Field and action potential recordings in heart slices: correlation with established in vitro and in vivo models

Abstract

BACKGROUND AND PURPOSE

Action potential (AP) recordings in ex vivo heart preparations constitute an important component of the preclinical cardiac safety assessment according to the ICH S7B guideline. Most AP measurement models are sensitive, predictive and informative but suffer from a low throughput. Here, effects of selected anti-arrhythmics (flecainide, quinidine, atenolol, sotalol, dofetilide, nifedipine, verapamil) on field/action potentials (FP/AP) of guinea pig and rabbit ventricular slices are presented and compared with data from established in vitro and in vivo models.

EXPERIMENTAL APPROACH

Data from measurements of membrane currents (hERG, I_Na), AP/FP (guinea pig and rabbit ventricular slices), AP (rabbit Purkinje fibre), haemodynamic/ECG parameters (conscious, telemetered dog) were collected, compared and correlated to complementary published data (focused literature search).

KEY RESULTS

The selected anti-arrhythmics, flecainide, quinidine, atenolol, sotalol, dofetilide, nifedipine and verapamil, influenced the shape of AP/FP of guinea pig and rabbit ventricular slices in a manner similar to that observed for rabbit PF. The findings obtained from slice preparations are in line with measurements of membrane currents in vitro, papillary muscle AP in vitro and haemodynamic/ECG parameters from conscious dogs in vivo, and were also corroborated by published data.

CONCLUSION AND IMPLICATIONS

FP and AP recordings from heart slices correlated well with established in vitro and in vivo models in terms of pharmacology and predictability. Heart slice preparations yield similar results as papillary muscle but offer enhanced throughput for mechanistic investigations and may substantially reduce the use of laboratory animals.

Introduction

Although action potential (AP) recordings in ex vivo heart preparations are not a requirement within the framework of the ICH S7B guideline for the cardiac safety of drugs (Anonymous, 2005), they still constitute an important component of the preclinical cardiac safety assessment of new chemical entities, because AP recordings provide the electrophysiologist with a wealth of information. While most AP measurement models (e.g. papillary muscle, Purkinje fibre, ventricular strips, etc.) are sensitive, predictive and informative, they usually suffer from a low throughput. One promising approach to overcome this limitation constitutes the synchronous recording of bioelectrical signals from multiple cardiac tissue slices, a technique that has been recently optimized (Bussek et al., 2009). Here, effects of selected class 1 to 4 anti-arrhythmics on field potential (FP) and AP of guinea pig and rabbit ventricular slices are presented and compared with established in vitro and in vivo models.

Tissue slices from brain, kidney, liver, lung and pancreas are well-established models for electrophysiological, biochemical and toxicological studies (Edwards et al., 1989; Parrish et al., 1995; Vickers and Fisher, 2004; Colbert, 2006). Compared with isolated cells (e.g. cardiomyocytes), tissue slices offer the advantages of preserved tissue structure, no enzymatic digestion, no selection of cells during the isolation procedure, no exposure to alien growth factors or serum and viability for considerable periods of time when maintained under appropriate conditions. While isolated cardiomyocytes allow numerous measurements in cells from a single heart, the lack of intercellular contacts or cell surface damage by enzymatic digestion may lead to erroneous conclusions about drug actions. On the other hand, pharmacological and electrophysiological experiments in Langendorff hearts, papillary muscle or Purkinje fibres provide evidence of drug action under conditions more or less close to in vivo physiology but are time-consuming and expensive, because only a limited number of drugs or drug concentrations can be tested in the heart of one animal. Heart slices combine the advantages of whole organ and isolated cells, because they exhibit intact tissue structure and cellular contacts, yet a large number of preparations can be obtained from a single heart. Furthermore, the slice thickness (350 µm) ensures maintained tissue oxygenation, which may become limiting in other in vitro heart preparations like papillary muscles (Barclay, 2005).

In this paper, we present integrated data on the electrophysiological effects of selected anti-arrhythmic drugs (flecainide, quinidine, atenolol, sotalol, dofetilide, nifedipine, verapamil) using complementary in vitro and in vivo approaches. The experimental techniques range from measurement of membrane currents (hERG, I_Na, I_Ca.L), to AP/FP in slices, Purkinje fibre APs and haemodynamic/ECG parameters in conscious, telemetered Beagle dogs and also include published data. The main result of this complementary approach is that FP/AP recordings from heart slices correlate well with established in vitro and in vivo models in terms of pharmacology and predictability. Heart slice preparations yield similar results as papillary muscles but offer enhanced throughput for mechanistic investigations and may substantially reduce the use of laboratory animals. This work has been presented previously at the annual meeting of the Safety Pharmacology Society (Sept. 2010, Boston, MA).

Methods

Drugs

Chemicals and drugs [flecainide acetate (Flecainid-Isis® tablets), quinidine hydrochloride monohydrate, atenolol, dofetilide, d,l-sotalol hydrochloride (Sotalol ratiopharm® tablets), nifedipine, verapamil hydrochloride] were provided by Bayer-Schering Pharma (Berlin, Germany) or obtained from commercial sources (Sigma-Aldrich, Taufkirchen, Germany; RBI, Natick, MA, USA; local pharmacy) and stored at room temperature or as appropriate. On each of the experimental days, drug solutions were freshly prepared using frozen stock solutions (in dimethyl sulfoxide) for in vitro studies. For in vivo studies (oral administration of gelatine capsules, 0.25 mL·kg⁻¹), drugs were formulated in water (verapamil), aqueous Tylose MH300 (0.5%; atenolol, dofetilide), ethanol/polyethylene glycol 400 (PEG400, 10:90; quinidine), PEG400 (nifedipine) or as pulverized tablets in gelatine capsules (flecainide, d,l-sotalol). Solutions and formulations were stored at room temperature.

Drug/molecular target nomenclature (e.g. receptors, ion channels, etc.) follows and conforms to BJP's Guide to Receptors and Channels (Alexander et al., 2011).

In vitro electrophysiology studies: hERG K⁺ current

HEK 293 cells stably transfected with cDNA encoding the hERG (human ether-a-go-go-related gene) K⁺ channel (Zhou et al., 1998) were cultured as previously described (Himmel, 2007). The single-electrode whole-cell voltage clamp method was applied using an EPC-9 amplifier and TIDA 5 software (HEKA Elektronik, Lambrecht, Germany) at room temperature as previously reported (Himmel, 2007).

The clamp protocol consisted of stepping the command voltage to +20 mV (duration 1000 ms) followed by a hyperpolarizing step to −120 mV (500 ms) and a step back to the holding potential of −80 mV (cycle length 12 s). The inward tail current elicited by stepping from +20 to −120 mV was used to quantify hERG K⁺ current.

The standard extracellular solution was composed of (in mM) NaCl 146.0, KCl 4.0, CaCl₂ 2.0, MgCl₂ 2.0, HEPES 10.0 (pH 7.4; 300–310 mosmol·L⁻¹?). The intracellular (electrode filling) solution was composed of (in mM) KCl 135.0, MgATP 2.0, HEPES 10.0, EGTA 10.0 (pH 7.4; 290–300 mosmol·L⁻¹).

The concentration-dependence of effects was modelled with a standard four-parameter logistic equation: effect = min + (max / (1 + 10^{[(logIC50 − logX) × nH]})), with minimal and maximal effects (min, max), half-maximal inhibitory drug concentration (IC₅₀), drug concentration (X) and Hill slope (n_H). Minimal and maximal effects were usually treated as constants (max = 100 and min = 0), and IC₅₀ and n_H as variables (GraphPad Prism 3, GraphPad Software Inc., La Jolla, CA, USA).

In vitro electrophysiology studies: hNav1.5 Na⁺ current

HEK 293 cells stably transfected with cDNA encoding the hNav1.5 Na⁺ channel were cultured under standard conditions (section 2.2; Himmel and Hoffmann, 2010). The clamp protocol (whole-cell voltage clamp, 22°C) consisted of stepping the command voltage to −120 mV (duration 500 ms) followed by a depolarizing step to −35 mV (20 ms) and a step back to the holding potential of −80 mV (cycle length 2 s). The peak inward current elicited by stepping from −120 to −35 mV was used to quantify hNav1.5 Na⁺ current. The bath solution consisted of (mM) NaCl 150, KCl 2, MgCl₂ 1, CaCl₂ 1.5, HEPES 10, glucose 5 (pH 7.4); the pipette solution contained (mM) KCl 35, CsF 105, EGTA 10, HEPES 10 (pH 7.4).

In vitro electrophysiology studies: rabbit Purkinje fibre AP

Purkinje fibres were carefully dissected from the ventricles of female rabbit hearts (strain: New Zealand White and Chinchilla; age 4–14 months; body weight 2–5 kg) and stored in high-K⁺ Tyrode solution at 37°C as previously described (Vormberge et al., 2006; Himmel, 2007). Purkinje fibres were mounted in a horizontal organ bath that was perfused in a non-circulating manner with oxygenated (95% O₂, 5% CO₂) standard recording Tyrode solution (3 mL·min⁻¹; 37°C). The recording Tyrode solution was composed of (in mM) NaCl 127.0, KCl 3.0, CaCl₂ 1.8, MgCl₂ 0.5, NaH₂PO₄ 0.36, NaHCO₃ 22.0, d(+)-glucose 5.5 (pH 7.4 ± 0.1 at 37 ± 1°C following equilibration with 95% O₂, 5% CO₂). The preparations were stimulated electrically by means of square wave pulses (duration: 2 ms; 50% above threshold) at a standard stimulation rate of 1 Hz (software ISO-2, MFK, Niedernhausen, Germany). Conventional borosilicate glass microelectrodes with tip resistances of 20–40 MΩ when filled with 3 M KCl were used in order to measure APs intracellularly with a microelectrode amplifier (model BA-1S, npi electronic, Tamm, Germany).

Each preparation was equilibrated in the bath for approximately 60 min. When the impalement was stable, the bath was perfused with drug solution at cumulatively increasing concentrations (30 min per concentration) followed by washout. The frequency-dependence of effects was assessed by varying the stimulation frequency (0.2, 1.0 and 2.5 Hz).

Digitally recorded (10 kHz) APs were analysed for resting membrane potential (RMP), AP amplitude (APA), maximal upstroke velocity (V_max), plateau potential at 35 ms after the upstroke and AP duration at 50% and 90% of repolarization (APD₅₀, APD₉₀). Furthermore, the ratio of APD₅₀ divided by APD₉₀ was calculated in order to quantify the potential AP triangulation.

In vitro electrophysiology studies: guinea pig papillary muscle AP

Papillary muscles were dissected from the ventricles of female guinea pig hearts (strain: Dunkin Hartley; body weight 200–250 g). Papillary muscles were mounted in a horizontal organ bath that was perfused in a non-circulating manner with oxygenated (95% O₂, 5% CO₂) Tyrode solution (35°C). Conventional borosilicate glass microelectrodes (20–40 MΩ, 3 M KCl) were used in order to measure APs at 1 Hz (software ISO-2, MFK) with a microelectrode amplifier (model BA-1S, npi electronic).

Each preparation was equilibrated in the bath for approximately 60 min. When the impalement was stable, the bath was perfused with d,l-sotalol or quinidine at cumulatively increasing concentrations (30 min per concentration). The frequency-dependence of effects was assessed by varying the stimulation frequency (0.3, 1.0 and 3.0 Hz).

Digitally recorded APs were analysed for RMP, APA, V_max and APD₃₀, APD₆₀, APD₉₀.

In vitro electrophysiology studies: guinea pig and rabbit left ventricular slice AP and FP

Ventricular heart slices were obtained from male guinea pigs (about 300 g body weight; Charles River, Sulzfeld, Germany) and female rabbits (strain: Chinchilla; 2.0–2.5 kg body weight; Charles River) as described previously (Bussek et al., 2009). Briefly, animals were anaesthetized (guinea pig: 70% CO₂, 30% O₂; rabbit: ketamine/xylazine 50/12 mg·kg⁻¹ i.m. and 10/5 mg·kg⁻¹ i.v.), and their hearts were quickly removed and perfused on a Langendorff apparatus with oxygenated (5% CO₂, 95% O₂) Tyrode's solution (composition in mM: NaCl 126.7, NaH₂PO₄ 0.4, NaHCO₃ 22, KCl 5.4, CaCl₂ 1.8, MgCl₂ 1.1, glucose 5, pH 7.4) for 1 min followed by a 1 min perfusion with high potassium (HK+) solution (composition in mM: NaCl 120, KCl 20, CaCl₂ 2, MgCl₂ 1, HEPES 10, glucose 10, pH 7.4) to inhibit electrical activity. Contractile activation was suppressed with 2,3-butanedione monoxime (BDM, 15 mM; Sellin and McArdle, 1994).

A tissue piece (10 × 4 mm) of the middle part of the left ventricle was glued directly to an agarose block with histoacryl tissue adhesive (Aesculap AG & Co. KG, Tuttlingen, Germany), with the tissue position corresponding to the slice direction. This procedure allows direct contact between the heart tissue and the superfusion solution in order to avoid the risk of inadequate oxygen supply. The block was then fixed to the cutting stage of a vibratome (Integraslice, Campden Instruments Ltd., Loughborough, UK). Vertical transmural slices (consisting of epi-, mid-myo- and endocardial layers) of 350 µm thickness were cut in cold (4°C) oxygenated HK+ solution containing 15 mM BDM. Cutting was done with a steel blade driven at a speed of 0.03 mm·s⁻¹, amplitude of 1 mm and vibration frequency of 51 Hz. Freshly prepared slices were stored in oxygenated HK+ solution at room temperature, and fixed by a grid (‘slice holder’, SDH-27N/15, Harvard Apparatus, Holliston, MA).

To characterize the electrophysiological parameters of guinea pig and rabbit cardiac slices, we recorded extracellular FPs and intracellular APs in the mid-myocardium at the centre of the slices. Myocardial slices were transferred to a four-channel submerged type recording chamber [Lohmann Research Equipment (LRE), Castrop-Rauxel, Germany] and continuously superfused with oxygenated Tyrode's solution (2 mL·min⁻¹) at 37°C. FPs were recorded simultaneously in up to four heart slices using the multiple slice evaluation system Synchroslice (LRE). Concentric bipolar stainless steel stimulation electrodes (LRE; diameter 1.5 mm), used in order to minimize effects of stimulation on neurotransmitter release, and tungsten platinum recording electrodes (Thomas Recording, Giessen, Germany) were advanced until contact with the slice surface using manually driven micromanipulators under visual control through a multiple CCD camera system. Data acquisition (sampling rate 10 kHz per channel, bandwidth 1 Hz–3 kHz), electrical stimulation (1 Hz) and application of drugs to the superfusion with an eight-channel Teflon valve system were controlled via automated software (SynchroHeart, LRE). The software was also used for offline analysis of single FP component amplitudes, slopes and latencies. In addition, the difference between the primary (positive or negative) and secondary (negative or positive) peak latencies, termed Q_Na, were analysed in order to detect changes in Na⁺ conductance.

Intracellular APs were recorded with conventional glass micropipettes (Bussek et al., 2009). Signals were accepted when the resting membrane potential was more negative than −75 mV. and the amplitude of the AP was larger than 115 mV. After 40 min under control conditions, drugs were cumulatively added (one concentration every 30 min) to the superfusion solution.

Cardiovascular function and ECG in conscious dogs in vivo

The animal care and experimental procedures were in accordance with the German Law on the Protection of Animals and were performed with the permission from the State Animal Welfare Committee. Furthermore, all dogs were examined before the experiments and found to be healthy. Beagle dogs of either sex (body weight: 11–20 kg, age: 2–8 years) were obtained from various commercial breeders. The dogs were identified by a tattooed ear number and a collar and housed in groups of two to three animals. Lights were on for 12 h per day from 6:00 h to 18:00 h, room temperature was 20–23°C and relative humidity 30–70%. The dogs were fed once daily with pelleted standard dog chow; drinking water was available ad libitum.

Briefly, under general anaesthesia, the dogs were implanted with a telemetry device [model TL11M2-D70-PCT; Data Science Inc. (DSI), St. Paul, MN] comprising a transmitter, a catheter and two electrodes. The body of the transmitter was inserted into a s.c. pouch made on the dog's flank. The catheter was passed s.c. to the femoral artery, inserted distal to the inguinal ligament and advanced approximately 20 cm upstream to measure blood pressure in the abdominal aorta. The two ECG electrodes were placed s.c. in a standard lead II configuration. After the implantation surgery, a post-operative recovery of at least 2 weeks was allowed.

For measurement of blood pressure and ECG, the dogs were separated into a cage equipped with two individual telemetry receivers (model RMC-1, DSI). The signals were captured using Ponemah P3 Plus software (V.4.10 including Dataquest OpenART, V.2.30, DSI). Collected data were averaged over a period of 15 min, and telemetry signals were analysed for systolic, diastolic and mean arterial blood pressure, heart rate and various ECG intervals (PQ, QRS, QT, QTc). QT intervals were corrected for heart rate according to Van de Water (QTcV; Van de Water et al., 1989).

Data collection began at least 90 min before administration of either of the test compounds. Following drug administration at 15 h 30 min, data were recorded overnight for a period of about 15 h. Drug effects were assessed as the changes versus pre-drug values as compared with those of the vehicle control group.

Determination of drug plasma concentrations in dogs

Blood samples for determination of drug plasma concentrations were taken from satellite animals of either sex without telemetry implant (n= 3–4 per group) via the jugular vein or cephalic vein into lithium-heparin-coated monovettes at 1, 3, 7 and 24 h post treatment. After centrifugation at ≤4°C, the plasma samples were stored at −20°C until analyis. Following protein precipitation with acetonitrile or acetonitrile/ammonium acetate including an appropriate internal standard, drug concentrations in plasma were determined by means of HPLC and tandem mass spectrometry detection. In the case of d,l-sotalol determination, plasma concentrations were measured after solid phase extraction using separation by HPLC and fluorescence detection. Lower limits of quantification were 0.25 µg·L⁻¹ for flecainide, quinidine, dofetilide and verapamil; 0.5 µg·L⁻¹ for nifedipine; 5 µg·L⁻¹ for atenolol; and 12 µg·L⁻¹ for d,l-sotalol.

Statistics

The data presented are rounded group mean values and corresponding SD of n experiments unless indicated otherwise. Statistical analysis was performed either as ordinary one-way or repeated-measures anova followed by Dunnett's multiple comparisons post hoc test or by applying an appropriate nonparametric test. Differences were considered statistically significant if P < 0.05. Statistical analysis and graphical presentation of data was done with GraphPad Prism (release 3) or Microsoft Excel.

Focused literature search

A literature search in PubMed (http://www.ncbi.nlm.nih.gov/pubmed/?) was conducted using the terms flecainide, quinidine, atenolol, sotalol, dofetilide, nifedipine, verapamil, anti-arrhythmic, hERG, calcium channel, sodium channel, AP, Purkinje fibre, papillary muscle, ECG, QT/QTc, QRS, plasma concentration either alone or in various combinations. Amongst the numerous retrieved articles, we focused on those presenting concentration–effect data in vitro and ex vivo, and dose-dependence of effects together with exposure data in vivo.

Results

Class 1 anti-arrhythmics flecainide and quinidine

As expected for class 1 anti-arrhythmics, both flecainide and quinidine were found to be inhibitors of the human cardiac sodium channel hNav1.5 with IC₅₀ concentrations of 11 and 12 µM, respectively (Table 1, Figures 1B and 2B), and 20% inhibition at around 3 µM. At lower concentrations, however, quinidine and flecainide (IC₅₀∼1.7 µM) also inhibited the hERG K⁺ channel (Table 1, Figures 1C and 2C) with an IC₂₀ level at about 0.5 µM.

Table 1.

Effects of selected anti-arrhythmics on hERG K⁺ current, hNav1.5 Na⁺ current and cardiac L-type Ca⁺⁺ current in vitro

	hERG K+ current		hNav1.5 Na+ current		L-type Ca++ current
	Own IC50a (µM)	Literature IC50 (µM)	Own IC50a (µM)	Literature IC50 (µM)	Literature IC50 (µM)
Flecainide	1.7	0.7–3.9b	11	2.5–7c	20, 41, 63d
Quinidine	N/A	0.41e	12	20f	10, 56g
Atenolol	>100	>1000h	N/A	N/A	N/A
d,l-sotalol	1200	69i	N/A	N/A	N/A
Dofetilide	0.036	0.158j	N/A	N/A	N/A
Nifedipine	315	>50, 275k	N/A	N/A	0.2–0.3l
Verapamil	0.68	0.143m	N/A	N/A	1.0n

^aLargely unpublished IC₅₀ values from the authors laboratory (d,l-sotalol IC₅₀ from Vormberge et al., 2006); N/A, not available.

^bPaul et al., 2002; Ducroq et al., 2007.

^cEstimated from Nitta et al., 1992; Ducroq et al., 2007.

^dScamps et al., 1989; Kihara et al., 1996; Hancox and Convery, 1997.

^ePaul et al., 2002.

^fEstimated from Ducroq et al., 2007.

^gSalata and Wasserstrom, 1988; Scamps et al., 1989.

^hKawakami et al., 2006.

ⁱNumaguchi et al., 2000; Ducroq et al., 2007.

^jDavie et al., 2004.

^kZhang et al., 1999; Zhabyeyev et al., 2000.

^lUehara and Hume, 1985; Charnet et al., 1987; Shen et al., 2000.

^mZhang et al., 1999.

ⁿZhang et al., 1999.

Figure 1.

(A) Examples of intracellular (AP) and extracellular (FP) recordings from guinea pig (left) and rabbit (right) ventricular slices. Shown are superimposed tracings before (pre-drug control) and following flecainide exposure (10 µM). Insets: enlarged view of primary and secondary peak of the FP, the latency difference of which is a measure of Na⁺ conductance (Q_Na). (B) Concentration-dependence of effects of flecainide on ΔV_max, ΔQ_Na and ΔQRS in the following preparations: rabbit Purkinje fibre and ventricular slices, guinea pig ventricular slices and papillary muscle (data from Hayashi et al., 2005) and conscious dog. Also depicted is a concentration–response curve for inhibition of the hNav1.5 Na⁺ current (dashed grey line) and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003). Please note that throughout all figures (i) solid symbols reflect data originating from the authors' laboratories, whereas data from the literature are depicted with unfilled symbols; and (ii) only mean values are depicted for clarity. (C) Concentration-dependence of effects of flecainide on ΔAPD₉₀, ΔFPD₉₀ and ΔQTcV in the following preparations: rabbit Purkinje fibre and ventricular slices, guinea pig ventricular slices and papillary muscle (data from Hayashi et al., 2005) and conscious dog. Also depicted is a concentration–response curve for inhibition of the hERG K⁺ current (solid grey line) and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003).

Figure 2.

(A) Examples of intracellular (AP) and extracellular (FP) recordings from guinea pig ventricular slices. Shown are superimposed tracings before (pre-drug control) and following quinidine exposure (30 µM). Insets: enlarged view of primary and secondary peak of the FP, the latency difference of which is a measure of Na⁺ conductance (Q_Na). (B) Concentration-dependence of effects of quinidine on ΔV_max, ΔQ_Na and ΔQRS in the following preparations: rabbit Purkinje fibre (data from Aubert et al., 2006), guinea pig ventricular slices and papillary muscle (own data and data from Hayashi et al., 2005), conscious dog and dog Purkinje fibre (data from Burke et al., 1986) and human subjects (data from Duff et al., 1985). Also depicted is a concentration–response curve for inhibition of the hNav1.5 Na⁺ current (dashed grey line) and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003). (C) Concentration-dependence of effects of quinidine on ΔAPD₉₀, ΔFPD₉₀ and ΔQTcV in the following preparations: rabbit Purkinje fibre (data from Aubert et al., 2006), guinea pig ventricular slices and papillary muscle (own data and data from Hayashi et al., 2005), conscious dog and dog Purkinje fibre (data from Roden and Hoffman, 1985) and human subjects (data from Baker et al., 1983). Also depicted are concentration–response curves for inhibition of the hERG K⁺ current (solid grey line) and the hNav1.5 Na⁺ current (dashed grey line), and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003).

Because of this dual ion channel inhibition, flecainide and quinidine may interfere with both depolarization and repolarization, and hence the maximum rate of depolarization, V_max, of the cardiac AP and its equivalent Q_Na in slices as well as APD and FPD are of particular interest. In guinea pig and rabbit ventricular slices (Table 3, Figures 1A,B and 2A,B), guinea pig papillary muscle (Table 2) and rabbit Purkinje fibre (Table 2, Figure 1B), V_max was reduced and Q_Na was increased by both flecainide and quinidine with a concentration-dependence similar to that required for Na⁺ current inhibition. When recording ECGs in vivo, Na⁺ channel block manifests itself as slowing of conduction (i.e. widening of the QRS complex in the ECG). The corresponding effect was detected in conscious dogs (Table 4, Figures 1B and 2B), with a concentration-dependence and to an extent similar to those required for Na⁺ current inhibition.

Table 3.

Effects of selected anti-arrhythmics on AP and FP parameters from guinea pig (A) and rabbit (B) ventricular slice preparations in vitro^§

		AP&				FP
	Concentration (µM)	Vmax (V·s−1 or Δ%)	APD90 (ms or Δ%)	Triangulation (ms or Δ%)	−Vmax (V·s−1 or Δ%)	QNa (ms or Δ%)	FPD (ms or Δ%)	Triangulation (ms or Δ%)
A: guinea pig
Flecainide (NAP= 12/5, NFP= 8/2)	0	162.7	180.2	36.8	N/A	0.9	190.8	N/A
	1	−6.2	+1.4	−0.3		+8.3	+2.7
	3	−2.9	+2.5	−0.5		+30.6	+2.7
	10	−15.1	−2.0	−3.3		+41.0	−5.8
Quinidine (NAP= 6/5, NFP= 6/2)	0	203.6	177.0	27.9	N/A	0.8	192.1	N/A
	1	−15.2	+0.3	+6.0		+1.6	+9.0
	10	−25.1	−5.2	+20.8		+16.3	+10.3
	30					+42.4	−10.2
	100	−56.6	−28.7	+11.2		+79.7	−41.0
Atenolol (NAP= 6/5, NFP= 6/2)	0	222.6	169.3	26.3	N/A	0.8	197.8	N/A
	1	−14.7	−11.3	+5.0		−1.2	+1.9
	10	−21.7	−15.9	+21.8		+7.8	+6.0
	50					+3.3	+5.2
d,l-sotalol (NAP= 11/9, NFP= 9/2)	0	173.6	187.5	40.3	N/A	0.8	193.1	N/A
	1	+2.2	+3.6	+1.6		−2.3	−0.7
	3	+0.7	+5.3	+3.6		+5.9	+3.5
	10	+3.5	+12.1	+14.3		+4.9	+12.0
	30	+3.4	+18.2	+28.4		−6.2	+22.7
	100	+12.5	+19.3	+48.7			+25.0
Dofetilide (NAP= 7/4, NFP= 7/2)	0	181.1	178.4	25.9	N/A	0.9	202.4	N/A
	0.001	+1.1	+2.4	+2.3			+9.2
	0.01	−0.4	+13.1	+17.3			+10.2
	0.1	−7.3	+22.7	+34.0		−1.2	+20.8
	1	−4.2	+21.7	+32.7		+4.2	+48.6
Nifedipine (NAP= 8/4, NFP= 8/2)	0	138.2	178.2	27.9	N/A	1.0	212.3	N/A
	0.1	−3.6	−2.8	−4.6		−1.2	−3.9
	1	−1.5	−16.9	−8.2		+4.2	−26.3
	10	+7.7	−30.9	−15.0		+5.8	−62.9
Verapamil (NAP= 7/3, NFP= 7/2)	0	220.4	173.2	25.9	N/A	0.9	193.9	N/A
	0.1	−15.3	−3.5	−1.3		−5.3	−5.6
	1	−6.3	−8.7	+2.1		−17.3	−6.9
	10	−7.0	−23.1	+11.7		−11.1	−22.4
B: rabbit?
Flecainide (NAP= 6/2, NFP= 4/2)	0	170.9	182.0	52.3	N/A	0.9	183.9	N/A
	0.1	−3.9	−2.8	0.2
	1	−9.6	−3.8	0.7		19.4	−3.2
	3					33.6	−2.2
	10	−16.9	−2.9	4.0		30.9	−3.1
	30	−54.7	−17.7	−0.5
Dofetilide (NAP= 4/2, NFP= 4/2)	0	139.0	215.8	46.4	N/A	0.7	192.6	N/A
	0.001	0.0	6.5	2.5		−5.0	4.1
	0.01	−6.6	45.1	40.8		−5.3	18.4
	0.1	−5.8	201.3	88.6		2.3	23.1
Nifedipine (NAP= 3/2, NFP= 4/2)	0	141.7	215.7	36.3	N/A	0.9	211.7	N/A
	0.1	11.7	−4.1	4.9		5.1	0.4
	1	3.1	−9.3	7.0		−3.4	−14.4
	10	−1.7	−19.6	42.5		−0.7	−43.4

^§Data are mean values from AP and FP measurements in guinea pig and rabbit ventricular slice preparations; N_AP= i/j refers to the number of slices used for AP measurements with i slices from j animals (the same applies to FP measurements). Nomenclature analogous to Table 2; Q_Na, latency difference between primary and secondary peak of the FP as measure for Na⁺ conductance.

^&Pre-drug control resting membrane potential in guinea pig and rabbit ventricular slices. Guinea pig: flecainide, −82.4 ± 1.5 mV (mean ± SEM, n= 12); quinidine, −88.3 ± 2.5 mV (n= 6); atenolol, −87.2 ± 2.5 mV (n= 6); d,l-sotalol, −86.9 ± 1.3 mV (n= 11); dofetilide, −88.4 ± 1.1 mV (n= 7); nifedipine, −88.2 ± 2.9 mV (n= 8); verapamil, −87.4 ± 1.8 mV (n= 7). Rabbit: flecainide, −84.9 ± 1.4 mV (n= 6); dofetilide, −91.7 ± 1.3 mV (n= 5); nifedipine, −90.2 ± 2.8 mV (n= 3). No major changes (i.e. >±3 mV) of resting membrane potential occurred throughout the experiments.

Table 2.

Effects of selected anti-arrhythmics on rabbit cardiac Purkinje fibre and guinea pig papillary muscle AP parameters in vitro^§

		Rabbit Purkinje fibre AP&				Guinea pig papillary muscle AP
	Concentration (µM)	Vmax (V·s−1 or Δ%)	APD90 (ms or Δ%)	Triangulation (ms or Δ%)	−Vmax (V·s−1 or Δ%)	Vmax (V·s−1 or Δ%)	APD90 (ms or Δ%)	Triangulation (ms or Δ%)
Flecainide (n= 4–6)	0	393.0	445.3	124.0	−0.783
	0.1	+5.0	+1.0	−2.9	−2.1
	1	−2.6	0.0	+44.5	−36.7
	3	−19.6	−4.8	+164.0	−65.7
	10	−40.9	−1.5	+257.1	−78.2
Quinidine (n= 6)	0					153.2	158.8	58.7
	1					−5.8	+14.5	+1.7
	10					−21.6	+26.0	+21.0
	100					−59.0	+3.5	+12.4
Atenolol (n= 5–6)	0	259.4	475.7	135.5	−0.682
	1	+17.3	+0.1	+4.1	−6.3
	10	+4.3	−1.6	+4.7	−6.7
	100	+8.5	−1.5	−0.8	−3.3
d,l-sotalol (n= 5–6)	0	273.4	331.8	116.2	−0.775	170.9	163.3	62.8
	1					−5.8	+11.5	−4.7
	3	−21.4	+6.1	−5.0	−4.9
	10	−23.4	+41.8	+15.5	−18.2	−7.8	+24.9	+1.0
	30	+2.9	+119.0	+127.9	−29.0	−7.4	+29.6	+36.6
	100					−8.0	+45.5	+44.9
Dofetilide (n= 3–6)	0	194.8	415.8	145.9	−0.772
	0.3	−7.5	+12.1	+16.7	−12.2
	1	−12.5	+13.6	+11.5	−15.0
	3	−24.3	+66.5	+81.1	−42.7
	10	−9.7	+105.4	+98.4	−57.7
Nifedipine (n= 5–7)	0	318.0	447.3	77.4	−0.810
	0.1	+17.1	−3.8	−10.8	−0.9
	1	−2.3	−8.9	+19.8	+8.5
	10	+7.1	−36.3	+2.3	+5.4
Verapamil (n= 4–6)	0	305.8	397.3	80.3	−0.689
	0.1	−1.7	−2.1	+12.8	−13.3
	1	+5.7	+11.5	+31.1	−25.5
	10	+10.0	+61.7	+236.7	−61.7

^§Data are mean values from AP measurements in rabbit Purkinje fibres and guinea pig papillary muscle; n refers to the number of preparations investigated (usually one preparation per animal). Pre-drug control values (concentration = 0) are absolute values in V·s⁻¹ or ms, whereas all other values are expressed as % change versus pre-drug control (Δ%). V_max, maximal depolarization velocity; triangulation, difference between APD₉₀ and APD₅₀ (Purkinje fibre) or APD₉₀ and APD₃₀ (papillary muscle); −V_max, maximal repolarization velocity.

^&Pre-drug control resting membrane potential in rabbit Purkinje fibres: flecainide, −88.7 ± 0.6 mV (mean ± SEM, n= 6) (depolarization to −80.1 ± 3.4 mV at 10 µM); quinidine, N/A; atenolol, −89.4 ± 0.5 mV (n= 5); d,l-sotalol, −91.1 ± 1.4 mV (n= 6); dofetilide, −90.5 ± 0.5 mV (n= 6); nifedipine, −87.6 ± 0.4 mV (n= 9); verapamil, −89.4 ± 0.5 mV (n= 8). No major changes (i.e. >±3 mV) of resting membrane potential occurred throughout the experiments.

Table 4.

Effects of selected anti-arrhythmics on haemodynamic and ECG parameters in conscious, telemetry device-implanted Beagle dogs in vivo^§

		BPS	BPD	HR	QRS	PQ	QT	QTcV	Cmax	Cmax.u
		mmHg Δ%	mmHg Δ%	Bpm Δ%	ms Δ%	ms Δ%	ms Δ%	ms Δ%	µg·L−1	µM
Flecainide (mg·kg−1)	BL	129	76	76	45	114	234	253	–	–
	3	+1.1	+3.9	+13.0	−1.4	−2.8	−2.9	+0.9	238	0.33
	10	−8.4	−1.0	+20.6	+8.4	+4.0	−5.1	−0.9	693	0.97
	30	+6.9	+23.7	+49.3	+24.5	+15.5	−8.7	−4.8	2700	3.78
Quinidine (mg·kg−1)	BL	127	75	78	46	116	234	253	–	–
	5	−14.4	−15.9	+24.9	+2.8	−9.9	+4.1	+6.8	1631	0.65
	15	−15.5	−13.8	+39.3	+6.4	−10.8	+2.7	+8.0	2538	1.02
	45	−21.5	−18.4	+58.4	+14.7	−11.7	+2.6	+15.1	4229	1.69
Atenolol (mg·kg−1)	BL	129	75	76	49	117	237	254	–	–
	0.3	−7.3	−9.3	−5.9	−0.4	+10.2	+0.4	−1.3	144	0.52
	1.5	−6.2	−6.1	−11.2	+1.3	+17.1	+5.3	+1.6	594	2.16
	5	−2.5	−5.1	−12.6	−1.5	+18.1	+1.7	−2.0	2388	8.70
	15	−6.9	−8.6	−10.1	+3.8	+17.1	+4.2	+0.9	8069	29.39
d,l-sotalol (mg·kg−1)	BL	134	83	85	51	110	234	259	–	–
	5	−4.5	−8.2	−12.6	+1.3	+14.9	+13.2	+10.7	3073	11.28
	10	−10.3	−9.9	+3.8	+3.2	+15.9	+4.2	+11.4	5533	20.31
	25	−18.9	−24.9	−10.9	+2.5	+18.8	+17.5	+17.2	12767	46.87
	50	−18.2	−15.5	+3.6	+2.1	+14.2	+16.3	+19.1	26302	96.56
Dofetilide (mg·kg−1)	BL	133	78	88	48	108	227	255	–	–
	0.01	+0.1	−3.1	+1.4	+1.7	−2.5	+4.9	+4.6	1.55	0.0016
	0.03	+9.8	+4.9	−13.5	+2.0	+3.2	+13.6	+8.6	4.24	0.0044
	0.1	−2.9	−2.1	−6.7	−0.3	+11.6	+16.5	+12.8	16.90	0.0176
Nifedipine (mg·kg−1)	BL	130	78	85	43	112	228	253	–	–
	0.3	−6.6	−9.8	−4.0	+2.6	+4.9	+1.4	+0.3	16	0.0002
	1	−15.7	−16.9	+36.1	+1.7	−6.5	−8.5	−1.0	53	0.0006
	3	−20.6	−24.0	+90.4	−6.2	−16.2	−18.2	−3.5	177	0.0020
Verapamil (mg·kg−1)	BL	127	78	82	48	117	233	255	–	–
	3	−3.4	−4.3	+8.9	+3.3	+5.5	−0.1	+0.7	16	0.005
	10	−4.4	−2.6	+17.4	+1.7	+20.8	−3.7	−1.3	149	0.049
	20	−12.5	−10.5	+36.1	−3.3	+30.8	−5.9	+0.5	872	0.288

^§Data are mean values (n= 5–6 per dose) of the following haemodynamic and ECG parameters: BPD, diastolic blood pressure; BPS, systolic blood pressure; HR, heart rate; QTcV, QT interval corrected for heart rate according to Van de Water. Pre-drug baseline values (BL) are absolute values in mmHg or bpm or ms, whereas all other values are expressed as % change versus pre-drug baseline levels (Δ%). Maximal drug plasma concentrations (C_max) were determined in satellite animals (n= 3–4 per dose) and corrected for protein binding (C_max.u). The following data (MW, molecular weight; fu, protein-unbound fraction) were used to calculate C_max.u values: flecainide: MW 414.3, fu 58% (human: Zordan et al., 1993); quinidine: MW 324.4, fu 13% (dog: Rakhit et al., 1984); atenolol: MW 266.3, fu 97% (human: Drug Information); d,l-sotalol: MW 272.4, fu 100% (dog: Schnelle and Garrett, 1973?); dofetilide: MW 441.6, fu 46% (dog: Smith et al., 1992); nifedipine: MW 346.3, fu 4% (dog: Bayer internal data); verapamil: MW 454.6, fu 15% (dog: Belpaire et al., 1989).

Although flecainide is a potent inhibitor of hERG (Table 1), it did not substantially alter APD₉₀ and/or FPD₉₀ in guinea pig and rabbit ventricular slices (Table 3, Figure 1C), guinea pig papillary muscle and rabbit Purkinje fibre (Table 2, Figure 1C), unless at concentrations ≥10 µM. Also, QTcV in conscious dogs in vivo remained essentially unchanged (Table 4, Figure 1C) at concentrations overlapping and exceeding the human therapeutic range. It should be pointed out, however, that flecainide dramatically altered the shape of APs, particularly in rabbit Purkinje fibres, where the maximal repolarization velocity was reduced and significant triangulation was observed (Table 2).

The results situation was more complex with quinidine, since there was considerable variation between types of tissue preparation and species. Concentration-dependent APD₉₀ prolongation was only observed in rabbit Purkinje fibres (Lu et al., 2001; Aubert et al., 2006; Ducroq et al., 2007), at concentrations inhibiting hERG K⁺ current (Figure 2C). In guinea pig papillary muscle, APD₉₀ was prolonged at ≤10 µM, while at ≥10 µM APD₉₀ returned to baseline (Table 2, Figure 2C; Hayashi et al., 2005). In guinea pig ventricular slices, FPD₉₀ was prolonged at ≤30 µM and shortened at higher concentrations, whereas APD₉₀ was only shortened at ≥10 µM (Table 3, Figure 2C). In conscious dogs, the application of quinidine was associated with QTcV prolongation (Table 4, Figure 2C).

Class 2 anti-arrhythmic atenolol

The class 2 anti-arrhythmic drug atenolol lacks any relevant interactions with cardiac ion channels (e.g. the hERG K⁺ channel) (Table 1), and this was also reflected by the fact that atenolol hardly alters the shape of cardiac action or FPs. This was demonstrated in guinea pig ventricular slices (Figure 3, Table 3) and in rabbit Purkinje fibre (Table 2, Figure 3). In the former preparation, however, the application of atenolol was associated with a concentration-dependent shortening of the APD₉₀ in the range of 5–15%. Finally, in vivo, atenolol prolonged the PQ interval (lowered heart rate) but was without effect on the duration of the QTc interval in conscious dogs (Table 4, Figure 3B) at concentrations overlapping and exceeding the human therapeutic range (Figure 3B).

Figure 3.

(A) Examples of intracellular (AP) and extracellular (FP) recordings from guinea pig ventricular slices. Shown are superimposed tracings before (pre-drug control) and following atenolol exposure (10 µM). (B) Concentration-dependence of effects of atenolol on ΔAPD₉₀, ΔFPD₉₀, ΔQTcV and ΔPQ in the following preparations: rabbit Purkinje fibre, guinea pig ventricular slices and conscious dog. Also depicted is the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003).

Class 3 anti-arrhythmics d,l -sotalol and dofetilide

The characteristic feature of class 3 anti-arrhythmics is their ability to delay cardiac repolarization as for instance by inhibition of the hERG K⁺ channel by d,l-sotalol and dofetilide, with IC₂₀ concentrations of about 20 µM and 10 nM respectively (Table 1, Figures 4B and 5B). In guinea pig ventricular slice preparations, the effect of d,l-sotalol was characterized by a concentration-dependent drug-mediated prolongation of APD₉₀ and FPD₉₀, the extent of which was approximately 20% at 30 µM (Table 3, Figure 4). A similar concentration-dependence of APD₉₀ prolongation was observed in guinea pig papillary muscle (Table 2; Figure 4B), whereas rabbit Purkinje fibres were more sensitive to d,l-sotalol-mediated APD prolongation (Table 2; Figure 4B). In vivo, d,l-sotalol prolonged the QT interval, as demonstrated in conscious and anaesthetized dogs at plasma concentrations, and to an extent that was similar to that observed in guinea pig ventricular slices and to hERG K⁺ current inhibition (Table 4, Figure 4B).

Figure 4.

(A) Examples of intracellular (AP) and extracellular (FP) recordings from guinea pig ventricular slices. Shown are superimposed tracings before (pre-drug control) and following d,l-sotalol exposure (30 µM). (B) Concentration dependence of effects of d,l-sotalol on ΔAPD₉₀, ΔFPD₉₀ and ΔQTcV in the following preparations: rabbit Purkinje fibre (own data and data from Aubert et al., 2006), guinea pig ventricular slices and papillary muscle, conscious dog and anaesthetized dog (data from Mortensen et al., 1991). Also depicted is a concentration–response curve for inhibition of the hERG K⁺ current (dashed grey line; data from Ducroq et al., 2007) and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003).

Figure 5.

(A) Examples of intracellular (AP) and extracellular (FP) recordings from guinea pig (left) and rabbit (right) ventricular slices. Shown are superimposed tracings before (pre-drug control) and following dofetilide exposure (100 nM). (B) Concentration-dependence of effects of dofetilide on ΔAPD₉₀, ΔFPD₉₀ and ΔQTcV in the following preparations: rabbit Purkinje fibre (own data and data from Dumotier et al., 1999) and ventricular slices, guinea pig ventricular slices and conscious dog. Also depicted are a concentration–response curve for inhibition of the hERG K⁺ current (solid grey line) and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003).

In a similar manner, the effect of dofetilide in both guinea pig and rabbit ventricular slice preparations was characterized by a concentration-dependent drug-mediated prolongation of APD₉₀ and FPD₉₀, the extent of which was approximately 20% at 10–30 nM (Table 3, Figure 5). Also here, rabbit Purkinje fibres were more sensitive to dofetilide-mediated APD or FPD prolongation (Table 2; Figure 5B). In vivo, dofetilide prolonged the QT interval in conscious dogs at plasma concentrations and to an extent that was similar to the findings in ventricular slice preparations and to hERG K⁺ current inhibition (Table 4, Figure 5B).

Class 4 anti-arrhythmics nifedipine and verapamil

The class 4 anti-arrhythmics nifedipine and verapamil inhibited L-type calcium channels at submicromolar concentrations (Table 1, Figures 6B and 7B). In addition, both drugs were inhibitors of the hERG K⁺ channel (Table 1), albeit with one important difference. Nifedipine had very low potency as an hERG blocker displaying an IC₅₀ of about three orders of magnitude higher than that for calcium channel inhibition, whereas verapamil inhibited both hERG K⁺ and L-type calcium channel at almost identical concentrations (Table 1, Figures 6B and 7B).

Figure 6.

(A) Examples of intracellular (AP) and extracellular (FP) recordings from guinea pig (left) and rabbit (right) ventricular slices. Shown are superimposed tracings before (pre-drug control) and following nifedipine exposure (10 µM). (B) Concentration-dependence of effects of nifedipine on ΔAPD₉₀, ΔFPD₉₀ and ΔQTcV in the following preparations: rabbit Purkinje fibre and ventricular slices, guinea pig ventricular slices and papillary muscle (data from Hayashi et al., 2005) and conscious dog. Also depicted are concentration–response curves for inhibition of the hERG K⁺ current (dashed grey line) and the L-type Ca²⁺ current (solid grey line; from Uehara and Hume, 1985) and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003). The decrease in diastolic blood pressure (BP) in conscious dogs is also shown.

Figure 7.

(A) Examples of intracellular (AP) and extracellular (FP) recordings from guinea pig ventricular slices. Shown are superimposed tracings before (pre-drug control) and following verapamil exposure (10 µM). (B) Concentration-dependence of effects of verapamil on ΔAPD₉₀, ΔFPD₉₀ and ΔQTcV in the following preparations: rabbit Purkinje fibre (own data and data from Dumotier et al., 1999), guinea pig ventricular slices and papillary muscle (data from Hayashi et al., 2005), conscious dog and human subjects (Echizen et al., 1988; Mikus et al., 1990). Also depicted are concentration–response curves for inhibition of the hERG K⁺ current (dashed grey line) and the L-type Ca²⁺ current (from Zhang et al., 1999) and the range of therapeutically effective protein-unbound drug plasma concentrations in humans (orange area; data from Redfern et al., 2003).

The effects of nifedipine on the shape of the cardiac AP and FP were characterized by concentration-dependent shortening of the APD₉₀ and/or FPD₉₀ in guinea pig and rabbit ventricular slices (Figure 6; Table 3), and rabbit Purkinje fibres (Table 2, Figure 6B) and a decrease in the AP plateau potential (Figure 6A); AP upstroke or Q_Na was unaffected (Tables 2 and 3). The extent of APD/FPD shortening was similar in guinea pig slices, rabbit slices and rabbit Purkinje fibres, and amounted to approximately 10–25% at 1 µM. In contrast, the diastolic blood pressure in conscious dogs (Table 4, Figure 6B) was decreased to a similar extent at concentrations approximately three orders of magnitude lower; at these low single-digit nanomolar concentrations, the QTcV interval in conscious dogs was not prolonged but rather slightly shortened (Table 4, Figure 6B).

In guinea pig ventricular slice preparations, the effect of verapamil was characterized by a concentration-dependent decrease in APD₉₀ and FPD₉₀ (Figure 7; Tables 2 and 3); this decrease was approximately 20% at 10 µM. In contrast, concentration-dependent verapamil-mediated APD₉₀ prolongation together with pronounced triangulation prevailed in rabbit Purkinje fibres (Table 2, Figure 7B). In conscious dogs, verapamil prolonged the PQ interval at concentrations that were about one order of magnitude lower than those associated with AP/FP duration changes in vitro, while the QTc interval duration was hardly affected within the same concentration range (Table 4, Figure 7B).

Discussion

Class 1 anti-arrhythmics flecainide and quinidine

Quinidine and flecainide have been used in past decades to suppress/prevent ventricular tachyarrhythmias in humans [quinidine (Halkin et al., 1979; Carliner et al., 1980; Rosen and Wit, 1983; Duff et al., 1985), flecainide (Duff et al., 1981; Hodges et al., 1982; Lui et al., 1982; Estes et al., 1984) ]. Later, the pro-arrhythmic potential of these class 1 anti-arrhythmic drugs came more into focus (Morganroth and Horowitz, 1984) and led to investigations of the underlying electrophysiological mechanism(s).

Both flecainide and quinidine are inhibitors of the human cardiac sodium channel hNav1.5 as demonstrated here (Table 1, Figures 1B and 2B) and as expected from the literature (Hondeghem and Katzung, 1977; Konzen et al., 1990; Snyders and Hondeghem, 1990; Nitta et al., 1992; Ducroq et al., 2007). At slightly lower concentrations, however, quinidine and flecainide also inhibit the hERG K⁺ channel (Table 1, Figures 1C and 2C; Paul et al., 2002; Ducroq et al., 2007), and at higher concentrations, both drugs interfere with other K⁺ channels (Slawsky and Castle, 1994) and with L-type Ca²⁺ channels (Table 1).

As expected for hNav1.5 blockers, the maximum rate of depolarization, V_max, of the cardiac AP was reduced by flecainide and quinidine with a concentration-dependence similar to that required for Na⁺ current inhibition. This has been demonstrated in numerous ex vivo AP measurements, for example, in preparations of rabbit Purkinje fibre (Table 2, Figures 1B and 2B; Konzen et al., 1990; Aubert et al., 2006), dog Purkinje fibre (Burke et al., 1986), guinea pig papillary muscle (Table 2, Figure 1B; Hayashi et al., 2005) and guinea pig ventricular slices (Table 3, Figures 1B and 2B). In an in vivo situation, Na⁺ channel block manifests itself as a slowing of conduction (i.e. widening of the QRS complex in the ECG). The respective effect was detected in dogs (Table 4, Figures 1B and 2B; Tashibu et al., 2005; Toyoshima et al., 2005) and in humans [flecainide (Duff et al., 1981; Hodges et al., 1982; Lui et al., 1982; Conrad and Ober, 1984; Estes et al., 1984), quinidine (Henning and Nyberg, 1973; Halkin et al., 1979; Carliner et al., 1980; Duff et al., 1985)] in the same range of drug concentrations that cause hNav1.5 inhibition and V_max decrease, and that overlap with the range of therapeutically effective protein-unbound drug plasma concentrations in humans (Redfern et al., 2003). As an equivalent to a prolonged QRS complex in the ECG following Na⁺ channel blockade, the increase in the latency difference between the primary and the secondary peak of the FP (Q_Na) indicates reduced Na⁺ conductance in cardiac slices (Table 3; Figures 1A,B and 2A,B). In fact, both flecainide and quinidine increased Q_Na at concentrations that induced QRS widening in vivo (compare Tables 3 and 4).

Drugs that are highly selective blockers of the hERG K⁺ channel are expected to prolong the duration of AP/FP and the QT interval, whereas with mixed ion channel blockers such as flecainide and quinidine the net effect on APD, FPD and QT is more difficult to predict. Despite being a potent inhibitor of hERG (Table 1, Figure 1C), flecainide did not substantially alter APD₉₀/FPD₉₀ in rabbit Purkinje fibre (Table 2, Figure 1C), guinea pig and rabbit ventricular slices (Table 3, Figure 1A,C) and papillary muscle (Hayashi et al., 2005) unless at concentrations ≥10 µM (Figure 1C; Ducroq et al., 2007). Then, flecainide prolonged APD in guinea pig ventricular slices but shortened it in rabbit ventricular slices; this might be explained by a differential contribution of L-type Ca²⁺ channels to the shape of the ventricular slice AP, contributing more current in guinea pig and less in rabbit. Also, QTcV in conscious dogs in vivo remained essentially unchanged (Table 4, Figure 1C) at concentrations overlapping and exceeding the human therapeutic range (Redfern et al., 2003). Finally, in rabbit Purkinje fibres, flecainide reduced the maximal repolarization velocity resulting in significant triangulation (Table 2), whereas in ventricular preparations, such a dramatic alteration of the shape of APs was not observed. The most likely reason for this difference is that the Purkinje fibre AP plateau depends to a greater extent on Na⁺ and Ca²⁺ influx than that of ventricular muscle.

Quinidine presents us with a more complex picture, since there was considerable variation between types of tissue preparation and species. Concentration-dependent APD₉₀ prolongation was only observed in rabbit Purkinje fibres (Lu et al., 2001; Aubert et al., 2006; Ducroq et al., 2007), at concentrations inhibiting hERG K⁺ current (Figure 2C). In guinea pig papillary muscle, APD₉₀ was prolonged at ≤10 µM, while at ≥10 µM, APD₉₀ returned to baseline (Table 2, Figure 2C; Jurevicious et al., 1991?; Hayashi et al., 2005). In guinea pig ventricular slices, FPD₉₀ was prolonged at ≤30 µM and shortened at higher concentrations, whereas APD₉₀ was only shortened at ≥10 µM (Table 3, Figure 2C). Overall, the prolongation of APD and FPD at lower quinidine concentrations in guinea pig ventricular preparations was consistent with the inhibition of repolarizing hERG K⁺ currents, while at higher quinidine concentrations, the additional inhibition of depolarizing L-type Ca²⁺ currents (Table 1) shifted the balance towards an APD/FPD shortening. In conscious dogs, the application of quinidine was associated with QTcV prolongation (Table 4, Figure 2C) occurring at a similar magnitude as APD/FPD increases in the range of therapeutically effective protein-unbound drug plasma concentrations in humans (Figure 2C; Baker et al., 1983; Redfern et al., 2003). In dog Purkinje fibres, however, quinidine has been reported to mediate concentration-dependent APD₉₀ prolongation (Figure 2C; Roden and Hoffman, 1985; Davidenko et al., 1989; Nemeth et al., 1997; Lu et al., 2001) as well as APD₉₀ shortening (Figure 2C; Burke et al., 1986) and even a biphasic effect (Wyse et al., 1993). These latter results cannot be explained based on differential inhibition of depolarizing versus repolarizing ion channels but may be due to differences in experimental protocols.

In general, the effects of multichannel blockers like flecainide or quinidine on AP shape are difficult to predict. For flecainide, blocking effects on I_Kr, I_Kur, I_to and I_Ca have been reported. Since the contribution of individual channels to a certain AP shape shows large variations depending on species and location within the heart of one species, the effects of drugs like flecainide and quinidine will depend on the predominating conductance(s) during the plateau or repolarization phase and how they are affected by overlapping active concentration ranges (Dumotier et al., 1999), thus resulting in APD prolongation as well as shortening.

Class 2 anti-arrhythmic atenolol

For several decades, atenolol and other class 2 anti-arrhythmics have been amongst the mainstay drugs for treatment of heart failure and hypertension, and are known to possess few adverse effects, particularly with regard to pro-arrhythmia, and to improve life expectancy. The rather benign profile of atenolol is based on the lack of relevant interactions with cardiac ion channels, including the hERG K⁺ channel (Table 1; Kawakami et al., 2006), and is also reflected by the fact that atenolol hardly alters the shape of cardiac APs. This has been demonstrated in rabbit Purkinje fibre (Table 2, Figure 3B) and papillary muscle (Manley et al., 1986), and in guinea pig ventricular slices (Figure 3, Table 3). In the latter preparation, however, atenolol was associated with a concentration-dependent shortening of the APD₉₀, which was most likely as a result of ‘rundown’ of the preparation. Finally, in the in vivo situation, atenolol prolonged the PQ interval (lowered heart rate) but was without effect on the duration of the QTc interval in conscious dogs (Table 4, Figure 3B; McAinsh and Holmes, 1983; Davies and McAinsh, 1986; Coppi et al., 1987; Kvetina et al., 1999) and human volunteers and patients (Coppi et al., 1987; Way et al., 1988; Demolis et al., 1997) and even in case of human intoxication (Snook et al., 2000).

Class 3 anti-arrhythmics d,l -sotalol and dofetilide

Class 3 anti-arrhythmics had been developed with the specific aim to prolong the refractory period by prolonging the AP duration, this being the main distinguishing feature compared to class 1 anti-arrhythmics. The electrophysiological basis of the delayed cardiac repolarization is inhibition of the hERG K⁺ channel by d,l-sotalol and dofetilide (Table 1, Figures 4 and 5; Numaguchi et al., 2000; Davie et al., 2004; Ducroq et al., 2007). This is in line with concentration-dependent drug-mediated prolongation of APD₉₀ and FPD₉₀ in the following ex vivo preparations: rabbit Purkinje fibre (Table 2, Figures 4B and 5B; Abrahamsson et al., 1993; Dumotier et al., 1999; Lu et al., 2001; 2002; Aubert et al., 2006; Ducroq et al., 2007) and papillary muscle (Manley et al., 1986; Abrahamsson et al., 1993), dog Purkinje fibre (Gwilt et al., 1991; Knilans et al., 1991; Wyse et al., 1993; Lee et al., 1996; Nemeth et al., 1997; Lu et al., 2001) and papillary muscle (Nemeth et al., 1997), guinea pig papillary muscle (Table 2, Figure 4B; Tande et al., 1990; Davie et al., 2004), and guinea pig ventricular slices (Table 3, Figures 4 and 5). In vivo, both d,l-sotalol and dofetilide prolonged the QT interval as demonstrated in conscious (Table 4, Figures 4B and 5B; Schneider et al., 2005) and anaesthetized dogs (Mortensen et al., 1991; Tashibu et al., 2005; Vormberge et al., 2006) as well as in human patients (Wang et al., 1986; Way et al., 1988; Redfern et al., 2003) or intoxications (Elonen et al., 1979; Neuvonen et al., 1981; Wang et al., 1986; Edvardsson and Varnauskas, 1987?).

Although the direction of drug-induced changes was consistent throughout the various models in general, there was considerable variation in terms of sensitivity when individual models and species were compared. The magnitude of effects as well as the d,l-sotalol sensitivity of models was comparable for inhibition of the hERG K⁺ current, prolongation of APD₉₀/FPD₉₀ in guinea pig papillary muscle and ventricular slices and prolongation of the QTc interval in anaesthetized and conscious dogs (Figure 4B); there, the concentrations producing a 20% effect were mostly between 10 and 100 µM (Figure 4B). In contrast, both the magnitude of the effect and sotalol sensitivity appeared to be greater in rabbit and dog Purkinje fibres and in humans, since concentration–response curves were steeper and shifted to approximately 10-fold lower concentrations (Figure 4B). The picture was similar for dofetilide (Figure 5B) where 20% prolongation of APD₉₀ in rabbit and dog Purkinje fibres occurred in the single-digit nanomolar range and overlapped with human therapeutically effective protein-unbound plasma concentrations. However, for hERG K⁺ current inhibition, APD₉₀/FPD₉₀ prolongation in rabbit, dog and guinea pig ventricular muscle, and QTcV prolongation in conscious dogs, double digit nanomolar dofetilide concentrations were required to reach the 20% effect level. In addition, rabbit appears to be the most and guinea pig the least sensitive species, regardless of whether this rank order was established for Purkinje fibre or for ventricular muscle preparations. Literature reports point to the same direction with regard to (i) a differential species sensitivity of Purkinje fibres towards delayed repolarization (e.g. Lu et al., 2001) and (ii) a higher sensitivity of Purkinje fibres versus ventricular muscle towards the APD shortening or prolonging effects of sodium channel or hERG K⁺ channel blockers [e.g. Abrahamsson et al., 1996 (rabbit); Lathrop, 1985 (dog) ] because of quantitative differences of the underlying ionic currents [e.g. Cordeiro et al., 1998 (rabbit) ]. Another important point of discussion concerns a potential bias introduced by sex-specific differences in ventricular repolarization (for review, see Cheng, 2006). Females are more sensitive towards APD prolongation than males, particularly at extremely low stimulation rates, as demonstrated in rabbit Purkinje fibre (Lu et al., 2000) and Langendorff heart (Liu et al., 1998), and in human patients and volunteers (Lehmann et al., 1999; Somberg et al., 2011), whereas this issue remains controversial in guinea pig ventricle (Brouillette et al., 2007; Hreiche et al., 2009) and in canine ventricle (Xiao et al., 2006). Also, inward currents vary with sex and cardiac region as shown for the L-type calcium current (Sims et al., 2008).

Class 4 anti-arrhythmics nifedipine and verapamil

Both nifedipine and verapamil share the common feature of class 4 anti-arrhythmics; that is, they inhibit L-type calcium channels at submicromolar concentrations (Table 1, Figures 6B and 7B; Uehara and Hume, 1985; Charnet et al., 1987; Zhang et al., 1999; Shen et al., 2000). However, since the potency of nifedipine is markedly increased at depolarized potentials, nifedipine targets predominantly the L-type calcium channels modulating vascular smooth muscle tone, whereas the main therapeutic effect of verapamil is to delay atrioventricular conduction. In addition, both drugs are inhibitors of the hERG K⁺ channel. However, nifedipine had a very low potency of about three orders of magnitude beyond that of calcium channel inhibition, thus physiologically irrelevant, whereas verapamil inhibited both hERG K⁺ and L-type calcium channel at almost identical concentrations (Table 1, Figures 6B and 7B; Zhang et al., 1999; Zhabyeyev et al., 2000?).

Since nifedipine is a fairly selective calcium channel blocker, its effects on the shape of cardiac APs were straightforward and characterized by concentration-dependent shortening of the APD₉₀/FPD₉₀ in rabbit Purkinje fibres (Table 2, Figure 6B; Dumotier et al., 1999), dog Purkinje fibres (Lee et al., 1996), cat and guinea pig papillary muscle (Bayer et al., 1977; Jurevicious et al., 1991; Hayashi et al., 2005) and guinea pig and rabbit ventricular slices (Table 3, Figure 6). This APD₉₀ shortening was observed at concentrations approximately two to three orders of magnitude higher than those required to decrease the diastolic blood pressure in conscious dogs (Table 4, Figure 6B) or the range of therapeutically effective protein-unbound drug plasma concentrations in man (Redfern et al., 2003). Due to the functional antagonism of hERG K⁺ and L-type calcium channel block and consistent with the in vitro/ex vivo findings, the QTcV interval in conscious and anaesthetized dogs was not prolonged but rather slightly shortened (Table 4, Figure 6B; Amlie et al., 1979; Tashibu et al., 2005). Whenever nifedipine-induced QTc prolongation is reported, the actual consensus opinion attributes this to an inadequate heart rate correction of the QT interval (e.g. Toyoshima et al., 2005).

The well-balanced mixed ion channel blocker verapamil presented a decidedly different picture, since the effect on APD₉₀/FPD₉₀ in ex vivo preparations ranged from prolongation to shortening, obviously depending on the type of preparation (Tables 2 and 3, Figure 7B). Concentration-dependent verapamil-mediated APD₉₀ prolongation prevailed in canine (Cranefield et al., 1974; Rosen et al., 1974) and rabbit Purkinje fibres (Table 2, Figure 7B; Dumotier et al., 1999), whereas in guinea pig papillary muscle (Zhang et al., 1997; Hayashi et al., 2005) or ventricular slices (Table 3, Figure 7), APD₉₀ shortening was observed. The desired effect of PQ interval prolongation occurred in conscious and anaesthetized dogs within the same concentration range as the therapeutic effect in man (Table 4, Figure 7B; Echizen et al., 1988; Mikus et al., 1990; Shiina et al., 2000; Redfern et al., 2003), while the QTc interval duration was hardly affected, adequate heart rate correction was provided (Table 4, Figure 7B; Shiina et al., 2000; Fossa et al., 2002; Schneider et al., 2005; Tashibu et al., 2005; Toyoshima et al., 2005).

Limitations and opportunities

Similar to other multicellular preparations, it is conceivable that neurotransmitters are also released due to field stimulation in cardiac tissue slices. In our preparation, we used a small concentric stimulation electrode, which is expected to minimize the effect of electrical stimulation on neurotransmitter release. This problem, however, has not yet been addressed systematically by inhibiting neurotransmitter release or by blocking neurotransmitter effects.

In cardiac slice preparations, it might reasonably be expected that drug actions on fibroblasts could be an area of considerable interest or matter of concern (Porter and Turner, 2009). Fibroblasts within cardiac tissue are assumed to play an important role in bridging electrical activity between myocytes. Since fibroblasts are coupled to myocytes by connexins (MacCannell et al., 2007), drug interaction with connexins but also interactions with fibroblast ion channels might influence drug effects on the cardiac AP. Investigating such an interaction was beyond the scope of our present work, and we are also not aware of publications addressing this question. We believe that these putative interactions have to be studied on isolated fibroblasts and in co-culture with cardiac myocytes. Although the tissue architecture in cardiac slice preparations is in principle not different from that of other multicellular preparations, cardiac slices may be more readily accessible for the study of specific modulation of fibroblast characteristics by gene transfer and/or knock-down methods.

Conclusion

In this paper, we have presented integrated data regarding electrophysiological effects of selected anti-arrhythmic drugs (flecainide, quinidine, atenolol, sotalol, dofetilide, nifedipine, verapamil) using complementary in vitro and in vivo study types. These complementary approaches range from measurement of membrane currents (hERG, I_Na, I_Ca.L), to AP/FP in slices, Purkinje fibre AP and haemodynamic/ECG parameters (conscious dog), and also include published data. The main result of this complementary approach is that FP and AP recordings from heart slices correlate well with established in vitro and in vivo models in terms of pharmacology and predictability. Heart slice preparations yield similar results as papillary muscle but offer enhanced throughput for mechanistic investigations and reduce the use of laboratory animals.

Glossary

AP

action potential

APA

AP amplitude

APD

AP duration

BDM

2,3-butanedione monoxime

FP

field potential

FPD

FP duration

hERG

human ether-a-go-go-related gene

HK+

solution, high potassium solution

I_Ca.L

L-type Ca²⁺ current

I_Na

Na⁺ current

PM

papillary muscle

Q_Na

latency difference between primary and secondary peak of the FP as measure for Na⁺ conductance

QTcV

QT interval corrected for heart rate according to Van de Water

RMP

resting membrane potential

V_max

maximal AP upstroke velocity

Author contributions

Major contributions by the authors to this work were as follows: (1) conception and design (HMH, HL, EW); (2) collection and assembly of data: voltage clamp experiments (HMH), rabbit Purkinje fibre AP recordings (HMH), guinea pig papillary muscle AP recordings (RB), ventricular slice AP and FP recordings (AB, EW, MS, HL), studies in conscious dogs (MH) and literature search (HMH); (3) data analysis and interpretation (HMH, AB, EW, MS, HL, MH, RB); and (4) manuscript writing (HMH).

Conflict of interest

The authors have no conflicts of interest that could inappropriately influence this work.