Antiarrhythmic agents that prolong the action potential duration and the refractory period via selective inhibition of the rapid component of the delayed rectifier potassium current (IKr), ie, the class III antiarrhythmic drugs, are associated with an increased risk of the polymorphic ventricular tachycardia torsades de pointes (for review, see Witchel and Hancox1). This is probably caused by an excessive prolongation of the ventricular repolarization, leading to the development of early afterdepolarizations (EADs), functional reentry, and, eventually, torsades de pointes.2 To avoid proarrhythmicity, the development of new antiarrhythmic drugs is focused on mixed sodium/potassium channel blockers and blockers of atrial specific ion channels, eg, Kv1.5. Several antiarrhythmic compounds have been shown to inhibit IKur (Kv1.5), eg, propafenone3-5 and quinidine6,7 in addition to their blockade of other cardiac ion currents eg, hERG and Ito.
Kv1.5 is encoded by the gene KCNA5, and much evidence suggests that it underlies the 4-aminopyridine-sensitive ultrarapid delayed rectifier current (IKur, Iso, or Isus) found in human atrial myocytes.8-11 IKur activates rapidly and inactivates slowly in response to membrane depolarization; hence, it persists during the action potential plateau phase and contributes to action potential repolarization.
The calcium-independent transient outward potassium current (Ito) is characterized by fast activation and fast inactivation kinetics and contributes to the early repolarization phase of the cardiac action potential (for review, see Oudit et al12); in human atria Kv4.3 is thought to underlie Ito.13-15 Coexpression of Kv4.3 and the auxiliary subunit KChIP2 has been shown to increase the current amplitude, slow the time course of inactivation, and increase the rate of recovery from inactivation.16 The enhanced recovery from inactivation, in particular, makes the hKv4.3/hKChIP2 current more similar to the native Ito than the hKv4.3 current alone.16
AZD7009 is a novel antiarrhythmic compound in early clinical development for conversion of atrial fibrillation and maintenance of sinus rhythm. It predominantly influences atrial electrophysiology and has high antiarrhythmic efficacy and low proarrhythmic potential in animals.17,18 Data from studies in dogs show that AZD7009 predominantly increases the atrial conduction time and excitability threshold, whereas the effects in the ventricle are minimal.17 AZD7009 has previously been shown to inhibit hERG and hNav1.5 currents expressed in CHO K1 cells,19 and it is most likely this combined current block that underlies the prolongation of the refractoriness and the low proarrhythmic activity demonstrated in animals in vivo. The predominant atrial effects can in part be explained by a more pronounced sodium current block in atrial tissue, but early observations indicate that AZD7009 also blocks hKv1.5 and hKv4.3 currents, actions that may further contribute to the predominant prolongation of atrial refractoriness reported in animals. The main objective of the present study was to characterize the inhibitory effects of AZD7009 on the hKv1.5 and the hKv4.3/hKChIP2.2 currents. Propafenone and R-propafenone were included in the study for comparative reasons and to validate the methodology adopted.
Transfection and Cell Culture
The ecdysone-inducible mammalian expression system (Invitrogen, Carlsbad, CA) was used to express a hKv1.5 and eGFP (enhanced Green Fluorescent Protein) fusion protein in CHO cells. The sequence of hKv1.5 was almost identical to the GenBank Accession No. M83254 with the following exceptions: C843G (silent), G844C (Val282Leu), GG insertion between positions 1734 and 1735 (leading to a longer 3′ end with the following sequence: GGCAGCTGCCCCCTAGAGAAGTGTAACGTCAAGGCCAAGAGCAACGTGGACTTGCGGAGGTCCCTTTATGCCCTCTGCCTGGACACCAGCCGGGAAACAGATTTGTGA. eGFP obtained from the pEGFP-NI vector (Clontech, Palo Alto, CA; GenBank Accession No. U55762) was moved into the ecdysone-inducible expression plasmid, pIND/Hygro vector (Invitrogen, Carlsbad, CA). EcR-CHO cells (Invitrogen, Carlsbad, CA) were stably transfected with pIND/Hygro/eGFP/hKv1.5. The cells were cultured in DMEM/nutrient mix F12 with Glutamax-1 (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA), 0.5 mg/mL hygromycin B (Roche Diagnostics GmbH, Mannheim, Germany), and 0.25 mg/mL zeocin (Cayla, Toulouse, France). Cells were plated out in small petri dishes 1-3 days before use, and expression of hKv1.5-eGFP fusion protein was induced by addition of 1 μg/mL ponasterone A (Invitrogen, Carlsbad, CA) 1 day before the voltage clamp experiments. The hKv4.3 gene (GenBank Accession No. AH009283 long splice variant AAF68177.1) was cloned into the PGEN-IRES neo vector (Clontech, Palo Alto, CA), and the KChIP2.2 gene (GenBank Accession No. AY026328) was cloned into the PGEN-IRES hygro vector (Clontech, Palo Alto, CA). CHO K1 cells were transfected with 1 μg phKv4.3 and 1 μg pKChIP2.2 per 1 mL of cell culture medium using lipofectamine plus™ reagent (Invitrogen, Carlsbad, CA); the incubation time was 4 hours. After selection in 0.6 mg/mL hygromycin B (Roche Diagnostics GmbH, Mannheim, Germany) and 1.1 mg/mL G-418 sulfate (Sigma-Aldrich Sweden AB, Stockholm, Sweden) for 8 days, single colonies were picked and tested for expression of Kv4.3 currents using the whole-cell configuration of the voltage-clamp technique.20 The stably transfected cells were cultured in HAM/nutrient mix F12 with Glutamax-1 (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA), 0.6 mg/mL hygromycin B, and 1.1 mg/mL G-418 sulfate. Cells used for voltage clamp experiments were plated out in petri dishes 1-3 days before the experiment. On the day of the experiment the cells were washed with Tyrode solution, detached with a cell scraper, and stored at room temperature for a maximum of 6 hours before use.
Voltage Clamp System
Recordings were made with an EPC9 amplifier (HEKA Elektronik, Lambrecht, Germany) using the whole-cell configuration of the single-electrode, continuous voltage-clamp technique.20 Electrodes were fabricated from thick-walled, filamented, borosilicate glass capillaries (Harvard Apparatus Ltd, Edenbridge, Kent, UK) with an inner diameter of 0.86 mm and outer diameter of 1.5 mm. The glass capillaries were pulled with a Flaming/Brown micropipette puller, model P-2000 (Sutter Instruments, Novato, CA) and had resistances of 1.5 to 4 MΩ when filled with electrode solution. The series resistance was maintained at less than 15 MΩ and compensated by 75% to 85%. The current was sampled at 10 or 20 kHz after filtering (2.9 kHz, EPC9 internal Bessel filters) using Pulse software v8.63 (HEKA Elektronik, Lambrecht, Germany). All experiments were carried out at room temperature (22°C). The holding potential was −80 mV. The voltage-clamp protocols used are described in detail in the appropriate results section. Solutions were applied to the cells using the Dynaflow™ technology for automated perfusion in a 16-channel chip (Dynaflow™ DF-16, Cellectricon AB, Göteborg, Sweden).
Solutions and Chemicals
Stock solutions of 25, 100, or 1000 mM AZD7009 in DMSO were prepared, stored at room temperature, and used throughout the experiments. Stock solutions of 100 mM propafenone and R-propafenone in DMSO were stored at room temperature and used throughout the experiments. Dilutions of the stock solutions in Tyrode solution were made on the day of use. The Tyrode solution contained (in mM): NaCl 140, Glucose 5, Hepes 10, KCl 5.4, CaCl2 1.8, MgCl2 1; pH was adjusted to 7.4 with NaOH. The electrode solution contained (in mM): potassium aspartate 110, KCl 20, NaCl 10, EGTA 10, Hepes 10, K2ATP 5, MgCl2 1, CaCl2 1; pH was adjusted to 7.4 with KOH. AZD7009 was synthesized at AstraZeneca R&D Mölndal, Sweden. Propafenone (used in hKv1.5 measurements) and R-propafenone hydrochloride (used in hKv4.3/hKChIP2.2 measurements) was purchased from Sigma Chemical Co, St Louis, MO, USA.
Data Analysis and Statistical Analysis
Pulsefit v8.63 (HEKA Elektronik, Lambrecht, Germany) was used for analysis of current amplitudes. Microsoft Excel (Microsoft Software Corporation), GraphPad Prism 4.0 (GraphPad Software, San Diego, CA), and IGOR Pro v4 (Wavemetrics Inc, Lake Oswego, OR) were used for further data processing and statistical analysis. Currents were leak-subtracted using the p/n-leak protocol in Pulse. The hKv1.5 current was measured as the current amplitude at the end of the 300-millisecond clamp step. The hKv4.3/hKChIP2.2 current was measured as the peak current amplitude or the current-time integral as specified in the results section. The peak current amplitude was measured as the current difference between the maximum peak current and the current after 1 second of depolarization. The current-time integral was measured as the area under the curve (AUC) during a 1-second depolarizing pulse.
Data are given as mean ± SEM of n experiments. Statistical differences were tested with Students t test or one-way ANOVA with Tukey or Dunnett posttest. Differences between data sets were considered statistically significant when P < 0.05.
Concentration and Frequency Dependence of hKv1.5 Current Block
Outward currents in hKv1.5-transfected cells were activated using a 300-millisecond depolarization to +20 mV at a stimulus frequency of 0.2 Hz; tail currents were recorded at −40 mV; the holding potential was −80 mV. Concentration-effect curves for AZD7009 and propafenone were then constructed by exposing cells to increasing concentrations of drug (0.3 to 1000 μM AZD7009 and 0.1 to 100 μM propafenone) (Fig. 1A,B). The time of exposure to each concentration was 3 minutes, which was sufficient to obtain maximal effect. Following application of AZD7009, the current activated, reached a peak, and rapidly decreased as the drug blocked the current. At the end of the 300-millisecond activating clamp step, the block of the current reached a steady state. Following application of propafenone, the decline of the current was more evenly distributed throughout the clamp step. Normalized steady-state currents were plotted versus the logarithm of concentration (Fig. 1E). Half-maximal block (IC50) for AZD7009 was obtained at 27.0 ± 1.6 μM, and the Hill slope was 1.28 ± 0.10 (n = 6). The IC50 for propafenone was 13.3 ± 2.5 μM and the Hill slope was 1.15 ± 0.15 (n = 6). In time-matched control measurements, the hKv1.5 current at the end of the protocol (24 minutes) was 104.3 ± 9.2% (n = 3) of the current in the beginning of the protocol.
Tail currents recorded on return to −40 mV following AZD7009 and propafenone are shown in Figure 1C,D. The tail current amplitude is decreased with increasing concentrations of both drugs, and the current deactivation is markedly slowed down compared with control, causing a crossover of the tail currents recorded in control conditions and following drug application. The tail current deactivation after application of AZD7009 was quantified by a double-exponential curve fit to the decay phase of the current (Table 1). In control conditions and following lower concentrations of AZD7009 (0.3 to 10 μM), the deactivating tail was reliably fit by a double-exponential function, but at higher concentrations reliable curve fits were not obtained in all the measurements. If the fast time constant was locked at the value of the fast time constant in control, reliable curve fits were obtained at all concentrations of AZD7009 except at 1000 μM. The slow time constant was increased (one-way ANOVA, P < 0.0001; Dunnet posttest, P < 0.05 for 30 μM and P < 0.01 for 100 and 300 μM), and the relative amplitude of the fast time constant was decreased (one-way ANOVA, P < 0.0001; Dunett posttest, P < 0.01 for concentrations between 3 and 300 μM) with increasing concentrations of AZD7009. The effect of AZD7009 on tail current deactivation was reversed on washout of the drug.
The frequency dependence of hKv1.5 current block by AZD7009 was examined using frequency changes between 0.2 and 5 Hz in 3-minute intervals. The current was activated by 100-millisecond clamp steps to +20 mV. Steady-state block by 25 μM AZD7009 at 0.2 Hz stimulation was achieved before the frequency changes were made. The intrinsic frequency dependence of hKv1.5 current in the control situation was pronounced (Fig. 2A). During the first and last 5-Hz stimulation trains, the current decreased to 47.2 ± 3.0% and 45.3 ± 2.3% of the current amplitude following the 0.2-Hz stimulation. When the frequency dependence of AZD7009 block was studied, the current was therefore normalized individually in every cell to the current amplitude at the same frequency and pulse number in control conditions (Fig. 2B). Following 0.2-Hz stimulation, a small but significantly larger block was seen as compared with both 5-Hz stimulation periods (one-way ANOVA, P = 0.0048; Tukey posttest, 0.2 Hz versus 5 Hzperiod1, P < 0.005, 0.2 Hz versus 5 Hzperiod2, P < 0.01, 5 Hzperiod1 versus 5 Hzperiod2, P > 0.05, n = 5).
Time Course of hKv1.5 Current Block by AZD7009
If AZD7009 accesses the binding site in hKv1.5 only after channel opening and dissociates from the binding site on channel closing, as described above, the development of block will be visible if it is slower than the activation process. Pulses to +60 mV were used to enable fast activation, and the time constant of activation (τact) in control conditions was 1.9 ± 0.2 milliseconds (n = 6). In control conditions the current activated and slowly declined with a monoexponential time course with a time constant (τinact) of 151.7 ± 9.8 milliseconds (n = 6). Following application of 10, 25, and 100 μM AZD7009, the current rapidly activated and then declined with a double-exponential time course; the fast time constant was concentration dependent, whereas the slow component was not significantly different from τinact measured in control conditions (Fig. 3A). The time constant of block development of AZD7009 (τblock) was therefore considered to be equal to the fast component of inactivation induced by AZD7009, and τblock was 21.6 ± 1.1 milliseconds, 14.0 ± 1.1 milliseconds, and 6.1 ± 0.5 milliseconds at 10 μM, 25 μM, and 100 μM AZD7009, respectively. An estimate of the apparent association (k) and dissociation (l) constants can be made from a linear fit of the inverse of τblock according to the equation:
The rate of block development (1/τblock) was plotted versus the concentration, and k and l were estimated in each individual cell by linear curve fit to the equation above; k was 1.3 ± 0.1 μm−1s−1 and l was 36.8 ± 3.2 seconds−1 (n = 6) (Fig. 3B). Assuming a bimolecular reaction, the apparent
which is close to the experimentally determined IC50 of 27.0 μM at depolarizations to +20 mV.
State Dependence of hKv1.5 Current Block by AZD7009
To further characterize the state dependence of block, washin and washout of 25 μM AZD7009 was made without stimulating pulses, and the amount of block and recovery from block were compared with that obtained after 3 minutes of 0.2-Hz stimulation to +20 mV (no figure is shown). After 3 minutes of washin of AZD7009 without stimulation, the current at the end of the 300-millisecond pulse had declined to 66.9 ± 4.9% (n = 7) of control; after an additional 3 minutes with continuous stimulation at 0.2 Hz, a small additional decrease of the current was seen, 62.4 ± 4.9% (n = 7). The time course of the current during the stimulating pulse did not change during 0.2-Hz stimulation as compared with the time course of the current during the first pulse after the 3-minute washin period at −80 mV. As expected, the channels were unblocked following the 3-minute washout period without stimulation. The current amplitude at the end of the 300-milliseconds pulse recovered to 104.7 ± 5.1% of control, after an additional 3-minute washout period with 0.2-Hz stimulation, the current amplitude was 99.6 ± 5.7% (n = 7). The effects of a 3-minute period without stimulation were also evaluated on the current amplitude in control conditions. The current after 3 minutes of 0.2-Hz stimulation was 96.0 ± 1.4% of the current following the 3-minute period without stimulation, indicating that small-amplitude changes occur in response to the 0.2-Hz stimulation train in the control situation as well. The small differences seen in the amount of block and recovery from block by AZD7009 directly after the period without stimulation and following the 3 minutes of 0.2-Hz stimulation are therefore likely to be explained by the frequency dependence of the hKv1.5 current itself.
Effects of AZD7009 on Voltage Dependence of Activation and Prepulse Inactivation of hKv1.5
The voltage dependence of activation of hKv1.5 was measured under control conditions and following steady-state block by 25 μM AZD7009. Steady-state block was achieved by a 3-minute drug application period during which the current was activated by repeated stimulation to +20 mV (0.2 Hz). Activation curves were recorded using a protocol with 300-millisecond clamp steps to potentials between −60 and +60 mV in 10-mV increments, frequency 0.2 Hz (Fig. 4A). Current-voltage (I/V) relationships were constructed by normalizing the peak current amplitude by the Goldman-Hodgkin-Katz equation.21 The resulting data were plotted versus clamp-step potential and fitted with a Boltzmann function (Fig. 4C). The potential for half-maximal activation (V50) was −9.6 ± 2.0 mV (n = 5) in control conditions and −15.6 ± 2.7 mV (n = 5) following application of 25 μM AZD7009. This shift was not significantly different from the shift seen in time-matched control measurements: in the time-matched control measurements the V50 was shifted by −5.5 ± 1.6 mV (n = 5). The slope was 6.2 ± 0.3 in control conditions and 5.5 ± 0.2 after AZD7009 (n = 5). The shift in the slope was not significantly different from the shift seen in time-matched control measurements.
Both in control conditions and following application of AZD7009, maximal activation was reached after activating clamp steps to +10 mV. At more positive potentials, the current amplitude was decreased, and the decrease was more pronounced after application of AZD7009. To quantify the voltage dependence of AZD7009 block, the relative current (IAZD7009/Icontrol) at the end of the 300-millisecond activating clamp step was calculated and plotted versus the potential (Fig. 4D). The block increased steeply between −20 and +10 mV, reflecting the voltage dependence of hKv1.5 current activation and indicating an open-channel block by AZD7009. The block continued to increase between +20 and +60 mV, but with a shallower voltage dependence. The decrease of the relative current from 50.4 ± 1.9% at +20 mV to 42.6 ± 1.3% at +60 mV was statistically significant (P < 0.0001, n = 5). The voltage dependence of block at potentials more positive than needed for maximal activation reflects the influence of the transmembrane electric field on the interaction between the binding site and the charged form of AZD7009 and gives an estimate of the binding site localization in the transmembrane electric field. Because >99% of the AZD7009 drug molecules are charged at pH 7.4 (pKa = 9.7), they will be influenced by the transmembrane electric field. To quantify this effect, the relative current between 20 and 60 mV was fit by
where [D] is the concentration of AZD7009, KD* is an estimate of the KD at the reference potential 0, E is the clamp step potential, δ is the fractional electric distance, z is the valence (+1 for cationic AZD7009), F is the Faraday constant, R is the gas constant, and T is the absolute temperature. The best fit to the relative current between 20 and 60 mV yielded δ = 0.20 and KD* = 30.2 μM (Fig. 4D).
The voltage dependence of channel availability was studied using 6-second prepulses to potentials between −120 and +20 mV followed by a 300-millisecond test pulse to +20 mV (Fig. 4B). The hKv1.5 current does not fully inactivate during the prepulses because of the very slow inactivation of the current. Therefore, it was not possible to measure a true steady-state availability relationship. To get an estimate of the prepulse potential for half-maximal inactivation (V0.5), the currents at the end of the clamp step were normalized and fitted to a Boltzmann sigmoidal function (Fig. 4C). In control conditions V0.5 was −17.9 ± 1.9 mV (slope 5.7 ± 0.3, n = 5), and following AZD7009, V0.5 was −20.0 ± 2.1 mV (slope 6.5 ± 0.4, n = 5). Neither the shift in the V50 nor the slope was significantly different from the shift seen in time-matched control measurements, where the V50 was shifted by −2.1 ± 1.3 mV (n = 5) and the slope by 0.8 ± 0.5 (n = 5).
Concentration and Frequency Dependence of hKv4.3/hKChIP2.2 Current Block
The effects of AZD7009 on hKv4.3/hKChIP2.2 currents were monitored during 1-second clamp steps to +40 mV at a frequency of 0.2 Hz. Concentration-effect curves were constructed after exposing cells to increasing concentrations of drug (0.01 to 500 μM), the duration of exposure to each concentration being 3.3 minutes (Fig. 5A). Because AZD7009 and R-propafenone exerted their major effect by speeding up the apparent inactivation kinetics and not by decreasing the peak current amplitude, AUC (current-time integral) was measured to give a more accurate estimation of the potency of the drug. The AUC in the presence of drug was normalized to the control current amplitude in the same cell, plotted versus the concentration and fitted to a sigmoid dose-response curve as described above (Fig. 5C). The IC50 for block of hKv4.3/hKChIP2.2 was 23.7 ± 4.4 μM, and the Hill slope was 0.98 ± 0.12 (n = 5). For comparison, the potency of R-propafenone was also determined: the IC50 was 1.7 ± 0.2 μM, and the Hill slope was 1.19 ± 0.11 (n = 5) (Fig. 5B,C). In time-matched control measurements, the hKv4.3/hKChIP2.2 current at the end of the protocol (∼24 minutes) was 95.8 ± 2.6% (n = 5) of the current in the beginning of the protocol.
The frequency dependence of hKv4.3 current block by AZD7009 was examined using frequency changes between 0.2 and 5 Hz in 3-minute intervals. The current was activated by 100-millisecond clamp steps to +40 mV (Fig. 6A). Steady-state block by 25 μM AZD7009 at 0.2-Hz stimulation was achieved before the frequency changes were made. The intrinsic frequency dependence of hKv4.3/hKChIP2.2 was pronounced. During the first and last 5-Hz stimulation trains, the AUC decreased to 63.9 ± 1.9% and to 67.4 ± 3.6% (n = 12) of the AUC following the 0.2-Hz stimulation. When the frequency dependence of AZD7009 block was studied, the AUC was therefore normalized to the AUC at the same frequency and pulse number in control conditions in the same cell (Fig. 6B). During the first 5-Hz period the AUC decreased to 24.1 ± 2.3% (n = 7) of that in control conditions; during the 0.2-Hz period the current increased to 40.3 ± 1.9% (n = 7); and following the second 5-Hz period the AUC decreased again to 22.4 ± 2.7% (n = 7). Following 0.2-Hz stimulation, a significantly smaller block was seen as compared with both 5-Hz stimulation periods (one-way ANOVA, P < 0.0001; Tukey posttest, 0.2-Hz versus 5-Hzperiod1, P < 0.001; 0.2-Hz versus 5-Hzperiod2, P < 0.001; 5-Hzperiod1 versus 5-Hzperiod2P > 0.05, n = 7). The change in the amount of block following the change in frequency was almost instantaneous. Following the frequency change from 5 Hz to 0.2 Hz, the block decreased with a time constant of 1.18 ± 0.09 seconds (n = 7), and following the frequency change from 0.2 Hz to 5 Hz, the block increased with a time constant of 0.21 ± 0.04 seconds (n = 7).
Time Course of hKv4.3/hKChIP2.2 Current Block by AZD7009
The time course of inactivation of hKv4.3/hKChIP2.2 during a test pulse to +40 mV was well fitted with a double-exponential function. The fast time constant (τfast) was 24.2 ± 2.5 milliseconds, and the slow time constant (τslow) was 110.0 ± 18.5 milliseconds (n = 5). The rate of inactivation of hKv4.3/KChIP2.2 was mainly determined by the fast component; the relative contribution of τfast was 76.4 ± 4.4%. During application of AZD7009, the apparent rate of inactivation was increased through a more rapid τfast as well as an increase in the relative contribution of τfast (Table 2). The apparent rate constants of binding (k) and unbinding (l) were calculated according to Hatano et al.22 The time constant of block (τblock) was calculated using an approximation of the channel-blocking kinetics of AZD7009 according to the equation
τinactivation was approximated by τfast in the absence of AZD7009, and τdecay was approximated by τfast in the presence of AZD7009. 1/τblock was plotted versus the concentration of AZD7009, and the data were individually fitted by linear regression according to the equation
where k was 0.33 ± 0.02 μm−1s−1, and l was 11.2 ± 5.9 seconds−1 (n = 5, Fig. 5D). Assuming a bimolecular reaction, the apparent
which is reasonably close to the experimentally determined IC50 of 23.7 μM.
State Dependence of hKv4.3/hKChIP2.2 Current Block by AZD7009
Washin and washout of 25 μM AZD7009 were made without stimulating pulses, and the amounts of block and recovery from block were compared with those obtained after 3 minutes of 0.2-Hz stimulation to +40 mV (no figure is shown). Following 3 minutes of application of 25 μM AZD7009 without stimulating pulses, the AUC at the first 300-millisecond activating clamp step was 54.2 ± 2.9% (n = 5) of that in control conditions. After an additional 3-minute application of drug with continuous stimulation at 0.2 Hz, the AUC was 50.6 ± 4.0% (n = 5) of that in control conditions. At the first depolarization following a 3-minute washout period without stimulating pulses, the AUC had recovered to 106.4 ± 4.0% of that in control conditions. After an additional 3-minute washout period with continuing 0.2-Hz stimulation, the AUC was not significantly changed (103.2 ± 2.3%, n = 5).
Effects of AZD7009 on Voltage Dependence of Activation and Prepulse Inactivation of hKv4.3/hKChIP2.2
The voltage dependence of activation of hKv4.3/hKChIP2.2 was measured under control conditions and following steady-state block of 25 μM AZD7009. Steady-state block was achieved by a 3-minute drug application period during which the current was activated by repeated stimulation to +40 mV (0.2 Hz). Activation curves were recorded using a protocol with 1-second clamp steps to potentials between −80 and +60 mV in 10-mV increments, frequency 0.2 Hz (Fig. 7A) Current-voltage (I/V) relationships were constructed by normalizing the peak current amplitude by the Goldman-Hodgkin-Katz Equation.21 The resulting data were plotted versus clamp-step potential and fitted with a Boltzmann function (Fig. 7C). The potential for half-maximal activation (V50) was −5.7 ± 0.9 mV (slope 7.3 ± 0.2, n = 7) in control conditions and −6.4 ± 0.8 mV (slope 7.4 ± 0.3, n = 7) following application of 25 μM AZD7009; the difference was not statistically significant. The voltage dependence of activation was also measured after a 3-minute period of washout in five out of seven cells; neither the V50 nor the slope was significantly changed as compared with paired control measurements.
The voltage dependence of prepulse inactivation of hKv4.3/hKChIP2.2 current was measured in control conditions and during steady-state block by 25 μM AZD7009 using a protocol with 6-second prepulses to potentials between −120 mV and +20 mV in 10-mV increments followed by a 1-second test pulse to +40 mV, frequency 0.1 Hz (Fig. 7B). Steady-state block was achieved by a 3-minute drug application period during which the current was activated by repeated stimulation to +40 mV (0.2 Hz). Peak current amplitudes before and after application of 25 μM AZD7009 were individually normalized to the peak current amplitude at a prepulse of −120 mV in control conditions, plotted versus the prepulse potential and fitted to a Boltzmann sigmoid function (Fig. 7C). V50 was −39.8 ± 0.6 mV (slope 4.6 ± 0.2, n = 7) before exposure to drug and −43.9 ± 1.1 mV (slope 7.2 ± 1.7, n = 7) during steady-state block by AZD7009. Washout measurements were obtained in five out of seven cells; the V50 was −42.2 ± 1.3 mV (slope 3.7 ± 0.9, n = 5). A one-way ANOVA analysis of the five cells with control, AZD7009, and washout measurements did not show any significant difference of either the V50 or the slope. A peculiar finding was that the amount of block by AZD7009 seemed to increase at potentials between −100 mV and −70 mV; this effect was fully reversible on washout.
The current-voltage relationship calculated according to the Goldman-Hodgkin-Katz equation reached a maximum at about +20 mV. To quantify the voltage dependence of AZD7009 block, the relative current-time integral (AUCAZD7009/AUCcontrol) was calculated and plotted versus the potential (Fig. 7D). The block increased steeply between −20 and 0 mV, reflecting the voltage dependence of hKv4.3/KChIP2.2 current activation. The block continued to increase between +10 and +60 mV, but with a shallower voltage dependence. The decrease of the relative AUC from 59.8 ± 3.2% at +20 mV to 47.6 ± 3.7% at +60 mV was statistically significant (P < 0.001, n = 7). The best fit to the relative current between 20 and 60 mV yielded δ = 0.32 and KD* = 48.1 μM.
AZD7009 is an investigational compound for management of atrial fibrillation, which predominantly influences atrial electrophysiology, demonstrates high antiarrhythmic efficacy, and has low proarrhythmic potential in animals.17,18 AZD7009 has previously been shown to inhibit hERG and hNav1.5 currents expressed in CHO K1 cells, effects that most likely underlie the prolongation of the refractoriness and the low proarrhythmogenic activity in animals.19 In the present study, AZD7009 is demonstrated to inhibit hKv1.5 and hKv4.3/hKChIP2.2 currents in a concentration-dependent manner.
The reference compounds used displayed blocking potencies similar to values given in the literature. The IC50 for propafenone on hKv1.5 (IC50 = 13.3 μM) are in agreement with previously reported values for block of hKv1.5 expressed in Ltk cells (4.4 μM at +60 mV and 6.1 μM at 0 mV)4 and IKur in human atrial myocytes (8.6 μM at +60 mV).4,5 The potency of R-propafenone to block the hKv4.3/hKChIP2.2 current (IC50 = 1.7 μM) is consistent with values reported for propafenone inhibition of Ito in rat ventricular myocytes (IC50 = 3.3 μM),23 cultured neonatal rat ventricular myocytes (IC50 = 2.1 μM),24 and human atrial myocytes (IC50 = 4.9 μM).5
The results obtained in the present study suggest that AZD7009 gains access to its binding site in hKv1.5 and hKv4.3/KChIP2.2 channels only after channel activation and that dissociation of the drug is necessary before the channels can close. Several findings support this view. First, AZD7009 does not block peak currents to the same extent as it blocks the current at the end of an activating clamp step. This can be explained by the necessity of channel opening before AZD7009 can access the binding site and block the current. The fast “inactivation” induced by AZD7009 can thus be explained not by speeding up the inactivation process itself but by AZD7009 binding to its site and blocking the current only after channel opening and thereby inducing a current decrease as the depolarization is prolonged. This was indeed seen on exponential curve fit to the inactivating phase of the hKv1.5 current: a second fast component was added following application of AZD7009 without any change in the time course of inactivation. Second, for hKv1.5 the tail current amplitude was decreased, and deactivation of the tail current was slower, following application of AZD7009 compared with control, producing a tail current crossover. Tail current crossover has been reported for many Kv1.5 blocking compounds, eg, propafenone4 and quinidine,6 and can be explained by dissociation of the drug from its binding site before closing of the channel can occur. When the drug dissociates, current conduction through the channel will be seen before the channel becomes blocked again or finally closes. The extra time needed before channels blocked by AZD7009 can close thus leads to a slowing of the deactivation of the current. Because the binding of AZD7009 to the channels is dependent on the concentration, this explains the concentration-dependent slowing of the tail current deactivation, although the rate of unbinding itself is not dependent on the concentration of the drug. It should be noted that another possible explanation for the tail current crossover might be that AZD7009 induces changes in the gating kinetics of the channel. Third, the steep voltage dependence of block seen at depolarizing clamp steps to potentials between −20 and +10 mV for both hKv1.5 and hKv4.3/hKChIP2.2 currents overlaps the voltage dependence of activation of the currents. Hence, development of block follows channel activation. Fourth, the results from washin and washout of AZD7009 at repolarized membrane potentials are also in accordance with a rapidly activated channel block mechanism for both hKv1.5 and hKv4.3/KChIP2.2. Because the development of block overlaps with the activation process, it is not possible to fully rule out a small contributing component of closed-channel block. The blocking mechanism for hKv1.5 and hKv4.3/hKChIP2.2 seems to be very similar. It is also similar to the blocking mechanism of hERG previously described for AZD7009, with the exception that AZD7009 was trapped in closed hERG channels until the channels were activated by a depolarizing pulse.19 The trapping of AZD7009 seen in hERG but not in hKv1.5 or hKv4.3/hKChIP2.2 can be explained by a difference in the size of the central cavity of the channels.
The frequency dependence of block differed between hKv1.5 and hKv4.3/hKChIP2.2 currents. For hKv1.5, a small but significant decrease of block was seen on increasing the stimulation frequency from 0.2 Hz to 5 Hz, whereas for hKv4.3/hKChIP2.2 current an increase of block was observed. This difference most likely depends on the difference between blocking kinetics and rate of channel inactivation. For Kv1.5 the slow inactivation and, hence, the relatively long time period spent in the open state(s) permits a rapid development of equilibrium block. The small decrease of current block at 5-Hz stimulation is then most likely because a larger amount of the hKv1.5 channels are driven into an inactivated state for which AZD7009 might have no or low affinity. The exact magnitude of this frequency-dependent relief in block would then depend on the balance between drug binding of kinetics from the open state and rate of entrance into inactivated or drug-bound inactivated states. In analogy with this, the rapid inactivation of hKv4.3/hKChIP2.2 would not permit full block to occur within the activation step. At higher activation frequencies the increased total amount of time spent in the open state and hence the increased total amount of time available for the drug to bind will then increase the amount of block.
The rate constants for association (k) and dissociation (l) determined for AZD7009 for hKv1.5 (k = 1.3 μm−1s−1, l = 36.8 seconds−1) and for hKv4.3/hKChIP2.2 (k = 0.33 μm−1s−1, l = 11.2 seconds−1) currents from the time constants of development of block at different concentrations of drug were well correlated to the experimentally determined IC50 values. Rate constants for association and dissociation have been reported in the literature for other hKv1.5 blockers, eg, propafenone (k = 8.9 μm−1s−1, l = 39.5 seconds−1)4 and quinidine (k = 4.5 μm−1s−1, l = 34 seconds−1).6 It is evident that the rate constant for association of AZD7009 is lower than those for these compounds, whereas the rate constant for dissociation is similar to that of both propafenone and quinidine, which explains the potency difference between these compounds. In a previous study, the apparent rate constants for association and dissociation of AZD7009 to the hERG potassium channel were determined (k = 0.37 μm−1s−1, l = 0.20 seconds−1).19 The major determinant of the potency difference among the block of hKv1.5 (IC50 = 27 μM), hKv4.3/hKChIP2.2 (IC50 = 24 μM), and hERG (IC50 = 0.56 μM) seems to be the difference in the rate constants of dissociation of AZD7009 from its binding site.
The fractional electric distance (δ) was estimated to 0.20 for hKv1.5 and 0.32 for hKv4.3/hKChIP2.2. The value for hKv1.5 corresponds well with estimated values for propafenone (δ = 0.17)4 and for quinidine (δ = 0.196 and δ = 0.23).25 The similarity between the values determined for the fractional electric distances of these compounds may indicate a common binding site for these drugs in the internal pore of hKv1.5 channels.
AZD7009 has previously been shown to inhibit hERG (IC50 = 0.56 μM) and hNav1.5 (IC50 = 4.3 μM at 10 Hz) currents more potently as compared with the inhibition of hKv1.5 (IC50 = 27 μM) and hKv4.3/hKChIP2.2 (IC50 = 24 μM) as reported in the present study.19 The combined inhibition of the hERG and the hNav1.5 currents is likely to be the principal mechanism behind the high antiarrhythmic efficacy demonstrated in animals in vivo.17,18 Data from studies in dogs show that AZD7009 predominantly increases the atrial refractoriness, conduction time, and excitability threshold, whereas the effects in the ventricle are much less pronounced.17 A more marked sodium channel blockade in the atria as compared with the ventricles might give an explanation for this. In addition, inhibition of the hKv1.5 and the hKv4.3/hKChIP2.2 currents might also contribute to the predominant effect in the atria. Although the IC50 values for blocking these currents were well above the plasma concentrations at which predominant atrial actions in vivo have been reported26 and clinical efficacy preliminary seen (conversion of atrial fibrillation to sinus rhythm has been noted at 2-2.5 μmol/L AZD7009; O. Almgren, personal communication), it is speculated that a minor degree of block of these channels may act synergistically with the more prominent block of IKr and INa (thus functionally resulting in a predominant action on the atria). Because the inhibition of hKv4.3/hKChIP2.2 current was increased at increased stimulation frequency, this current in particular is expected to be blocked in the fibrillating atria. Interestingly, Blaauw and co-workers recently demonstrated that the IKur/Ito blocker AVE0118 and the IKr blocker ibutilide synergistically increased the atrial refractoriness in the goat.27 In addition, the rate of the hKv1.5 tail current deactivation was decreased at relatively low concentrations of AZD7009 with the relative amplitude of the fast time constant being significantly decreased at 3 μM. A decreased rate of deactivation of hKv1.5 might help provide electrical stability in the atria between action potentials.
AZD7009 blocks both the hKv1.5 current and the hKv4.3/hKChIP2.2 current in a concentration-dependent manner. The block of the hKv4.3/hKChIP2.2 current increased with increasing stimulation frequency whereas the amount of block of the hKv1.5 current was slightly decreased with increasing stimulation frequency. The high antiarrhythmic efficacy and low proarrhythmic activity of AZD7009 seen in animals in vivo is most likely attributable to the inhibition of the hERG and the hNav1.5 currents as previously reported, but a contributing effect by the inhibition of the hKv1.5 and the hKv4.3/hKChIP2.2 currents can not be excluded. In addition, the inhibition of the hKv1.5 and the hKv4.3/hKChIP2.2 currents are likely to contribute to the predominant effects on atrial electrophysiology, possibly by the synergistic action of combined hKv1.5, hKv4.3/hKChIP2.2 and hERG blockade.
1. Witchel HJ, Hancox JC. Familial and acquired long QT syndrome and the cardiac delayed rectifier potassium current. Clin Exp Pharmacol Physiol
2. Volders PGA, Vos MA, Szabo B, et al. Progress in the understanding of early afterdepolarisations and torsades de pointes: time to revise current concepts. Cardiovasc Res
3. Gross GJ, Castle NA. Propafenone inhibition of human atrial myocyte repolarizing currents. J Mol Cell Cardiol
4. Franqueza L, Valenzuela C, Delpon E, et al. Effects of propafenone and 5-hydroxy-propafenone on hKv1.5 channels. Br J Pharmacol
5. Seki A, Hagiwara N, Kasanuki H. Effects of propafenone on K currents in human atrial myocytes. Br J Pharmacol
6. Snyders DJ, Knoth KM, Roberds SL, et al. Time-, voltage-, and state-dependent block by quinidine of a cloned human cardiac potassium channel. Mol Pharmacol
7. Singarayar S, Bursill J, Wyse KR, et al. Extracellular acidosis modulates drug block of Kv4.3
currents by flecainide and quinidine. J Cardiovasc Electrophysiol
8. Tamkun M, Knoth K, Walbridge J, et al. Molecular cloning and characterization of two voltage-gated K+
channel cDNAs from human ventricle. FASEB J
9. Fedida D, Wible B, Wang Z, et al. Identity of a novel delayed rectifier current from human heart with a cloned K+
channel current. Circ Res
10. Wang Z, Fermini B, Nattel S. Sustained depolarization-induced outward current in human atrial myocytes. Evidence for a novel delayed rectifier K+
current similar to Kv1.5 cloned channel currents. Circ Res
11. Amos GJ, Wettwer E, Metzger F, et al. Differences between outward currents of human atrial and subepicardial ventricular myocytes. J Physiol
12. Oudit GY, Kassiri Z, Sah R, et al. The molecular physiology of the cardiac transient outward potassium current (Ito
) in normal and diseased myocardium. J Mol Cell Cardiol
13. Dixon JE, Shi W, Wang H-S, et al. Role of the Kv4.3
channel in ventricular muscle: a molecular correlate for the transient outward current. Circ Res
14. Wang Z, Feng J, Shi H, et al. Potential molecular basis of different physiological properties of the transient outward K+
current in rabbit and human atrial myocytes. Circ Res
15. Bertaso F, Sharpe CC, Hendry BM, et al. Expression of voltage-gated K+
channels in human atrium. Basic Res Cardiol
16. Decher N, Uyguner O, Scherer CR, et al. hKChIP2 is a functional modifier of hKv4.3 potassium channels: cloning and expression of a short hKChIP2 splice variant. Cardiovasc Res
17. Goldstein RN, Khrestian C, Carlsson L, et al. AZD7009: A new antiarrhythmic drug with predominant effects on the atria effectively terminates and prevents reinduction of atrial fibrillation and flutter in the sterile pericarditis model. J Cardiovasc Electrophysiol
18. Carlsson L, Duker G, Linhardt G. AZD7009-A new atrial antifibrillatory compound that markedly delays atrial refractoriness and possesses low proarrhythmic activity in the anaesthetised rabbit. Eur Heart J
19. Persson F, Carlsson L, Duker G, et al. Blocking characteristics of hERG, hNav1.5 and hKvLQT1/hminK after administration of the novel antiarrhythmic compound AZD7009. J Cardiovasc Electrophysiol
20. Hamill OP, Marty A, Neher E, et al. Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pflugers Arch
21. Clay JR. Determining K+
channel activation curves from K+
channel currents. Eur Biophys J
22. Hatano N, Ohya S, Muraki K, et al. Dihydropyridine Ca2+
channel antagonists and agonists block Kv4.2, Kv4.3
and Kv1.4 K+
channels expressed in HEK293 cells. Br J Pharmacol
23. Slawsky MT, Castle NA. K+
channel blocking actions of flecainide compared with those of propafenone and quinidine in adult rat ventricular myocytes. J Pharmacol Exp Ther
24. Cahill SA, Kirshenbaum LA, Gross GJ. Transient outward inhibition by propafenone and 5-hydroxypropafenone in cultured neonatal rat ventricular myocytes. J Cardiovasc Pharmacol
25. Wang Z, Fermini B, Nattel S. Effects of flecainide, quinidine, and 4-aminopyridine on transient outward and ultrarapid delayed rectifier currents in human atrial myocytes. J Pharmacol Exp Ther
26. Duker G, Björkman JA, Linhardt G, et al. AZD7009-A new atrial antifibrillatory compound that markedly delays atrial refractoriness in the anaesthetised dog. Heart Rhythm
27. Blaauw Y, Goegelein H, Duytschaever M, et al. Synergistic class III action of blockade of IKur
(AVE0118) and IKr
(dofetilide/ibutilide) in electrically remodeled atria of the goat. Circulation
Keywords:© 2005 Lippincott Williams & Wilkins, Inc.
atrial fibrillation; antiarrhythmic drug; potassium channels; whole-cell voltage clamp; Kv1.5; Kv4.3