Effects of premature stimulation on HERG K⁺ channels

Abstract

1. The unusual kinetics of human ether-à-go-go-related gene (HERG) K⁺ channels are consistent with a role in the suppression of arrhythmias initiated by premature beats. Action potential clamp protocols were used to investigate the effect of premature stimulation on HERG K⁺ channels, transfected in Chinese hamster ovary cells, at 37 °C.

2. HERG K⁺ channel currents peaked during the terminal repolarization phase of normally paced action potential waveforms. However, the magnitude of the current and the time point at which conductance was maximal depended on the type of action potential waveform used (epicardial, endocardial, Purkinje fibre or atrial).

3. HERG K⁺ channel currents recorded during premature action potentials consisted of an early transient outward current followed by a sustained outward current. The magnitude of the transient current component showed a biphasic dependence on the coupling interval between the normally paced and premature action potentials and was maximal at a coupling interval equivalent to 90% repolarization (APD₉₀) for ventricular action potentials. The largest transient current response occurred at shorter coupling intervals for Purkinje fibre (APD₉₀– 20 ms) and atrial (APD₉₀– 30 ms) action potentials.

4. The magnitude of the sustained current response following premature stimulation was similar to that recorded during the first action potential for ventricular action potential waveforms. However, for Purkinje and atrial action potentials the sustained current response was significantly larger during the premature action potential than during the normally paced action potential.

5. A Markov model that included three closed states, one open and one inactivated state with transitions permitted between the pre-open closed state and the inactivated state, successfully reproduced our results for the effects of premature stimuli, both during square pulse and action potential clamp waveforms.

6. These properties of HERG K⁺ channels may help to suppress arrhythmias initiated by early afterdepolarizations and premature beats in the ventricles, Purkinje fibres or atria.

Ventricular arrhythmias are a major cause of morbidity and mortality, accounting for over 300 000 deaths per annum in the United States (Zipes & Wellens, 1998). Dissection of the molecular genetics of long QT syndrome (Keating & Sanguinetti, 1996) has implicated many ion channels, including human ether-à-go-go-related gene (HERG) K⁺ channels, in the pathophysiology of cardiac arrhythmias (Curran et al. 1995; Sanguinetti et al. 1995). There has therefore been considerable interest in understanding how HERG K⁺ channels normally provide protection against arrhythmias (Miller, 1996; Smith et al. 1996; Vandenberg et al. 2001). HERG K⁺ channels exhibit the property, unusual for depolarization-activated K⁺ channels, of inward rectification (Sanguinetti et al. 1995; Smith et al. 1996) due to rapid C-type inactivation (Smith et al. 1996; Spector et al. 1996). This has led to the suggestion that HERG K⁺ channels have a specific role in suppressing arrhythmias initiated by premature beats (Smith et al. 1996).

The voltage dependence of ion channel activation and inactivation has most frequently been characterized using square-pulse voltage clamp techniques. The behaviour of channels during physiological voltage changes, such as those observed during ventricular action potentials, has thence been reconstructed by mathematical modelling (Noble et al. 1998; Viswanathan et al. 1999). A more direct alternative approach has been to use action potential clamp protocols to record ionic currents in response to voltage clamp waveforms resembling in vivo action potentials (Doerr et al. 1990; Shimoni et al. 1992). This approach has the advantage that it is independent of assumptions that are inherent in any modelling approach (Noble et al. 1998). Zhou et al. (1998) and Hancox et al. (1998) have recently shown that the current-voltage relationship of HERG K⁺ channels, heterologously expressed in epithelial cell lines, is significantly different for ventricular action potential waveforms compared with square wave voltage commands. We hypothesized that the complex pattern of activation, inactivation and recovery from inactivation during repolarization of the cardiac action potential would also influence the effects of premature stimulation on HERG K⁺ channel activity.

Therefore the aims of this study were to use physiological voltage waveforms to study the effects of premature stimulation on HERG K⁺ channels and to compare these results with those predicted by a Markov model of HERG K⁺ channels (Kiehn et al. 1999).

METHODS

Cell culture

Chinese hamster ovary (CHO) cells were cultured in Iscove's DMEM (Dulbecco's modified Eagle's medium) with 10 % fetal bovine serum and maintained at 37 °C in 5 % CO₂. Cells, 60-80 % confluent, were cotransfected with 1.5 μg HERG in pIRES-neo (kindly donated by Dr Gail Robertson) and 0.3 μg enhanced green fluorescent protein (eGFP) in pCAGGS (kindly donated by Dr Jun-ichi Miyazaki) using lipofectamine (Gibco BRL). After transfection (36-48 h), coverslips were transferred to a recording chamber on a Nikon Diaphot TMD inverted microscope. Transfected cells were identified by detection of eGFP fluorescence (488 nm excitation, > 503 nm emission) immediately prior to electrophysiological recordings.

Patch clamping

Cells were superfused with normal Tyrode solution (mm): 129 NaCl, 5 sodium pyruvate, 5 sodium acetate, 4 KCl, 1 MgCl₂, 1.8 CaCl₂, 11.1 glucose and 5 Hepes (titrated to pH 7.4 with NaOH) at 37 °C. Patch pipettes with filled resistances of 1-2 MΩ were pulled from borosilicate glass tubing (Clark Electromedical, Reading, UK) using a horizontal puller (Sutter P87, USA) and fire-polished. The internal solution contained (mm): 140 KCl, 1 MgCl₂, 5 MgATP, 10 EGTA and 5 Hepes (pH 7.4 with KOH). The liquid-liquid junction potential between the internal and external solutions was -3 mV, which has not been corrected for. Experiments were conducted at 37 ± 1 °C. The temperature at the cell was maintained by preheating the perfusion solution and by heating the microscope oil immersion lens with a water jacket.

Current recordings were made using an Axopatch 200A or Axopatch 200B amplifier (Axon Instruments, Union City, CA, USA). Capacitance current transients were electronically subtracted. Series resistance was typically 3-4 MΩ which was compensated by ∼70 %. The largest currents were ∼5 nA, which would have resulted in a maximum voltage error of ∼5 mV. Current signals were filtered at 2 kHz and sampled at 5 kHz. All current traces were leak-subtracted off-line, assuming a linear leak in the range from -120 to +60 mV. Acquisition and analysis of data were performed using pCLAMP 6 software (Axon Instruments). Voltage protocols are illustrated in the insets of relevant figures. Action potential waveforms were derived from Oxsoft Heart 4.8 (Noble, 1999). Ventricular and atrial action potentials were scaled to give a resting membrane potential of -85 mV, a maximum overshoot potential of +40 mV and a maximum upstroke velocity of 125 V s⁻¹ before being converted to pCLAMP 6 format for use as voltage clamp waveforms using the DacFile command in pCLAMP 6.

All summary data were analysed using Microsoft Excel and values are expressed as means ±s.e.m.

Modelling

A Markov state model based on that developed by Kiehn et al. (1999) was used to fit the experimental data. The continuous-time Markov state model was phrased as a system of non-autonomous ordinary differential equations along with an algebraic equation representing the conservation of states property of Markov chains. Each state transition is described by a forward rate: α =α₀ exp[z_αV_m/(RT/F)] and a backward rate: β =β₀exp[-z_βV_m/(RT/F)], where R is the universal gas constant, T is the absolute temperature and F is Faraday's constant. The values of the parameters α₀, z_α, β₀ and z_β are given in Fig. 6B. Numerical time integration was computed using a differential algebraic solver (Brenan et al. 1996) by means of a custom program implemented in C++ language on a Microsoft Windows-based personal computer. Initial conditions were computed by solving the steady-state problem for the Markov chain with a direct matrix solution method. The reverse rate of the C₂-I transition was constrained based on all the other rates in the C₂-O-I loop to ensure microscopic reversibility was satisfied in the same manner as in Clancy & Rudy (2001) and Mazhari et al. (2001).

Figure 6. Markov state model for HERG K⁺ channel gating.

A, the model scheme where the rate constants for each transition are of the format: α =α ₀exp[z_αV_m/(RT/F)] and β =β₀exp[-z_βV_m/(RT/F)], as described in Methods. κ_f and κ_b are voltage-independent transition rates. B, rate constants for each equation used in the model. C, model predicted response to double-pulse protocols (as illustrated in Fig. 1A). D, paired endocardial AP stimuli (as illustrated in Fig. 4A).

RESULTS

Response of HERG K⁺ currents to double-pulse protocols

HERG K⁺ channels have unusual kinetics, characterized by slow activation and deactivation but rapid inactivation and recovery from inactivation. The response of HERG K⁺ channels to double-pulse protocols clearly illustrates the significance of the unusual kinetics of HERG K⁺ channels in terms of their ability to pass large transient outward currents in response to premature re-excitation (see Fig. 1A). The amplitude of the transient current following a second depolarization step to +20 mV showed a biphasic dependence on the duration of the interpulse interval, peaking at an interpulse interval of 4 ms (Fig. 1B).

Figure 1. Response of HERG K⁺ channels to double-pulse voltage steps.

A, effect of varying interpulse interval (between 5 and 100 ms; inset between 1 and 9 ms) on reactivation of HERG K⁺ channels. The zero current level is indicated by the dotted line. B, mean ±s.e.m. data for the magnitude of the peak transient outward current response for each interpulse interval (n = 11).

The initial increase in the magnitude of the transient current response as the coupling interval was increased predominantly reflects the rate of recovery from inactivation at -80 mV whereas the subsequent decline in the magnitude of the transient current response predominantly reflects the rate of deactivation at -80 mV. The rates of deactivation and recovery from inactivation are voltage dependent (see Fig. 2), which suggests that the response of HERG K⁺ channels to premature square-pulse stimuli will be determined by the interpulse voltage as well as the interpulse duration. The situation in the intact heart, however, is more complicated due to the complex shape of action potential waveforms as well as the variation in the duration of action potentials in different regions of the heart.

Figure 2. Voltage dependence of the rates of deactivation and recovery from inactivation at 37 °C.

A, time constants for recovery from inactivation and for deactivation were measured from tail currents in the range from -30 to -120 mV following a 3 s depolarization to +20 mV. Rates of recovery from inactivation were measured by fitting single exponential functions to the increasing portion (†) and rates of deactivation by fitting bi-exponential functions to the decaying portion (*) of the tail currents. B, time constants for recovery from inactivation (mean ±s.e.m.; n = 5). The time constants of the rate of recovery from inactivation increased in a sigmoidal manner from 0.43 ms at -100 mV to 1.6 ms at -40 mV. C, time constants for deactivation in the range from -120 to -40 mV (fast: □, n = 5; slow: ▪, n = 5). The time constant of the initial rapid phase of deactivation increased approximately exponentially from 4.3 ± 0.7 ms at -120 mV to 447 ± 54 ms at -40 mV. Similarly, the slow time constant of deactivation increased from 17 ± 3 ms at -120 mV to 2485 ± 309 ms at -40 mV. The percentage of current decay occurring via the fast pathway was 65 ± 13 % at -120 mV, 76 ± 9 % at -100 mV, 77 ± 9 % at -80 mV and 78 ± 11 % at -60 mV.

HERG K⁺ channel activity during cardiac action potential stimuli

HERG-expressing CHO cells showed gradual increases in outward current during the depolarization phase of voltage clamp waveforms whether they were derived from ventricular, Purkinje fibre or atrial action potentials (Fig. 3A). The currents reached a maximum towards the end of repolarization. The voltage at which the current peaked was similar for endocardial, epicardial and Purkinje waveforms (-42.7 ± 0.8, -39 ± 2.4 and -41.5 ± 2.2 mV, respectively; n = 4 for each group). However, the peak current occurred at a significantly higher voltage, -32.6 ± 2.4 mV (n = 4), for the atrial action potential (AP) waveforms (P < 0.05 compared to all the others). The decrease in current following this peak is due to a combination of channel deactivation and a decrease in the electrochemical driving force for K⁺ with recovery of the membrane potential. These two effects were separated by plotting conductance as a function of membrane potential (Fig. 3C). Such an approach removes the effect of a decrease in electrochemical driving force. For ventricular (epicardial and endocardial) action potentials there was a continuous increase in conductance between +30 and -80 mV followed by an abrupt decrease at more negative potentials. It is difficult to interpret this abrupt decrease, as the values are close to the reversal potential for K⁺. In contrast, during Purkinje and atrial action potentials conductance gradually decreased between -65 and -80 mV.

Figure 3. Response of HERG K⁺ channels during different cardiac AP waveforms.

A, typical current responses (continuous current traces) from the same transfected CHO cell for single endocardial, epicardial, Purkinje and atrial AP waveforms (dotted traces). The zero current level was at the -85 mV level in all panels. B, current-voltage relationships for each AP waveform. C, instantaneous conductance-voltage (G-V) relationships for each AP waveform.

HERG K⁺ channel activity during premature cardiac action potential stimuli

The different shapes of the conductance curves for different action potential waveforms suggest that the effect of premature stimuli upon HERG K⁺ channels will vary with action potential time course and hence in different regions of the heart. We examined this hypothesis by using paired action potential clamp protocols using waveforms as illustrated in Fig. 3. For the endocardial, epicardial, Purkinje fibre and atrial waveforms, paired action potentials were delivered at 10 s intervals with the second action potential introduced at coupling intervals ranging from the point of 90 % repolarization of the first action potential duration (APD₉₀) - 100 ms to APD₉₀+ 200 ms (see Fig. 4). The voltage protocols are illustrated as insets in each panel of Fig. 4, although for purposes of image clarity only 10 of the 31 waveforms used are illustrated in each panel.

Figure 4. The effect of premature stimuli on HERG K⁺ channels under action potential clamp conditions.

Typical examples of family of currents recorded when the second AP was introduced at times corresponding to APD₉₀ - 100 ms up to APD₉₀+ 200 ms for endocardial (A), epicardial (B), Purkinje (C) and atrial (D) AP waveforms. The AP waveforms were the same as in Fig. 3. Insets show representative voltage protocols for each AP waveform. Dotted lines indicate the zero current level.

During the premature action potential there was a large transient outward current followed by a sustained current similar to the current recorded during a single action potential (see Fig. 4). The magnitude of the transient outward current during the second action potential showed a biphasic dependence on the coupling interval. The largest transient outward current response occurred for stimuli delivered prior to complete repolarization, i.e. when the second action potential started at the APD₉₀ of the first action potential for epicardial and endocardial action potentials, at APD₉₀ - 20 ms for Purkinje fibre action potentials and at APD₉₀ - 30 ms for atrial action potentials (see Fig. 5A). It is also notable that the magnitude of the transient current response at short coupling intervals, i.e. in the region APD₉₀ - 100 ms to APD₉₀ - 50 ms, is considerably higher for Purkinje fibre action potentials than for epicardial, endocardial or atrial action potentials (see Fig. 5A).

Figure 5. Summary of the effect of premature stimuli on the transient current and sustained current components.

Mean ±s.e.m. (n = 5) of the relative amplitude for the transient current component (A) and peak sustained current component (B) for endocardial, epicardial, Purkinje fibre and atrial APs plotted against coupling interval. Transient current responses have been normalized to the maximum transient current in each experiment and the peak sustained current responses during the premature beat have been normalized to the peak sustained current response during the first action potential.

The sustained current response during premature action potentials was similar in shape to that observed during the first action potential (see Fig. 4); however, the magnitude of the current during the second action potential varied with coupling interval (see Fig. 5B). This increase in current was smallest for ventricular action potential waveforms; ∼10 % higher (Fig. 5B), compared to Purkinje fibre action potentials, ∼30 % higher at the shortest coupling intervals examined, and ∼60 % higher for atrial action potentials at the shortest coupling intervals examined (Fig. 5B).

Modelling the effects of premature stimulation on HERG K⁺ currents

To investigate the possible state dependence of the response of HERG channels to premature stimulation, various computational models were simulated with the Markov chain model of Wang et al. (1997) as a starting point. We also included transitions from the pre-open closed state to an inactivated state as originally formulated by Kiehn et al. (1999) (see Fig. 6A). The values for the rate constants that gave the best fits to the experimental data (see Fig. 6C and D) are shown in Fig. 6B. The model was able to reproduce both the response to square-pulse stimuli (compare Fig. 6C with Fig. 1A) and the response to premature action potential clamp stimuli (compare Fig. 6D with Fig. 4A).

DISCUSSION

It has been suggested that HERG K⁺ channels play a particular role in suppressing arrhythmias initiated by premature beats (Miller, 1996; Smith et al. 1996). In this study we have combined the advantages of using heterologous expression systems to study ion channel function (Snyders, 1999) with the action potential clamp technique to investigate the physiological and possible pathophysiological role of HERG K⁺ channels in response to premature stimuli.

The kinetics of HERG K⁺ channels are very temperature sensitive (Zhou et al. 1998). As the main aim of this study was to determine the possible physiological and pathophysiological relevance of the unusual kinetics of HERG K⁺ channels, it was important to perform experiments at 37 °C. The rates of deactivation and recovery from inactivation (see Fig. 2) as well as the rates of activation and inactivation (data not shown) that we measured at 37 °C in CHO cells are broadly similar to those reported previously for studies at physiological temperature (Hancox et al. 1998; Zhou et al. 1998). Our rates, however, are slightly faster, especially for the rates of inactivation and recovery from inactivation when compared to the results obtained by Zhou et al. (1998), in human embryonic kidney (HEK 293) cells, measured at 35 ± 1 °C. This probably reflects in part the slight difference in temperature. Another possibility, however, may be that there are endogenous subunits present either in HEK 293 or CHO cells that subtly alter the kinetics in one compared to the other. Nevertheless, the most important finding is that all the kinetic rate constants are markedly faster at 37 °C compared to room temperature and this was of particular importance for modifying the model we used to reconstruct our experimental data (see later).

HERG K⁺ currents during cardiac action potential waveforms

In theory, given the voltage dependence of the rates of activation, deactivation, inactivation and recovery from inactivation it should be possible to predict the ionic current flow during any voltage waveform. In practice, however, the mathematical formulations used in most models entail the use of assumptions that may or may not be valid and so may lead to errors (Noble et al. 1998). An alternative and more direct approach is to use action potential clamp studies of channels transfected in heterologous expression systems (Hancox et al. 1998).

Action potential waveforms in cardiac cells vary between the different chambers of the heart as well as within the chambers, e.g. epicardial versus endocardial ventricular myocyte action potentials. To investigate the effect that different action potential shapes have on HERG K⁺ current, we used action potential waveforms that broadly reflect the range of action potentials observed in a human heart (see Fig. 3). By using transfected cell lines we were able to compare the response to different voltage waveforms in the same cell and hence with the same level of HERG K⁺ channel expression. Under these circumstances the magnitude of current was highest for ventricular waveforms > Purkinje waveforms > atrial waveforms. Furthermore, the maximal HERG channel conductance (which in this experiment reflects the open probability) was approximately eight times higher during an epicardial action potential waveform than during an atrial action potential waveform (see Fig. 3C). This reflects the greater extent of activation of HERG channels during the plateau of the ventricular action potential (∼+20 mV for ∼200 ms) compared to < 0 mV for Purkinje fibre action potentials and < -20 mV for atrial action potentials. In the heart, the contribution that HERG currents will make to repolarization in any given cell type will also depend on the level of channel expression in that cell. To date there have been no systematic studies of the levels of HERG K⁺ channel expression in different chambers of the heart. However, recent studies (Gintant, 2000) suggest that in canine heart there is an approximately 50 % higher density of the rapidly activating delayed rectifier current (I_Kr) in atrial compared to ventricular myocytes.

HERG currents during premature action potential stimuli

There are two important components of HERG K⁺ currents following premature stimulation: a transient outward current component (Smith et al. 1996) and a sustained current component. The magnitudes of both the transient current response and the sustained current response varied both with the coupling interval of the two action potential stimuli and with the shape and duration of the action potential waveforms used.

Transient outward current response to premature stimuli

The instantaneous increase in HERG K⁺ channel current following a premature stimulus reflects the increase in electrochemical driving force for K⁺ ions and that many channels are still in the open state as a result of rapid recovery from inactivation and slow deactivation. The transient nature of this initial large increase in outward current reflects the rapid rate of inactivation of channels at depolarized potentials. The magnitude of the transient outward current is proportional to the number of channels that are in the open state, which in turn is a complex function of recent voltage history (see Fig. 4). Thus, the biphasic dependence of the transient current response on interpulse coupling interval during square-pulse stimuli reflects the relative rates of recovery from inactivation (τ= 0.98 ± 0.07 ms at -80 mV) and deactivation (τ_f= 41 ± 8 ms at -80 mV). This pattern of response was also clearly seen during the action potential clamp studies (see Fig. 4), although the coupling interval at which the maximum transient outward current response occurred varied with action potential waveform. The maximum transient outward current occurred at intervals equivalent to APD₉₀ for ventricular action potential waveforms but at APD₉₀ - 20 ms for Purkinje fibre waveforms and at APD₉₀ - 30 ms for atrial action potential waveforms. The timing of the peak transient current response correlates closely with the point of maximum instantaneous conductance during the different action potential waveforms (see Fig. 3). This confirms that the instantaneous conductance curves accurately reflect the number of channels in the open state at any given time.

Sustained current response to premature stimuli

The magnitude of the sustained current response during premature ventricular action potentials was very similar to that recorded during normally paced action potentials (see Fig. 4). Conversely, the sustained current response during premature Purkinje fibre and atrial action potential waveforms was significantly higher than during normally paced beats. Furthermore, this increase in current persisted for premature stimuli at long coupling intervals. These results suggest that during ventricular action potentials the plateau potential is high enough (∼+20 mV) and the duration long enough (∼200 ms) to achieve full activation (as would be expected from the rates of activation (Zhou et al. 1998)), and therefore during a premature action potential the same maximum is reached. Conversely, at the lower plateau potentials of Purkinje fibre action potentials (< 0 mV) and atrial action potentials (< -20 mV) there is insufficient time for channels to reach maximum activation during the first action potential. Thence, following a premature stimulus, there will already be some channels in the open state and so the total number of channels that are activated during the second action potential will be larger than during the first action potential. However, at longer coupling intervals all the channels that opened during the first action potential will have deactivated and so the sustained current response will again be similar to that recorded during normally paced action potentials.

Modelling

Many models have been developed for HERG K⁺ channel kinetics. In recent years most workers have used a Markov chain model that includes three closed states, an open state and an inactivated state with transitions between the pre-open closed state and the inactivated state. This model was able to reproduce both the square pulse and action potential clamp data (see Fig. 6). In this study the rate constants were constrained by activation, deactivation, inactivation and double-pulse data recorded in CHO cells at 37 °C while those in Clancy & Rudy (2001) were fitted primarily to activation data from guinea-pig ventricular myocytes at physiological temperatures and those in Mazhari et al. (2001) were fitted primarily to activation, deactivation and inactivation for HERG recorded at 22 °C and then scaled assuming a Q₁₀ of 3.3. These different approaches can explain the slight differences in rate constants used in each model.

Physiological implications

The increase in HERG current following a premature stimulus is as rapid as the change in electrochemical driving force as the channels are already in the open state. This indicates that the HERG current following a premature stimulus will oppose the depolarization of the membrane potential and therefore tend to suppress sodium channel activation, thereby contributing to the normal refractoriness of cardiac tissue. It is interesting to note that the peak amplitude of the transient current occurs at ∼APD₉₀, which is approximately the time of the ventricular effective refractory period. The increase in the sustained current following premature stimuli, atrial > Purkinje > ventricular action potentials, suggests that HERG currents will also contribute to shortening of action potential duration following a premature stimulus, particularly in the atria.

Pathophysiological implications

The results presented here suggest at least two possible anti-arrhythmic roles for HERG K⁺ channels. Firstly, the maximum HERG K⁺ current occurs in the voltage range of calcium channel window currents (> -45 mV; Pelzmann et al. 1998). Therefore HERG K⁺ currents will tend to oppose the reactivation of L-type Ca²⁺ channels which are thought to be a major contributor to early afterdepolarizations and initiation of torsade de pointes arrhythmias (January & Riddle, 1989). Secondly, the initial large transient outward current following premature stimulation will help to suppress premature beats, as first suggested by Smith et al. (1996).