Ionic mechanisms limiting cardiac repolarization reserve in humans compared to dogs

Abstract

The species-specific determinants of repolarization are poorly understood. This study compared the contribution of various currents to cardiac repolarization in canine and human ventricle. Conventional microelectrode, whole-cell patch-clamp, molecular biological and mathematical modelling techniques were used. Selective I_Kr block (50–100 nmol l⁻¹ dofetilide) lengthened AP duration at 90% of repolarization (APD₉₀) >3-fold more in human than dog, suggesting smaller repolarization reserve in humans. Selective I_K1 block (10 μmol l⁻¹ BaCl₂) and I_Ks block (1 μmol l⁻¹ HMR-1556) increased APD₉₀ more in canine than human right ventricular papillary muscle. Ion current measurements in isolated cardiomyocytes showed that I_K1 and I_Ks densities were 3- and 4.5-fold larger in dogs than humans, respectively. I_Kr density and kinetics were similar in human versus dog. I_Ca and I_to were respectively ∼30% larger and ∼29% smaller in human, and Na⁺–Ca²⁺ exchange current was comparable. Cardiac mRNA levels for the main I_K1 ion channel subunit Kir2.1 and the I_Ks accessory subunit minK were significantly lower, but mRNA expression of ERG and KvLQT1 (I_Kr and I_Ksα-subunits) were not significantly different, in human versus dog. Immunostaining suggested lower Kir2.1 and minK, and higher KvLQT1 protein expression in human versus canine cardiomyocytes. I_K1 and I_Ks inhibition increased the APD-prolonging effect of I_Kr block more in dog (by 56% and 49%, respectively) than human (34 and 16%), indicating that both currents contribute to increased repolarization reserve in the dog. A mathematical model incorporating observed human–canine ion current differences confirmed the role of I_K1 and I_Ks in repolarization reserve differences. Thus, humans show greater repolarization-delaying effects of I_Kr block than dogs, because of lower repolarization reserve contributions from I_K1 and I_Ks, emphasizing species-specific determinants of repolarization and the limitations of animal models for human disease.

Key points

• Cardiac repolarization, through which heart-cells return to their resting state after having fired, is a delicate process, susceptible to disruption by common drugs and clinical conditions.

• Animal models, particularly the dog, are often used to study repolarization properties and responses to drugs, with the assumption that such findings are relevant to humans. However, little is known about the applicability of findings in animals to man.

• Here, we studied the contribution of various ion-currents to cardiac repolarization in canine and human ventricle.

• Humans showed much greater repolarization-impairing effects of drugs blocking the rapid delayed-rectifier current I_Kr than dogs, because of lower repolarization-reserve contributions from two other important repolarizing currents (the inward-rectifier I_K1 and slow delayed-rectifier I_Ks).

• Our findings clarify differences in cardiac repolarization-processes among species, highlighting the importance of caution when extrapolating results from animal models to man.

Introduction

Many drugs can affect transmembrane K⁺ currents and thereby cause therapeutically useful (Honhloser & Woosley, 1994; Brendorp et al. 2001) or harmful (Surawicz, 1989; El-Sherif, 1992) effects. Blocking cardiac K⁺ channels prolongs repolarization and refractoriness, producing Class III antiarrhythmic effects both in ventricles and atria (Sing & Vaughan-Williams, 1970). Excessive lengthening of repolarization may induce life-threatening ventricular tachyarrhythmias like torsades de pointes (Hondeghem & Snyders, 1990; El-Sherif, 1992). Predicting the risk of such serious side effects is a major challenge in cardiac safety pharmacology. Torsade-risk estimation is hampered by a lack of easily usable methods and by incomplete understanding of the repolarization process in both experimental animals and humans.

Repolarization is controlled by two major inward currents (Na⁺ and Ca²⁺) and four major outward K⁺ currents (rapid and slow delayed-rectifier (I_Kr and I_Ks), transient-outward (I_to) and inward-rectifier (I_K1) currents), as well as other less well-characterized currents, electrogenic pumps and exchangers (Nerbonne & Kass, 2005). According to the concept of ‘repolarization reserve’ (Roden, 1998), normal repolarization is accomplished by multiple different potassium channels providing a strong safety reserve for repolarization. Thus, in normal cardiac tissue the pharmacological block or impairment of a single type of potassium channel does not necessarily lead to marked QT interval prolongation. However, in situations where the density of one or more types of potassium channels is decreased by congenital disorders or remodelling, inhibition of other potassium channels may lead to unexpectedly augmented action potential duration (APD) prolongation resulting in proarrhythmic reactions. In genetic channelopathies certain potassium channels, which normally contribute to repolarization, can attenuate the capability of the heart to repolarize (Biliczki et al. 2002; Jost et al. 2005).

Transmembrane ion currents flow through channel complexes composed of α- and β-subunit proteins including ERG (encoded by KCNH2), minK (KCNE1), MiRP1–4 (KCNE25), KvLQT1 (KCNQ1), Kv4.3 (KCND3), Kv1.4 (KCNA4), KChIP2 (KCNIP2) and Kir2.1–2.4 (KCNJ2, KCNJ12, KCNJ4, KCNJ14). These proteins are abundantly expressed in mammalian hearts, but their relative contributions vary considerably among species (Varróet al. 2000; Zicha et al. 2003). Differential K⁺ current expression causes interspecies differences in the response to K⁺ channel blocking drugs, affecting predictive value for their effects in humans (Nerbonne & Kass, 2005). Despite the very common use of the dog in evaluating long-QT risk in man, there is little quantitative information available about the relative responses of human versus canine hearts to QT-prolonging interventions or regarding underlying differences in ionic currents. Here, we compared the contribution of three particularly important K⁺ currents, I_Kr, I_K1 and I_Ks, to repolarization in dog and human hearts, studied the molecular basis of differences observed, and analysed their importance with a mathematical model.

Methods

For methodological details, please see Supplemental Methods.

Ethical approval and species

Patients

Hearts were obtained from organ donors whose non-diseased hearts were explanted to obtain pulmonary and aortic valves for transplant surgery. Before cardiac explantation, organ donors did not receive medication apart from dobutamine, furosemide, and plasma expanders. The investigations conformed to the principles of the Declaration of Helsinki. Experimental protocols were approved by the University of Szeged and National Scientific and Research Ethical Review Boards (Nos. 51-57/1997OEj and 4991-0/2010-1018EKU (339/PI/010)). After explantation, each heart was perfused with cardioplegic solution (for contents see Online Data Supplement) and kept cold (4–6°C) for 2–4 h prior to dissection.

Animals

All experiments complied with the Guide for the Care and Use of Laboratory Animals (NIH publication No 85-23, revised 1985). The protocols were approved by the Review Board of the Department of Animal Health and Food Control of the Ministry of Agriculture and Rural Development, Hungary (XII./01031/000/2008 and XIII./1211/2012). Adult mongrel dogs of either sex weighing 8–16 kg were anaesthetized with pentobarbital (30 mg kg⁻¹ i.v.). Hearts were removed through right lateral thoracotomies and rinsed in modified Locke's solution containing (mmol l⁻¹): Na⁺ 140, K⁺ 4, Ca²⁺ 1.0, Mg²⁺ 1.0, Cl⁻ 126, HCO₃⁻ 25 and glucose 11; pH 7.35–7.45, 95% O₂-5% CO₂, 37°C.

Action potential measurements

Action potentials (APs) were recorded in right ventricular trabeculae and papillary muscle preparations (<2 mm diameter), from 15 non-diseased human donor hearts (9 male and 6 female, age = 44.6 ± 5.9 years) and 25 dogs, with conventional microelectrode techniques, as described in detail previously (Varróet al. 2000; Biliczki et al. 2002; Jost et al. 2005).

Transmembrane current measurements

Cell isolation

Ventricular cardiomyocytes were enzy-matically dissociated from the left ventricular midmyocardial free wall of 10 additional non-diseased human donor hearts (5 male and 5 female, age = 43.4 ± 5.3 years) and 21 dog hearts with previously described procedures (Varróet al. 2000; Biliczki et al. 2002; Jost et al. 2005).

Experimental protocol

Rod-shaped, striated cardiomyocytes were placed in a recording chamber on the stage of inverted microscopes Olimpus, IX51 (Olympus Ltd, Tokyo, Japan) and Nikon TMS (Nikon Ltd, Tokyo, Japan) and allowed to adhere. The solutions, equipment and voltage-clamp protocols (see Supplemental Methods) were as previously detailed for K⁺ currents (Varróet al. 2000; Biliczki et al. 2002; Jost et al. 2005) and for L-type Ca²⁺ current (I_CaL) and Na⁺–Ca²⁺ exchanger (NCX) current (Hobai et al. 1997; Birinyi et al. 2005).

Molecular biology

Reverse transcription (RT) quantitative polymerase chain reaction (qPCR)

Left ventricular midmyocardial free-wall samples were obtained from eight human (7 male and 5 female, age = 45.2 ± 3.7 years) and eight dog hearts, and snap-frozen in liquid N₂. RNA was isolated with the Qiagen RNase Tissue kit (Amersham). Reverse transcription (RT) was performed with Superscript-II RNase H-Reverse Transcriptase (Invitrogen). QPCR was performed on a RotorGene-3000 instrument (Corbett Research, Australia) with gene-specific primers (Supplemental Table 1) and SybrGreen. Expression values were normalized to β-actin. Triplicate standard curves were run for each experiment. Data analysis was performed with the Pfaffl method (Pfaffl, 2001), correcting for amplification efficiency differences.

Western blot

Membrane proteins were obtained from the same samples used for qPCR. Samples were suspended in lysis buffer, dounced and centrifuged (2000 ×g, 10 min, 4°C). The supernatant was resuspended in lysis buffer containing 2% Triton X-100. After 1.5 h incubation on ice, samples were ultracentrifuged (100 000 ×g, 35 min, 4°C), supernatants collected and stored at −70°C. Protein concentration was measured by the Lowry method and samples diluted in loading buffer for SDS–polyacrylamide gel electrophoresis. Fractionated proteins were transferred onto polyvinylidine difluoride (PVDF) membranes, blocked in Tris buffer supplemented with Tween-20 (TBST) and 10% non-fat milk (BioRad, USA), then incubated overnight (4°C) with rabbit polyclonal primary antibodies against Kir2.1, Kir2.2, Kir2.3, ERG, minK and KvLQT1, goat anti-Kir2.4 (Santa Cruz Biotechnology) or mouse monoclonal anti-α-sarcomeric actin (DAKO). Bound primary antibodies were detected with anti-rabbit, anti-goat or anti-mouse secondary antibodies conjugated to horseradish peroxidase. Immunoreactivity was visualized with enhanced chemoluminescence and analysed with ImageJ™. All values were quantified relative to internal controls on the same samples (α-actin for Kir2.x, KvLQT1 and minK, GAPDH for ERG).

Immunohistochemistry

Isolated dog (n= 6) and human (3 male, 1 female, age = 48.3 ± 4.7 years) left ventricular midmyocardial free-wall ventricular cardiomyocytes on glass coverslips were fixed with acetone. Samples were rehydrated with calcium-free phosphate-buffered saline (PBS) and blocked for 2 h with PBST (PBS with 0.01% Tween) containing 1% BSA at room temperature. Incubation with the primary polyclonal rabbit antibody for 1.5 h at room temperature was followed by 1 h incubation with secondary antibodies (Alexafluor 448-conjugated goat anti-rabbit IgG). Control samples were incubated only with secondary antibody.

Fluorescence images were obtained with an Olympus FV1000 confocal laser-scanning microscope and standardized parameter settings. Images were quantified in greyscale TIFF format with ImageQuant™ software. On each image, three to five random strips were selected and fluorescence profiles plotted. Baseline pixels were identified and subtracted from total profile area.

Statistics

Results are expressed as means ± SEM. Statistical significance was determined by two-tailed Student's t tests and ANOVA with Bonferroni-corrected post hoc t tests as appropriate. Results were considered significant for P < 0.05.

Results

Current densities

I_K1 was recorded with 300 ms 0.33 Hz test pulses from a holding potential of −80 mV (Fig. 1A) and quantified based on end-pulse amplitude. I_K1 was significantly larger in dog than human cardiomyocytes (Fig. 1B). Maximum outward current density at −60 mV was almost 3-fold greater in dog versus human (1.72 ± 0.07 pA pF⁻¹ vs. 0.65 ± 0.1 pA pF⁻¹, n= 21–28, Fig. 1C).

Figure 1. Inward-rectifier potassium current (I_K1) in human and dog ventricular cardiomyocytes.

A, original I_K1 recordings in a human (top traces) and a dog (bottom traces) ventricular myocyte. Voltage protocol shown above traces. B, mean ± SEM I_K1 density–voltage relations. C, mean ± SEM I_K1 density at −60 mV (left) and −140 mV (right) membrane potentials. *P < 0.05, **P < 0.01 dog versus human. n= number of experiments.

Mean I_Kr and I_Ks data are shown in Fig. 2. I_Kr data are shown in panels A–C and I_Ks data in panels D–F. Examples of original I_Kr recordings are in the top row, and I_Ks recordings in the middle row. I_Kr tail current at −40 mV after 1000 ms test pulses (0.05 Hz) did not differ significantly between species (Fig. 2C). In contrast, I_Ks tail current at −40 mV after 5000 ms test pulses (0.1 Hz) was about 4.5-fold larger in dog versus human (Fig. 2F).

Figure 2. I_Kr and I_Ks in human and dog ventricular cardiomyocytes.

A and B, original I_Kr recordings from a human (A) and a dog (B) ventricular cardiomyocyte. C, mean ± SEM I_Kr tail current density–voltage relations. D and E, original I_Ks recordings from a human (A) and a dog (B) ventricular cardiomyocyte. F, mean ± SEM I_Ks tail current density–voltage relations. n= number of experiments. *P < 0.05, **P < 0.01 and ***P < 0.001.

To estimate the magnitude of I_K1, I_Kr and I_Ks activated during the cardiac action potential, we compared the amplitudes of the BaCl₂-sensitive (I_K1), E-4031-sensitive (I_Kr) and L-735,821-sensitive (I_Ks) currents during ‘action potential’ test pulses. These test pulses were obtained by digitizing representative right ventricular human and canine action potentials recorded with conventional microelectrodes (Fig. 3A). Under these conditions, the BaCl₂-sensitive I_K1 difference current flowing during the AP was substantially larger in dog than in human (Fig. 3B), while the E-4031-sensitive I_Kr difference current was similar (Fig. 3C). The L-735,821-sensitive I_Ks during the action potential plateau phase was very small and not clearly different between the two species (Fig. 3D).

Figure 3.

A, currents recorded with action potential voltage-clamp waveforms, obtained by recording typical normal human or canine ventricular action potentials with a conventional microelectrode in a multicellular papillary muscle preparation. B–D, original BaCl₂ (I_K1, purple recordings, B), E-4031 (I_Kr, red recordings, C) and L-735,821 (I_Ks, green recordings, D) sensitive currents obtained by digitally subtracting currents elicited by action potential test pulses in the presence of the blocker from current in the same cell prior to the blocker in human (left panels) and dog (middle panels) ventricular myocytes. Right panels represent corresponding mean amplitudes of drug-sensitive I_K1, I_Kr and I_Ks currents in 4–13 cells per measurement. Arrows indicate the points at which current amplitudes were determined. Bars represent means ± SEM; corresponding n values are provided for each current and species.

The activation and deactivation kinetics of I_Kr and I_Ks measured at the whole range of activating and deactivating membrane potentials are shown in Fig. 4. The I_Ks kinetics of human and dog are quite similar (Fig. 4A and B). I_Kr deactivation (Fig. 4C) at voltages (−70 and −60 mV) relevant to physiological current deactivation (i.e. near the resting potential) consisted predominantly of a rapid phase with a time constant of 200–400 ms, not significantly different between human and dog. At more positive voltages, the kinetics became more clearly biexponential. The rapid-phase time constants were similar at all voltages for human and dog. At voltages negative to −30 mV, the slow-phase time constant was also similar, whereas at more positive voltages the slow-phase time constant was greater in dog.

Figure 4. The voltage dependence of the activation and deactivation kinetics of human and canine I_Kr and I_Ks.

A, voltage dependence of activation kinetics. I_Kr and I_Ks were activated by test pulses with durations from 10 to 5000 ms, to test potentials ranging from 0 to 50 mV; then the cells were clamped back to −40 mV. The amplitudes of tail currents as a function of the duration of the depolarization were well fitted by single exponentials. B, the voltage dependence of I_Ks deactivation kinetics was determined by activating I_Ks with 5000 ms test pulses to 50 mV from a holding potential of −40 mV. Then the cells were clamped back for 2 s to potentials ranging from −50 to 0 mV (pulse frequency 0.1 Hz) and the deactivation time course of the tail current was fitted by a single exponential function. C, the voltage dependence of I_Kr deactivation kinetics was determined by activating I_Kr with 1000 ms test pulses to 30 mV from a holding potential of −40 mV. Then the cells were clamped for 16 s to potentials ranging from −70 to 0 mV (pulse frequency 0.05 Hz) and the deactivation time course of the tail current was fitted by a double exponential function. The left panel shows the voltage dependence of slow and fast time constants. An expanded version of the results for voltage dependence of the fast time constants is provided in the right bottom panel. The right top panel shows the relative amplitudes of the fast and slow components at different voltages in dog (black) and human (red) ventricular myocytes.

Species-dependent contributions of I_K1, I_Kr and I_Ks to repolarization

The contribution of I_K1, I_Kr and I_Ks to repolarization was investigated (Fig. 5) by selectively blocking these currents with BaCl₂ (10 μmol l⁻¹), dofetilide (50 nmol l⁻¹) and HMR-1556 (1 μmol l⁻¹), respectively. We previously reported that 10 μmol l⁻¹ BaCl₂ blocks over 70% of I_K1 without affecting I_Kr, I_Ks and I_to (Biliczki et al. 2002). In human ventricular muscle, selective inhibition of I_K1 only marginally prolonged AP duration (APD, by 4.8 ± 1.5%), while it caused significant APD prolongation in dog (17.9 ± 2.1%, P < 0.05 vs. human, n= 7–11). In contrast, selective inhibition of I_Kr caused markedly greater APD prolongation in humans (56.3 ± 8.4%) compared to the dog (21.7 ± 2.5%, P < 0.05, n= 17–20). The differential response was due to differences in maximal effects and not drug sensitivity per se, as shown by similar dofetilide IC₅₀ values between species (Supplemental Fig. 1). I_Ks block did not significantly alter APD in either studied species.

Figure 5. Effect of selective I_K1 (10 μm BaCl₂), I_Kr (50 nmol l⁻¹ dofetilide) or I_Ks (1 μmol l⁻¹ HMR-1566) block on APs measured with standard microelectrode techniques in canine and human right papillary muscles.

A, recordings (at 1 Hz) before and after 40 min superfusion with BaCl₂ (left), dofetilide (middle) or HMR-1566 (right). Corresponding mean±SEM values for controls (C) and drug (D) effects are given under each action potential recordings. B, mean ± SEM AP duration at 90% of repolarization (APD₉₀) under each condition. n= number of experiments, **P < 0.01 and ***P < 0.001.

Contributions to repolarization reserve

We then studied the role of I_K1 and I_Ks differences in contributing to the larger APD increases produced by I_Kr block in human versus canine cardiomyocytes. Tissues were exposed to dofetilide in the absence or presence of 10 μmol l⁻¹ BaCl₂ to inhibit I_K1 (Fig. 6A) or HMR-1566 to block I_Ks (Fig. 6B). The change in APD (relative to BaCl₂-free control) caused by dofetilide alone indicates the effect of the drug with repolarization reserve intact, whereas the change caused in the presence of BaCl₂ (dofetilide + BaCl₂ vs. BaCl₂ alone) indicates the effect with I_K1 suppressed, i.e. the contribution of I_K1 to repolarization reserve. In human cells, dofetilide increased APD by 59 ± 5% in the presence of BaCl₂, versus 44 ± 4% in the absence of BaCl₂. The relative increase from 44% prolongation with I_K1 intact to 59% prolongation with I_K1 removed indicates a 34% increase in I_Kr blocking effect with I_K1 suppressed. For dog cells, dofetilide increased APD by 25 ± 2% in the presence of BaCl₂, versus 16 ± 2% in the absence of BaCl₂, indicating a 56% increase in I_Kr blocking effect with I_K1 suppression. This result confirms a larger contribution of I_K1 to repolarization reserve in the dog versus man. For I_Ks (Fig. 6B), dofetilide increased APD by 63 ± 4% in the absence of HMR-1566-induced I_Ks block in humans, versus 73 ± 2% in the presence of HMR-1566, an increase of 16% attributable to the loss of the I_Ks contribution. In the dog, dofetilide prolonged APD by 29 ± 5% in the absence of HMR-1566, versus 43 ± 4% in its presence, indicating a 49% enhancement attributable to loss of I_Ks. Thus, the larger I_Ks of canine tissues also contributes to greater repolarization reserve versus humans.

Figure 6. Effect of combined I_Kr+I_K1 and I_Kr+I_Ks inhibition in human and dog ventricular muscle preparations (endocardial impalements).

A, representative APs at baseline (circle), following exposure to 10 μmol l⁻¹ BaCl₂ (triangle), 50 nmol l⁻¹ dofetilide (diamond), and combined 10 μmol l⁻¹ BaCl₂+ 50 nmol l⁻¹ dofetilide (rectangle) in human (top traces) and dog (bottom traces) ventricular muscle. Brackets show average differences between conditions indicated. B, representative APs at baseline (circle), following exposure to 1 μmol l⁻¹ HMR-1566 (triangle), 50 nmol l⁻¹ dofetilide (diamond), and combined 1 μmol l⁻¹ HMR-1566 + 50 nmol l⁻¹ dofetilide (rectangle) in human (top traces) and dog (bottom traces) ventricular muscle. Brackets show average differences between conditions indicated.

Ion channel subunit expression

To assess the potential molecular basis for the observed differences in I_K1 and I_Ks densities, qPCR was applied for subunits underlying I_K1, I_Kr and I_Ks. Gene expression values for I_K1-encoding subunits are shown in Fig. 7A. Kir2.1-encoding mRNA (KCNJ2) was >2-fold more abundant in the dog than the total mRNA level for Kir2.1, Kir2.2, Kir2.3 and Kir2.4 combined in the human. The KCNH2 gene encoding I_Kr was equivalently expressed in canine and human ventricle (Fig. 7B). KCNQ1 gene expression was not significantly different between human and dog (Fig. 7C), but the KCNE1 gene encoding the I_Ksβ-subunit protein minK was ∼6-fold more strongly expressed in dog. Examples of Western blots for Kir2.x, ERG, KvLQT1 and minK proteins are shown in Fig. 7D–F. Mean data are provided in Table 1. In agreement with qPCR-findings, Kir2.1 was significantly stronger in canine than human hearts, whereas Kir2.2 was stronger in humans. ERG was detected as two larger molecular mass bands (Fig. 7E) corresponding to ERG1a (∼150 and 165 kDa) and two smaller bands corresponding to ERG1b (∼85 and 95 kDa). ERG1a was less abundant in human samples, while ERG1b band intensities were not significantly different from dogs. The very similar expression of ERG1b, in agreement with physiological data (Figs 2C and 3), is consistent with recent evidence for a particularly important role of ERG1b in forming functional I_Kr (Sale et al. 2008) and with a recent study of Purkinje fibre remodelling with heart failure (Maguy et al. 2009). MinK bands were also stronger in dog hearts, whereas KvLQT1 band intensity was greater in human.

Figure 7. Expression of I_K1-related (Kir2.x), I_Kr pore-forming (ERG) and I_Ks-related subunits (KvLQT1 and minK).

A–C, mean ± SEM mRNA levels of Kir2.x (A), ERG (B) and KvLQT1/minK (C) subunits in left ventricular human (n= 6–8) and dog (n= 816) preparations. *P < 0.05, **P < 0.01 and ***P < 0.001. n= number of experiments. D–F, representative Western blots for Kir2.x (D), ERG (E) and KvLQT1/minK (F) in human and dog left ventricular preparations.

Table 1.

Protein expression data for ion channel subunits in human versus dog ventricular tissues

Currents/subunits	Subunit†	Human†	Dog
IK1 subunits	Kir2.1 (n= 4/4)	0.22 ± 0.01	0.45 ± 0.06*
	Kir2.2 (n= 4/4)	0.64 ± 0.03**	0.37 ± 0.02
	Kir2.3 (n= 4/4)	0.10 ± 0.01	0.09 ± 0.007 (P= NS)
	Kir2.4 (n= 4/4)	0.01 ± 0.002	0.20 ± 0.009**
IKr subunits	ERG1a (n= 5/4)	0.30 ± 0.16	0.97 ± 0.27**
	ERG1b (n= 5/4)	0.71 ± 0.05	0.73 ± 0.07 (P= NS)
IKs subunits	KvLQT1 (n= 4/4)	0.15 ± 0.01**	0.05 ± 0.003
	MinK (n= 4/4)	0.31 ± 0.01	0.40 ± 0.05*

Mean ± SEM data. *P < 0.05, **P < 0.01, ***P < 0.001. n designates number of samples from humans/dogs.

†All values are expressed as arbitrary optical density units, quantified relative to an internal control on the same sample (α-actin for Kir2.x, KvLQT1 and minK, GAPDH for ERG).

We also performed immunohistochemical analyses on isolated cardiomyocytes (Fig. 8), with identical image settings for human versus canine cells. Examples are shown in Fig. 8A. Anti-Kir2.1 showed significantly stronger staining for canine cells (Fig. 8B), and Kir2.3 staining was also slightly but significantly greater for dog. In contrast, ERG staining was comparable for the two species (Fig. 8C). KvLQT1 staining was modestly but significantly greater for human cells (Fig. 8D), but in keeping with the qPCR data, mink staining was much greater (∼5-fold) for dog cells versus human. Supplemental Fig. 2 presents negative controls for immunostaining measurements.

Figure 8. Immunofluorescence confocal microscope image analysis for I_K1-related (Kir2.x), I_Kr pore-forming (ERG) and I_Ks-related (KvLQT1 and MinK) subunits in left ventricular cardiomyocytes.

A, representative immunofluorescence images of human (left) and dog (right) cardiomyocytes. Dark-field images of typical human and dog ventricular cardiomyocytes are shown at the bottom. B–D, mean ± SEM fluorescence intensities for various subunits in human versus dog cardiomyocytes. Results are shown for Kir2.x (B), ERG (C) and KvLQT1 and minK (D) subunits. n= number of experiments. *P < 0.05 and ***P < 0.001 for dog versus human.Constant image-settings were maintained for each construct for all cells studied.

Other ionic current differences and in silico assessment

The functional, pharmacological, and biochemical data described above all point to reduced repolarization reserve due to smaller I_Ks and I_K1 expression in human hearts as the basis for their larger APD prolonging response to I_Kr inhibition. To assess the potential role of other ionic current differences, we compared several other currents between canine and human hearts. I_to, recorded as the difference between peak and end-pulse current during 300 ms depolarizing pulses from −90 mV (0.33 Hz), was smaller in human versus dog (Fig. 9A). I_CaL evoked by 400 ms test pulses from −40 mV was ∼30% larger in human (Fig. 9B). Recovery kinetics of I_to (Supplemental Fig. 3A) and I_Ca (Supplemental Fig. 3B) currents were not statistically different in myocytes from human and dog ventricle. Ni²⁺ (10 mmol l⁻¹)-sensitive NCX current was not significantly different between species (Fig. 9C and D).

Figure 9.

A, I_to current–voltage density (I–V relationship) relation obtained with the inset protocol. *P< 0.05 and +P < 0.05 for human versus dog. I–V relationships for I_to are determined and depicted as peak current (open circles and squares) and as sustained current (closed circles and squares) as well. B, I_CaL current–voltage density relation obtained with the insetprotocol. *P < 0.05 for human vs. dog. I–V relationships for I_Ca are determined and depicted as peak current (open circles and squares) and as sustained current (closed circles and squares) as well. C, ramp protocol was applied to measure current before and after application of Ni²⁺ (10 mmol l⁻¹) under conditions to isolate NCX. Representative Ni²⁺-sensitive difference currents from dog and human cells are shown below. D, mean inward (at −80 mV) and outward (at +50 mV) NCX current density values.

To assess the contribution of ionic current components to repolarization reserve in human versus canine hearts, we initially adapted the Hund–Rudy dynamic (HRd) canine ventricular AP model (Hund & Rudy, 2004). We then adjusted the current densities in the dog model according to the experimentally observed differences in humans, to obtain ‘humanized’ APs (see Supplemental Methods). Supplemental Fig. 4 shows the resulting simulations: APD₉₀ at 1 Hz in the dog model was 209 ms, versus human 264 ms, close to experimentally determined values (APD₉₀ at 1 Hz: dog 227 ms, human 270 ms). I_Kr block increased APD₉₀ by 26% in the human AP model (Supplemental Fig. 4A) versus 15.5% in the dog model (Supplemental Fig. 4B), qualitatively consistent with experimental findings (56%, 22% respectively). I_Kr inhibition increased human APD₉₀ by 71.2% in the presence of I_K1 block, indicating a 173.8% increase in I_Kr blocking effect with the I_K1 contribution to repolarization reserve suppressed (Supplemental Fig. 4A). For the canine model (Supplemental Fig. 4B), I_Kr block increased APD₉₀ by 45.4% in the presence of I_K1 block, indicating a 193.5% increase in I_Kr blocking effect when I_K1 is decreased. This result is consistent with experimental data suggesting a larger contribution of I_K1 to repolarization reserve in the dog. I_Kr block prolonged human APD₉₀ by 29.4% (Supplemental Fig. 4C) in the presence of I_Ks inhibition, an increase of 14.6% attributable to the loss of I_Ks contribution to repolarization reserve. For the dog AP model (Supplemental Fig. 4D), I_Kr block prolonged APD by 23.8% in the presence of I_Ks inhibition, indicating a 53.6% enhancement attributable to loss of the repolarization reserve effect of I_Ks. Thus, the model also confirms the importance of larger I_Ks to greater repolarization reserve in dogs. Finally, we used the model to explore the contributions of I_CaL and I_to differences. Supplemental Fig. 5 shows the APD changes induced by I_Kr inhibition in canine (panel A) and human (panel B) models. The effect of I_Kr inhibition in the human model was then verified with I_CaL (panel C) or I_to (panel D) modified to canine values. APD₉₀ increases in the human model resulting from I_Kr inhibition were minimally affected by substituting canine I_to in the human model. Substituting canine I_CaL into the human model enhanced the I_Kr blocking effect on APD, whereas if canine I_CaL contributed to the larger repolarization reserve in the dog it should reduce the APD prolonging effect. These results indicate that I_CaL and I_to differences do not contribute to the enhanced repolarization reserve in the dog.

To assess further the contribution of ionic current components to repolarization reserve in human versus canine hearts, we performed the analysis in a reverse fashion, with the more recently published O’Hara–Rudy dynamic (ORd) human ventricular AP model (O’Hara et al. 2011, see Supplemental Methods). Figure 10 shows the resulting simulations: APD₉₀ at 1 Hz in the canine and human models were 210 ms and 271 ms (versus experimental APD₉₀ at 1 Hz: dog 227 ms, human 270 ms). I_Kr block increased APD₉₀ by 42.4% in the human versus 29.4% in the dog model, consistent with experimental findings (56%, 22% respectively). With the human ionic model (Fig. 10A), I_Kr block increased APD by 58.7% in the presence of I_K1 block, versus 42.4% in the absence of I_K1 block. These results indicate a 38.3% increase in I_Kr blocking effect on APD with I_K1 blocked. For the dog ionic model (Fig. 10B), I_Kr block increased APD by 45.8% in the presence of I_K1 block, versus 29.4% in the absence of I_K1 block, indicating a 55.7% increase in I_Kr blocking effect when I_K1 was decreased. This result confirms the notion based on our experimental data, indicating a larger contribution of I_K1 to repolarization reserve in the dog compared to man. I_Kr block increased APD by 42.4% in the absence of I_Ks block in the human model (Fig. 10C), versus 50.3% in the presence of I_Ks block, an increase of 18.5% attributable to the loss of I_Ks contribution to repolarization reserve. In the dog ionic model (Fig. 10D), I_Kr block prolonged APD by 29.4% in the absence of I_Ks block, versus 46.9% in its presence, indicating a 59.4% enhancement attributable to loss of the repolarization reserve effect of I_Ks. Thus, the model also confirms the importance of larger I_Ks to greater repolarization reserve in dogs. Finally, we also used this modelling approach to explore the contributions of I_CaL and I_to differences, and found no evidence that they contribute to the differences in I_Kr blocking effects between human and dog (Supplemental Fig. 6).

Figure 10. Simulations of effect of combined I_K+I_K1 and I_Kr+I_Ks inhibition on human and dog ventricular muscle APs by applying the O’Hara dynamic (ORd) canine ventricular AP model.

A, simulated human APs at control, following I_K1 block (70% reduction), I_Kr block (50% reduction), and combined I_K1+I_Kr block. B, corresponding data for dog I_K1+I_Kr block. C, simulated human APs at control, following I_Ks block (50% reduction), I_Kr block (50% reduction), and combined I_Ks+I_Kr block. D, corresponding data for dog I_Ks+I_Kr block.

Discussion

In this study, we found that I_Kr inhibition causes substantially greater APD prolongation in human versus canine ventricular muscle, indicating reduced repolarization reserve in man. Ionic current measurements showed larger I_K1 and I_Ks densities in canine versus human hearts and APD studies with selective blockers indicated larger repolarization reserve in canine hearts due to stronger I_K1 and I_Ks contributions. Expression studies suggested that the ionic current differences are due to species-related differences in mRNA expression of underlying subunits.

Experimental model considerations

We compared experimental data between non-diseased human donor hearts and canine hearts. There is a potential difference in relative maturity/age between the humans and dogs that provided our tissue samples, which were essentially impossible to control, other than by virtue of the fact that both study populations comprised adult and not senescent individuals. Important transmural and regional differences in ion channel subunit protein expression and current densities exist within the heart. Extrapolation of our findings to the whole heart must therefore be cautious. We were careful to perform all measurements in corresponding regions of canine and human hearts to ensure comparability. Current and mRNA/protein densities were measured from the left ventricular midmyocardial free-wall, but APs were recorded from right ventricular subendocardial tissue. This was done both for technical reasons (standard microelectrode recordings from left ventricular tissue were difficult to obtain and more likely to be contaminated by subendocardial Purkinje fibres) and to maximize data from each human heart by using all available tissues. We had to optimize the information obtained from each human heart, because functional measurements were greatly limited by the unpredictable and infrequent availability of human donor tissue and because of the short time window for meaningful functional measurement after tissue procurement. Of note, our patch-clamp/biochemical results in left ventricular free-wall were fully compatible with our AP data from right ventricular tissues, indicating that at least for these two widely separated regions the observations are consistent.

Relationship to previous studies of repolarizing currents and repolarization reserve

Our data suggest important expression differences in Kir2.x channel mRNA expression between human and dog ventricle. Kir2.1 expression was about 3-fold greater in the dog than human, but Kir2.2 and Kir2.4 levels were negligible in dogs. In human hearts, we found Kir2.3 mRNA expression comparable with that of Kir2.1, generally considered the principal subunit underlying I_K1 (Dhamoon & Jalife, 2005). Significant Kir2.3 protein expression in human ventricle was also detected by Western blot (Fig. 7D). Kir2.1 currents display strong inward rectification, whereas Kir2.3 inward rectification is incomplete and negative slope conductance is less steep (Dhamoon et al. 2004). In our study, the current–voltage relation of I_K1 in dog strongly resembles that previously reported for Kir2.1 channels, but in human cells resembles better a mixture of Kir2.1 and Kir2.3 properties (Dhamoon et al. 2004) corresponding to mRNA data.

Protein quantification showed lesser ERG1a abundance in human compared to dog tissue while expression of ERG1b was not different. A higher ERG1b:ERG1a expression ratio in humans suggests the possibility of different channel subunit stoichiometry in human tissue versus dog. This difference might have two functional consequences. First, partially due to the accelerated activation kinetics of heteromeric channels compared to homomeric channels consisting of ERG1a only, the relative contribution of I_Kr to the repolarization reserve is expected to be higher in humans (Sale et al. 2008; Larsen & Olesen, 2010). Secondly, ERG1a–ERG1b subunit stoichiometry could also affect drug binding affinity of dofetilide to I_Kr channels, as slightly higher IC₅₀ values were obtained for ERG1a–1b heteromeric channels as compared to ERG1a homomer channels (150 nm vs. 100 nm, respectively; Abi-Gerges et al. 2011). We have not detected any significant difference in the kinetic behaviour of I_Kr in humans versus dogs and dofetilide affinity was not different based on concentration–response curves (Supplemental Fig. 1). Thus, relative expression on Western blots may not reflect accurately relative local subunit expression in ion channels.

Relatively little information is available about the molecular basis of differential repolarization patterns among species. APD prolongation and early afterdepolarization formation upon exposure to I_Kr blocking drugs varies widely, with rabbits being the most sensitive, guinea-pigs, swine and sheep the least, and dogs intermediate (H. R. Lu et al. 2001). Guinea-pigs have particularly large, and rabbits particularly small, I_Ks (Z. Lu et al. 2001). This difference results from weaker mink expression in the rabbit, despite stronger KvLQT1 expression in rabbits (Zicha et al. 2003). Interestingly, this expression difference resembles what we observed for human versus dog in the present study, with dogs having much larger minK, but smaller KvLQT1, expression than humans, along with considerably larger I_Ks density. Dumaine & Cordeiro (2007) also observed larger I_K1 and I_Ks, along with similar I_Kr, for dog compared to rabbit. MinK, on the other hand, has also been found to modulate hERG and Kv4.3 current densities and gating of the channels (Pourrier et al. 2003). Therefore, other currents in addition to I_Ks, such as I_Kr and I_to might be potentially influenced by the relatively lower minK expression level in human ventricles we found in this study.

Possible implications

Larger APD prolongation in human tissues versus dog in response to I_Kr blockade, despite similar I_Kr, is a novel finding that may have important implications. Based on the present results, despite observations that the properties of individual K⁺ channels in dog resemble those of humans (Varróet al. 2000; Jost et al. 2005), the reserve to repolarization-delaying drugs may differ substantially between the two species. Consequently, the clinical repolarization-delay potential of drugs with I_Kr/HERG blocking properties could be underestimated based on experiments in dogs, and using dogs in safety-pharmacology studies to estimate QT-lengthening liability could be misleading. However, there is greater similarity of individual currents in human and dog, and better heart rate correlation between human and dog than human and rabbit (Lengyel et al. 2001) and the similar relative profile of rabbit versus dog K⁺ currents in the Dumaine–Cordeiro study (Dumaine & Cordeiro, 2007) to the human versus dog results in the present work raise the issue of whether the commonly used, simpler and cheaper rabbit model might be more predictive. QT prolongation by non-cardiovascular drugs is a major problem and considerable resources are expended to optimize QT-liability drug screening in drug development (Vargas, 2008). Our findings have potentially important implications for the optimization of drug screening.

Based on our data, I_K1 block or downregulation/ mutation would not necessarily lead to substantial QT prolongation in humans, unlike in the dog, but a reduction of repolarization reserve would be expected (Roden, 1998; Biliczki et al. 2002; Silva & Rudy, 2005; Roden, 2006). Therefore, an I_K1 (Kir2.x) channel defect due to ion channel mutations or drug-induced malfunction may not significantly prolong human QT intervals, but could produce excess QT prolongation and life-threatening torsades de pointes in the face of additional repolarization impairment.

The present study is, to our knowledge, the first detailed analysis of the molecular and ionic determinants of repolarization reserve in the human heart, and the first to compare these determinants with those of an animal species commonly used as a model for human cardiac electrophysiology. Our results therefore provide novel fundamental insights into this clinically crucial process.

Potential limitations

I_K1 flows through a variety of channel subtypes that may be constituted by different alpha-subunits including Kir2.1, Kir2.2, Kir2.3, Kir2.4, TASK and TWIK (Wang et al. 1998; Lopatin & Nichols, 2001; Melnyk et al. 2002; Dhamoon et al. 2004). The latter two-pore channels do not rectify (Lesage & Lazdunski, 2000) and were not studied in our experiments, although their contribution to I_K1 cannot be ruled out. Previous reports indicate important species and regional differences in relative expression of Kir2.x proteins (Wang et al. 1998; Melnyk et al. 2002; Dhamoon & Jalife, 2005). The densities of I_K1 and distribution of Kir2.x proteins differ in atria versus ventricles (Melnyk et al. 2002; Dhamoon & Jalife, 2005). In the present study, we focused on ventricular tissue exclusively. Kir2.2 has been reported absent in rabbit ventricle but present in human (Wang et al. 1998) and dog (Melnyk et al. 2002) ventricles. Kir2.x proteins not only form homomeric channels, but can also show heteromeric co-assembly (Zobel et al. 2003), complexifying interpretation. Heteromeric assembly of Kir2.1 and Kir2.3 proteins produces I_K1 channels with lower conductance than homomeric Kir2.1 assembly (Yan et al. 2005; Fang et al. 2005). Since the mRNA expression of Kir2.1 and Kir 2.3 in human ventricle was relatively similar, unlike the dog, heteromeric Kir2.1–2.3 channels may be more likely in the human than in the dog ventricle, contributing to the lower I_K1 density that we observed in humans. Indirect evidence indeed points to a significant role for heteromeric Kir2.x channels in human I_K1 (Schram et al. 2003).

All of our human samples were stored in cardioplegic solution following harvesting during transportation to our facility. In preliminary studies in which we stored canine heart samples in cardioplegic solution and then recorded ionic currents and APs, we did not observe any electrophysiological effects of cardioplegic storage. Donors received haemodynamic support with dobutamine prior to heart explantation, a ubiquitous practice in cardiac transplantation. We cannot exclude possible effects of dobutamine infusion on the properties of explanted hearts.

The effects of pharmacological blockade on canine APs vary among different laboratories. For example, the Antzelevitch laboratory has reported larger increases in canine ventricular APD with K⁺ channel blockade (Shimizu & Antzelevitch, 1999; Tsuboi & Antzelevitch, 2006) than we observed in the present study. The discrepancies are likely to relate to differences in experimental conditions. For this reason, it is important that comparative studies between species responses are produced within a single laboratory rather than comparing changes observed for one species in one laboratory with those for another species in a different laboratory.

The Na⁺–Ca²⁺ exchanger current (NCX) current was defined and measured as Ni²⁺-sensitive current. This approach has limitations, because it cannot be excluded that Ni²⁺ blocks other ionic currents. However, for the measurement of the NCX, we blocked other ionic currents (including K⁺, Na⁺ and Ca²⁺ currents, along with Na⁺–K⁺ pump current) according to the experimental approach described by Hobai et al. (1997), which is a relative widely used method for studying NCX current (Tóth et al. 2009).

The Na⁺–K⁺ pump is critically dependent on extra- and intracellular Na⁺ and K⁺ concentrations, voltage, subcellular space and cAMP levels, and is not well explored in dog and human cardiomyocytes (Fuller et al. 2013). Since we have no experimental data regarding this current, we cannot exclude a contribution of species difference in the function of Na⁺–K⁺ pump currents to repolarization reserve discrepancies.

Although I_Ks is several-fold larger under square-wave voltage-clamp conditions in dog than man (Fig. 2), there was no significant difference under AP-clamp conditions (Fig. 3). We believe that the apparent discrepancy is due to the fact that during the normal AP, cells spend very little time at potentials for which there is a significant difference in I_Ks (positive to +20 mV; Fig. 2).

The enhanced density of I_Ks in canine versus human heart appears to be due, at least in part, to stronger expression of minK in the dog. However, there is a discrepancy between the Western blot results, showing a 33% greater expression level in the dog (Table 1), and the immunofluorescence results (Fig. 8), showing an approximately 5-fold greater expression in canine cardiomyocytes. In addition, if minK overexpression were responsible for greater I_Ks in the dog, kinetics should have differed markedly between the species, which they do not. Therefore, while differences in minK may be involved in the species differences in I_Ks, other factors are likely involved and should be addressed in future work.

Conclusions

Human ventricular cardiomyocytes have reduced repolarization reserve compared to dog. The differential response occurs despite similar I_Kr densities, due to lower I_K1 and I_Ks densities in human hearts. The underlying molecular basis appears to be differential expression of Kir2.x and minK subunits between human and canine hearts. These results suggest that the protection afforded by I_K1 and I_Ks against repolarization stress is limited in humans, making humans susceptible to excess repolarization impairment from I_Kr blocking agents. Animal models are widely used to study cardiac pathophysiology and pharmacological responses. Our findings highlight the importance of caution when extrapolating results from animal models to man, even from species as apparently similar in ionic current mechanisms as dogs.

Glossary

AP

action potential

APD

action potential duration

I_CaL

L-type Ca²⁺ current

I_K1

inward rectifier K⁺ current

I_Kr

rapid delayed-rectifier K⁺ current

I_Ks

slow delayed-rectifier K⁺ current

I_to

transient-outward current

N_CX

Na⁺–Ca²⁺ exchanger current

Additional information

Conflict of interest

None declared.

Author contributions

Conception and design of the experiments: N.J., L.V., J.Gy.P., A.V., S.N.; collection, analysis and interpretation of data: N.J., L.V., P.C., B.Ö., V.Sz., Gy.S., M.B., Zs.K., I.K., N.N., T.Sz., J.M., M.K., L.G.P., Cs.L., A.V., S.N.; drafting the article and revising it critically for intellectual content: N.J., L.V., P.C., B.Ö., V.Sz., Gy.S., J.M., L.G.P., E.W., U.R., P.P.N., J.Gy.P., A.V., S.N. All authors approved the final version of the manuscript.

Funding

This work was supported by grants from the Hungarian Scientific Research Fund (CNK-77855, K-82079 and NK-104331), the National Office for Research and Technology-Baross Programmes (REG-DA-09-2-2009-0115-NCXINHIB), the National Development Agency and co-financed by the European Regional Fund (TÁMOP-4.2.2/B-10/1-2010-0012; TÁMOP-4.2.2.A-11/1/KONV-2012-0035; TÁMOP-4.2.2A-11/1/KONV-2012-0073 and TÁMOP-4.2.2.A-11/1/KONV-2012-0060), the Hungarian Academy of Sciences, the Canadian Institutes for Health Care Research (MOP 68929), German Hungarian Research Cooperation DFG Grant (436 UNG 113/176/0-1) HU-RO Cross-Border Cooperation Programmes (HURO/0802/011_AF-HURO_CARDIOPOL and HURO/1001/086/2.2.1_ HURO-TWIN,) and the Hungarian Academy of Sciences.

Supplementary material

Supplemental Methods

Supplemental Table 1

Supplemental Fig. 1

Supplemental Fig. 2

Supplemental Fig. 3A

Supplemental Fig. 4

Supplemental Fig. 6