Second Department of Internal Medicine, University of Occupational and Environmental Health Japan, 1-1 Iseigaoka, Yahatanishi-ku, Kitakyushu 807-8555, Japan. E-mail: email@example.com
1Miconazole, an imidazole antifungal agent, is associated with acquired long QT syndrome and ventricular arrhythmias. Miconazole increases the plasma concentration of QT-prolonging drugs by inhibiting the hepatic cytochrome P450 metabolic pathway, but whether it has direct effects on cardiac ion channels has not been elucidated.
2To determine the mechanism underlying these clinical findings, we investigated the effect of miconazole on human ether-a-go-go-related gene (HERG) K+ channels.
3HERG channels were heterologously expressed in human embryonic kidney 293 (HEK293) cells and whole-cell currents were recorded using a patch-clamp technique (23°C).
4Miconazole inhibited HERG peak tail current in a concentration-dependent manner (0.4–40 μM) with an IC50 of 2.1 μM (n=3–5 cells at each concentration, Hill coefficient 1.2). HERG block was not frequency-dependent. It required channel activation, occurred rapidly, and had very slow dissociation properties.
5The activation curve was shifted in a negative direction (V1/2: −9.5±2.3 mV in controls and −15.3±2.4 mV after 4 μM miconazole, P<0.05, n=6). Miconazole did not change other channel kinetics (activation, deactivation, onset of inactivation, recovery from inactivation, steady-state inactivation).
6The S6 domain mutation, F656C, abolished the inhibitory action of miconazole on HERG current indicating that miconazole preferentially binds to an aromatic amino-acid residue within the pore-S6 region.
7Our findings indicate that miconazole causes HERG channel block by binding to a common drug receptor, and this involves preferential binding to activated channels. Thus, miconazole prolongs the QT interval by direct inhibition of HERG channels.
rapidly activating component of the delayed rectifier K+ current
slowly activating component of the delayed rectifier K+ current
human embryonic kidney
MinK-related peptide 1
Miconazole is a widely prescribed imidazole antifungal drug available in topical, vaginal and parenteral formulas. Azole antifungals (e.g. fluconazole, ketoconazole) are associated with the acquired long QT syndrome and ventricular arrhythmias (Moss, 1999; Tonini et al., 1999; Wassmann et al., 1999; Dorsey & Biblo, 2000; Tamargo, 2000). Since azole antifungals inhibit multiple cytochrome P450 enzymes in the liver and gastrointestinal tract, the mechanism of arrhythmogenesis is considered to be a rise in the plasma concentration of QT-interval prolonging drugs (e.g., class III antiarrhythmic drugs, some H1-receptor antagonists and antibiotics) that use the same metabolic pathway (Albengres et al., 1998; Venkatakrishnan et al., 2000; Roden, 2001). However, ventricular arrhythmia can occur in the absence of other QT-interval prolonging drugs (Coley & Crain, 1997; Wassmann et al., 1999), suggesting that azole antifungal drugs may have direct effects on the heart. In fact, ketoconazole has been reported to directly block human ether-a-go-go-related gene (HERG or KCNH2) encoded K+ channels (Dumaine et al., 1998). Although miconazole is known to inhibit multiple cytochrome P450 enzymes (Venkatakrishnan et al., 2000; Zhang et al., 2002), it is not known whether miconazole exerts direct effects on HERG channels.
In mammalian cardiac ventricular cells, the principal repolarising currents activated during the action potential plateau are the rapidly (IKr) and slowly (IKs) activating components of the delayed rectifier K+ current (Sanguinetti & Jurkiewicz, 1990; Carmeliet, 1993), with IKr encoded by HERG (Curran et al., 1995; Sanguinetti et al., 1995; Trudeau et al., 1995). In addition to class III antiarrhythmic drugs, a remarkable array of structurally diverse therapeutic agents that cause acquired long QT syndrome block HERG or IKr channels (Viskin, 1999; Drici & Barhanin, 2000; Witchel & Hancox, 2000). One possible explanation for this is that these drugs bind to a common drug receptor within the pore of HERG channels. Using alanine scanning mutagenesis of the pore-S6 region of HERG channel, aromatic amino-acid residues (Y652 and particularly F656) were reported to be key determinants of drug binding to HERG channels (Mitcheson et al., 2000a).
The aim of the present study was to investigate the effect of miconazole on HERG K+ channels heterologously expressed in a human cell line. We found that miconazole potently inhibited HERG K+ current by preferential binding to activated channels with ultra-slow dissociation kinetics and that the binding required a common drug receptor within the pore-S6 region.
DNA constructs and transfection of HEK293 cells
HERG cDNA was subcloned into BamHI/EcoRI sites of the pCDNA3 vector (Invitrogen, San Diego, CA, U.S.A.). HERG wild-type channels were stably expressed in a human embryonic kidney (HEK293) cell line as described previously (Zhou et al., 1998). The HERG F656C (phenylalanine to cysteine at position 656) mutation was generated by site-directed mutagenesis of wild-type HERG cDNA with the use of the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, U.S.A.), and the integrity of the construct was verified by DNA sequencing. HEK293 cells were transfected with this construct using the lipofectamine method (Invitrogen). Stably transfected cells generated through G418 antibiotic selection were subcloned to achieve a uniform HERG expression level. The cells were cultured in Dulbecco's Modified Eagle's Medium (D-MEM: Life Technologies, Gaithersburg, MD, U.S.A.) supplemented with 10% fetal bovine serum, 100 U ml−1 penicillin and 100 μg streptomycin. For electrophysiological analysis, cells were harvested from the culture dish by trypsinisation, washed with D-MEM, and stored in this medium at room temperature for later use. Cells in culture dishes were easily detached by trypsinisation and studied within 8 h of harvest.
HERG channel current was recorded using the whole-cell patch-clamp technique. Transfected cells were transferred to a bath mounted on the stage of an inverted microscope (Diaphot, Nikon, Japan). The bath was perfused with 4-(2-hydroxyethyl)-1-piperazineethanesulphonic acid (HEPES)-buffered Tyrode solution containing (in mM) 137 NaCl, 4 KCl, 1.8 CaCl2, 1 MgCl2, 10 glucose, and 10 HEPES (pH 7.4). The volume of the bath was 0.6 ml and the external solution in the bath was almost completely exchanged within 1 min at a perfusion rate of 1.25 ml min−1. The internal pipette solution contained (in mM) 130 KCl, 1 MgCl2, 5 ethylene glycol bis (beta-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA), 5 Mg-ATP, and 10 HEPES (pH 7.2). Electrodes were constructed from borosilicate glass using a micropipette puller (P-87, Sutter Instrument Co., Novato, U.S.A.) and heat-polished with a microforge (MF-83, Narishige, Tokyo). The final resistance of the electrode was 3–5 MΩ when filled with the pipette solution. Membrane currents were recorded with an Axopatch 200B amplifier (Axon Instruments, Union City, U.S.A.) and digitised at 2–10 kHz with an analogue-to-digital converter (DigiData 1200B; Axon Instruments). Computer software (pCLAMP Ver. 8.1, Axon Instruments) was used to generate voltage clamp protocols, acquire data and analyse current traces. All experiments were performed at room temperature (23±1°C).
Miconazole was kindly provided by the Mochida Pharmaceutical Co. (Tokyo) and was dissolved in dimethyl sulphoxide (DMSO) to prepare a stock solution (40 mM). Final drug concentrations were prepared by diluting stock solution with Tyrode solution. The highest concentration of DMSO used in this study was 0.1%. In preliminary experiments, HERG tail current amplitudes expressed in HEK293 cells were not significantly changed after 3 min of DMSO application at 0.1% (n=4).
Data are expressed as mean±s.e.m. Where applicable, n represents the number of cells studied. Statistical significance was analysed using a two-tailed Student's t-test (paired or unpaired). P-values <0.05 were considered statistically significant. Curve fitting was performed using multiple nonlinear least-squares regression analysis (pCLAMP Ver. 8.1, Axon Instruments or Sigma Plot Ver. 7.0, SPSS Science, Chicago, IL, U.S.A.).
Miconazole inhibits HERG channels expressed in HEK293 cells
The effect of miconazole on the HERG current–voltage (I–V) relationship is shown in Figure 1. HERG current was elicited from a holding potential of −80 mV by 4-s long depolarising steps to between −70 and 60 mV applied in 10 mV increments every 15 s. Tail current was recorded with a step to −50 mV for 6 s. Representative currents recorded in control conditions and 3 min after application of miconazole (4 μM) in the same cell are shown in Figure 1a. Miconazole reduced HERG current amplitude during the depolarising step as well as tail current. The averaged normalised I–V relationships for HERG current measured at the end of the depolarising steps and for peak tail current are shown in Figure 1b and c, respectively. Control HERG current during the depolarising steps was maximal at about 0–10 mV, with tail current fully activated following steps to 10–20 mV. Miconazole (4 μM) reduced the steady-state current amplitude to 25.2±1.9% (n=6) of control at 10 mV, and tail current peak amplitude to 27.8±2.4% (n=6) of control at 20 mV, respectively. Voltage-dependence of activation was also evaluated by plotting normalised tail current as a function of voltage. Data were fitted with a Boltzmann function: I=Imax × [1+exp(V1/2−V)/κ]−1, where Imax is maximum amplitude, V1/2 and κ are half-activation voltage and the slope factor, respectively. In this protocol, the half-activation voltage was significantly shifted in a negative direction from −9.5±2.3 mV in controls to −15.3±2.4 mV after miconazole (P<0.05, n=6) and the slope factor was significantly decreased from 8.7±0.7 in controls to 6.9±0.4 after miconazole (P<0.05, n=6).
Figure 2a shows the effects of miconazole (40 μM) on the HERG current during a drug wash-in protocol. HERG current was elicited from a holding potential of −80 mV by a 4-s depolarising step to 20 mV followed by a repolarising step to −50 mV for 6 s, and the protocol was applied every 15 s. After obtaining the control record, miconazole was applied to the bath. HERG channel block occurred rapidly to reach a steady-state level within 1–2 min (Figure 2b). The blocking effect on HERG channels was not reversed during washout of up to 10 min.
The concentration–response relationship of miconazole block was assessed by changes in the amplitude of HERG current measured at the end of depolarising step to 20 mV and the peak tail current after attaining steady-state drug block. Only a single drug concentration was tested on each cell. miconazole reduced HERG current in a concentration-dependent manner (Figure 2c). The IC50 value and Hill coefficient for HERG current measured at the end of depolarising step to 20 mV were 2.2 μM and 1.0, respectively, and for tail current peak amplitude were 2.1 μM and 1.2, respectively.
Attenuation of miconazole block by F656C mutant
Aromatic amino-acid residue, F656, located in the S6 transmembrane domain of HERG channels seem to play a major role in drug binding properties as reported previously (Mitcheson et al., 2000a). In order to gain additional information on the molecular basis of miconazole binding to HERG channels, we investigated the effects of miconazole on HERG F656C mutant channels. The affinity of miconazole to HERG wild-type and mutant channels was compared by measuring the effects of miconazole at 40 μM (Figure 2d). Voltage protocols were applied as described in Figure 2a to record currents under control and after application of miconazole. In wild-type HERG channels, the tail current peak amplitudes were reduced to 5.0±0.3% (n=3) of the control current amplitude (Figure 2c). The inhibitory effect of miconazole on HERG current was almost completely abolished in the F656C mutant channels. The mean relative peak amplitude of the tail current measured after 3 min of miconazole application yielded 98.2±0.6% (n=7) of the control current amplitude.
State-dependence of HERG channel block by miconazole
To determine the state-dependence (closed, open and inactivated) of channel block, we used a protocol modified from Kiehn et al. (1996) and Thomas et al. (2001). From a holding potential of −80 mV, the cell was first depolarised to 80 mV for 100 ms to maximally activate channels from the closed state, and then the cell was repolarised to 0 mV for 300 ms to allow for maximal channel recovery from inactivated to the open state (see Figure 1). The cell was then either held at 0 mV (protocol 1) or a second step to 80 mV was applied for 300 ms to increase channel inactivation before step-back to 0 mV (protocol 2; Figure 3a). This approach of employing steps to very positive potentials was designed to determine whether maximal drug block occurred when channels were predominantly in the inactivated or open states. Figure 3a shows example control records with these protocols. With protocol 2 there was less current during the 300-ms long step to 80 mV because of increased channel inactivation, whereas following step-back to 0 mV, the current with both protocols was similar. Figure 3b shows the effect of miconazole (4 μM). For this experiment, cells were held continuously at −80 mV for 3 min during drug wash-in to keep the channels in the closed state. The data show that after drug wash-in, with the initial depolarisation a large amplitude HERG current was present consistent with minimal drug binding to the closed state, and this was followed by a decline in HERG current as drug block developed (‘first pulse block’). The time-dependent decline in HERG current amplitude at 0 mV exhibited a single exponential process (dashed line). The 300-ms long step to 80 mV (protocol 2) interrupted this exponential time course and the current amplitude at the beginning of the second step to 0 mV exceeded the extrapolated predicted current amplitude by 7.9±1.3% (n=5). The maximum current amplitude at the second step to 0 mV and the extrapolated predicted current amplitude at the same time point were 691±184 and 645±177 pA, respectively (n=5). The difference was statistically significant by the paired t-test (P<0.01). These results suggest that blockade by miconazole is weaker when the HERG channels are predominantly in the inactivated state. Thus, the block of HERG current by 4 μM miconazole requires channel activation from the closed state, and block is greater using protocols that maximise availability of open channels.
We examined in greater detail the possibility of closed state block. Using the protocol shown in Figure 3b, currents for control conditions and with first pulse block with 4 or 12 μM miconazole (Figure 3c and d, respectively) were superimposed for the initial step to 80 mV and the following step to 0 mV. With 4 μM miconazole, the initial current during the step to 80 mV was not different from the control current, and with the step to 0 mV block gradually developed, suggesting minimal closed state block (Figure 3c). When the drug concentration was increased to 12 μM (Figure 3d), a small reduction in the current amplitude was observed at the initial phase of the depolarisation step to 80 mV and there was greater block with the step to 0 mV, suggesting minor closed state block. However, the rapid development of open-channel blockade was not excluded, because drug binding might be accelerated at higher drug concentration. The peak current amplitudes evaluated at the initial 80 mV step-pulse were decreased by 5.4±2.9% (n=5) and 25.4±7.8% (n=3) after application of 4 and 12 μM miconazole, respectively (P<0.05). These results suggest that miconazole is predominantly an activated state blocker with higher affinity for the open state, and lowest affinity for the closed state.
Frequency dependence of HERG current block
The frequency dependence of HERG current block was investigated by applying 30 repetitive pulses (see inset, Figure 4) at 0.2 and 1 Hz, after holding the cell at −80 mV for 3 min during wash-in of 4 μM miconazole. HERG current was first rapidly activated by a 100 ms step to 80 mV, which was followed by a 400 ms step to 20 mV. As block of HERG current develops quickly after channel activation (see Figure 3), HERG current was measured as the maximum outward current amplitude during the step pulse at 20 mV, and was normalised to the control current measured prior to drug exposure. For control conditions, the HERG current amplitude during the pulse train decreased only slightly (<4%). Following exposure to 4 μM miconazole, application of the pulse train at either 0.2 or 1.0 Hz decreased current amplitude by 50.3±5.8% (n=5) at 0.2 Hz and by 51.1±2.4% (n=5) at 1 Hz. The time constants of the fast and slow components of the time course of block were not significantly different between the 0.2 and 1.0 Hz frequencies (n=5 cells at each pulsing rate). Thus, neither the development of block nor the amount of steady-state block was frequency-dependent.
Effects of miconazole on kinetics of channel gating
Drug block of ion channels often exerts apparent effects on channel kinetics. Using voltage clamp protocols identical to those of Zhou et al. (1998), we measured activation, deactivation, inactivation and recovery from inactivation. After obtaining the control record, step pulses as illustrated in Figure 2a were applied in the presence of 4 μM miconazole for 3 min and then HERG current was recorded with identical protocols. To evaluate the time course of HERG channel activation, the cell was clamped to 20 mV from a holding potential of −80 mV for varying durations followed by a repolarising step to −50 mV to elicit tail current. The time constant obtained by fit to the envelope of tail currents was used for evaluation of the HERG current activation (Figure 5a and b). The time constants for activation did not significantly change after application of 4 μM miconazole (152±16 ms in control vs 141±17 ms after miconazole, n=8). Deactivation current was elicited by 10 mV step pulses from −100 mV to −20 ms after a 1-s depolarisation pulse to 60 mV from a holding potential of −80 mV (Figure 5c). Deactivation time constants were obtained by fit with double exponential function to the decay phase of tail current. The two time constants describing current decay were not significantly changed by 4 μM miconazole (n=5, Figure 5d).
To test the effects of miconazole on recovery from inactivation, the cells were depolarised to 60 mV for 200 ms to rapidly inactivate HERG channels and then repolarised to between −100 and −20 mV to elicit tail currents (Figure 6a). The rising phase of the tail current represents the rapid recovery of HERG channels from inactivated to open states. After fitting a double exponential function to the tail current trace at each potential, the time constant for the rising phase was used to measure recovery from inactivation. There was no significant difference before and after 4 μM miconazole application (n=5, Figure 6c).
We also studied the effects on the onset of inactivation with 4 μM miconazole. The cells were depolarised to 60 mV for 200 ms to promote inactivation and hyperpolarised to −100 mV for 10 ms, which is a sufficient time for recovery from inactivation as indicated in Figure 6c without significantly deactivating the channels (Figure 6b). Then, channels that recovered were forced to reinactivate at potentials between −20 and 60 mV. The time constant of inactivation was measured by fitting a single exponential function to the decaying current trace. The onset of inactivation was accelerated at strongly depolarised potentials (50 and 60 mV), but there were no significant differences between control and drug application in wide range of voltages (n=5, Figure 6c). Thus, miconazole has little effect on the rates of inactivation and recovery from inactivation of HERG channels.
Steady-state inactivation was measured using a double-pulse protocol with varying interpulse repolarisation levels (Figure 6d). After the first 200 ms pulse to 60 mV promoting inactivation, 10 ms interpulses were applied to potentials between −100 and 20 mV, followed by a second step to 60 mV. Peak currents measured 3 ms after the onset of the second step to 60 mV were plotted as a function of the proceeding repolarising potential and then fitted with a Boltzmann function: I=Imax × [1−exp(V−V1/2)/κ]−1, where I is current amplitude, Imax is maximum amplitude, V1/2 and κ are half-inactivation voltage and the slope factor, respectively. Data from six cells under control and 4 μM miconazole were normalised to the asymptotic values of their respective Boltzmann fits (Figure 6e). The half-inactivation voltage and slope factor obtained by fit with Boltzmann equation were −64.8±2.2 mV and −17.2±1.3 in controls (n=6), and −65.5±1.4 mV and −17.2±1.7 after 4 μM miconazole (n=6), respectively. Both parameters were not statistically different. Thus, miconazole does not exert specific effects on channel inactivation. Otherwise unbinding upon repolarisation, which allows channels to become available for opening, does not occur because of the slow unbinding of miconazole from the channels and unchanged kinetics might reflect the kinetics of miconazole unbounded channels.
In this study, we demonstrated that the azole antifungal drug miconazole potently inhibited HERG current heterologously expressed in a human cell line. As with most drugs, the block was state-dependent. Under our experimental conditions, closed (rested) state block was minimal and was noted only at high drug concentrations. In contrast, HERG channel block by miconazole required channel activation, and the greatest block appeared to be with the open state of the channel. As shown in Figure 3, evidence to support preferential open state block was derived from experiments showing a reduction in drug block at voltages (80 mV) where the occupancy of the inactivated state was greatest. Nevertheless, the experiments could not clearly distinguish between open or inactivated state block.
When trains of depolarising steps were applied, miconazole caused a progressive decrease in current, as expected for a drug with state-dependent block. The time course of development of block and the amount of steady-state block were not significantly different between 1.0 and 0.2 Hz (Figure 4). This lack of frequency-dependence can be accounted for by a very slow dissociation of the drug from the channel. With sufficiently slow drug dissociation the longer recovery interval between pulses at slower rates does not cause appreciable additional recovery over the time scale studied in our experiments. The negligible current recovery after drug washout also supports the notion of very slow channel dissociation of this drug. The pattern of activated state block and slow dissociation is consistent with an activation gate trapping mechanism, as also reported for other HERG channel blocking drugs such as the methanesulphonanilide compounds (Carmeliet, 1992; Mitcheson et al., 2000b).
The structural basis of HERG channel block by several chemically unrelated compounds have recently been studied in detail. Aromatic amino-acid residues, particularly F656, located in the S6 transmembrane domain of HERG channels were proposed to be the most important molecular determinants of drug binding to HERG channels (Lees-Miller et al., 2000; Mitcheson et al., 2000a; Kamiya et al., 2001; Sanchez-Chapula et al., 2002; Scholz et al., 2003; Thomas et al., 2004). In the present study, F656C mutant channels markedly attenuated HERG current inhibition by miconazole. The result clearly shows that miconazole predominantly binds to a drug receptor within the pore-S6 region.
It is of interest to compare our findings with miconazole to previous findings with ketoconazole (Dumaine et al., 1998). Ketoconazole was reported to block HERG channels predominantly by binding to the closed state of HERG channels to cause tonic block, with lower affinity binding to the open state. In addition, HERG current block by ketoconazole was reversible upon drug washout. Our data show that block of HERG channels by miconazole required channel activation from the closed state, and the drug block was not easily reversed by drug washout. Thus, miconazole appears to act like most drugs previously studied on HERG channels and IKr, where they bind preferentially to the channel following its activation from the closed state. The IC50 values for block of HERG tail current peak amplitude also differed (49 μM in ketoconazole (Dumaine et al., 1998), and 2.1 μM in miconazole). These differences may be attributed in part to differences in the expression systems used (Xenopus oocytes versus HEK293 cells), experimental conditions, and channel kinetics.
Clinical implications of HERG channel blockade by miconazole and limitations of the study
Although miconazole is associated with cardiac arrhythmias, the underlying mechanism(s) remains speculative. Since miconazole is known to inhibit multiple cytochrome P450 enzymes in the liver and gastrointestinal tract (Zhang et al., 2002), warnings have been issued against the concomitant administration of miconazole with QT interval-prolonging drugs that use the same metabolic pathway (Albengres et al., 1998; Venkatakrishnan et al., 2000; Roden, 2001). The present study demonstrates that miconazole also directly inhibits HERG channels. The maximal therapeutic concentrations of miconazole in patients ranges from 0.96 to 3.5 μg ml−1 when administrated intravenously, corresponding to 2.3–8.4 μM. Since the plasma protein binding for miconazole is reported to be approximately 90% (Stevens et al., 1976), the IC50 values determined in the present study may exceed slightly the expected free-drug concentration range in plasma. Thus, the apparent low incidence of QT prolongation by miconazole might in part be accounted for by its high protein binding. Nonetheless, our findings provide a direct cellular mechanism for miconazole-induced QT interval prolongation and cardiac arrhythmias. Direct inhibition of the HERG current may be of greater concern when miconazole is used in patients with congenital long QT syndrome, in patients with electrolyte abnormalities such as hypokalemia, or in patients receiving other QT interval prolonging drugs whether acting directly through HERG channel block or inhibition of drug metabolism through the P450 system.
The present study has certain limitations. We used HERG channels heterologously expressed in a human cell line. The possible effects of miconazole on other ion channels as well as receptors in vivo cannot be excluded. Secondly, minK-related peptide 1 (MiRP1), encoded by KCNE2, coassembles with the pore-forming HERG subunit and probably reconstitutes native IKr (Abbott et al., 1999). Compared to channels formed by the HERG subunit alone, HERG/MiRP1 complexes show altered channel properties and increase the potency of channel block by E-4031. Moreover, the mutant forms of MiRP1 demonstrated diminished potassium currents or increased channel blockade by drugs (Sesti et al., 2000). Therefore, it is possible that miconazole sensitivity might be different between HERG and HERG/MiRP1 channels, although wild-type MiRP1 does not markedly influence the drug sensitivity of the channels (Numaguchi et al., 2000; Kamiya et al., 2001; Scherer et al., 2002; Friederich et al., 2004).
We thank Drs Zhengfeng Zhou, Qiuming Gong, Brian P. Delisle and Blake D. Anson for expert technical assistance and advice. We thank Dr Shetuan Zhang for his review of the manuscript. This work was supported by Astrazeneca Research Grant (to T.N.), and by UOEH Research Grants for Promotion of Occupational Health (to T.N. and H.A.).