Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Ketoconazole is an antifungal agent used for treatment of disorders ranging from toe nail fungus to dandruff. Members of the conazole family inhibit the C-14 -demethylase enzyme that converts lanosterol to ergosterol, an important component of fungal cell membranes (Kimura et al., 1992). Ketoconazole, however, is a potent inhibitor of hepatic P450 enzymes (Honig et al., 1993; Jurima-Romet et al., 1994), the most frequently reported side effects being related to endocrine physiology because P450 enzymes are involved in the synthesis of adrenal and gonadal steroid hormones (Santen et al., 1983; Loose et al., 1983; Pont et al., 1997). In spite of these side effects, ketoconazole is widely used in long-term treatment of blastomycosis, histoplasmosis and coccidioidomycosis (Simons and Simons, 1997) because of its low cost (Como and Dismukes, 1994).

Severe warnings are given against the concomitant administration of conazoles and compounds that use the same CYP3A4-P450 metabolic pathway. Among them, terfenadine (Seldane), historically one of the most widely prescribed antihistamines, has been linked to prolonged Q-T intervals in the EKG. Q-T prolongation may be associated with polymorphic ventricular arrhythmias such as torsade de pointes and death (Davies et al., 1989). The buildup of terfenadine to high plasma concentrations (~ 100 nM) (Davies et al., 1989) after reduction of its metabolism by conazoles seems to be the cause of its cardiotoxicity. A similar effect is also observed in patients who have taken macrolide antibiotics, such as erythromycin. Other causes of toxic plasma concentrations include excessive intake and impaired liver function.

Recently we reported that terfenadine blocks currents produced by two human cardiac potassium channel clones, HERG and Kv1.5, as expressed in Xenopus oocytes. Kv1.5 appears to be responsible for I_Kur in human atrium, and block by terfenadine has an IC₅₀ of 3 µM (Daneshmend and Warnock, 1983; Rampe et al., 1993; Roy et al., 1996). HERG, the human ether-a-go-go-related channel, is responsible for I_Kr in ventricle (Sanguinetti et al., 1995; Curran et al., 1995) and is blocked by terfenadine with an IC₅₀ of 300 nM (Roy et al., 1996; Suessbrich et al., 1996). HERG appears to be the primary target for terfenadine (Roy et al., 1996) for two reasons. First, HERG is clearly present in human ventricle, and HERG mutations result in Q-T prolongation, torsade de pointes and sudden cardiac death, whereas Kv1.5 protein has been primarily localized in the atrium (Amos et al., 1994). Second, HERG is blocked by nanomolar concentrations attained during terfenadine toxicity, whereas micromolar concentrations are required to block Kv1.5 currents.

The cardiac ion channel gene products that are targets for ketoconazole are unknown. However, previous studies by Woosley and Chen (Woosley and Chen, 1994; Woosley, 1996) showed that the delayed rectifier tail current and I_to in cat ventricle were strongly reduced by ketoconazole, resulting in a 15% prolongation of the action potential plateau. We therefore tested for a possible block of HERG and Kv1.5 heterologously expressed in Xenopus oocytes. We found that both HERG and Kv1.5 are blocked by ketoconazole, with micromolar IC₅₀ values. The blocks by terfenadine and ketoconazole are additive. By using a computer simulation, we found that concomitant administration of the two drugs led to an additive block of I_K, prolongation of the cardiac action potential and, by inference, Q-T prolongation. Ketoconazole potentiates terfenadine cardiotoxicity in two ways: by a direct effect on HERG and Kv1.5 channels and by an indirect metabolic effect on terfenadine pharmacokinetics.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Molecular biology. The HERG clone was a gift from Dr. M. T. Keating. The full-length cDNA was cloned into the pSP64 transcription vector as previously described (Sanguinetti et al., 1995). Amplification of cDNA was obtained by transformation into E. coli and overnight incubation at 37°C. HERG cDNA was linearized by digestion with EcoR1 for runoff transcription with the SP6 mMessage mMachine in vitro transcription kit (Ambion, Austin, TX). Kv1.5 was previously cloned in our laboratory (Fedida et al., 1993) and subcloned into A⁺-pCRII (Taglialatela et al., 1994). The final cRNA product was resuspended in 0.1 M KCl and stored at 80°C. The cRNA was diluted to the desired concentration (5-150 pg/nl) immediately before oocyte injection. Stage V to VI Xenopus oocytes were defolliculated by collagenase treatment (2 mg/ml for 1.5 h) in a Ca⁺-free buffer solution containing (in mM): NaCl 82.5, KCl 2.5, MgCl₂ 1, HEPES 5, gentamicin 100 mg/ml; pH 7.6 (NaOH-HCl). The defolliculated oocytes were injected with 46 nl of cRNA solution (in 0.1 M KCl) and incubated at 19°C in culture medium containing (in mM): NaCl 100, KCl 2, CaCl₂ 1.8, MgCl₂ 1, HEPES 5, pyruvic acid, 2.5, gentamicin 100 mg/ml; pH 7.6. Electrophysiological measurements were made 3 to 7 days after cRNA injection. All experiments were conducted according to institutional animal care committee regulations.

Electrophysiology. Whole-cell currents were recorded from Xenopus oocytes using the conventional two-microelectrode voltage-clamp technique. Beveled microelectrodes were filled with a 0.3% agarose solution containing 3 M KCl, 10 mM HEPES and 10 mM EGTA, pH 7.4 (TRIS), to give tip resistance of 0.2 to 0.5 M. Oocytes were placed in a chamber and perfused with Ringer's solution containing (in mM): NaCl 120, KCl 2.5, CaCl₂ 1.1, EGTA 1.0, HEPES-acid 10; pH 7.2 (NaOH). Stock solutions of ketoconazole (5 mM) and terfenadine (5 mM) were prepared by diluting an appropriate amount of compound in DMSO. The portion of DMSO in the perfusing solutions never exceeded 0.2% (v/v) to avoid artifactual effects. The drugs were applied on the oocytes with a maximum of two concentrations, typically one low and the other high, per experiment. The amplitude of the currents was monitored until no changes in the current amplitude (steady-state level) could be recorded for 3 to 5 min. After the monitoring period, test pulses were applied as described in the text. In control experiments, we found that 0.2% DMSO reduced HERG current by 4 ± 0.03% (n = 5) but did not have any effect on Kv1.5 current. The effects were rapidly reversible after washout. We added an equal amount of DMSO (0.2%) to the control perfusion solution during the experiments with high concentrations of ketoconazole (500 µM) to avoid the contribution of the solvent to the block.

Cardiac action potential modeling. Modeling of the ventricular cardiac action potential was done using GPVentricle, from the Oxsoft Heart program (Oxsoft Ltd. Oxford, England. V 4.4). Maximal contribution of I_Kr to I_K was set to 50%, as shown by Sanguinetti and Jurkiewicz (Sanguinetti and Jurkiewicz, 1990) and Zeng et al. (Zeng et al., 1995). Simulations used concentrations of terfenadine associated with toxicity and therapeutic concentrations of ketoconazole. The extent of block on HERG (I_Kr) was taken from our steady-state dose-response measurements for ketoconazole (this study) and terfenadine block (Roy et al., 1996) and computed as a corresponding reduction of maximal conductance in the model.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Although therapeutic concentrations of terfenadine (nM) have been shown to block HERG and I_Kur in human atrial myocytes, as well as Kv1.5 current expressed in transfected human embryonic kidney cells or Xenopus oocytes (Daneshmend and Warnock, 1983; Rampe et al., 1993; Roy et al., 1996), no direct effects of ketoconazole on Kv1.5 or HERG currents have been reported. We therefore heterologously expressed human Kv1.5 and HERG in Xenopus oocytes to compare the block by ketoconazole with that of terfenadine in the same expression system.

HERG and Kv1.5 currents were reduced by ketoconazole in a dose-dependent manner and reached a steady-state level within 10 min when perfusing the oocytes at a rate of 3 ml/min. The external solution in the chamber was completely exchanged within 2 min as measured from the effects of changes of potassium concentration on holding and peak currents (not shown). We monitored the onset of the block by applying a standard 800-ms pulse to 0 mV from a holding potential of 80 mV every 30 s. After an initial perfusion of 1 min without stimulation, block was clearly observed, which showed that blockade does not require that channels be activated. Measurements were done once the block reached a steady state, usually within 10 min of perfusion with the drug. The block on HERG and Kv1.5 was fully reversible within 20 min for the lower doses and partially reversible (80%) in that time frame with 500 µM.

Ketoconazole reduced HERG current elicited by depolarization from a holding potential of 80 mV in a concentration-dependent manner, 100 µM ketoconazole blocking the HERG steady-state current by 60% (fig. 1A). Ketoconazole similarly reduced HERG maximal tail currents (fig. 1C), with respective 50% inhibition concentrations (IC₅₀) for steady-state current and tail current of 49 ± 13 µM and 31 ± 2 µM, respectively, as fitted by 1:1 binding isotherm (fig. 1C).

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Dose response of Kv1.5 and HERG currents expressed in Xenopus oocytes to ketoconazole. A) Representative family of currents from two-electrode voltage-clamp recordings generated using the pulse protocol shown. The oocyte was pulsed from a holding potential of 80 mV to test potentials from 60 to +40 mV, in 10-mV increments for 800 ms. In the same oocyte, 100 µM ketoconazole reduced steady-state and tail currents by 60% and 70%, respectively, within 7 min. B) Kv1.5 currents elicited by 350-ms depolarizing pulses from 40 to 50 mV (5-mV increments) were blocked by 60% after addition of 100 µM ketoconazole to the perfusion solution. C) HERG currents recorded at 0 mV in the presence of varying concentrations of ketoconazole were normalized to the control amplitude and plotted as a concentration of the drug to yield the dose response. Steady-state currents (filled circle) were measured at the end of the stimulating pulse (800 ms), and peak tail currents (filled down triangle) were measured upon return to the holding potential. Kv1.5 currents (open triangle) were measured at the end of the stimulating pulse and normalized in a similar fashion. The dose-response curves for both HERG steady-state and tail currents yielded IC50 values of 31 ± 2 µM and 49 ± 13 µM, respectively (n = 4) using a 1:1 binding isotherm to fit the data. We obtained 50% reduction of Kv1.5 current with 107 ± 5 µM (n = 5). Data are expressed as mean ± S.E.M.

Kv1.5 currents elicited by the application of 350-ms depolarizing steps in 5-mV increments from a holding potential of 60 mV (fig. 1B) were as previously described (Daneshmend and Warnock, 1983; Rampe et al., 1993; Roy et al., 1996; Fedida et al., 1993; Wang et al., 1993), with activation developing between 20 and 20 mV (Fedida et al., 1993; Wang et al., 1993) and an ohmic increase in current at more depolarized potentials. As seen with HERG, ketoconazole reduced the Kv1.5 steady-state current in a dose-dependent fashion with 55% block at a concentration of 100 µM. We found an IC₅₀ value of 107 ± 5 µM (fig. 1C).

Absence of voltage-dependent block. HERG currents recorded using the two-electrode voltage-clamp technique exhibited voltage-dependent properties as previously reported (Roy et al., 1996; Sanguinetti et al., 1995). Briefly, the current activated at potentials greater than 40 mV, reached a peak at 0 mV and then decreased at more positive potentials (fig. 1A; 2A) giving the steady-state I/V relationship its typical bell-shaped appearance (fig. 2A). The "tail" current, generated after the stimulating pulse has been completed, increased with voltage and then plateaued for test potentials positive to +10 mV as previously reported (Roy et al., 1996; Sanguinetti et al., 1995).

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Ketoconazole block of HERG steady-state and tail current is not voltage-dependent. A) Relative current-voltage relationships for steady-state HERG current in control (n = 8), 20 µM (n = 4) and 100 µM (n = 6) ketoconazole. B) Tail currents are similarly plotted. C and D) Fraction of tail and steady-state control currents blocked at every given potential. The differences between the control and the treated current amplitudes were normalized to the control value. In each case, the current reduction at 40 mV was not significantly different from the block at 20 mV. E) Averaged activation curves, as calculated from the normalized peak tail current amplitudes under control, 2 µM, 20 µM and 100 µM (n = 3) ketoconazole. No significant shift of the mid-activation potential was apparent.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Ketoconazole blocks Kv1.5 as heterologously expressed in Xenopus oocytes. A) Current-voltage relationship of Kv1.5 in the presence of 2, 20 and 100 µM. Currents were normalized to the maximum current recorded during a 350-ms pulse to 50 mV from a holding potential of 60 mV to yield the relative current amplitude. B) The fraction of current blocked was obtained from the difference between the amplitude of the control and the drug test current normalized to the amplitude of the control current for every given potential. The block followed the voltage dependence of the activation and plateaued when the channels were fully activated. C) Control currents during a 350-ms pulse to 35 mV were reduced by 56% after application of 100 µM ketoconazole (*). D) The traces from C were normalized to their respective maximal steady-state values and superimposed. There was no significant change in the activation kinetics of the current after application of ketoconazole.

View larger version (9K): [in this window] [in a new window]   Fig. 4.   Time-dependent block of HERG. Ketoconazole potentiates the current decay in a dose-dependent fashion after extracellular application of 20 µM and 100 µM concentrations of the drug. Currents were recorded during a depolarizing step to 40 mV from a holding potential of 80 mV.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Ketoconazole did not change the fast-inactivation parameters. A) Time course of recovery from inactivation was measured by fitting exponential functions to the tail currents elicited after repolarization from a 1-s conditioning pulse to 40 mV from a holding of 80 mV. A sum of two exponential functions was necessary at potentials between 120 and 50 mV, and a single exponential function was used at more depolarized potentials. Shown are tail currents from test potentials to 120, 110, 100, 90, 80, 70, 50 and 30 mV. After recovery from inactivation, channels deactivated slowly, as shown by the traces at more hyperpolarized test potentials (120 to 60 mV), and the fast component of the fit was taken as the time constant of recovery. B) Onset of inactivation was measured by fitting the current time course during the second (test) pulse of a double-pulse protocol. The first 1-s pulse to 20 mV promoted inactivation, and the 30-ms interpulse to the resting potential enabled the channels to recover quickly without significantly deactivating. Channels that recovered were forced to "reinactivate" during the second pulse. Time constants for the onset were obtained by fitting a single exponential function (thin lines) to the current traces during the test potential. C) Time constants for the recovery (filled symbols) and onset (open symbols) of inactivation are plotted against the membrane potential. Data were obtained from the experiments described in panels A and B from control (squares) and after extracellular application of 50 µM ketoconazole (circles, Keto.). There were no significant differences between the two sets of data when compared using an ANOVA or a double-sided paired Student's t test. Values expressed are mean ± S.E.M. D) Steady-state inactivation was measured using a double-pulse protocol with varying interpulse repolarization levels. Duration of the pulses is as presented in panel B. The membrane potential during the interpulse interval determined the number of channels to recover from the inactivating (first) pulse and, therefore, the steady-state inactivation of the channels. To avoid any contribution from the capacitance of the cell, current traces were fitted in a region where the capacitive artifact was completely resolved, and the fit was extrapolated back to the end of the interpulse (thin lines). The maximal value from the extrapolation was taken as a measure of the availability (inactivation) of the channels. The capacitive artifact (blanked) was fully resolved within 2.5 ms from the beginning of the pulse at all tested potentials. E) Maximal currents obtained from data in panel D in control conditions and after application of 50 µM ketoconazole. A Boltzmann distribution fitted to the data (solid line) returned mid-inactivation potential values of 60 mV and 60.4 mV and asymptotic values of 19.7 µA and 11.4 µA for control and 50 µM ketoconazole conditions, respectively. F) Data from five cells were normalized to the asymptotic value of their respective Boltzmann fit, obtained as in panel E, and then averaged to yield the normalized current. A Boltzmann distribution fitted to the normalized currents gave averaged mid-inactivation potentials of 57 ± 5 mV and 54 ± 5 mV for control and 50 µM ketoconazole conditions, respectively, in the five cells tested. Capacitive artifacts were blanked from all the current traces, and data are expressed as mean ± S.E.M.

View larger version (32K): [in this window] [in a new window]   Fig. 6.   Full activation time course of HERG current as obtained from a tail current envelope protocol. A) Representative current traces obtained after a series of consecutive steps to 40 mV in 50-ms increments, followed by a 500-ms return to 50 mV to maximize the tail current (protocol). Holding potential was 80 mV. The tail current after repolarization at 50 mV gives the maximal amount of channels activated during the test pulse to 40 mV. The rapid recovery from inactivation ensures its minimal contribution to the tail current. The last pulse protocol of the train and its corresponding current trace are highlighted. Corresponding membrane potentials are indicated. B) The peak tail current after the return to 50 mV was plotted as a function of the test pulse duration. A monoexponential fit to the control data yielded the time constant  = 62.9 ms. Application of 50 µM ketoconazole introduced a slow decay in the full activation of the channels and yielded time constants of 63.0 ms for the onset of activation and 480 ms for the time course of the block when data were fitted by a sum of two exponential functions (solid line). There were no significant changes in the onset of activation. C) The peak tail currents were normalized to their respective maximal values. The decaying phase of the normalized tail current plot shows the onset of the extra-block appearing after the opening of the channels, without contamination from the fast inactivation. The same phenomenon was observed in five cells (see text). Ctrl.: control, Keto.: ketoconazole, t: varying pulse duration.

Absence of use-dependent block. Use-dependent block was assessed by applying the monitoring pulse at frequencies of 0.1 to 1 Hz, after a steady-state block was attained. At 1 Hz, a small residual portion of activated channels was carried from one pulse to the other because of the slow deactivation of HERG channels, but the relative increase in active current was the same with or without 100 µM ketoconazole in the bath. Similar results were obtained for Kv1.5 in the same conditions. Kv1.5 current usually had a higher amplitude but did not deactivate as slowly as HERG, so we observed a decrease in the amplitude of the current at high frequencies (1 Hz) possibly as a result of the increased mean time at depolarized potentials or changes of the potassium reversal potential. Thus we did not observe use-dependent block of the two channels by ketoconazole in the range of frequencies studied. Because the cardiotoxicity of the terfenadine-ketoconazole interaction is presently linked to pharmacokinetics effects, namely competitive effects on the CYP3A4-P450 metabolic pathway, we next tested for competitive block by terfenadine and ketoconazole on HERG (I_Kr).

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Additive block by ketoconazole and terfenadine. Current recordings after an 800-ms depolarizing pulse to 20 mV from a holding potential of 80 mV. Ketoconazole in concentrations close to its IC50 (49 µM) blocked 51% of the control current. Cumulative addition of 300 nM terfenadine reduced the residual current by 49% as expected from the IC50 for terfenadine (350 nM). This shows that adding ketoconazole does not shift the IC50 value for terfenadine, a result expected for an additive rather than a competitive blockade by the two drugs. The experiments were repeated in 12 cells, and all yielded similar results.

Simulated prolongation of the cardiac action potential by ketoconazole. Chen and Woosley (1993) reported a 15% prolongation of the cat ventricular AP by ketoconazole, and Hey et al. (1996) reported a 5% prolongation by terfenadine alone in guinea pig. The contribution of I_Kr (HERG) to the ventricular AP is difficult to estimate primarily because of its overlap with I_Ks. Furthermore, the possibility of having other endogenous currents blocked by ketoconazole makes it difficult to determine experimentally the contribution of HERG blockade to the prolongation of the AP. Therefore, we simulated the effects of the block of I_Kr (HERG) by ketoconazole alone and with terfenadine, using the ventricular AP routine of the Heart Oxsoft v. 4.4 program (Oxsoft, Oxford, England) and compared these values with experimental data previously reported. We first reduced the conductance of I_Kr (HERG) by 20% to mimic the block by 100 nM terfenadine (Roy et al., 1996). We set the contribution of I_Kr to I_K to 50%, for the duration of the AP plateau on the basis of previous reports (Sanguinetti and Jurkiewicz, 1990; Zeng et al., 1995). The block of the delayed rectifier current by terfenadine was therefore 10% (fig. 8B), a value in agreement with the inhibition of I_K in cardiac human ventricle as reported by Berul and Morad (1995). On the basis of our results with ketoconazole, we also simulated its effects on the ventricular AP waveform after a 30% block of HERG (15% of I_K, fig. 8B). This value corresponds to a circulating concentration between 10 and 12 µM, similar to the plasma concentration reported by von Moltke (von Moltke et al., 1994) during normal use (~10 µM).

View larger version (18K): [in this window] [in a new window]   Fig. 8.   Ketoconazole and terfenadine prolong the cardiac action potential (AP) in an additive fashion. A) Action potential simulated in control conditions and in the presence of therapeutic concentrations of ketoconazole (10 µM) and terfenadine (100 nM) with the Heart Oxsoft program (see text for details). Prolongation of the action potential plateau was measured at a repolarization potential of 20 mV (arrows). The base line (0 mV) is indicated by a dotted line, and the AP resulting from the additive effect of terfenadine and ketoconazole is indicated by Terf. + Keto. B) AP associated IK. The currents generated by the OxSoft Heart program had their conductances reduced by 10% and 15% to simulate the block of IKr by terfenadine and ketoconazole, respectively. The values for the block of HERG by both compounds were obtained from Roy et al. (1996) and this study. The maximal contribution of IKr to IK at the end of the pulse was set to 50% as previously reported (Sanguinetti and Jurkiewicz, 1990; Zeng et al., 1995).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

We found that HERG and Kv1.5 currents were reduced by ketoconazole with IC₅₀ values of 49 µM and 107 µM, respectively. These results were initially surprising, because there have been no formal reports in the literature of cardiac problems or side effects with ketoconazole treatment in human patients, except during concomitant administration with terfenadine or macrolide antibiotics. In one report analyzing the terfenadine-ketoconazole interactions, the investigators mention that such a study may be desirable, although they speculate that the effects of ketoconazole on EKG would be small (Honig et al., 1993). In a previous paper (Roy et al., 1996), we showed that 100 nM terfenadine, a physiologically relevant concentration (Clusin, 1983; Sorkin and Heel, 1985) is likely to block 20% of I_Kr (HERG). Interestingly, a comparison with the block of HERG by therapeutic concentrations (10 µM) of ketoconazole (von Moltke et al., 1994) shows a similar reduction in the steady-state outward current. Chen and Woosley (1993) showed that ketoconazole blocked the tail current of I_K with an IC₅₀ of 2.5 µM in cat ventricle, a value slightly lower than the one we report here for HERG. The prolongation of the action potential by terfenadine in our simulation was 3.5%, a value well within the indirect clinical observations (EKG) of 1% to 6% reported by Pratt et al. (1996) for normal use in human. The results of Hey et al. (1996), obtained in vivo from guinea pig heart, showed an increase of 5% (15 ms/275 ms) in the Q-T interval corrected for heart rate (Q-Tc), a value in agreement with our simulation. Ketoconazole alone prolonged the simulated AP plateau by 5.1%, slightly more than terfenadine, whereas the cumulative block of I_Kr by both compounds led to a prolongation of only 9.5%. Hey et al. (1996) however, showed an 18% increased of the Q-Tc interval in guinea pig.

Some of the discrepancies between our results with AP simulations and the experimental measurements might be explained by imperfections in the model. Alternatively, the oocyte expression system used in our study may give a higher IC₅₀ value when compared with measurements in native cells or the mammalian expression system. Hydrophobic drugs such as ketoconazole often show a lower affinity for ionic channels heterologously expressed in frog oocytes. The difference in IC₅₀ when compared with mammalian cells is strongly dependent on two factors: the hydrophobicity of the blocker and the lipid composition of the cytosolic membrane; the later influences the solubility of some blockers and their access to the ionic channel. This means that we may have underestimated the affinity of ketoconazole for HERG in native cells in our measurements. Because the IC₅₀ we measured in oocytes is within the range of therapeutic plasma concentrations, a toxic effect of the drug might therefore be seen at a lower concentration in clinical settings. In this context, our results should be interpreted as an upper limit for the toxicity threshold and may in part explain the discrepancies observed with our AP simulation. Species differences may also result in different ratios of I_Kr and I_Ks in the cardiac ventricular wall (Liu and Antzelevitch, 1995). Moreover, isoforms of HERG with different affinities for the drug may also be found in other species. It is known that the RNA message for Kv1.5 (Amos et al., 1994; Tamkun et al., 1991; Brahmajothi et al., 1997) is present in the cardiac ventricles, but so far, no current in native ventricular cells has been specifically attributed to this gene. Given the propensity of voltage-dependent potassium channels to form heteromultimers, there is a possibility that Kv1.5 is part of a channel involved in the repolarization of the heart ventricle. Such a channel may therefore share some of the Kv1.5 sensitivity to ketoconazole and may also help to explain the discrepancies we found.

Kv1.5 has been linked to I_Kur (Fedida et al., 1993; Wang et al., 1993), a component of I_K involved in the repolarization of the atrium. Significant blockade of I_Kur will appear at higher concentrations of ketoconazole when compared with the block in HERG. Disturbance of sinus rhythm is unlikely to be the dominant effect of the drug, and the differences in the IC₅₀ values may help explain the ventricular origin of the arrhythmogenic effects of the two drugs during light overdose.

Our results showed that ketoconazole produces a tonic block on HERG, which accounts for most of the current reduction. Because this block does not require previous activation and is not voltage-dependent, this suggests that most of the tonic block is due to blockade in the closed (resting) states or to a fast open-state block. Ketoconazole also introduces a small time-dependent decrease in HERG current, which accounts for up to 10% of the peak current during an 800-ms pulse at strongly depolarized potentials. We found that this time-dependent block was not due to potentiation of the intrinsic inactivation of the channel but was probably due to a slowly developing block of the open channels. This last result suggests a lower affinity of the drug when the channels are in the open state. Blockade of HERG by ketoconazole is therefore composed of a high-affinity block, possibly in the resting (closed) states responsible for the tonic block, and a lower affinity block in the open states, primary responsible for the small time-dependent block.

In contrast, application of ketoconazole on Kv1.5 did not change the current time course, an indication that there was no change of affinity for the open state. But the block was voltage-dependent, increased with activation at weak depolarizing potentials and saturated during full activation at strongly depolarized potentials. The increase of the block in a range of potentials where the channel rectifies (30 and +20 mV) suggests that a transition rate during the activation process is acting as a limiting factor for the binding of the drug between 30 and +20 mV.

Our observations indicate that the interaction between the two drugs is not limited to the competitive pharmacokinetic effects at the level of the CYP3A4-P450 but may also encompass a pharmacodynamic effect of direct additive block on I_Kr.

Cardiac Kv1.5 and HERG ion channels were blocked by ketoconazole at concentrations reported in serum, and cardiovascular effects would ultimately depend on the amount of ketoconazole present in the heart tissues. Patients treated with ketoconazole may be using the drug over an extended period of time, and they should be monitored during treatment, particularly in cases of compromised liver function or simultaneous use of a Class III antiarrhythmic medication or CYP3A4-P450-requiring compounds.