In vivo electrophysiological effects of a selective slow delayed-rectifier potassium channel blocker in anesthetized dogs: potential insights into class III actions

Abstract

Objectives: This study evaluated the in vivo electrophysiological effects of a highly selective slow delayed-rectifier K⁺-current blocker, HMR 1556, to gain insights into the consequences of selectively inhibiting the slow delayed-rectifier current in vivo. Methods: Atrial and ventricular effective refractory periods, sinus node recovery time, Wenckebach cycle-length, atrial fibrillation duration and electrocardiographic intervals were measured before and after intravenous HMR 1556. Results: HMR 1556 increased atrial and ventricular refractory periods (e.g. by 6±4% and 27±6% at cycle lengths of 360 and 400 ms, respectively), QT intervals and sinus-node recovery times. Beta-adrenoceptor blockade with nadolol abolished all effects except those on ventricular refractoriness and changed positive use-dependent effects on refractoriness to reverse use-dependent ones. In the presence of dofetilide to block rapid delayed-rectifier current, HMR 1556 effects were potentiated (e.g. atrial and ventricular refractory periods increased by 26±3% and 34±3% at cycle lengths of 360 and 400 ms, respectively). HMR 1556 reduced vagal atrial fibrillation duration from 1077±81 to 471±38 s, an effect abolished by nadolol and greatly potentiated by dofetilide (duration 77±30 s). HMR 1556 increased Wenckebach cycle length only in the presence of dofetilide. Conclusions: Slowed delayed-rectifier current inhibition affects atrial repolarization, sinus node function and atrial fibrillation in vivo, but only in the presence of intact beta-adrenergic tone, and delays ventricular repolarization even when beta-adrenoceptors are blocked. The slow delayed-rectifier current is particularly important when rapid delayed-rectifier current is suppressed, illustrating the importance of repolarization reserve.

Antiarrhythmic agents, Arrhythmia (mechanisms), Ion channels, K-channel, Long QT syndrome

This article is referred to in the Editorial by M.J. Curtis (pages 651–652) in this issue.

1 Introduction

The delayed-rectifier K⁺-current (I_K) is a key current in cardiac repolarization and consists of rapid (I_Kr) and slow (I_Ks) components [1,2]. I_Kr has been the target of most class III antiarrhythmic agents to date, but the use of I_Kr-blockers has been limited by the risk of acquired long QT syndrome (LQTS)-related proarrhythmia. The slower kinetics of I_Ks led to the suggestion [3], for which experimental evidence was provided [3,4], that I_Ks-blockers may have a more preferable profile of rate-dependent action than I_Kr-blockers and may therefore be useful antiarrhythmic agents. One of the great limitations in studying the role of I_Ks in cardiac electrophysiology has been a lack of sufficiently selective pharmacological probes. Several drugs described initially as promising I_Ks-blockers [5–7] were subsequently found to have significant effects on other cardiac currents [8–12]. Perhaps in part because of the deficiencies in the available tools, the functional role of I_Ks in cardiac electrophysiology has remained controversial [13,14].

Recent work has resulted in the production of more selective I_Ks-blockers, including benzodiazepine [15] and chromanol [16,17] derivatives. The chromanol derivative HMR 1556 has approximately 1000-fold selectivity for I_Ks over I_Kr, the transient outward current (I_to) and L-type Ca²⁺ current (I_Ca) and has been found to increase QT interval in conscious dogs but to have no detectable effect on action potential duration (APD) in normal canine ventricular muscle in vitro [18,19]. The present studies were designed to take advantage of the strong selectivity of HMR 1556 to study the potential contribution of I_Ks to canine electrophysiological function and atrial fibrillation (AF) maintenance in vivo. In addition, we sought to explore the importance of background β-adrenergic and intact I_Kr for the consequences of I_Ks inhibition and, by inference, for the I_Ks contribution to cardiac electrophysiological function.

2 Methods

2.1 Animal preparation

Experiments were performed in accordance with Canadian Council on Animal Care guidelines and conformed to principles enunciated by the National Institutes of Health. Adult mongrel dogs of either sex weighing 25.9±3.0 kg were anesthetized with morphine (2 mg/kg sc) and α-chloralose (120 mg/kg iv, followed by 29.25 mg/kg/h iv) and ventilated mechanically. The femoral artery and both femoral veins were cannulated for blood-pressure monitoring and drug administration. Body temperature was maintained at 37 °C. The surface ECG and blood pressure were continuously monitored.

The heart was exposed via median sternotomy and a pericardial cradle was created. Two bipolar electrodes each were hooked into the right (RA) and left (LA) atrial appendages for stimulation and recording, and an additional electrode was hooked into the right ventricle (RV) for stimulation. A programmable stimulator (Digital Cardiovascular Instruments, Berkeley) was used to deliver 2-ms, twice diastolic-threshold current pulses. A ventricular demand pacemaker (GBM 5880, Medtronic, Minneapolis) was used to stimulate the RV at 80/min when the ventricular rate became excessively slow during vagal stimulation.

The vagus nerves were isolated and divided in the neck. Bipolar hook electrodes were inserted within each vagus nerve. Bilateral vagal stimulation (0.1-ms square-wave pulses, 5 V, 60% of frequency for asystole) was applied continuously during periods of electrophysiological study and AF induction, as previously described [4]. The bradycardic effect of vagal stimulation was noted, and vagal stimulation frequency readjusted over the course of experiments to maintain the same bradycardic effect as under control conditions.

2.2 Electrophysiological study

The atrial effective refractory period (AERP) was measured with 10 basic stimuli (S1) at various basic cycle lengths (BCLs), followed by a premature stimulus (S2) with 5-ms decrements. The ventricular effective refractory period (VERP) was measured during vagal stimulation to allow for measurement over a wide range of cycle lengths. The longest S1–S2 interval that failed to capture the atrial or ventricular response defined the ERP. The sinus-node recovery time was measured by pacing the RA at a 200-ms cycle length for 1 min and then recording the interval from the last paced beat to the first sinus beat. Wenckebach cycle length was measured by decreasing the RA-pacing cycle length by 20-ms decrements until failure of 1:1 conduction to the ventricles. Once the approximate value was established, the measurement was repeated with 10-ms cycle-length decrements to increase its precision.

AF was defined as a rapid (>500/min), irregular atrial rhythm with varying electrogram morphology. AF duration was quantified in the absence and presence of vagal nerve stimulation under control conditions and then in the presence of each drug used. AF induction was performed 10 times if AF duration was <5 min, 5 times if AF continued between 5 and 20 min. When AF persisted ≥20 min without spontaneous termination, which was defined as sustained AF, AF induction was conducted 2 times. AF cycle length was measured by counting the number of AF cycles (based on atrial electrograms) during a 1-s AF period recorded at 100 mm/s paper speed.

The heart-rate slowing effect of vagal stimulation at 1, 3 and 5 Hz vagal stimulation was determined by measuring heart rate during the last 10 s of a 40-s vagal stimulation period at each frequency. This was obtained before and after drugs to exclude antivagal actions. No significant changes in vagal frequency–response relations were seen with any of the drugs used.

2.3 Experimental protocol

Three series of studies were performed. In all series, baseline AF duration and electrophysiological parameters were obtained in the absence and presence of vagal stimulation under control and each drug infusion condition. In the first series (Group 1, n = 8 dogs), HMR 1556 (loading dose 1 mg/kg iv, iv maintenance infusion 1 mg/kg/h) was begun during sustained vagal AF under control conditions. If the drug terminated AF, measurements were directly repeated during drug administration. If AF was not terminated, vagal stimulation was discontinued to allow for AF termination and then measurements were repeated. Group 2 dogs (n = 7) underwent the same procedures, but all measurements (both control and HMR 1556) were performed in the continuous presence of the β-adrenoceptor blocker nadolol at doses (initial dose 0.5 mg/kg iv, then 0.25 mg/kg iv every 2 h) that we previously showed to produce strong and stable β-blockade. In Group 3 dogs (n = 7), dofetilide (loading dose 0.08 mg/kg iv, iv maintenance infusion 0.008 mg/kg/h) was administered following the acquisition of baseline measurements, and measurements were repeated. HMR 1556 was then administered during continued dofetilide infusion and the responses of AF and electrophysiology were noted. Information from Group 1 indicated the effect of HMR 1556 with intact β-adrenergic tone, reflecting the drug-induced loss of the contribution of I_Ks, whereas Group 2 studies indicated HMR 1556 effects with β-adrenergic effects prevented by β-blockade (i.e. I_Ks effects exerted in the absence of β-adrenergic tone). Group 3 experiments indicated the role of I_Kr (dofetilide vs. control) as well as the role of I_Ks in the absence of I_Kr (HMR 1556+dofetilide vs. dofetilide alone).

2.4 Plasma concentration measurements

Blood samples were obtained from Group 1 and 2 dogs at the end of the loading infusion and then hourly for 3 h. Plasma was removed following centrifugation and frozen for subsequent drug concentration measurement. Samples (0.5 ml) were extracted into 5-ml dichloromethane. After centrifugation, the upper aqueous phase was sucked off, and the remaining organic phase was pipetted into a conically tapered centrifuge tube and then evaporated to dryness in a heated water bath under a stream of N₂ gas. The remaining residue was redissolved in 50-μl toluene, 1 μl of which was injected into a gas chromatograph for subsequent nitrogen-specific or mass-selective detection. The limit of quantification was 0.025 μg/ml.

2.5 Data analysis

Group data are expressed as mean±S.E.M. Paired t-tests were used for single comparisons and multiple-group comparisons were made with ANOVA. A two-tailed P<0.05 was considered statistically significant.

3 Results

3.1 Electrophysiological effects of I_Ks-blockade with intact β-adrenergic tone

Fig. 1A and B shows the effects of HMR 1556 on AERP in the presence and absence of vagal nerve stimulation, with intact β-adrenergic tone. HMR 1556 significantly increased AERP, with effects being somewhat greater at shorter cycle lengths. Effects on VERP are shown in Fig. 2. In the absence of nadolol, HMR 1556 significantly increased VERP (Fig. 2A) with no reverse use dependence, e.g. by 21.8±6.5% at cycle length 1000 ms vs. 27.3±5.7% at cycle length 400 ms. ECG intervals were measured during RA pacing with 1:1 atrioventricular conduction. PR and QRS intervals did not change after HMR 1556. However, QT interval was significantly prolonged (Fig. 2B), consistent with effects on VERP. HMR 1556 significantly increased sinus-node recovery time (Fig. 3A), indicating a role for I_Ks in SA node function when β-adrenergic tone is intact. However, administration of HMR 1556 alone did not affect AV nodal function as reflected by Wenckebach cycle length (Fig. 4A).

Fig. 1

Effects of HMR 1556 (HMR) on AERP in RA (top) and LA (bottom), in the presence (V) and absence of vagal stimulation, and the presence [Nadolol (+)] and absence [Nadolol (−)] of β-adrenoceptor blockade (CTL=pre-HMR 1556 control).

Fig. 2

Effects of HMR 1556 (HMR) on RV ERP (top) and QT interval (bottom) in the presence [Nadolol (+)] and absence [Nadolol (−)] of β-adrenoceptor blockade.

Fig. 3

Effects of HMR 1556 (HMR, A) and dofetilide (Dof, B) on sinus-node recovery time. Effects of HMR 1556 were measured in the presence [Nadolol (+)] and absence [Nadolol (−)] of β-adrenoceptor blockade.

Fig. 4

Effects of HMR 1556, dofetilide (Dof) and the combination (Dof+HMR) on Wenckebach cycle length. Effects of HMR 1556 were measured in the presence [Nadolol (+)] and absence [Nadolol (−)] of β-adrenoceptor blockade.

3.2 Electrophysiological effects of HMR 1556 during β-adrenoceptor blockade

The elimination of β-adrenoceptor-mediated effects by nadolol strongly altered the actions of HMR 1556. While vagal effects on AERP were not perceptibly different in the presence of nadolol (Fig. 1C and D) vs. in its absence (Fig. 1A and B), HMR 1556 had no effect on AERP in the presence of nadolol, despite the clear HMR 1556-induced AERP prolongation in the absence of β-blockade. Similarly, the prolongation in sinus-node recovery time induced by HMR 1556 in the absence of β-blockade (Fig. 3A) disappeared when studies were performed in the presence of nadolol. In contrast to effects on atrial refractoriness and sinus-node recovery, which were clearly dependent on intact β-adrenergic tone, HMR 1556 continued to significantly increase VERP (Fig. 2C) and QT interval (Fig. 2D) in the presence of nadolol. β-Blockade did seem to modify the frequency dependence of the HMR 1556 effect on VERP: whereas under control conditions the drug's actions showed slightly positive use dependence, in the presence of β-blockade, reverse use dependence was seen (e.g. VERP increased by 17.1±4.4% at cycle length 400 ms vs. by 23.0±5.1% at cycle length 1000 ms).

3.3 Effects of I_Ks-blockade in the presence of I_Kr inhibition

Dofetilide significantly prolonged both AERP and VERP (Figs. 5 and 6). The effects of HMR 1556 were clearly greater in the presence of dofetilide. To compare quantitatively HMR 1556 effects in the absence of significant I_Kr with those in its presence, we calculated the percentage increase in AERP and VERP for HMR 1556 vs. control (effects with I_Kr intact) as well as the percentage increase for HMR plus dofetilide vs. dofetilide alone (effect on a background lacking I_Kr). AERP increases caused by HMR 1556 were clearly enhanced when the drug was given in the presence of dofetilide (Fig. 5B). Similarly, HMR 1556-induced increases in VERP were greater at cycle lengths 600 and 1000 ms when the drug was given on a background of I_Kr blockade (Fig. 6A, bottom). Dofetilide alone significantly increased QT interval (Fig. 6B) and HMR 1556 in the presence of dofetilide produced dramatic QT prolongation at a cycle length of 350 ms. No data for the combination are available at a 250-ms cycle length because 1:1 conduction to the ventricles could not be maintained at that cycle length in the presence of both drugs. Dofetilide significantly increased both sinus-node recovery time (Fig. 3B) and Wenckebach cycle length (Fig. 4B). Potential interactions between HMR and dofetilide could not be evaluated at the cycle length used to measure sinus-node recovery (200 ms) because 1:1 atrial capture could not be obtained in the presence of both drugs. In the presence of dofetilide, HMR 1556 had a strong effect on Wenckebach cycle length (Fig. 4B), in contrast to its lack of effect in the absence of dofetilide (Fig. 4A).

Fig. 5

(A) RA (left) and LA (right) ERP recorded under control (CTL) conditions and then in the presence of dofetilide (Dof) and dofetilide plus HMR 1556 (Dof+HMR). (B) Percentage increase in RA (left) and LA (right) ERP caused by HMR 1556 alone (HMR vs. CTL) and caused by HMR 1556 against a background of I_Kr-blockade (HMR+Dof vs. Dof alone).

Fig. 6

(A) RV-ERP recorded under control (CTL) conditions and then in the presence of dofetilide (Dof) and dofetilide plus HMR 1556 (Dof+HMR). Bottom: Percentage increase in RV ERP caused by HMR 1556 alone (HMR vs. CTL) and caused by HMR 1556 against a background of I_Kr-blockade (HMR+Dof vs. Dof alone). (B) Top: QT recorded during atrial pacing under control (CTL) conditions and then in the presence of dofetilide (Dof) and dofetilide plus HMR 1556 (Dof+HMR). Bottom: Percentage increase in QT interval caused by HMR 1556 alone (HMR vs. CTL) and by HMR 1556 against a background of I_Kr-blockade (HMR+Dof vs. Dof alone). Note that no QT data are available for HMR+Dof at a cycle length of 250 ms because 1:1 conduction through the AV node could not be maintained at this cycle length for several dogs in the presence of both agents.

3.4 Effects on AF

In the absence of vagal stimulation, sustained AF was not seen. Mean AF duration decreased after HMR 1556 administration (e.g. 25.0±8.3 s pre-drug in Group 1 dogs vs. 4.9±2.2 s after, P = 0.05). In the presence of vagal stimulation and intact β-adrenergic tone, mean AF duration was significantly reduced by HMR 1556 administration (Fig. 7A). However, when β-adrenergic receptor stimulation was prevented by nadolol, HMR 1556 did not affect AF duration. HMR 1556 terminated AF that was sustained in the presence of vagal stimulation in 2/8 (25%) dogs studied without β-blockade and in no dogs studied in the presence of nadolol. Dofetilide alone had no significant effect on AF duration in the presence of vagal stimulation (Fig. 7B). However, the effect of HMR 1556 on AF duration was greatly enhanced in the presence of dofetilide. In 3/7 dogs (whose results could not be used to calculate AF duration in the presence of dofetilide and HMR 1556), no AF could be induced during vagal stimulation in the presence of both drugs. HMR 1556 was administered to 7 dogs with sustained vagal AF in the presence of dofetilide and terminated AF in 5/7 (83.8%).

Fig. 7

Effects of HMR 1556 and dofetilide (Dof) and the combination (Dof+HMR) on AF duration (DAF). Effects of HMR 1556 were measured in the presence [Nadolol (+)] and absence [Nadolol (−)] of β-adrenoceptor blockade.

HMR 1556 significantly increased AF cycle length during vagal stimulation in the absence of β-blockade by 34.1±4.3% (82.7±2.3 to 110.5±2.3 ms, P<0.05). In the presence of nadolol, AF cycle length averaged 88.9±2.2 ms before and 95.0±1.3 ms (P = NS) after HMR 1556, respectively. Dofetilide also increased AF cycle length, which averaged 87.8±2.9 ms before and 121.0±2.3 ms (P<0.01) after the drug. The combination of dofetilide and HMR 1556 produced striking AF cycle-length prolongation, from 121.3±2.3 to 185.0±14.0 ms (P<0.001).

3.5 Plasma concentration measurements

Total plasma concentrations measured by gas chromatography are provided in the Table 1, along with free plasma concentration calculated based on measurements showing that 75% of HMR 1556 is bound to plasma proteins in the dog. Drug concentrations were relatively stable over time and estimated free-drug concentrations averaged ∼1 μM.

Table 1

Plasma concentrations of HMR 1556

	End-load	1 h	2 h	3 h
Nadolol-free dogs (Group 1)
Total (free) conc.a	5.6±1.1 (1.4±0.3)	3.8±0.4 (0.9±0.1)	4.4±0.5 (1.1±0.1)	4.1±0.5 (1.0±0.1)
Nadolol-treated dogs (Group 2)
Total (free) conc.	3.2±0.5 (0.7±0.1)	4.0±0.4 (1.0±0.1)	4.0±0.4 (1.0±0.1)	4.8±0.4 (1.2±0.1)

^a

Conc.=measured concentration in μM; free concentration=total × unbound fraction (25%).

4 Discussion

In this study, we evaluated the response to the highly selective I_Ks-blocker HMR 1556 to assess the potential role of I_Ks in canine in vivo electrophysiology and AF maintenance, along with interactions with β-adrenergic tone and I_Kr-block. Our main findings are that I_Ks appears to contribute to atrial and ventricular repolarization, as well as to sinus node function and maintenance of vagal AF, in the presence of intact β-adrenergic tone. β-Adrenoceptor blockade eliminates the contribution of I_Ks to atrial refractoriness, SA node function and AF maintenance, but I_Ks continues to contribute to the determination of ventricular refractoriness in the absence of β-adrenergic stimulation. The contribution of I_Ks is particularly important when I_Kr is suppressed, reflecting the important contribution of I_Ks to repolarization reserve.

4.1 Comparison with previous findings regarding the functional role of I_Ks

Jurkiewicz and Sanguinetti [3] first suggested that I_Ks-blockers might produce selective APD prolongation at rapid rates, a more favourable profile of action than for I_Kr-blockers. Subsequently, ambasilide and azimilide, drugs with significant I_Ks-blocking action, were found to increase atrial APD without reverse use dependence and to be more effective against vagotonic AF than dofetilide [4,20]. However, both agents block other channels in addition to I_Ks, making it difficult to know which actions are responsible for the difference between their effects and those of pure I_Kr-blockade [9,10,21,22].

With the development of more selective I_Ks-blockers, it became possible to assess the role of I_Ks more directly. Chromanol 293B inhibits I_Ks with ∼20-fold selectivity over I_to and with virtually no effect on I_Kr[11,23]. Bosch et al. [11] noted that chromanol 293B-induced APD increases in guinea pig and human ventricular cardiomyocytes were comparable at different frequencies, in contrast to the clear reverse use-dependent effects produced by dofetilide. Shimizu and Antzelevitch [24] observed substantial canine ventricular APD prolongation in arterially perfused wedge preparations in vitro. However, a role of I_to-block in chromanol 293B effects on APD cannot be totally excluded, and the drug appears to be much less potent in multicellular superfused preparations than on isolated cells [14]. With the development of even more selective benzodiazepine- [15] and chromanol-based [16,17,25] I_Ks-blockers, it has become possible to evaluate the role of I_Ks more clearly. However, the results have been somewhat contradictory and controversial.

Studies with the benzodiazepine I_Ks blocker L-735,821 and chromanol 293B suggested no significant role for I_Ks in canine [26] or rabbit [27] ventricular repolarization in isolated myocytes and papillary muscle preparations. Volders et al. [19] observed important effects of HMR 1556 on QT intervals in conscious dogs but no change in APD in multicellular ventricular preparations studied in vitro. They attributed the discrepancy to the absence of β-adrenergic tone in the isolated preparation because they found that the compound partly reversed isoproterenol-induced APD abbreviation in isolated preparations, producing an APD increase compared to the isoproterenol, HMR-free baseline. Stengl et al. [18] showed reverse use-dependent effects of HMR 1556 on canine APD in the presence of isoproterenol in superfused preparations. They pointed out the potential difficulty in extrapolating from the in vitro to the in vivo situation and indicated the need for in vivo studies [18]. Varro et al. [26] used 10-μM chromanol 293B and 100-nM L-735,821 to probe the role of I_Ks in canine ventricular myocardium. I_Ks-block produced small increases in APD (<7%) with I_Kr intact and much larger increases in the presence of 1 μM E-4031 to block I_Kr. In vivo administration of 293B as a single 1-mg/kg bolus did not alter QT interval, but plasma concentrations were not measured. Their findings agree with ours in terms of the interaction between I_Kr- and I_Ks-blockade, but disagree in terms of the extent of repolarization change with I_Ks-blockade. The discrepancies may be related to the specific drugs and the models (in vitro superfused preparation vs. intact heart) used.

In the present paper, we report the in vivo effects of HMR 1556 at a dose that produced relatively stable, ∼1-μM free plasma concentrations. Since drug bound to plasma proteins is not available for diffusion into the interstitial fluid, it is the free drug concentration that equilibrates with the extracellular space and is relevant to actions studied with in vitro systems. Thomas et al. [17] reported that 100-nM HMR 1566 decreases I_Ks in isolated canine cardiomyocytes by >90% and Bosch et al. [25] reported that 1-μM HMR 1566 reduces I_Ks in isolated atrial myocytes by ∼95%. At 10–50 μM concentrations, HMR 1556 inhibits other currents, including I_Kr, I_to and I_Ca; however, at 1-μM concentrations, the drug has no measurable effect on I_Kr, I_to, I_K1 or I_Ca[17,25]. Our findings therefore suggest involvement of I_Ks in sinus node, atrial and AV node electrical function, with the sinus node and atrial actions being dependent on intact β-adrenergic tone and the AV nodal effect being manifest only when I_Kr is inhibited. In addition, I_Ks appears to contribute to AF maintenance in the presence of intact β-adrenergic tone and to contribute to ventricular repolarization in the presence and absence of intact β-adrenoceptor function.

4.2 Novelty and potential importance of our observations

The responses we observed to HMR 1556 point to a significant role of I_Ks in cardiac electrical function in vivo. Effects on atrial refractoriness were appreciable and reflected in AF maintenance; however, atrial actions were smaller than ventricular and were abolished with β-adrenoceptor blockade. The discrepancy between the contribution of I_Ks to atrial vs. ventricular repolarization is likely due to the different voltage–time profiles of their respective action potentials, with atrial action potentials having shorter plateaus at more negative potentials [28], and therefore spending much less time at voltages associated with I_Ks activation [2,3,29].

Our observation that I_Ks plays a β-adrenergic stimulation-dependent role in sinus-node function in vivo is consistent with recent results suggesting that I_Ks contributes to pacemaker function in isolated SA node cells only in the presence of β-adrenoceptor stimulation [30]. Tachycardia-dependent refractoriness prolongation would be a desirable property for an antiarrhythmic agent. There has been controversy about the frequency dependence of the I_Ks contribution to repolarization. Whereas Stengl et al. [18] noted I_Ks accumulation at rapid rates in the presence of isoproterenol, HMR 1556-induced APD prolongation showed reverse use dependence. We found that the APD-prolonging effect of HMR 1556 in vivo was not reverse use-dependent in the absence of β-blockade. However, in the presence of β-blockade, a reverse use-dependent profile emerged (Fig. 2B, top). This observation suggests that the stable or increased HMR 1556 effect on refractoriness observed at rapid rates in vivo may be due to increased β-adrenergic tone (and consequently I_Ks activation) in response to tachycardia, rather than to an intrinsic property of I_Ks-induced APD prolongation.

The major contribution of our study is to evaluate in detail the result of I_Ks-blockade in vivo as a function of heart rate, the presence or absence of β-adrenergic tone and the intactness of I_Kr function. As such, our results are complementary to those of in vitro studies. In vitro systems allow for the evaluation of detailed cellular mechanisms, including changes in action potential properties and ionic currents. However, in vitro systems are subject to a number of potential limitations: alterations in ionic currents and membrane signaling systems by cell isolation, changes in intracellular second messengers and metabolism with the cellular dialysis that is inevitable with tight-seal patch-clamp, the death of deeper cell layers in superfused multicellular systems and a lack of normal tissue oxygenation and microcirculatory dynamics upon perfusion with crystalloid solutions. In vivo studies provide results directly relevant to those in intact organisms, but do not address underlying cellular mechanisms.

Our findings indicate a role of I_Ks in a variety of electrophysiological functions, but with the exception of ventricular refractoriness, the contribution of I_Ks is limited in the absence of β-adrenergic tone, which is well known to enhance I_Ks substantially [29]. Our results also illustrate the important principle of repolarization reserve. The effects of HMR 1556 on atrial and ventricular refractoriness were clearly greater in the presence of dofetilide, indicating the important role that I_Ks plays in maintaining repolarization when I_Kr is compromised. In the case of AV nodal function, a contribution from I_Ks could only be demonstrated when I_Kr was inhibited.

Previous papers have suggested that I_Ks-selective blockers may constitute an interesting new class of drugs for AF therapy [4,20]. The present study of a highly selective I_Ks blocker shows statistically significant, but modest, efficacy of I_Ks-blockade in the suppression of vagal AF when β-adrenergic tone is intact. These relatively small effects, along with the substantial concomitant delaying effect on ventricular repolarization, suggest that I_Ks per se may not be a very desirable target in general for AF prevention. The demonstration of an apparent primary role for enhanced I_Ks in a familial type of genetically determined AF [31] may indicate the presence of specific forms of AF in which I_Ks-blockers could be particularly useful.

4.3 Potential limitations

Our experiments were performed in dogs. Species differences in repolarizing K⁺-currents are well recognized [32] and may cause species-related discrepancies in the results of I_Ks-blockade [33]. Therefore, extrapolation of our results to other species, including man, should be cautious. In addition, our results were obtained in an anesthetized, open-chest preparation, which will clearly affect autonomic tone. Although the HMR 1556 concentrations we studied appear to be highly selective for I_Ks over I_Kr, I_to, I_K1 and I_Ca[17,25], no drug is absolutely specific, state-dependent actions could produce different effects in vivo from those studied under voltage-clamp conditions in vitro and HMR 1556 has not been evaluated on other currents, such as those carried by Na⁺ and Cl⁻.

Our results indicate significant ventricular refractoriness-prolonging actions, which could translate into efficacy against ventricular reentrant arrhythmias. These may be particularly interesting in situations, like acute myocardial infarction, in which β-adrenergic stimulation plays an important role since, in contrast to the effects of I_Kr-blockers, whose effects are reduced by β-adrenergic stimulation, the effects of I_Ks-blockers appear to be enhanced [34]. However, the significant ventricular repolarization delays that we saw with HMR 1556 may also confer a risk of Torsades de Pointes arrhythmias, which would be of particular concern when I_Kr is also reduced. Therefore, any potential development of I_Ks-blockers as antiarrhythmic agents should include careful safety screening for excess QT prolongation and ventricular proarrhythmia.

Acknowledgements

The authors thank the Canadian Institutes of Health Research (CIHR) and the Quebec Heart and Stroke Foundation for providing operating funding, Aventis Pharmaceuticals for providing HMR 1556, Nathalie L'Heureux and Chantal Maltais for technical assistance and France Thériault and Diane Dubois for excellent secretarial help. Hideko Nakashima is a CIHR Research Fellow.