OUP user menu

Ultra-rapid delayed rectifier channels: molecular basis and therapeutic implications

Ursula Ravens, Erich Wettwer
DOI: http://dx.doi.org/10.1093/cvr/cvq398 776-785 First published online: 15 December 2010

Abstract

The ultrarapid delayed rectifier channels have attracted considerable interest as targets for ‘atrial-selective’ antiarrhythmic drugs because they contribute to atrial but not to ventricular repolarization. Thus, IKur channel blockers are expected to prolong selectively the atrial effective refractory period without inducing proarrhythmic effects due to excessive ventricular action potential prolongation. Here we provide an overview of the properties of IKur channels in expression systems and native cardiomyocytes. The ion conducting pore of the channel is formed by four Kv1.5 α-subunits, whereas the ancillary β-subunits Kvβ1.2, Kvβ1.3, and Kvβ2.1 control channel trafficking and plasma membrane integration as well as activation and inactivation kinetics. Investigation of IKur channel blockers in cardiomyocytes is complicated (i) by substantial overlap of IKur with other currents, notably the transient outward current Ito, (ii) by lack of drug selectivity, and (iii) by disease-induced regulation of IKur. Some new compounds developed as IKur blockers are described and their efficacy in treatment of atrial fibrillation (AF) is discussed. Current evidence suggests that pure IKur channel block may not be sufficient to suppress AF.

  • Atrial fibrillation
  • Electrical remodelling
  • Kv1.5
  • IKur
  • Atrial selective drugs

1. Introduction

The cardiac action potential (AP) differs from that of other excitable tissues by a prominent plateau phase resulting from fine tuning of depolarizing inward and repolarizing outward currents. Inward current is predominantly conducted via L-type Ca2+ channels (ICa,L), whereas multiple K+ channels contribute to outward current. For instance, Kv4.3, Kv4.2, and Kv1.4 conduct the transient outward currents (Ito), and Kv7.1 (KvLQTS), Kv11.1 (hERG), and Kv1.5 conduct the slowly, rapidly, and ultrarapidly activating outward rectifying currents IKs, IKr, and IKur, respectively (see ref.1 for review). The individual ion channels have been cloned from cardiac tissue and expressed in heterologous cell systems for the study of their electrophysiological properties. Overlapping K+ currents are usually distinguished on the basis of these properties or by use of selective blockers; however, in native cardiomyocytes, it is difficult to unequivocally differentiate between all outward current components. For example, during short voltage clamp steps, transient outward K+ current typically activates and inactivates rapidly, though not completely. The ‘pedestal’ steady-state current could either represent incompletely inactivating Ito or an underlying extra current with rapid activation and very slow inactivation kinetics as described in rat atrial myocytes.2 It is distinctly different from the slowly activating, delayed rectifier current IK in rat ventricle.3

Outwardly rectifying K+ currents with rapid activation kinetics and little or no inactivation were identified in atrial myocytes from several species including man. These kinetics were reflected in the early current nomenclature, such as steady-state current Iss,2 sustained current Isus,46 sustained outward current Iso,7 or ultrarapid K+ current IKur.6,8 The latter is now the generally accepted name of the current, because its characterization as ‘sustained’ or ‘non-inactivating’ may be misleading since substantial and fast current inactivation can be expected under physiological conditions. Kv1.5 channels are the molecular correlate of IKur in human cardiomyocytes.

Here we will review the biophysical properties of Kv1.5 channels and their β-subunits in expression systems, provide an overview of procedures for separating IKur in native cardiomyocytes from overlapping currents notably transient outward current Ito, and briefly discuss some new compounds developed as IKur blockers for treatment of atrial fibrillation (AF).

2. Composition of channels for IKur

2.1 α-Subunit

The HK2 channel cloned from human ventricle9 and atrium10 is highly homologous with the Shaker K+ channel in Drosophila11 and represents the human isoform of Kv1.5. Despite mRNA and protein expression of Kv1.5 in both atria and ventricles,12 the IKur current is confined to atrial myocytes and virtually absent in ventricular myocytes of most species. Kv1.5 channels are also expressed in many other organs, including pulmonary arteries,13 rat brain,14 skeletal muscle,15 and have been associated with cell cycle regulation.16

The channel protein complex for the ultrarapidly activating outward current is made up of four α-subunits encoded by KCNA5, various ancillary subunits, and additional anchoring and/or scaffolding proteins.

Based on the crystal structure of the Kv1.2 channel (Figure 1A and B),17,18 several groups have constructed three-dimensional homology models of the Kv1.5 channel.19,20 Each Kv1.5 α-subunit consists of six transmembrane spanning segments (S1–S6) with intracellular amino- and carboxy-termini. The extracellular linker between S5 and S6 dips back into the membrane. Together with the linkers from the other three α-subunits, this P-region lines the centrally located hydrophilic pore that is responsible for ion selectivity and permeation (Figure 1A–C). Voltage-dependent gating is associated with the movement of positive charges within the S4 segments (four arginine residues) in response to changes of the electrical field across the membrane. The ‘voltage sensor’ of the ion channel consists of part of the S3 and S4 α-helices that form a hairpin-like structure called the paddle motif. At rest, the ‘paddle’ is thought to push mechanically on the linker (‘L’ in Figure 1D) between S4 and S5 which acts as a lever to keep the activation gate in a closed conformation.17 After membrane depolarization, however, the S4 segment rearranges within the paddle, relieving its constraint on the activation gate and allowing the channel to open. Recent findings suggest that a minimum of two gating charges is required for the voltage sensor to function (Figure 1D).21

Figure 1

Ribbon representation of K channel structures and scheme of interaction between voltage sensor and pore domain. (A) Side view of the four α-subunits, each in a different colour, of the Kv1.2 channel, with their transmembrane spanning segments (TM), including the voltage sensor and the pore, the T1 domain, and a β-subunit tetramer. (B) Kv1.2 channel viewed from the extracellular side of the pore. (C) Side view of two opposing subunits of the KcsA channel depicting the selectivity filter and the inner cavity of the ion channel pore of the KcsA channel from Streptomyces lividans. (D) Schematic representation of the voltage sensor in the Kv1.2 channel. The voltage sensing domain consists of a helix-turn-helix hairpin motif between segments S3 and S4 (green-blue loop) that form the ‘paddle’ and are connected to the pore domain by a short lever (marked L; left). Ribbon presentation of the voltage sensor structure (middle). Orange spheres identify the individual gating charges (R1–R4). Sequential truncations and the minimal voltage-sensing domain (right). Truncation of more than the top two gating charges leads to voltage-independent channels. Images reproduced from ref.17 in (A),18 in (B),22 in (C), and109 in (D), with permission of the publishers.

From the crystal structure of Kv1.2 channels, it is concluded that the N-terminus part of the tetramer link to the transmembrane spanning region (TM in Figure 1A) by structures that form four side pores by which ions can enter and leave the inner cavity of the pore.17 Inactivation of Kv1 channels can occur as N-type inactivation, where one inactivation peptide at the N terminus occludes the ion permeation pathway of the channel (e.g. Kv1.4, Figure 2A). Kv1.5 channels have a much slower time course of inactivation. Inactivation of Kv1.2 (and Kv1.5) is of the C-type which consists of a conformational change of the tetramer around the outer mouth of the pore in close vicinity to the selectivity filter (Figure 2B).22

Figure 2

Schematic illustration of inactivation mechanisms of Kv1 channels. (A) N-type K1 channel inactivation where the amino terminus of each of the α-subunits contains an inactivation peptide (‘ball- and-chain’). The inactivation ball of α-subunit drawn into the inner cavity of the channel pore is sufficient to obstruct ion passage and inactivate the channels. (B) C-type inactivation of Kv1.5 is thought to be due to conformational changes around the selectivity filter and is inhibited by elevation of extracellular K+. (C) Schematic representation of N-type inactivation conferred by co-expressing Kvβ1.3 with Kv1.5. The inactivation peptide of the β-subunit enters the inner cavity of the channel pore through one of the side pores formed between the transmembrane-spanning segments S1–S6 and the T1 domain. Redrawn from ref.22,110

Co-expression with β-subunits confers N-type inactivation onto Kv1.5 channels and converts cloned channels with delayed rectifier properties into channels that rapidly inactivate and resemble IKur in native cells (see below).23

2.2 Ancillary subunits

Interaction of Kvβ subunits with Kv1.5 controls channel trafficking and integration into the plasma membrane, and modulates activation and inactivation kinetics of the current.24 Kvβ subunits are localized in the cytosol and consist of ∼300 AS with a conserved carboxy-terminus and a variable amino-terminus.

The major group of ancillary subunits that associate with Kv1.5 channels are Kvβ subunits (for review see ref.25), but KChIP2 also modifies channel properties.26 In the heart, Kvβ1.2, Kvβ1.3, and Kvβ2.1 are most abundantly expressed.27,28 When co-expressed with Kv1.5, they shift the steady-state activation and inactivation curves to more negative potentials.

The conserved C-terminus of the Kvβ-subunit interacts with conserved regions in the N-terminus of the α-subunits. The N-termini of Kvβ1.2 and Kvβ1.3 form an inactivation peptide which can confer an N-type-like inactivation by entering the inner cavity of the channel pore through one of the side pores formed between the transmembrane-spanning segments S1–S6 and the T1 domain (Figure 2C).22,29 One inactivation ball from one Kvβ1 subunit is sufficient for inactivation. Although this mechanism resembles N-type inactivation of K+ channels, Kvβ1 subunit-induced inactivation is not complete like N-type inactivation.23 Interestingly, interaction of Kvβ2.1 with Kv1.5 does not induce fast inactivation.30 In addition, the electrophysiological consequences of the interactions appear to be controlled by post-transcriptional regulation of the α- and β-subunits.

2.3 Expression, trafficking, assembly, turnover

The Kvα and Kvβ subunits of the Kv1.5 channel complexes are assembled at an early stage of channel synthesis in the endoplasmic reticulum. Channel density at the cell surface is a function of the equilibrium between trafficking to the surface and internalization. A fraction of Kv1.5 channels on the cell membrane is rapidly internalized with a half time of ∼10 min and some of the channels return to the surface membrane with a half time of ∼30 min.31 The recycling is controlled by specific Rab-GTPase-dependent processes.

The fate of internalized Kv1.5 channels can be controlled by post-translational modifications including phosphorylation,3235 sumoylation,36 thioacylation,37 or modulation by PKA and PKC.38,39 Internalization of channels with possible permanent degradation of channels after long-term treatment may also contribute to IKur reduction in therapy with conventional antiarrhythmic drugs, e.g. quinidine.40

2.4 Part of macro-complexes?

The α-subunits of Kv1.5 are stabilized at the cell surface not only by the Kvβ subunits but also by scaffolding proteins that attach the K+ channel complex to the cytoskeleton. Membrane-associated guanylate kinase (MAGUK) proteins function as partners for targeting ion channels and receptors to special domains on the cell surface. The synapse-associated protein 97 (SAP97) is such a MAGUK protein, to which Kv1.5 binds.41 SAP97 is strongly expressed in the heart and retains Kv1.5 in the plasmamembrane.42 In co-immunoprecipitation experiments, it was demonstrated that SAP97 and Kv1.5 co-localize in the lateral membrane of atrial cardiomyocytes and in the intercalated disk near to connexins and cadherin. It is not known why the greatest density of ion channels is found in this location, but it has been speculated that the region of cell-to-cell contact is well suited for anchoring channel proteins to the cytoskeleton.12 Another related protein, Four-and-a-half Lim protein (FHL-1), was shown to be a binding partner for KCNA5 in human atrium.43 Its increased expression in pacing-induced AF has been interpreted as compensatory against the downregulation of Kv1.5 associated with this arrhythmia.44

The plasmamembrane lipids are arranged in specialized cholesterol-rich areas, the lipid rafts and caveolae, that contain macro-complexes of signalling molecules.45,46 Notably, Kv1.5 co-localizes with caveolin, a protein associated with caveolae formation, on the cell surface and also redistributes with caveolin following microtubule disruption suggesting close functional interrelations.47

2.5 Regulation by phosphorylation and nitrosylation

The amino acid sequences of Kv 1.5 and Kvβ subunits possess multiple sites for PKA and PKC phosphorylation. In human atria, phosphorylation induced by stimulation of β-adrenoceptors and activation of PKA increases, whereas PKC-mediated phosphorylation decreases IKur amplitude.48 Expression experiments indicate that coexpression of Kvβ1.3 with Kv1.5 is required for the PKA-mediated increase in K+ current.33 The enhancing effect on IKv1.5 inactivation by coexpression of Kv1.5 with Kvβ1.3 is attenuated when Kvβ1.3 is phosphorylated at serine 24 of its NH2 terminus, explaining the β-adrenoceptor-stimulated current increase.

Stimulation of PKC has little effect on expressed Kv1.5 channels alone; however, co-assembly of Kvβ1.2 with Kv1.5 enhances the response of the channel to PKC activation with a reduction in the K+ current.39 These results suggest that phosphorylation sites for PKC are probably located in the unique amino-terminus of Kvβ1.2.

Kv1.5 channels are also regulated by NO via a cGMP/PKG-dependent pathway. NO donors such as SNAP (S-nitroso-N-acetylpenicillamine) and l-arginine concentration-dependently decrease IKv1.5 in an expression system, and elevate the plateau in mouse ventricular cells (which unlike human ventricular cardiomyocytes do endogenously express Kv1.5).49 Based on molecular modelling, these authors showed that S-nitrosylation of two cystein residues in S2 (Cyr331 and Cys 346) are involved in this effect.

3. Physiological properties of IKur

In 1987, Escande et al.50 provided evidence that two distinct components of voltage-activated outward currents contribute to early repolarization in human atrial myocytes, consisting of brief outward current (Ibo) that persists in 3 mM 4-aminopyridine, and long lasting outward current (Ilo) that is sensitive to block with 0.5 mM 4-aminopyridine. While the former current corresponds to Ito, the latter is part of the delayed rectifying K+ current IKsus,6 IKsus, or IKur because of its ultrarapid activation kinetics is distinctly different from IKr and IKs which are also present in human atria.51

3.1 Current in heterologous expression systems

The biophysical properties of ion channels are conveniently analysed in heterologous expression systems that lack endogenous channels. The first comprehensive biophysical description of Kv1.5 (H2) stably expressed in mouse Ltk cells was provided by Snyders et al.52: A single population of channels gives rise to robust outward current at half-maximum activation potential of −14.5 mV with K+ as charge carrier. At room temperature, current activates rapidly at potentials between 0 and +60 mV (τact 10 to <2 ms) and inactivates only partially, i.e. by ∼20% after 250 ms at +60 mV. Inactivation follows a bi-exponential time course with time constants τinact of ∼240 and 2700 ms. The current is blocked by low concentrations of 4-aminopyridine, but is rather insensitive to tetraethylammonium or dendrodotoxin, and does not respond to selective blockers of the rapidly activating outward rectifying K+ current IKr.53

Though the Kv1.5 current can be regarded as a ‘sustained’ current at room temperature, inactivation proceeds much faster at more physiological temperature.6,7,52 Moreover, the time course of recovery from inactivation is also faster at higher temperatures.

3.2 Study of IKur in native cardiomyocytes: problems of separation

For biophysical characterization, Kv1.5 channels are cloned into cell lines that do not endogenously express the channel. This approach is not possible when channel properties are to be studied in relation to health and disease. In such cases, native cardiomyocytes must be investigated where quantitative analysis is complicated by multiple overlapping currents in addition to outward currents. The current of interest is usually dissected by varying the ionic compositions of the superfusion and pipette solutions, by applying drugs, or by exploiting differences in steady-state kinetics. Nevertheless, analysis of the current of interest may be substantially biased by time- and voltage-dependence of the procedures for eliminating the underlying currents. Such difficulties are also encountered when attempting to characterize Ito and IKur in human atrial myocytes because the two currents reveal similar voltage ranges of activation and IKur reveals considerable inactivation at physiological temperature.

Figure 3 illustrates some published separation procedures. In the initial efforts to identify outward current related to Kv1.5 in human atrial myocytes, room temperature was deliberately used in order to slow down ultrarapid activation.6 During a test step from −50 to +50 mV, an outward current rapidly activates and then inactivates to a constant steady-state level within a 1s long pulse (Figure 3A, left). The transient current is Ito, the current at the end of the test steps represents either non-inactivation Ito or ‘sustained’ outward current. Since at room temperature, Ito recovers very slowly from inactivation, a brief 25 ms clamp step back to holding potential followed by a second test pulse to +50 mV allows to analyse activation of sustained current without interference from Ito (pre-pulse inactivation of Ito).6 Notably, at room temperature and due to the very long prepulse, inactivation of IKur is almost lost so that the current does indeed appear as a rapidly activating, hardly inactivating outward current, but not so at physiological temperature (see below). Since the Kv1.5 current is highly sensitive to block by 4-aminopyridine,52 sustained currents using the prepulse protocol—albeit with shorter prepulses—were studied before and after addition of 1 mM of the blocker. As seen in Figure 3B, the amplitude of IKur measured after prepulse inactivation of Ito was not completely suppressed by 4-aminopyridine,54 suggesting additional background current of unidentified origin. In our own studies of IKur (Iso),7 we have used a holding potential of −20 mV for steady-state inactivation of Ito (data not shown). At physiological temperature, pre-pulse inactivation of Ito is no longer effective because Ito recovers too rapidly from inactivation so that 25 ms are sufficient for partial recovery (Figure 3C), whereas recovery of IKur is considerably slower (own data, unpublished results). The transmembrane charge carried by Ito is calculated from the area under the curve during the initial 50 ms of the second test pulse (AUC50 ms) and serves as a suitable parameter for studying drug effects on Ito.55 We have recently suggested yet another approach for estimating the contribution of Ito and IKur to total outward current in human atrial myocytes by exploiting their differences in time constants of inactivation.55 At physiological temperature, IKur is by no means a ‘sustained’ current, but exhibits considerable inactivation. In stably transfected mouse fibroblasts, IKv1.5 inactivates with time constants of 18 ms (τinact,fast) and 297 ms (τinact,slow).55 Fitting a multiexponential function to outward current in native human atrial myocytes from patients in sinus rhythm (Figure 3D) yields three exponential current components with time constants of 5.7, 25.4, and 347 ms, in addition to a non-inactivating component. The latter two time constants are in reasonable agreement with the τinact,fast and τinact,slow of expressed Kv1.5 current, the most rapid inactivation time constant corresponds to Ito. Interestingly, in cardiomyocytes from patients in chronic AF, the amplitudes of inactivating current components were significantly smaller than in sinus rhythm; however, inactivation time constants representing IKur as well as the amplitude of the non-inactivating current component were not different.55

Figure 3

Schematic drawing of different methods to separate Ito from IKur in human atrial myocytes from non-fibrillating atria. (A) Ito and IKur at 25°C. Currents during a 1s long test pulse (left tracing) and during a 100ms long test pulse (right tracing). The right tracing was preceded by a 1 s prepulse to +50 mV 20 ms before the test pulse. For clarity, only current during a test step to +50 mV has been redrawn. Ito and IKur as indicated, see text for further explanation (adapted from ref.6). (B) Effect of 4-aminopyridine (4-AP, 1 mM) on IKur and 4-AP-sensitive current (red). Similar protocol as in (A) but shorter prepulse (adapted from ref.54). (C) Ito and IKur at 36°C. Current tracing during two 500ms long test steps to +50 mV, separated by 25 ms at the holding potential of −60 mV. Ito recovers rapidly at this temperature and can be estimated by the area under the curve during the first 50 ms (AUC50ms) of the second test pulse (redrawn from ref.55). (D) Ito and IKur in response to a 500 ms test pulse from −60 to +50 mV at 36°C. A tri-exponential curve was fitted to the current trace, where τA is the time constant of inactivation for Ito and τB and τC are time constants for fast and slow of inactivation, respectively, of IKur.55

The nature of the non-inactivating current is unclear. Contribution of the Kv2.1 channel current is unlikely, because expression of this delayed rectifier K+ channel in human atria is minute.28 Other less well-investigated channels that could contribute are two-pore-domain potassium channels,56 transient potential receptor channels,57 and Ca2+-activated K+ channels of small conductance,58 but these have not been well characterized in human atria and are discussed elsewhere.59

The advantage of the double-pulse protocol by Wang et al.6 is the rather complete separation of IKur from Ito, albeit at unphysiologically low temperature that affects other channel properties as for instance kinetics. In addition, the long prepulse substantially inactivates also IKur. This disadvantage is overcome by the shorter pre-pulse protocol which nevertheless is still performed at room temperature.54 In our own laboratory, we have used separation procedures by curve fitting for the study of human atrial cardiomyocytes at physiological temperatures to measure drug action.55 However, less complete separation may be achieved because at high temperature, a small fraction of IKur may have recovered from inactivation at the beginning of the second clamp step (Figure 3C), as extrapolated from data in expression systems.

4. IKur and heart disease

4.1 Genetics: Kv1.5 channel mutations can cause AF

Although AF is not primarily considered an inheritable disease, recent epidemiological studies have provide evidence that gene polymorphisms may substantially enhance disease susceptibility.6062 A few cases of familial AF have been associated with genetic abnormalities relating to mutations in genes encoding for ion channels.63 Patients with ‘lone’ AF lack heart disease or risk factors. Screening these patients for genetic aberration revealed a nonsense mutation in KCNA5 in one patient with familial AF, which is absent in several hundred unrelated control individuals. When this mutation is heterozygously expressed, the defective gene encodes for a truncated Kv1.5 channel that fails to generate any IKur.64 Absence of IKur may excessively prolong atrial APD with an enhanced risk of early after-depolarizations that can trigger and/or maintain AF. Three further KCNA5 loss-of-function mutations have been recently reported in 4 of the 120 unrelated AF families.65

4.2 Remodelling in AF

Experimental evidence for remodelling of Kv1.5/IKur in AF is conflicting, with reports of reduced IKur density5,54 or no change.4,66,67 Our own data indicate a small, but significant decrease in IKur density, yet total outward current density is only slightly smaller in AF than in SR cardiomyocytes suggesting that an additional background current not present in SR could have developed during remodelling.55 mRNA and Kv1.5 protein are either not changed or downregulated. Expression of Kvβ subunits in atrial tissue from patients with AF due to valvular heart disease is not different from patients in sinus rhythm and coronary artery disease.68

5. Role in cardiac action potential: modelling

The contribution of various ion channels on the shape of atrial APs is studied by ion channel blockers though possibly impeded by insufficient selectivity of the selected blocker, or in computer models by mathematically eliminating the respective current.69 As indicated in Figure 4A, IKur is responsible for the negative potential range of the atrial plateau phase in ‘normal’ APs and block of IKur in the model strongly elevates the plateau potential. However, in APs remodelled due to AF with their characteristic triangular shapes,4,67,70 block of IKur is much less effective in elevating the plateau potential but prolongs AP duration at 90% of repolarization (APD90) and effective refractory period. Interestingly, in human native atrial myocytes, IKur block by very low concentrations of 4-aminopyridine (4-AP 5 µM) mimics the effect on AP plateau predicted in the mathematical model, but reveals shortening of APD90 in atrial cardiomyocytes from SR patients.50,71,72 Similar effects, i.e. shortening of APD90 in SR but prolongation in chronic AF APs, were observed for another IKur blocker, the investigational drug AVE0118 (Figure 4B) despite the compound's known additional effects on Ito and IK,ACh55 or with Xention D-0101 (own data, unpublished results).

Figure 4

Human atrial action potentials simulated by a computer model (A) and recorded in isolated atrial trabeculae from patients in sinus rhythm and in atrial fibrillation (B). NAP, normal action potential; AFAP, atrial fibrillation action potential; ΔAPD-60, difference in action potential duration at −60 mV; AVE0118, 10 µM. Reproduced from refs55,69 with kind permission of the publishers.

Physiologically, rapid activation of IKur in the positive potential range following the AP upstroke may offset depolarizing ICa,L and hence lead to the less positive plateau phase in atrial compared with ventricular cardiomyocytes.73 Conversely, block of IKur produces a more pronounced spike-and-dome configuration and therefore shifts the potential into a more positive range where ICa,L activation enhances systolic Ca2+ influx during a free-running AP.72 Such an indirect effect on ICa,L should be shared by all IKur blockers and is expected to result in a positive inotropic effect. Indeed, 4-aminopyridine and AVE0118 significantly increase atrial contraction amplitude in a concentration-dependent manner;55,72,74 however, Schotten et al.74 associate increased contractility with enhanced reverse-mode Na+/Ca2+ exchange. Although the positive inotropic effect is much smaller in AF, it is considered to be clinically interesting, because atrial contractility is already impaired in AF.75 Moreover, many drugs used either for rhythm control (e.g. flecainide, propafenone, amiodarone) or for rate control (e.g. β-adrenoceptor blockers or calcium channel blockers) have a negative inotropic effect.

6. Drug target

Pharmacological studies of K+ currents in native myocytes are notoriously complicated. The outward current of interest is isolated from contaminating inward Na+ and Ca2+ currents by ion substitution, ion channels blockers, or voltage-dependent inactivation. However, these experimental procedures themselves may distort characteristic properties of the channels under investigation. For example, the bivalent cation Cd2+ that blocks L-type Ca2+ current will modify voltage dependence of Ito gating properties.76 Moreover, for correct kinetic analysis of the native current of interest in the presence of a pharmacological modulator, block of other currents by drugs must be voltage- and time-independent.

6.1 Binding sites for drugs

Drugs but also endogenous compounds like the endogenous endocannaboid anandamide directly interact with Kv1.5 channels.77,78 Two mechanisms of block have been described: compounds may bind to the external vestibule of the channel thereby interfering with closure of the activation gate (‘foot-in-the-door’ phenomenon79) or they can function as an open channel blocker with binding sites in the inner cavity of the pore and access from the intracellular side (see below).

Kv1.5 channels are considered promising targets for atrial selective therapy of AF because IKur is absent in the human ventricle. Numerous selective IKur blockers have been developed80 and drug binding to the channel has been studied in some detail. Inactivation of Kv1.5 conferred by Kvβ1 subunits requires distinct N-terminal residues to bind to multiple residues within the S6 domains of Kv1.5 (Figure 5). Drug binding in the inner cavity of the Kv1.5 channel pore can compete with binding of β-subunits because there is partial overlap in the residues involved in either binding.81 Homology models based on the crystal structure of the KcsA channel,82 incorporating a predicted bending of the S6 domain,8385 and site-directed mutagenesis studies that allow to identify the interacting amino acid residues by a mutation-induced decrease in extent of inactivation caused by Kvβ1.3 or a decrease in drug-induced block provide the concept of drug–channel interaction (Figure 5): Drugs such as AVE0118 and S0100176 (Figure 5), for instance, enter the channel and compete for binding with the inactivation peptide of the Kvβ1.3 subunit at several residues in the S6 segments that line the inner cavity of the channel pore.81 Indeed, the N-terminus of Kvβ1.3 decreases drug affinity of Kv1.5 for local anaesthetic and antiarrhythmic drugs (e.g. bupivacaine;86,87 AVE0118;84 vernakalant88).

Figure 5

Binding sites for Kvβ1.3 and drugs on the S6 domain of the Kv1.5 channel. The S5–S6 domains of a single Kv1.5 subunit are depicted with important residues for interaction with Kvβ1.3 (left panel), AVE0118 (middle panel), and S0100176 (right panel). Reproduced from ref.81 with kind permission of the publisher.

Interestingly, screening the coding region of the KCNA5 gene in different ethnic populations has revealed two polymorphisms (P532L and R578K, both in the C-terminus) that when expressed in Chinese hamster ovary cells exhibit normal gating, but are resistant to block by the prototypical inhibitor quinidine.89 If the C-terminus of KCNA5 is truncated, the channel generated possesses drug sensitivity similar to the wild-type channel. These findings suggest that P532 is not directly involved in drug binding but that the altered structure due to the P532L polymorphism may impair access of the drug to a binding site within the inner cavity of the channel pore.90

6.2 Brief description of new drugs

Numerous new antiarrhythmic compounds have been targeted against Kv1.5 channels with the hopeful expectation to develop effective and safe drugs against AF (for a comprehensive overview, see ref.91). However, most of the compounds also affect other channels as well (for detailed discussion, see ref.59).

The biphenyl derivative AVE0118 blocks IKur atrial myocytes with additional effects on Ito and IK,ACh in a similar concentration range.84,92 AVE0118 effectively reduces IKur in myocytes from patients in SR, whereas in myocytes from patients in chronic AF, IKur becomes resistant to the compound, despite absence of reduced current amplitude or downregulation of mRNA expression of Kv1.5.55 In goat models of AF, AVE0118 effectively prevents inducibility of AF episodes93 and also converts chronic AF to sinus rhythm.94 Besides blocking IKur, the analogue compound AVE1231 also blocks IK,ACh and IKr albeit at higher concentrations.95 AVE1231 is less effective than AVE0118 in the goat model of AF.96

Notably, myocytes from patients in AF exhibit a strong AVE0118-resistant fraction of what appears IKur55 If this peculiar resistance of IKur to block is not confined to AVE0118 and presents as a more general phenomenon, IKur could be of limited value as a drug target for treatment of chronic AF. However, nothing is known about the time course of development of channel resistance towards AVE0118.

The two compounds S9947 and S20951 possess good selectivity of IKur block over IK1 and Ito blocks, and reveal strong frequency-dependent block.97 In an in vivo model of anaesthetized pigs, S9947 and S20951—like AVE0118 and AVE1231—almost completely suppress left atrial vulnerability against AF induced by pre-mature stimuli.98

The diphenylphosphine oxide DPO-1 blocks IKur with decreasing concentrations the higher the stimulation frequency.99 Like 4-AP, DPO-1 elevates the plateau potential, leading to APD shortening (sinus rhythm) or prolongation (AF) in human atrial tissue, but is devoid of any significant effect in human ventricular APs. In vivo studies in primates show that DPO-1 increases atrial ERP only.100

Vernakalant (RSD1235) is of special interest among the new compounds that were developed for IKur block, because it has recently been approved by the European drug agencies for intravenous conversion of recent onset of AF.101,102 It produces a strong, positive frequency-dependent IKur block with an IC50 of 13 µM at 1 Hz in stably expressed hKv1.5 channels, whereas three-fold higher concentrations are required albeit at 0.25 Hz for open-channel block of expressed hNav1.5 channels.103 However, the IC50 for block of hNav1.5 by vernakalant is reduced to 9 µM at the high activation frequency of 20 Hz, and Na+ channel block is more sensitive to vernakalant at depolarized membrane potentials.103 These features of Na+ channel block provide atrial selectivity of vernakalant action, especially in AF.104

Block of IKur is larger than block of Ito or IKr, when studied in expression systems.103 Our own preliminary data in human atrial trabeculae suggests that unlike selective Kv1.5 blocking concentrations of 4-AP, vernakalant does not elevate plateau potential or rhythm-specifically modulate APD, neither in preparations from SR nor AF patients.105 The reported rapid and safe conversion of AF into sinus rhythm by vernakalant is probably due to Na+ channel block.101

Whether pure IKur block can effectively suppress AF has yet to be demonstrated.106 Since IKur may be downregulated in chronic AF5,54 and current amplitude is markedly reduced at high activation rates as encountered in AF,8 IKur blockers may lose effectiveness in AF. However, IKur block slightly prolongs APD90 and effective refractory period only in AF-remodelled APs.55,72

7. Outlook: are Kv1.5 channels useful drug targets in AF?

Since the ultrarapidly activating K+ current, IKur, is prominent in atria and negligible in ventricle, it has been proposed as a safe ‘atrial-selective’ drug target in supraventricular tachyarrhythmias.107,108 Indeed, many drug companies have developed selective IKur blockers for safe conversion of AF into SR without much effect on ventricular AP duration that could cause torsades de pointes arrhythmias.80 Despite these convincing theoretical considerations of an ideal drug target for treatment of AF, clinical data on successful cardioversion of AF into SR by IKur blockers are sparse. Possibly, AF-induced remodelling which decreases IKur amplitude and may alter its sensitivity towards block55 can further complicate the final assessment of the clinical usefulness of this principle. The recently recognized contribution of drug-induced channel internalization to the ion channel blocking effects of antiarrhythmic drugs highlights new avenues for future drug development.

Funding

The authors gratefully acknowledge funding by Fondation Leducq (07CVD03, ENAFRA); and EU FP7-Health-2010-single-stage (Grant No. 261057, EUTRAF).

Acknowledgements

The authors thank Torsten Christ for critical comments.

Conflict of interest: the authors have received funding for experimental research from and are consulting Xention, Merck, Cardiome, and Sanofi-Aventis.

Footnotes

  • This article is part of the Review Focus on: New Insights into the Molecular Basis of Atrial Fibrillation

References

View Abstract