Transient outward K⁺ current reduction prolongs action potentials and promotes afterdepolarisations: a dynamic-clamp study in human and rabbit cardiac atrial myocytes

Abstract

Human atrial transient outward K⁺ current (I_TO) is decreased in a variety of cardiac pathologies, but how I_TO reduction alters action potentials (APs) and arrhythmia mechanisms is poorly understood, owing to non-selectivity of I_TO blockers. The aim of this study was to investigate effects of selective I_TO changes on AP shape and duration (APD), and on afterdepolarisations or abnormal automaticity with β-adrenergic-stimulation, using the dynamic-clamp technique in atrial cells. Human and rabbit atrial cells were isolated by enzymatic dissociation, and electrical activity recorded by whole-cell-patch clamp (35–37°C). Dynamic-clamp-simulated I_TO reduction or block slowed AP phase 1 and elevated the plateau, significantly prolonging APD, in both species. In human atrial cells, I_TO block (100%I_TO subtraction) increased APD₅₀ by 31%, APD₉₀ by 17%, and APD_{−61 mV} (reflecting cellular effective refractory period) by 22% (P < 0.05 for each). Interrupting I_TO block at various time points during repolarisation revealed that the APD₉₀ increase resulted mainly from plateau-elevation, rather than from phase 1-slowing or any residual I_TO. In rabbit atrial cells, partial I_TO block (∼40%I_TO subtraction) reversibly increased the incidence of cellular arrhythmic depolarisations (CADs; afterdepolarisations and/or abnormal automaticity) in the presence of the β-agonist isoproterenol (0.1 μm; ISO), from 0% to 64% (P < 0.05). ISO-induced CADs were significantly suppressed by dynamic-clamp increase in I_TO (∼40%I_TO addition). ISO+I_TO decrease-induced CADs were abolished by β₁-antagonism with atenolol at therapeutic concentration (1 μm). Atrial cell action potential changes from selective I_TO modulation, shown for the first time using dynamic-clamp, have the potential to influence reentrant and non-reentrant arrhythmia mechanisms, with implications for both the development and treatment of atrial fibrillation.

Key points

• The shape of the cardiac atrial action potential is influenced by the flow of a transient outward K⁺ current (I_TO) across atrial muscle cell membranes.

• Whether changes in I_TO could alter atrial cell action potentials in ways that could affect mechanisms of abnormal heart rhythms (arrhythmias) is unclear, because currently available I_TO blocking drugs are non-selective.

• We used the ‘dynamic-clamp’ technique, for the first time in atrial cells isolated from patients, and from rabbits, to electrically simulate selective changes in I_TO during action potential recording.

• We found that I_TO decrease prolonged atrial cell action potential duration and, under β-adrenergic-stimulation, provoked abnormal membrane potential oscillations (afterdepolarisations) that were preventable by I_TO increase or a β-blocker.

• These results help us better understand the contribution of I_TO to atrial cell action potential shape and mechanisms of arrhythmia, with potential implications for both the development and treatment of atrial fibrillation.

Introduction

The cardiac transient outward K⁺ current (I_TO) activates rapidly upon action potential (AP) initiation. Its large magnitude in atrial cells of many species, including human, and in ventricular cells of some rodents contributes to the characteristic spiky AP shape in these cells, with fast early repolarisation (phase 1) and small plateau (Xu et al. 1999; Workman et al. 2000; Workman et al. 2001; Sah et al. 2002; Madhvani et al. 2011). In such cells, I_TO reduction can markedly alter AP shape. For example, in atrial (Xu et al. 1999) and ventricular (Sah et al. 2002) cells from mice genetically engineered for reduced I_TO, the plateau was elevated and AP duration at late (90%) repolarisation (APD₉₀) markedly prolonged. I_TO reduction has the potential, therefore, to influence various electrophysiological mechanisms of cardiac arrhythmias. Plateau elevation may cause afterdepolarisations or abnormal automaticity, particularly during adrenergic elevation (Workman, 2010), and APD₉₀ increase may inhibit reentry (Workman et al. 2011).

Atrial fibrillation (AF) is the most common cardiac arrhythmia, and atrial I_TO is markedly reduced in chronic AF and in several predisposing pathologies, by electrophysiological remodelling (Le Grand et al. 1994; Van Wagoner et al. 1997; Nattel et al. 2007; Christ et al. 2008; Workman et al. 2008; Workman et al. 2009; Schotten et al. 2011; Dobrev et al. 2012). Furthermore, some drugs used in the treatment of AF also inhibit I_TO (Wang et al. 1995; Varro et al. 1996; Gross & Castle, 1998; Marshall et al. 2012). The rapid activation of this current means that its reduction, by affecting membrane voltage (V_m) early during repolarisation, has knock-on (secondary) effects on other voltage-gated currents, e.g. L-type Ca²⁺ current (I_CaL) and resultant Ca²⁺-induced Ca²⁺ release (CICR) and Na⁺/Ca²⁺-exchanger current (I_Na/Ca) (Sah et al. 2003) with inevitable complex, species- and pathology-dependent influences on the plateau and APD₉₀.

However, it is unclear what effect I_TO reduction has on atrial APs in humans and most other mammals, because of the non-specificity of the best available I_TO blocker, 4-aminopyridine (4-AP). This drug additionally blocks ultra-rapid delayed rectifier K⁺ current (I_Kur) (Wang et al. 1993), and although atrial I_Kur may be small in some species, e.g. rabbit (Giles & Imaizumi, 1988; Lindblad et al. 1996), its large magnitude in human atrium (Wang et al. 1993; Van Wagoner et al. 1997; Christ et al. 2008) necessitates the use of other tools for studying effects of I_TO reduction. Mathematical modelling is a powerful alternative, but yielded variable results depending on the algorithm used and intrinsic AP shape, and on constraining electrophysiological parameters (Courtemanche et al. 1998; Nygren et al. 1998; Zhang et al. 2005; Grandi et al. 2011; Marshall et al. 2012). Moreover, such models can only incorporate known experimental data. The dynamic-clamp technique (Wilders, 2006), however, provides the opportunity to reduce I_TO selectively in live atrial cells. A voltage- and time-dependent current based on a mathematical model of I_TO is calculated in real-time as a function of the cell's V_m, and injected into the cell with opposite polarity during AP recording, thus cancelling out defined fractions of I_TO. Dynamic-clamping has previously been used to alter ventricular I_TO or I_CaL (Sun & Wang, 2005; Dong et al. 2006; Dong et al. 2010; Madhvani et al. 2011) and a stretch-activated current in rat atrial cells (Wagner et al. 2004), but not so far to alter I_TO in atrial cells. No ion current has been studied previously using dynamic-clamping in human atrial cells.

The aim, therefore, was to investigate effects of selective, graded changes in I_TO on AP shape and APD, and on afterdepolarisations or abnormal automaticity in the presence of β-adrenergic-stimulation, using the dynamic-clamp technique in human and rabbit atrial cells.

Methods

Ethical approval

Procedures and experiments involving human atrial cells were approved by the West of Scotland Research Ethics Service (REC: 99MC002). Written, informed consent was obtained from all patients. The study conformed with the Declaration of Helsinki. Procedures and experiments involving rabbit atrial cells (UK Project Licence No. PPL60/40206) were approved by Glasgow University Ethics Review Committee. All procedures and experiments conformed to the principles of UK regulations, as described in Drummond (2009).

Atrial cell isolation

Cells were isolated from rabbit and human atrial tissues, as described previously (Workman et al. 2000; Workman et al. 2001). Rabbits (New Zealand White, male, age 17.2 ± 1.0 weeks, weight 3.1 ± 0.1 kg; n = 24) were humanely killed by intravenous injection of anaesthetic (100 mg kg⁻¹ pentobarbital sodium) and removal of the heart, which was retrogradely perfused via the aorta for enzymatic dissociation of left atrial cells (Workman et al. 2000). Right atrial appendage tissues were obtained from 14 adult patients (59.4 ± 2.6 years; 13 male/1 female) undergoing cardiac surgery (79%: coronary artery bypass graft; 21%: aortic valve replacement). All patients were in sinus rhythm with no history of AF; 79% had angina, 57% had hypertension, 43% had myocardial infarction, 21% had mild/moderate left ventricular systolic dysfunction (LVSD), 79% had no LVSD. Patients were undergoing treatment with the following cardiac drugs: β-blocker (36%), ACE inhibitor/angiotensin receptor-blocker (36%), Ca²⁺ channel blocker (29%), statin (79%), nitrate (36%), nicorandil (29%). Cells were isolated by the chunk technique (Workman et al. 2001), and stored for ≤9 h at ∼20°C in a low [Ca²⁺] solution containing (mm): NaCl (140), KCl (4), CaCl₂ (0.18), MgCl₂ (1), glucose (11) and Hepes (10); pH 7.4.

Conventional whole-cell-patch clamp technique

Cells were superfused at 35–37°C with the above solution except that [Ca²⁺] was 1.8 mm. Microelectrodes (1.5–5 MΩ) contained (mm): potassium aspartate (130), KCl (15), NaCl (10), MgCl₂ (1), Hepes (10), EGTA (0.1); pH 7.25. Membrane currents and APs were stimulated and recorded by whole-cell-patch clamp, with an AxoClamp 2B amplifier (Axon Instruments, Union City, CA, USA) and WinWCP or WinEDR software (J. Dempster). I_TO was stimulated by voltage-clamping with the activation and inactivation protocols in Fig. 1A and C. Pulse frequency was 0.1 Hz for rabbit, 0.33 Hz for human. I_TO amplitude was calculated as peak outward minus end-pulse current, thus avoiding the slowly inactivating I_Kur and other slowly or non-inactivating current components that contribute to total outward current in human atrial cells (Christ et al. 2008). Extracellular Cd²⁺ (0.2 mm) blocked I_CaL. Trains of APs were stimulated at 50–600 beats min⁻¹ by current-clamping in bridge mode with 3–5 ms pulses, 1.2–1.5× threshold. Most human atrial cells required a small (0.87 ± 0.19 pA pF⁻¹, n = 18) constant hyperpolarising current to gain ∼−80 mV resting V_m, as previously (Le Grand et al. 1994; Bénardeau et al. 1996; Workman et al. 2001).

Figure 1. Comparison between live and simulated atrial I_TO in rabbit and human.

Open symbols: rabbit data; filled symbols: human. Continuous lines: live atrial cell data; dashed lines: current generated by dynamic-clamp I_TO models. A–D, steady-state voltage dependence: I_TO activation, using voltage-clamp protocol shown in A, resulting in currents in Ba–d; I_TO inactivation with protocol shown in C, resulting in currents shown in Da–d. E–H, voltage dependence of I_TO activation and inactivation in live atrial cells (n = 8–15, 5–6 rabbits; 10–12 cells, 6–7 patients) or from single simulation runs. I_TO steady-state activation (E) and inactivation (F) curves constructed by dividing peak I_TO at each voltage by driving force to give conductance, G, and dividing G by G_max. I_TO activation (G) and inactivation (H): time constants estimated from exponential functions fitted to activation and inactivation time courses; half-activation and inactivation potentials, V_0.5, and slopes, minimum and maximum time constants, τ_min and τ_max, estimated by fitting Boltzmann functions; values shown with corresponding curve labels.

Dynamic-clamp technique

The dynamic-clamp technique was used in model-clamp configuration (Wilders, 2006) to simulate systematic changes in peak I_TO conductance, in rabbit and human live atrial cells. I_TO was modelled as a simple Hodgkin–Huxley voltage-activated current based upon an activating parameter, m, and an inactivating parameter, h, using the following equations, based on (Nygren et al. 1998).

(1)

(2)

(3)

(4)

(5)

(6)

(7)

The voltage dependence of the steady-state activation and inactivation parameters (m_∞, h_∞), activation and inactivation time constant (τ_m, τ_h) and peak I_TO conductance (G_max) were derived from the average peak current and steady-state activation and inactivation curves, measured in groups of rabbit and human atrial cells (Fig. 1).

The dynamic-clamp was custom-built using an Analog Devices ADuC 7024 analog microcontroller (Analog Devices Inc., Norwood, MA, USA), incorporating a 40 MHz ARM7 microprocessor and 12 bit A/D and D/A converters. The I_TO model, executing on the microcontroller, was written in C and the device controlled from a host computer via an RS232 interface. V_m was monitored and dynamic-clamp current output injected into the AxoClamp 2B stimulus current pathway at precisely timed 100 μs intervals.

Statistics

Electrophysiological data are expressed as means ± SEM. Continuous data were compared using two-sided, two-sample paired Student's t test; categorical data using a χ²-test with Yates's correction. P < 0.05 was regarded as statistically significant. All statistical and curve fitting analyses were done using Graphpad Prism 5.00 (GraphPad Software Inc., La Jolla, CA, USA).

Results

Comparison of live and simulated atrial I_TO characteristics

I_TO showed characteristic rapid activation and decay (Fig. 1Ba for rabbit; Fig. 1Bc for human). A large non-inactivating current was present in human atrial cells, reflecting mainly I_Kur (Wang et al. 1993). Mean voltage dependence of I_TO steady-state activation (Fig. 1E) and inactivation (F) differed between rabbit and human. I_TO G_max was 24.4 nS (0.34 nS pF⁻¹) in rabbit, and 9.9 nS (0.12 nS pF⁻¹) in human atrial cells. The time courses of I_TO activation and inactivation were mono-exponential, and mean τ-voltage relationships also differed between species (Fig. 1G and H). Traces in Fig. 1B and D (dashed symbols) are dynamic-clamp (i.e. simulated) currents injected into a simple resistor (500 MΩ)–capacitor (33 pF) circuit (Patch-1 model cell, Axon Instruments) during voltage-clamp with the protocols in Fig. 1A and C. I_TO was modelled as a fully inactivating current (Dong et al. 2006), so Fig. 1Bd and Dd do not contain the non-inactivating, I_Kur component. There was generally good agreement between live and simulated I_TO in both species for I_TO waveform, voltage and time dependence curves, and curve fit values (Fig. 1E–H). The largest difference was in the steady-state voltage dependence of I_TO inactivation in rabbit (Fig. 1F), reflected by a 3.9 mV difference in V_0.5.

Effects of changing I_TO by dynamic-clamp on rabbit atrial cell action potentials

Action potentials from rabbit atrial cells had prominent phase 1 repolarisation and low amplitude plateau (Fig. 2Aa; control). Partial I_TO inhibition, by dynamic-clamp subtraction of a small (5 or 10 nS) peak I_TO conductance (Fig. 2Ab) slowed phase 1, elevated the plateau, and moderately increased APD (Fig. 2Aa). Twenty nanosiemens I_TO subtraction, reflecting ∼80%I_TO block, markedly slowed phase 1 and elevated the plateau, and prolonged APD, particularly at early-to-mid repolarisation (APD_30-50). Late repolarisation (e.g. at ≥100 ms; labelled) was also prolonged by I_TO subtraction, despite negligible or zero current at ≥100 ms. At approximately full I_TO block (25 nS), APs had a prolonged, ‘spike & dome’ shape (Fig. 2Aa). Further (supra-physiological) I_TO subtraction (30 nS) produced no extra APD increase. Addition of I_TO in this cell (downward deflections in Fig. 2Ab) enhanced phase 1 and depressed the plateau (e.g. 20 nS addition in Fig. 2Aa), and moderately shortened APD at all levels of repolarisation. Mean data (Fig. 2B) revealed a non-linear relationship between I_TO and APD, particularly APD_20-50 due to prominent plateau elevation by I_TO subtraction. Action potential maximum upstroke velocity (V_max) was unaffected by I_TO change (Fig. 2B). Full I_TO block (24.4 nS I_TO subtraction) significantly increased mean APD_20-90, and APD_{−64 mV} (reflecting cellular effective refractory period, ERP; Workman et al. 2000), and had no effect on V_max (Fig. 2C). Absolute APD increase was greatest at APD₅₀ (Fig. 2D), consistent with the plateau elevation. Nevertheless, APD₉₀ and APD_{−64 mV} were also increased, by ∼20 ms. Relative (%) APD increase by I_TO block was markedly greater at mid than late repolarisation (Fig. 2E).

Figure 2. Effects of decreasing and increasing I_TO conductance by dynamic-clamp on rabbit atrial cell APs.

A, superimposed action potentials (APs; a) and corresponding dynamic-clamp (D-C) currents (b) recorded from a single representative atrial cell, without D-C (Control), or with D-C subtraction or addition of I_TO over a range of peak conductances (labelled). D-C subtracted I_TO is +ve, because +ve charge is injected into the cell to compensate for the loss of +ve charge via native I_TO. Stimulation rate = 75 beats min⁻¹ to permit sufficient time for I_TO reactivation (Giles & Imaizumi, 1988; Workman et al. 2000). P: current pulse (3 ms, ∼1.2× threshold) to stimulate APs. B, relationship between I_TO conductance (–ve Δ peak I_TO = D-C subtraction of I_TO) and APD_20-90, APD_{−64 mV} and V_max; n = 16 cells, 6 rabbits. Vertical arrow indicates 24.4 nS I_TO subtraction, reflecting full I_TO block. Effect of full I_TO block on APDs and V_max shown by comparing control (open bars) with 24.4 nS I_TO subtraction (filled; *P < 0.05 vs. control, NS: not significant) (C), or as absolute (D) or relative (E) APD change.

Effects of changing I_TO by dynamic-clamp on human atrial cell action potentials

Human atrial cell APs were typically type 3 (low- or no-dome), with prominent phase 1 in control (Fig. 3Aa). A 50% inhibition of I_TO, by 5 nS dynamic-clamp subtraction (Fig. 3Ab), slowed phase 1 and moderately elevated the plateau and prolonged late repolarisation; APD_70-90 (Fig. 3Aa). Full I_TO block (10 nS) elevated the plateau further (Fig. 3Aa), but less so than in rabbit (Fig. 2Aa) and further increased APD. The APD increase at ≥150 ms (labelled) occurred despite negligible or zero current at that time. Extra I_TO subtraction (15–20 nS; supra-physiological) caused additional increase in late repolarisation and, by contrast with rabbit (Fig. 2A), little further plateau elevation. Dynamic-clamp addition of I_TO accelerated phase 1, lowered the plateau and decreased late APD (Fig. 3A). Mean APD changes indicated that within the physiological range of peak I_TO conductance (vertical dashed line in Fig. 3B), I_TO subtraction lengthened APD, particularly APD_70-90 and APD_{−61 mV} (reflecting cellular ERP; Workman et al. 2001), and I_TO increase had the opposite effects (Fig. 3B). V_max was unaltered by I_TO change (Fig. 3B). Figure 3C shows that full I_TO block (9.9 nS subtraction) significantly increased APD at each repolarisation level. This was most prominent in absolute terms (Fig. 3D) for APD_70-90 and APD_{−61 mV}; relative (Fig. 3E) APD increase was similar at each level.

Figure 3. Effects of decreasing and increasing I_TO by dynamic-clamp on human atrial cell APs.

A, superimposed APs (a) and corresponding D-C currents (b) recorded from a single representative human atrial cell, without D-C (Control), or with D-C subtraction (sub) or addition of I_TO over a range of peak conductances. Rate = 75 beats min⁻¹. P: current pulse (5 ms, ∼1.5× threshold) to stimulate APs. B, relationship between I_TO conductance and APDs and V_max; n = 18 cells, 6 patients. Vertical dashed arrow indicates 9.9 nS I_TO subtraction, reflecting full I_TO block. Effects of full I_TO block shown by comparing control (empty bars) with 9.9 nS I_TO subtraction (filled; *P < 0.05 vs. control) (C), or as absolute (D) or relative (E) APD change.

I_TO reduction prolongs late repolarisation mainly by elevating action potential plateau

Figure 4Aa and b shows that interrupting the subtraction of I_TO late during phase 3 (∼APD₇₀), i.e. when current was small or zero but APD increase by I_TO reduction was large, caused negligible deviation in the course of subsequent repolarisation: compare trace F with I, and see Fig. 4Aa inset. This was confirmed by subtracting I from F (Fig. 4Ac) and indicated that any residual (non-inactivated) I_TO present after APD₇₀ contributed little to the APD increase over control (I-C; Fig. 4Ad) during that time. When I_TO subtraction was interrupted at the end of phase 1 (Fig. 4B), i.e. early during repolarisation but when the highest amplitude and a large portion of I_TO had already occurred (current leftwards of vertical line in Fig. 4Bb), repolarisation rapidly deviated from its course of a slowed phase 1, to follow control: Fig. 4Ba and inset. This was confirmed by the small magnitude I-C (Fig. 4Bd) and indicated that the marked plateau elevation and APD₇₀ increase caused by I_TO reduction (F vs. C in Fig. 4Ba) was not a consequence of the I_TO reduction during phase 1. Interrupting I_TO subtraction midway through the plateau (∼APD₆₀), i.e. at a point (vertical line in Fig. 4Ca) before which I_TO reduction had markedly elevated the plateau, caused a moderate deflection in the course of subsequent repolarisation. The magnitude of F-I rightwards of the vertical line in Fig. 4Cc was larger than in 4Ac, indicating that a residual I_TO block at ∼APD₆₀ contributed, albeit to a small degree, to the APD increase during subsequent repolarisation. However, the substantially elevated plateau and prolonged APD observed well beyond the point of interrupting I_TO subtraction, i.e. between APD_60-90 (I-C rightwards of vertical line in Fig. 4Cd), showed that the major influence of I_TO to prolong APD during that time had occurred earlier, during the plateau and after the end of phase 1, i.e. at around APD_50-60.

Figure 4. I_TO reduction prolongs late repolarisation mainly by elevating AP plateau.

Effect on human atrial AP late repolarisation of interrupting D-C subtraction of I_TO (by temporarily switching off D-C current injection) during late repolarisation (AP phase 3; A), early repolarisation (phase 1; B), and mid repolarisation (phase 2; C), in a single cell. Rate = 75 beats min⁻¹. Top panels (a) show superimposed, representative APs recorded without D-C subtraction of I_TO (control: C), or with 9.9 nS I_TO subtraction for the full (F) duration of the AP, or with D-C subtraction of I_TO interrupted (I) at the point indicated by vertical dashed lines. Insets show magnified portion of APs close to point of interruption. Panels b show corresponding superimposed D-C subtraction currents. Inset in B, example of non-superimposed D-C-interrupted trace (* Point of interruption). Panels c and d, subtraction of AP trace I from F, and C from I (as in corresponding top panels).

Production of spontaneous depolarisations in rabbit atrial cells: evidence for DAD component

Spontaneous APs or subthreshold transient depolar isations >3 mV (‘cellular arrhythmic depolarisations’; CADs; Redpath et al. 2006), which represent afterdepolarisations and/or abnormal automaticity (Redpath et al. 2006), occurred during 6 s pauses between AP trains, in rabbit atrial cells (Fig. 5). At BCL 800 ms (used to investigate I_TO change on APD), such post-train CAD(s) occurred in 5/12 control cells and in 12/12 cells exposed to the β-receptor agonist isoproterenol (ISO) or ISO+elevated [Ca²⁺]_o (P < 0.05). In a representative cell in which post-train CADs occurred with ISO (Fig. 5A), rapid stimulation (BCL 100 ms) resulted in a high number and frequency of these CADs. Progressive reduction in stimulation rate resulted in fewer post-train CADs, of lower frequency at each rate, and an increase in the time between the last stimulated AP and the first CAD (coupling interval). The lowest rate resulted in a subthreshold post-train CAD. Figure 5B shows a similar coupling interval increase in a different cell. Mean data (Fig. 5C) show a linear relationship between BCL and both coupling interval (panel a) and amplitude (b) between BCL 100–400 ms, indicating a delayed afterdepolarisation (DAD) component to these CADs.

Figure 5. Production of spontaneous depolarisations in rabbit atrial cells: evidence for DAD component.

A, APs (upper trace) recorded from a rabbit atrial cell during stimulation with trains of current pulses (lower trace) delivered at BCL shown, in the presence of isoproterenol (ISO) 0.1 μm. *Subthreshold CAD. B, APs in a different cell at magnified time scale at short BCLs using same protocol as A. C, BCL vs. mean (± SEM) coupling interval (a) and amplitude (b) of post-train AP or CAD with 0.1 μm ISO or 0.1 μm ISO+3.6 mm[Ca²⁺]. n = 10–12 cells, 5 rabbits.

I_TO reduction potentiates β-stimulation-induced CADs in rabbit atrial cells

Effects of dynamic-clamp subtraction of I_TO on CADs, in the presence of ISO, were investigated in rabbit atrial cells (Fig. 6). In 24 cells in which no CAD occurred in control (e.g. Fig. 6A, top trace), ISO produced CAD(s) within the pauses between AP trains in 13 cells (e.g. Fig. 6A, lower trace), thus significantly increasing the incidence of post-train CADs (Fig. 6A, bar graph). ISO also elevated the AP plateau, including that of the CAD (Fig. 6A). ISO had no effect on peak I_TO: 36.3 ± 4.8 pA pF⁻¹ at +50 mV in control vs. 34.7 ± 4.3 pA pF⁻¹ in ISO (P > 0.05; n = 8 cells, 4 rabbits). In the cells in which ISO did not produce post-train CADs, I_TO subtraction (10 nS; ∼40%I_TO block) during continued ISO elevated the plateau further than ISO alone, and significantly increased the incidence of post-train CADs (Fig. 6B). Furthermore, full I_TO block produced post-train CADs in half of the cells in which partial I_TO block did not. ISO also produced CADs within AP trains, typically stabilising as rapid, continuously and irregularly firing threshold depolarisations (Fig. 6C). In the 15/24 cells in which these did not occur with ISO, subsequent I_TO reduction or block significantly increased the incidence of within-train CADs, again associated with marked plateau-elevation (Fig. 6D).

Figure 6. I_TO reduction potentiates β-stimulation-induced CADs in rabbit atrial cells.

Effects of ISO 0.1 μm with or without D-C subtraction of I_TO on the incidence of CADs. Upper traces of trace pairs: AP trains (75 beats min⁻¹) followed by 2 s pauses. Lower traces: corresponding currents: AP stimulus pulses with or without D-C reduction of I_TO (of labelled conductance and duration). Bar graphs: incidences of CADs following (post-train) or during (within-train) AP trains; values within bars: cell n (≤7 rabbits), *P < 0.05. A, effect of ISO 1.5 min (max. time needed for stable effect on APs and CADs) on post-train CADs, without D-C subtraction of I_TO. B, in the cells in which ISO did not produce CADs in A, effects of 10 nS I_TO subtraction (3rd trace pair) and/or 24.4 nS I_TO subtraction (bar graph) on post-train CADs. C and D, corresponding changes to within-train CADs.

Production of EADs by I_TO reduction plus β-stimulation

Transient, small amplitude depolarisations occurred during the plateau of several APs under I_TO subtraction plus ISO, in 1/7 cells (Fig. 7) in which ISO alone did not produce post-train CADs but subsequent 10 nS I_TO subtraction did. These plateau phase depolarisations occurred in 0/24 cells treated with ISO alone, appeared ∼20 s after I_TO subtraction (Fig. 7Ad), persisted during each AP train (e.g. Fig. 7Ae–h), and disappeared upon ceasing I_TO subtraction (Fig. 7Ah). One occurred on the plateau of a post-train CAD (Fig. 7Ag). Over the 2 min period of 10 nS I_TO subtraction plus ISO, 117 APs were stimulated in this cell. Of those, 47 (40%) had a plateau phase depolarisation, and no CAD (threshold or subthreshold) occurred between stimulated APs (that could suggest DADs or abnormal automaticity), identifying these depolarisations as early afterdepolarisations (EADs). Figure 7B shows the EADs as distinct from the AP dome. Full (24.4 nS) I_TO subtraction caused a double EAD and a ‘late phase’ EAD (Fig. 7Bb).

Figure 7. Production of EADs by I_TO reduction plus β-stimulation.

A, upper traces of trace pairs: AP trains (75 beats min⁻¹) followed by 2 s pauses; lower traces: corresponding currents, recorded in a rabbit atrial cell before (a) and after (b) ISO 0.1 μm and during subsequent D-C subtraction of I_TO (c–h). •: EAD. S, subthreshold CAD; T, threshold CAD. B, expanded portions of panel A traces: a, 4th, 5th and 6th APs from Ad; b, 1st, 2nd and 3rd APs from Ag.

Suppression of CADs: by interrupting I_TO reduction, by adding I_TO, or by β₁-antagonism

Suppression or abolition of CADs that were caused by ISO or concurrent I_TO reduction was demonstrated using three interventions (Fig. 8). First, interrupting dynamic-clamp subtraction of I_TO in the presence of ISO caused a rapid reduction in the number and frequency of CADs in most cells (e.g. Fig. 8A, 2nd trace pair), no further CADs with continued ISO, and accompanying plateau-depression (e.g. Fig. 8A, bottom trace pair). Second, increasing I_TO (by ∼40%), in cells in which ISO alone produced CADs, typically rapidly and continuously suppressed these CADs, accompanied by plateau depression (e.g. Fig. 8B, bottom trace pair), significantly reducing their incidence (Fig. 8B, bar graph). Reversibility was assessed in four cells: interrupting I_TO addition caused CADs to return in each. Third, β₁-receptor antagonism with atenolol abolished CADs caused by decreased I_TO in the presence of ISO (Fig. 8C). In each of four cells in which ISO did not produce CADs but subsequent I_TO subtraction did (e.g. Fig. 8C, top 4 trace pairs) atenolol, in the continued presence of ISO and I_TO reduction, abolished these CADs (e.g. Fig. 8C, bottom trace pair), significantly decreasing their incidence (Fig. 8C, bar graph).

Figure 8. Suppression of CADs by interrupting I_TO reduction, by adding I_TO, or by β₁-antagonism.

Upper traces of trace pairs: AP trains (75 beats min⁻¹) followed by 2 s pauses; lower traces: corresponding currents, recorded in rabbit atrial cells. Bar graphs show incidences of CADs (combined post- and within-train); values within bars: cell n, *P < 0.05. A, effect of interrupting D-C subtraction of I_TO on CADs in cells (3 rabbits) in which ISO did not produce CADs but concurrent D-C subtraction of I_TO (10 or 24.4 nS) did. B, effect of D-C addition of I_TO on CADs in cells (4 rabbits) in which ISO produced CADs. C, effect of atenolol (1 μm) on CADs in cells (3 rabbits) in which ISO did not produce CADs but concurrent D-C subtraction of I_TO did.

Discussion

The key findings are that selective I_TO reduction, simulated in live atrial cells for the first time to our knowledge using dynamic-clamping, elevated the action potential plateau and prolonged late repolarisation (APD_70-90) in human and rabbit, and promoted afterdepolarisations or abnormal automaticity under β-adrenergic-stimulation.

Plateau elevation by I_TO subtraction was most prominent in rabbit, and terminal repolarisation (APD₉₀) was increased by ∼20% in both species. It was previously unclear what effect I_TO reduction would have on human atrial APD₉₀, because the best available I_TO blocker, 4-AP, inhibits I_Kur at ≥40-fold lower concentration than I_TO (Wang et al. 1993), human atrial I_Kur is large (Wang et al. 1993; Van Wagoner et al. 1997; Christ et al. 2008), and selective I_Kur reduction affects human atrial APD₉₀ (Wang et al. 1993; Wettwer et al. 2004). However, I_Kur is notoriously difficult to separate completely from I_TO, and we cannot exclude the possibility that our I_TO model included a component of rapidly inactivating I_Kur, as reported by Christ et al. (2008). Nevertheless, we were able to exclude the large, slowly inactivating I_Kur and other non-inactivating current components (Christ et al. 2008) and, therefore, use the dynamic-clamp as a novel intervention to improve our understanding of the contribution of I_TO to human atrial cell APD and arrhythmia mechanisms.

Studies of effects of selective I_Kur block have shown some similarities to, and important differences from, the present effects of I_TO block. Thus, in atrial trabeculae from patients in sinus rhythm, micromolar 4-AP elevated the plateau, but, by contrast with I_TO block, here, shortened APD₉₀, probably from secondary increase in delayed rectifier K⁺ currents (Wettwer et al. 2004). Increase in both plateau and APD₉₀, however, have also been reported in human atrial cells (Wang et al. 1993). In canine atrial cells, C9356, an I_Kur(d) blocker, increased the plateau, APD₂₀ and APD₅₀, but with negligible effect on APD₉₀ (Fedida et al. 2003). Combined I_TO/I_Kur block in human atrial cells, with millimolar 4-AP, increased both plateau and APD₉₀, by ∼30% (Workman et al. 2001). The novel I_Kur/I_TO/I_KACh blocker, AVE0118, had negligible effect on APD₉₀ (Wettwer et al. 2004; Schotten et al. 2007; Christ et al. 2008), but elevated the plateau (Wettwer et al. 2004; Schotten et al. 2007; Christ et al. 2008) and increased systolic [Ca²⁺]_i and contractility (Schotten et al. 2007). Furthermore, Schotten et al. (2007) showed, using various AP-clamping techniques, that such change in the AP shape was responsible for this drug's positive inotropic action. Thus, in the absence of AVE0118, atrial cell fractional shortening was markedly increased either by elevating the plateau amplitude or by prolonging its duration (without changing APD₉₀), or indeed by imposing the AP waveform recorded under AVE0118.

Previous dynamic-clamp studies of I_TO used ventricular cells, from dog or guinea-pig. In epicardial cells, which have a moderate phase 1 ‘notch’ due to I_TO, an I_TO subtraction sufficient to eliminate this notch did not affect subsequent repolarisation (Sun & Wang, 2005). In endocardial cells, which had small or no I_TO or notch, I_TO addition sufficient to produce a notch also lacked effect on repolarisation (Dong et al. 2006; Dong et al. 2010). However, ventricular AP shape, some ion currents, and [Ca²⁺]_i handling differ markedly from atrial, including in human (Grandi et al. 2011), and the relatively large atrial I_TO and phase 1 in human (Grandi et al. 2011) and rabbit (Giles & Imaizumi, 1988) are likely to account for the observed relatively strong atrial APD response to I_TO decrease.

The magnitude of the human atrial APD₉₀ increase with I_TO subtraction (i.e. 9% with 50%I_TO subtraction) and block (17% with 100%I_TO subtraction) is consistent with several mathematical models. Thus, in the Grandi model (Grandi et al. 2011), 50 and 100%I_TO decrease lengthened APD₉₀ by 9 and 19%, respectively. In the Nygren model (Nygren et al. 1998), 90%I_TO decrease also lengthened APD₉₀, by 18%. By contrast, in the Courtemanche–Ramirez–Nattel (CRN) model (Courtemanche et al. 1998), 50 or 90%I_TO decrease shortened APD₉₀. Another study, based on the CRN model, also reported APD₉₀ shortening (Zhang et al. 2005). However, the CRN model generates a type 1 (spike and dome) AP, whereas the present APs and those generated by the other models (Nygren et al. 1998; Grandi et al. 2011) are type 3. Moreover, modification of the CRN model to generate type 3 s (Marshall et al. 2012) reversed the effect of I_TO decrease: a 41% decrease lengthened APD₉₀ by 9% (Marshall et al. 2012), again consistent with present data.

The increased APD₉₀ and APD_{−61 mV} by I_TO reduction has implications for atrial reentry. APD_{−61 mV} estimates cellular ERP, because −61 mV is the mean V_m from which partially refractory APs took off in human atrial cells (Workman et al. 2001). Since leading circle reentry wavelength (λ) = ERP × conduction velocity (θ), then increased APD_{−61 mV}, with the unchanged V_max (a determinant of θ), has the potential to increase λ and thus inhibit reentry (Workman et al. 2011). Furthermore, I_TO reduction may terminate spiral wave reentry, by plateau elevation alone, according to a CRN-based mathematical model of human atrial two-dimensional reentry (Pandit et al. 2005). Atrial electrophysiological remodelling in chronic AF and predisposing pathologies such as atrial dilatation and LVSD invariably involves I_TO reduction (Le Grand et al. 1994; Van Wagoner et al. 1997; Nattel et al. 2007; Christ et al. 2008; Workman et al. 2008; Workman et al. 2009; Schotten et al. 2011; Dobrev et al. 2012). However, several other currents remodel also, which are likely to have more bearing on atrial ERP than I_TO, despite secondary ionic effects of I_TO reduction, making the consequences of I_TO reduction difficult to predict. For example, increased inward rectifier K⁺ current (I_K1 and I_KACh) and altered [Ca²⁺]_i handling are likely to be the principle mechanisms of ERP-shortening in chronic AF (Dobrev et al. 2005; Cha et al. 2006; Workman et al. 2008; Dobrev et al. 2012; Voigt et al. 2012). Human atrial ERP increase, by chronic β-blockade (Workman et al. 2003), was associated with decreased I_TO and I_K1 (Marshall et al. 2012), and mathematical modelling suggested a small contribution from I_TO (Marshall et al. 2012). It is conceivable that ERP increase by some other drugs used to treat AF, e.g. flecainide, propafenone and amiodarone, could also involve their observed inhibition of I_TO (Wang et al. 1995; Varro et al. 1996; Gross & Castle, 1998), in addition to their recognised effects on other ion currents and [Ca²⁺]_i handling (Dobrev et al. 2012). Since only I_TO decrease was detected in patients with LVSD, associated with atrial ERP shortening (Workman et al. 2009), the present data indicate that other, as yet unidentified, ionic mechanisms were responsible for the ERP shortening in that study (Workman et al. 2009).

Critically timed interruption of I_TO subtraction during human atrial repolarisation revealed that the present APD₉₀ prolongation was not caused by I_TO reduction during phase 1, nor by reduction of any substantial residual I_TO flowing during late repolarisation (APD₇₀ or beyond), but rather by the effect of I_TO reduction between these points, i.e. mid plateau (APD_50-60), to elevate the plateau. The mid plateau-to-APD₉₀ region is substantially maintained by forward mode (depolarising) I_Na/Ca from Ca²⁺ extrusion after CICR: I_Na/Ca block with Li⁺ approximately halved APD_{−60 mV} (Bénardeau et al. 1996). In mouse ventricular cells, which also have type 3 APs, genetic I_TO reduction slowed I_CaL inactivation and increased net Ca²⁺ influx via I_CaL, and markedly increased systolic [Ca²⁺]_i (Sah et al. 2002). The increased net I_CaL in that study is likely to have contributed substantially to the increased [Ca²⁺]_i, but increased reverse-mode I_Na/Ca (Ca²⁺ influx) may also contribute, since plateau elevation may shift V_m towards E_Na/Ca (Sah et al. 2003). Indeed, support for such a contribution from reverse-mode I_Na/Ca was provided by the AP-clamp studies in canine atrial cells from Schotten et al (2007). They showed that, in control cells, imposing the AP waveform recorded under AVE0118 (a drug which markedly elevated the plateau by inhibiting I_Kur and I_TO, and increased systolic [Ca²⁺]_i and contractility) had negligible effect on net Ca²⁺ influx via I_CaL despite also increasing contractility, yet inhibition of reverse-mode I_Na/Ca (using KBR7943 or elimination of pipette Na⁺) abolished this AP-clamp-induced positive inotropic effect (Schotten et al. 2007). Irrespective of the mechanism of systolic [Ca²⁺]_i increase resulting from selective reduction in I_TO, a consequent increase in depolarising I_Na/Ca is a likely contributory mechanism for the effect of I_TO decrease to prolong APD₉₀ in the present study. Furthermore, I_Na/Ca is the major current responsible for DADs and abnormal automaticity, especially during adrenergic stimulation, which increases I_CaL amplitude, sarcoplasmic reticular Ca²⁺ content, and thus systolic [Ca²⁺]_i (Workman, 2010).

Afterdepolarisations (DADs and EADs) and abnormal automaticity have complex, partially shared mechanisms of generation and are difficult to distinguish, hence their grouping here as CADs (Redpath et al. 2006). We found a DAD component in rabbit atrial cells under β-stimulation, since post-train CADs showed a rate dependence of coupling interval and amplitude characteristic of DADs (Johnson et al. 1986). Nevertheless, ISO produced a variety of spontaneous activities, during and after trains, so we did not restrict investigation of I_TO change to DADs. ISO was used at E_max concentration for I_CaL (Redpath et al. 2006), and had no effect on I_TO. When ISO alone did not produce CADs, it nevertheless elevated the plateau. Subsequent I_TO subtraction produced immediate (within 1 beat) further plateau elevation, and CADs followed either within a few beats or after several AP trains, their incidence increasing with strength of I_TO subtraction. It is reasonable to suppose that these CADs resulted from the marked plateau elevation from concurrent I_TO subtraction and β-stimulation, each of which increases [Ca²⁺]_i (Sah et al. 2002; Dong et al. 2010; Workman, 2010), with consequent [Ca²⁺]_i loading sufficient to produce DADs and/or abnormal automaticity via spontaneous sarcoplasmic reticular Ca²⁺ release and subsequent inward I_Na/Ca. Furthermore, in one cell, ISO plus I_TO subtraction produced EADs, distinguished by their consistent, regular appearance during phase 2, with no DADs or abnormal automaticity. The V_m from which they arose (−25 to −15 mV) centred on ‘window I_CaL’ (where I_CaL activation and inactivation curves overlap; −30 to −10 mV in rabbit atrium; Lindblad et al. 1996), consistent with a recognised mechanism of EAD generation: APD increase at window I_CaL voltages. The contribution of window I_CaL to EADs was studied by dynamic-clamping I_CaL in rabbit ventricular cells (Madhvani et al. 2011). EAD formation was modified by only small changes in I_CaL voltage dependence that did not change [Ca²⁺]_i (Madhvani et al. 2011). However, effects of I_TO change on window I_CaL remain to be studied. Atrial DADs and/or EADs occur in several animal models of congestive heart failure (CHF)-induced AF (Nattel et al. 2007; Schotten et al. 2011; Dobrev et al. 2012). Since these models also feature decreased atrial I_TO and increased APD and/or I_Na/Ca (Nattel et al. 2007; Schotten et al. 2011; Dobrev et al. 2012), it is conceivable that I_TO decrease may contribute to atrial afterdepolarisation formation in CHF. CADs were suppressed, in the present study, by removal of I_TO subtraction, or by dynamic-clamp addition of I_TO. The associated AP (or CAD) plateau depression in each case is consistent with suppression of CADs by opposition of [Ca²⁺]_i loading. CADs were also suppressed by atenolol, used here at therapeutic concentration, 1 μm (de Abreu et al. 2003), probably involving antagonism of [Ca²⁺]_i loading produced by β-stimulation of I_CaL and CICR (Workman, 2010).

Study limitations

(1) We did not study cells from patients with AF, and since chronic AF remodels atrial ion currents and [Ca²⁺]_i handling (Schotten et al. 2011; Dobrev et al. 2012), it may be expected to alter APD responses to I_TO change, as shown for some other currents. For example, block of I_Kur (Wettwer et al. 2004) or I_Kur/I_TO/I_KACh (Wettwer et al. 2004; Christ et al. 2008) increased APD₉₀ in atrial trabeculae from patients with chronic AF, but not sinus rhythm. Chronic AF also increased the propensity for atrial DADs, by increasing diastolic sarcoplasmic reticular Ca²⁺ leak and I_Na/Ca (Voigt et al. 2012), and so could also alter effects of I_TO change on DADs. Whether chronic AF could affect the propensity to EADs, and hence their response to I_TO change, is unclear. (2) Whilst the modelled I_TO generally conformed well to the live data, with a maximum V_0.5 discrepancy of <4 mV, an influence of such a difference on the dynamic-clamped AP cannot be excluded. (3) Rabbit and human atrial I_TO is carried by different pore-forming α-subunits: Kv1.4 and Kv4.3, respectively (Wang et al. 1999), which confer markedly different reactivation kinetics and rate-dependence of I_TO between these species. At the stimulation rate (75 beats min⁻¹) and temperature (35–37°C) used for all the dynamic-clamp experiments, human atrial I_TO would reactivate fully between beats, whereas rabbit atrial I_TO would be ∼50% reactivated (Fermini et al. 1992) due to its relatively slow reactivation kinetics. We did not model this slow reactivation, and our rabbit atrial I_TO model therefore incorporated more ‘human-like’ reactivation kinetics than occur in vivo. Nevertheless, a subtraction of ∼40% of the fully reactivated I_TO significantly increased the incidence of afterdepolarisations in the rabbit atrial cells. (4) The holding current required in human (not rabbit) atrial cells to gain ∼−80 mV resting V_m, although kept constant in each cell and independent of dynamic-clamp current, is nevertheless recognised to shorten APD and depress the plateau. Such effects could lead to an underestimation of the potential for I_TO reduction to produce arrhythmogenic activity in human atrial cells in vivo, but an overestimation of the potential for I_TO addition to prevent such activity. (5) Whole-cell-patch clamp uses fixed intracellular and extracellular solutions, and the isolated cells also lack electrotonic influence from other cells. (6) The dynamic-clamp was used here to mimic the electrical effects of I_TO reduction. However, it is recognised that this technique cannot mimic local [K⁺] changes (Wilders, 2006).

In conclusion, changes in atrial cell action potential shape and durations from selective I_TO modulation, shown here for the first time using dynamic-clamp, have the potential to influence reentrant and non-reentrant arrhythmia mechanisms, with implications for both the development and treatment of AF.

Glossary

4-AP

4-aminopyridine

AF

atrial fibrillation

APD

action potential duration

APD_x

action potential duration at x% repolarisation

BCL

basic cycle length

CAD

cellular arrhythmic depolarisation

CHF

congestive heart failure

CICR

Ca²⁺-induced Ca²⁺ release

CRN

Courtemanche–Ramirez–Nattel

DAD

delayed afterdepolarisation

EAD

early afterdepolarisation

ERP

effective refractory period

h_∞

steady-state inactivation parameter

I_CaL

L-type Ca²⁺ current

I_K1

inward rectifier K⁺ current

I_KACh

acetylcholine-activated K⁺ current

I_Kur

ultra-rapid delayed rectifier K⁺ current

I_Na/Ca

Na⁺/Ca²⁺ exchanger current

ISO

isoproterenol

I_TO

transient outward K⁺ current

I_TOG_max

peak I_TO conductance

LVSD

left ventricular systolic dysfunction

m_∞

steady-state activation parameter

V_0.5

voltage of half activation or inactivation

V_m

membrane voltage

V_max

action potential maximum upstroke velocity

τ_h

inactivation time constant

τ_m

activation time constant

Author contributions

Conception and design of the experiments: A.J.W., J.D., G.L.S., A.C.R. Collection, analysis and interpretation of data: A.J.W., J.D., G.E.M., A.C.R., G.L.S. Drafting the article or revising it critically for important intellectual content: A.J.W., J.D., A.C.R., G.L.S., G.E.M. Final approval of the published version: A.J.W., G.E.M., A.C.R., G.L.S., J.D. All experiments were performed in the laboratory of the British Heart Foundation (BHF) Glasgow Cardiovascular Research Centre, Institute of Cardiovascular and Medical Sciences, University of Glasgow. The authors have no disclosures.