CABG	6 (31.6)	5 (23.8)	0.7271
PFO closure	1 (5.3)	0 (0.0)	0.4750
Ventricular aneurism plication	0 (0.0)	0 (0.0)
Thrombectomy	0 (0.0)	1 (4.8)	1.0000
LA appendage resection	0 (0.0)	1 (4.8)	1.0000
ASD repair	1 (5.3)	0 (0.0)	0.4750
Medications, n (%)
β-Blockers	10 (52.6)	16 (76.2)	0.1856
Coumadin	2 (10.5)	15 (71.4)	0.0001*
Diuretics	6 (31.6)	12 (57.1)	0.1253
Statins and other cholesterol meds	9 (47.4)	11 (52.4)	1.0000
Digoxin	3 (15.8)	9 (42.9)	0.0888
ASA	11 (57.9)	8 (38.1)	0.3419
ACE inhibitor	7 (36.8)	6 (28.6)	0.7378
AT2 receptor antagonists	7 (36.8)	6 (28.6)	0.3121
Amiodarone	1 (5.3)	3 (14.3)	0.6071
Ca2+ channel blockers	2 (10.5)	3 (14.3)	1.0000
Na+ channel blockers	0 (0.0)	0 (0.0)
Diabetes medications	1 (5.31)	0 (0.0)	0.4750
Antihistamine	3 (15.8)	4 (19.0)	1.0000
Antacid/proton pump inhibitor	5 (26.3)	3 (14.3)	0.4420
Antidepressant/hypnotic	3 (15.8)	2 (9.5)	0.6544
Gout medications	3 (15.8)	0 (0.0)	0.0981
Corticosteroids	3 (15.8)	2 (9.5)	0.6544
Plavix	1 (5.3)	2 (5.7)	1.0000
α-Blockers/BPH medications	3 (8.3)	3 (14.3)	0.6071

Values are means ± SE or number of subjects (n) with percentages.

^*Significant difference.

Table 3.

Patient characteristics: Western blot analysis of an age-matched subset

Characteristic	NSR Group	AF Group	P Value
n	13	10
Age, yr	60.6 ± 1.9	59.6 ± 1.8	0.7068
Male, n (%)	10 (76.9)	7 (70.0)	1.0000
Female, n (%)	3 (23.1)	3 (30.0)
Ejection fraction, %	53.7 ± 3.3	45.3 ± 3.7	0.1008
LA diameter, cm	4.74 ± 0.26	6.06 ± 0.44	0.0207*
Surgery performed, n (%)
Maze	0 (0.0)	6 (60.0)	0.0021*
MVR	13 (100)	9 (90.0)	0.4348
AVR	0 (0.0)	1 (10.0)	0.4348
TVR	0 (0.0)	3 (30.0)	0.0678
CABG	4 (30.8)	3 (30.0)	1.0000
PFO closure	1 (7.7)	0 (0.0)	1.0000
Ventricular aneurism plication	0 (0.0)	0 (0.0)
Thrombectomy	0 (0.0)	1 (10.0)	0.4348
LA appendage resection	0 (0.0)	0 (0.0)
ASD repair	1 (7.7)	0 (0.0)	1.0000
Medications, n (%)
β-Blockers	5 (38.5)	7 (70.0)	0.2138
Coumadin	1 (7.7)	7 (70.0)	0.0059*
Diuretics	4 (30.8)	6 (60.0)	0.2215
Statins and other cholesterol meds	5 (38.5)	6 (60.0)	0.4136
Digoxin	2 (15.4)	3 (30.0)	0.6175
ASA	8 (61.5)	5 (50.0)	0.6850
ACE inhibitor	4 (30.8)	5 (50.0)	0.4173
AT2 receptor antagonists	5 (38.5)	1 (10.0)	0.1790
Amiodarone	0 (0.0)	1 (10.0)	0.4348
Ca2+ channel blockers	2 (15.4)	2 (20.0)	1.0000
Na+ channel blockers	0 (0.0)	0 (0.0)
Diabetes medications	1 (7.7)	0 (0.0)	1.0000
Antihistamine	3 (23.1)	1 (10.0)	0.6036
Antacid/proton pump inhibitor	3 (23.1)	0 (0.0)	0.2292
Antidepressant/hypnotic	2 (15.4)	0 (0.0)	0.4862
Gout medications	3 (23.1)	0 (0.0)	0.2292
Corticosteroids	2 (15.4)	1 (10.0)	1.0000
Plavix	1 (7.7)	0 (0.0)	1.0000
α-Blockers/BPH medications	2 (15.4)	0 (0.0)	0.4862

Values are means ± SE or number of subjects (n) with percentages.

^*Significant difference.

Data analysis and statistics.

Data were analyzed using pCLAMP 8.0 (Axon) and Origin 6.0 (Microcal) and are presented as means ± SE. For each myocyte, TASK-1 current was obtained as the difference between the average of the traces at steady state under control conditions and in the presence of methanandamide. All recordings have been corrected for the junction potential (−9.8 mV). Student's t-test and ANOVA (with Bonferroni or Dunn's multiple-comparisons tests, as appropriate) were used to compare data. P values of <0.05 were considered statistically significant.

RESULTS

Chronic AF alters TASK-1 current in human atrial myocytes.

TASK-1 current was readily measured as the methanandamide-sensitive current by patch-clamp recording in control atrial myocytes from both human and canine tissue and is typically expressed as a difference current. Figure 1 shows a typical current-voltage relation for the net current before and after methanandamide exposure recorded from a single human atrial myocyte isolated from a subject in NSR (Fig. 1A). TASK-1 current (calculated as the methanandamide difference current) is also shown (Fig. 1B). The shape and reversal potential are appropriate for this K⁺-selective current recorded in a high-K⁺ solution (predicted reversal potential: −27.6 mV, calculated reversal potential: −28.9 ± 2.8 mV, n = 10). Similar results were obtained in atrial myocytes isolated from a control canine heart (reversal potential: −28.5 ± 1.6 mV, n = 26).

Using this method, we next compared the TASK-1 current in atrial myocytes from subjects who had been in NSR with those in chronic AF. The TASK-1 current, which was 0.35 ± 0.08 pA/pF (Fig. 1C, open circles; 10 cells from 6 subjects, measured at +30 mV) in atrial myocytes from subjects in NSR was absent in cells isolated from subjects in chronic AF (Fig. 1D, open circles; −0.002 ± 0.04 pA/pF, 9 cells from 5 subjects, P < 0.001).

Because we had previously observed a phosphorylation-dependent loss of TASK-1 current in atrial myocytes during perioperative AF in dogs (15), we wondered whether a similar phenomenon was occurring in chronic AF in humans. We attempted to rescue TASK-1 current by introducing the nonspecific serine-threonine phosphatase PP2A into the patch pipette. When PP2A (1 U/ml) was dialyzed into atrial cells from human hearts in NSR, there was no effect on TASK-1 current (Fig. 1C, solid circles; current density in the presence of PP2A: 0.32 ± 0.08 pA/pF, 8 cells from 5 subjects). However, intracellular application of PP2A to atrial myocytes from patients in AF restored TASK-1 current to levels that were comparable with those seen in myocytes from normal hearts (Fig. 1D, solid circles; 0.45 ± 0.17 pA/pF, 13 cells from 4 subjects).

Chronic AF alters TASK-1 current in canine atrial myocytes.

The phosphorylation-dependent loss of TASK-1 current in atrial myocytes from human subjects with chronic AF is striking. However, since patients with AF are pathologically and therapeutically distinct from patients in NSR, it was unclear whether the loss of current was caused by the arrhythmia or by some other shared pathology. Therefore, we repeated these analyses in atrial myocytes from a more controlled model, canine rapid pacing AF. Atrial myocytes isolated from normal dogs in sinus rhythm exhibited a TASK-1 current of 0.39 ± 0.09 pA/pF at +30 mV (Fig. 2A, left; n = 21 cells from 8 dogs). This did not differ from TASK-1 current measured in myocytes from dogs that had undergone a sham operation (Fig. 2A, middle; 0.44 ± 0.10 pA/pF, n = 17 cells from 4 dogs, NS). However, there was significantly less TASK-1 current in myocytes isolated from animals in chronic AF (Fig. 2A, right; 0.11 ± 0.05 pA/pF measured at +30 mV, n = 13, P < 0.001).

Fig. 2.

TASK-1 current is missing in myocytes from canine AF hearts and can be rescued by intracellular application of PP2A. Atrial myocytes were isolated from unoperated control (normal), sham-operated (sham), or AF dogs. TASK-1 current was recorded, and the average current density ± SE is shown for each group of cells as described in Fig. 1. A: data from control (left; n = 21 cells from 8 dogs), sham (middle; n = 17 cells from 4 dogs), and AF (right, n = 13 cells from 2 dogs) dogs. In a separate set of animals, TASK-1 current was recorded in the presence or absence of the serine/threonine phosphatase PP2A, and the average current density ± SE is shown for each group of cells as described in Fig. 1. B: PP2A had no effect on current density in cells prepared from normal canine ventricles (left; n = 11 cells from 4 dogs in the control group and 9 cells from 4 dogs in the PP2A-treated group) or from sham-operated hearts (middle; n = 3 cells from 1 dog in the control group and 6 cells from 1 dog in the PP2A-treated group). The TASK-1 I–V curve differed significantly from control after PP2A treatment (right, n = 5 cells from 3 dogs in the control group and 10 cells from 3 dogs in the PP2A-treated group, P < 0.0001 by ANOVA).

The phosphorylation dependence of TASK-1 inhibition was likewise tested in canine myocytes. In an independent set of cells, PP2A had no effect on current density at +30 mV in myocytes from normal dogs (Fig. 2B, left; control: 0.31 ± 0.07 pA/pF, 11 cells from 4 dogs, and PP2A: 0.21 ± 0.03 pA/pF, 9 cells from 4 dogs, NS), nor did it have any effect on current density at +30 mV in myocytes from sham-operated dogs (Fig. 2B, middle; control: 0.25 ± 0.03 pA/pF, 3 cells from 1 dog, and PP2A: 0.17 ± 0.04 pA/pF, 6 cells from 1 dog, NS). However, the addition of PP2A to the patch pipette rescued TASK-1 current in myocytes from dogs in chronic AF (Fig. 2B, right; control: 0.005 ± 0.05 pA/pF, 5 cells from 3 dogs, and PP2A: 0.11 ± 0.04 pA/pF, 10 cells from 3 dogs, P < 0.05).

TASK-1 and pThr³⁸³ TASK-1 protein levels in atria from hearts in NSR or chronic AF.

Our previous work demonstrated that the phosphorylation-dependent loss of TASK-1 current in perioperative AF was associated with increased phosphorylation of TASK-1 at Thr³⁸³. To determine whether a similar phenomenon was occurring in the chronic condition, we performed Western blot analysis to quantify TASK-1 and pThr³⁸³ TASK-1 protein in human atrial tissue taken from subjects in NSR versus chronic AF. The characteristics of these subjects are shown in Table 2. Equal protein was loaded, and total and pThr³⁸³ TASK-1 signals were measured (Fig. 3A). There was no apparent difference in total TASK-1 between NSR and AF subject groups (8,250 ± 694 vs. 8,091 ± 627 pixels; Fig. 3B). In addition, the ratio of pThr³⁸³ TASK-1 to total TASK-1 signals (an approximation of the fraction of channels phosphorylated at Thr³⁸³) was not significantly different between subjects in NSR and subjects with AF (1.79 ± 0.12 vs. 1.53 ± 0.10; Fig. 3C).

Fig. 3.

Chronic AF in human subjects is not associated with changes in total TASK-1 protein levels or Thr³⁸³ phosphorylation. Crude membrane fractions of left atrial tissue taken from human valve repair patients in NSR or with AF were resolved by SDS-PAGE and blotted to nitrocellulose. Protein was visualized using an antibody to total TASK-1 or to phosphorylated Thr³⁸³ (pT383) TASK-1, and band densities were quantified. TASK-1 levels were measured either two or three times for each subject, and the average was plotted in a scatterplot (○). Means ± SE for all subjects are shown (●). A: representative Western blots of tissue from subjects in NSR or with AF immunoblotted for total TASK-1 (top) and pT383 TASK-1 (bottom). B: there was no difference in average total TASK-1 band intensity of tissue from all subjects in NSR (n = 21) versus all subjects with AF (n = 19). C: there was also no difference in the ratio of pT383 to total TASK-1 band intensity in tissue from all subjects in NSR versus all subjects in AF. D and E: in an age-matched subset of subjects, there was likewise no difference between subjects in NSR versus subjects with AF in total TASK-1 band intensity (D; n = 13 in the NSR group and n = 10 in the AF group) or in the ratio of pT383 to total TASK-1 band intensity (E).

Not surprisingly, there were significant age and sex differences between NSR and AF patient populations, and it was possible that either characteristic could be a confounding factor, masking a potential difference in total or pThr³⁸³ TASK-1 signals between groups. Therefore, a separate analysis was performed on an age- and sex-matched cohort. The characteristics of this subset are shown in Table 3. In this age-matched subset, there was still no difference in total TASK-1 pixel density between the two groups (8,954 ± 832 pixels in the NSR group vs. 7,494 ± 607 pixels in the AF group; Fig. 3D), nor was there any difference in the ratio of pThr³⁸³ to total TASK-1 signals in NSR subjects and those with chronic AF (1.80 ± 0.10 vs. 1.57 ± 0.17; Fig. 3E).

The same Western blot analyses were performed on atrial tissue from dogs with chronic pacing AF compared with sham-operated and unoperated controls (Fig. 4A). There was no difference in the total TASK-1 signal between tissue from dogs with AF (8,991 ± 532 pixels, n = 7) and sham paced dogs (8,601 ± 299 pixels, n = 6), although both groups had a significantly higher level of TASK-1 than tissue from unoperated control dogs (6,556 ± 624 pixels, n = 5, P < 0.05; Fig. 4B). Despite the change in total TASK-1 levels, we found no significant difference in the ratio of pThr³⁸³ to total TASK-1 (n = 7 in the AF group, 6 in the sham group, and 5 for the control group; Fig. 4C), suggesting that phosphorylation at Thr³⁸³ is not responsible for the phosphorylation-dependent loss of current in chronic AF.

Fig. 4.

Chronic AF in canines is not associated with changes in total TASK-1 protein levels or Thr³⁸³ phosphorylation. Crude membrane fractions of right atrial tissue taken from unoperated control, sham paced, and chronic pacing AF dogs were resolved by SDS-PAGE and blotted to nitrocellulose. Protein was visualized using an antibody to total TASK-1 or to pT383 TASK-1, and band densities were quantified. TASK-1 levels were measured either two or three times for each dog, and the average was plotted in a scatterplot (○). Means ± SE for all animals are shown (●). A: representative Western blots of tissue taken from control dogs, sham-operated dogs, and dogs with AF immunoblotted for total TASK-1 (top) and pT383 TASK-1 (bottom). B: average total TASK-1 band intensity of tissue from control dogs (n = 5), sham-operated dogs (n = 6), and dogs with AF (n = 7). TASK-1 levels in tissue from sham-operated dogs and dogs with AF were significantly higher than in tissue from unoperated control dogs (P < 0.05). C: average pT383 band intensity, normalized to total TASK-1 band intensity, for the three groups. There were no differences in the pT383-to-TASK-1 ratio between control dogs, sham-operated dogs, and dogs with AF.

DISCUSSION

The heart undergoes extensive remodeling during chronic AF (18, 25). This remodeling plays an important role in the pathophysiology of AF and contributes to both the initiation and maintenance of the arrhythmia. Functional alterations in ionic currents, including Na⁺, K⁺, and Ca²⁺ currents, have been observed in chronic AF, but studies that have quantified the corresponding protein levels of these channels remain controversial (4, 6, 8, 9, 31, 32, 36). Nevertheless, the cumulative balance of all of these alterations results in the shortening of the effective refractory period and of the action potential duration and the loss of rate adaptation. Since TASK-1 is a background or leak current, its function is important for the maintenance of the resting membrane potential, electrical stability, and input resistance. Heterogeneity of the resting potential induced by changes in leak currents can alter atrial responsiveness, changing both the action potential duration and effective refractory period, creating the molecular substrate for atrial arrhythmias.

There are several reasons to think that leak currents such as those carried by TASK-1 or other K_2P channels might play an important role in modifying cardiac rate and rhythm. Inhibition of TASK-1 is sufficient to alter repolarization of isolated canine or murine myocytes and has been suggested to contribute to the generation of tachyarrhythmias in both mammalian and invertebrate models. Lalevée et al. (21) have shown that the Drosophila homolog of TASK-1, ORK1, plays a key role in regulating cardiac function in the fly. Mutant strains that do not express the channel have significantly higher heart rates, whereas cardiac overexpression of ORK1 leads to hyperpolarization of the myocyte resting potential followed by heart block. The beating of the heart can be restored with electrical stimulation, indicating that the myocytes are still excitable. A predominantly atrial phenotype was observed in zebrafish with TASK-1 knockdown. In addition to significant bradycardia, TASK-1 inactivation caused atrial dilation (22). The authors furthermore identified several loss of function TASK-1 mutations associated with AF in human patients. The TASK-1 knockout mouse has a more modest phenotype but nevertheless presents with an increased basal heart rate and prolongation of the rate-corrected QT interval as well as a broadened QRS complex. Action potential durations were prolonged when measured from isolated ventricular myocytes or from ventricular free wall recordings of isolated hearts (7, 10). Our own studies of TASK-1 function in ventricular myocytes clearly showed that inhibition of TASK-1 current disrupts repolarization of isolated cells and leads to abnormalities, including early afterdepolarizations and arrest at the plateau (2, 3). Similar observations were made in human atrial myocytes from subjects in NSR (23). Furthermore, recent work has also suggested that remodeling of both TASK-1 and the related K_2P channel TREK-1 occurs concomitantly with a chronic pacing-induced AF model in the pig (28, 29). Decreased TASK-1 and TREK-1 mRNA and decreased TREK-1 protein were observed in the RA 7 and 21 days after the onset of AF induced by rapid burst pacing. Unfortunately, the authors did not measure changes in TASK-1 protein or determine the functional level of either current, so it is unclear to what extent these changes contribute to electrical remodeling in this pathological state. It is noteworthy that in our studies, there was a significant increase in TASK-1 protein in the atria of all surgical animals, both sham and paced. However, the level of current ranged from normal in myocytes from sham atria to near zero in myocytes from paced hearts. Furthermore, while the current in myocytes from paced hearts can be rescued by inclusion of PP2A in the patch pipette, the rescue does not bring the current completely back to control levels. This suggests that some or much of the excess channel protein may not be at the plasma membrane and further suggests that channel trafficking may play an important role in regulating the level of functional channel expressed on myocytes under stress.

TASK-1 channel function can be modulated by environmental factors such as low pH (11) or hypoxia (5, 19), both of which may be prominent pathophysiological states associated with AF. However, our data support a more active regulation of the channel during AF that involves a phosphorylation/dephosphorylation cycle. The expression and function of both kinases and phosphatases are known to be altered during AF. For example, PKC-ε protein expression has been shown to be upregulated in human hearts with chronic AF (33). In addition, expression and activity of protein phosphatases have been found to be elevated in chronic AF (6, 14), although others have found them to be unchanged (20). Surprisingly, El-Armouche et al. (13) discovered that the elevation of PP2A, a nonspecific serine/threonine phosphatase, does not always correlate with a reduction in the phosphorylation level of some target proteins in the AF heart. Their study showed that while phosphorylation of myosin-binding protein-C was reduced, phosphorylation of phospholamban was increased. Others have noted an increase in the phosphorylation of troponin T in AF hearts (12). This lack of homogeneity in the response was presumed to be due to changes in the phosphorylation levels of phosphatase regulatory proteins as well as alterations in subcellular localization of these key enzymes.

The alterations in kinase and phosphatase activities noted in AF can have variable effects on cardiac ion channels (for a review, see Ref. 16). For example, increased PP2A activity has been found to reduce basal L-type Ca²⁺ current (6), whereas increased PKC-ε expression is linked to an increase in a constitutively active ACh-dependent K⁺ current (33). Our results suggest that there is an increase in the phosphorylation of TASK-1 during chronic AF that leads to a loss of current. While we have previously demonstrated that PKC-ε-dependent phosphorylation of TASK-1 on Thr³⁸³ increases in response to inflammation associated with perioperative AF (15), this mechanism appears not to be relevant in the chronic pacing model. Our data show that TASK-1 is inhibited through a different phosphorylation-dependent mechanism, although the details of this remain to be determined. The difference in regulation is not surprising since there are likely to be differences in signaling between the inflammation-driven perioperative atrium (that is largely preremodeled) and chronically paced atrium (that is postremodeled). The two diseases are not on a continuum. Rather, perioperative AF is typically a self-limiting event that does not lead to persistent AF, whereas chronic AF more often leads to a persistent form of the condition that is difficult to control. It is interesting, however, that the loss of TASK-1 current should be part of a final common pathway in the two situations. Our results are not enough to say that these changes are either necessary or sufficient to induce the arrhythmia. It is unclear from the available data thus far whether TASK-1 inhibition contributes to the initiation of the disease or whether it is a response to other pathological changes that are observed in AF. It is interesting to note, however, that knockdown of TASK-1 in the zebrafish caused atrial dilation in the absence of an arrhythmia (22). These data suggest that inhibition of TASK-1 can contribute to both the physical and electrical changes associated with AF and may not be strictly a consequence of the arrhythmia. Further work is required to determine which TASK-1 sites are targeted by which signaling cascades in chronic AF, but our results clearly demonstrate the loss of TASK-1 current in myocytes from human hearts in AF. Coupled with in vitro studies showing that the loss of TASK-1 current alone induces repolarization abnormalities (2, 23), these data are enough to suggest that this channel may be a fruitful target for therapy of this arrhythmia.

GRANTS

This work was supported by National Institutes of Health (NIH) Grants R01-HL-70105 (to S. J. Feinmark) and HL-67101 (to M. R. Rosen) as well as by NIH Grants TL1-TR-000082 and TL1-RR-024158. This work was also supported by a fellowship grant from the PhRMA Foundation (to E. Harleton) and a career development award from the Thoracic Surgery Foundation for Research and Education (to M. Argenziano).

Portions of this work were also supported by a Strategic Research Alliance with the Institut de Recherches Servier.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: E.H., A.B., P.C., P.D.J., and T.S.R. performed experiments; E.H., A.B., P.C., P.D.J., T.S.R., M.R.R., M.A., R.B.R., and S.J.F. analyzed data; E.H., A.B., P.D.J., M.R.R., M.A., R.B.R., and S.J.F. interpreted results of experiments; E.H., A.B., and S.J.F. prepared figures; E.H., R.B.R., and S.J.F. drafted manuscript; E.H., A.B., P.C., P.D.J., T.S.R., M.R.R., M.A., R.B.R., and S.J.F. approved final version of manuscript; M.R.R., M.A., R.B.R., and S.J.F. conception and design of research; M.R.R. and M.A. edited and revised manuscript.