Abstract
Tamoxifen is an antiestrogen drug commonly used to treat breast cancer and has been shown to cause prolongation of the electrocardiographic QT interval in humans. Because QT prolongation could influence cardiac arrhythmias, we sought to determine the electrophysiologic mechanism(s) underlying the tamoxifen action. The whole-cell patch-clamp technique was used to study the effect of tamoxifen on the delayed rectifier (IKr), the inward rectifier (IK1), the transient outward current (Ito), and the inward L-type calcium current (ICa) in rabbit ventricular myocytes. By switching to the current-clamp mode, the effect of tamoxifen on action potential duration (APD) was also studied. Tamoxifen blocked IKr in a time-, concentration- and voltage-dependent fashion. IKr tail currents were completely blocked by 10 μmol/l tamoxifen with no recovery after 15 min of washout. At +50 mV, tamoxifen 1 and 3.3 μmol/l blocked IKr by 39.5 ± 1.7% (P < .01) and 84.8 ± 1.3% (P < .01) respectively, while no significant block of IK1 or Ito was observed. Significant block of ICa by tamoxifen was also observed at concentrations greater than 1 μmol/l, with almost complete inhibition at 10 μmol/l. Tamoxifen showed no significant effect on APD at concentrations up to 3.3 μmol/l. We conclude that tamoxifen potently blocks both IKr and ICa at clinically relevant concentrations. The observed QT prolongation by tamoxifen in humans may be a result of its predominant effect on IKr. Inhibition of IKr, in conjunction with other QT-prolonging factors in patients could increase their risk of developing torsades de pointes-type cardiac arrhythmias.
Tamoxifen is one of the most effective and frequently prescribed drugs to treat breast cancer (Jordan, 1992). A recent clinical observation indicated that tamoxifen prolongs the QT interval in human subjects (Trumpet al., 1992). Prolongation of cardiac repolarization and the QT interval has clinical importance because under certain situations it may confer a class III antiarrhythmic effect; however, it has also been shown to increase the risk of developing a complex form of potentially lethal ventricular arrhythmia known as TdP (Ben-David and Zipes, 1993; Carlsson et al., 1990; Roden et al., 1986).
Crucial to generation of TdP is prolongation of the QT interval and APD that permits EADs to occur (Zeng and Rudy, 1995). Because potassium currents are major determinants of cardiac repolarization and because shortening of the action potential suppresses EADs in isolated myocytes (Bouchard et al., 1995), modulation of cardiac repolarizing potassium channels may be critical in modulating EADs. Clinically, drugs that prolong the QT interval by blocking cardiac potassium channels, especially the rapid component of the delayed rectifier current, IKr, are often associated with acquired TdP arrhythmias (Carlsson et al., 1990; Roden et al., 1986; Follmer et al., 1990; Woosley, 1996). In the present study, we evaluated the potential cardiac electrophysiological effects of tamoxifen at clinically relevant concentrations (Murphyet al., 1987) using the whole-cell patch-clamp technique in rabbit ventricular myocytes.
Methods
Isolation of ventricular cells.
Rabbit (3–4 months old, weight 3–3.5 kg; HRP Inc., Denver, PA) ventricular myocytes were isolated using a modification of the method previously described (Giles and Imaizumi, 1988). Briefly, rabbit hearts were removed and mounted on the Langendorff perfusion system. The following procedure was used: (1) perfusion with normal Tyrode’s solution for 10 min; (2) perfusion for ∼20 min with a Ca++-free Tyrode’s solution; (3) perfusion for 30 to 35 min with Tyrode’s solution containing 40 U/ml of collagenase II (Worthington Biochemical, Freehold, New Jersey) and 50 μmol/l CaCl2. The ventricles were then minced and gently stirred in Tyrode’s solution containing 100 μmol/l CaCl2. After stirring for 5 to 10 min, a large number of single ventricular cells was obtained. The resulting cell suspension was then filtered through a nylon mesh. Cells were collected by centrifugation at 50×g and then resuspended in Tyrode’s solution containing 250 μmol/l Ca++. The cells were again collected by centrifugation, and then resuspended in Tyrode’s solution containing 1 mmol/l Ca++. After a third centrifugation, the cells were resuspended in DMEM culture medium supplemented with 10% bovine serum (Hyclone Labs, Logon, UT). The myocytes were immediately seeded onto laminin-coated glass microcoverslips at a density of 104rod-shaped cells/cm2 and allowed to attach. The cells were stored in a humidified incubator in 5% CO2, 95% air at 37°C. Only cells within the first 36 hr after isolation were used.
Whole-cell patch-clamp.
The patch-clamp technique was used to record the membrane currents and action potentials in single ventricular myocytes. Command voltage pulses were generated using PCLAMP 6.0.2 Software connected to an interface (Axon Instruments, Foster City, CA), an IBM-compatible Pentium computer, and an Axopatch 200A amplifier. Membrane potentials and current signals were monitored on an oscilloscope (Model 5103, Tektronix, Beaverton, Oregon) and stored in the lab computer. Pipettes with tip resistance of 1 to 4 MΩ were pulled from borosilicate glass (World Precision Instruments, Sarasota, Florida) and filled with an intracellular solution containing (mmol/l) KCl 125, NaCl 10, CaCl2 1, Mg-ATP 5, EGTA 14, HEPES 10, cAMP 0.1, adjusted with KOH to pH 7.2. A holding potential of –40 mV was used to inactivate fast sodium and T-type calcium currents. The external solution was Tyrode’s solution containing (mmol/l) NaCl 137, KCl 5.4, HEPES 10.0, MgCl2 1.0, CaCl2 2.0, glucose 10.0, and was adjusted with NaOH to pH 7.4 (NaOH). Cd++ (0.2 mmol/l) was used to block the L-type calcium channel (ICa) and to shift the I-V relationship of Ito and IKrto more positive potentials (Daleau et al., 1997; Aguset al., 1991). This allows 1) separation of IKr and the outward portion of IK1, especially at membrane potentials positive to –30 mV and 2) marked increase of Itoavailability at a holding potential of –40 mV. For action potential recording, Cd++ was omitted from the extracellular solution and the voltage-clamp mode was switched to current clamp mode.
Myocytes adhering to glass coverslips were placed in a small chamber mounted on the stage of an inverted microscope and superfused at 1.5 ml/min at room temperature (22–24°C). Only rod-shaped single cells that were quiescent and exhibiting well defined, cross-striations were studied. After establishment of whole-cell configuration and measurement of cell capacitance, series resistance (≤10 MΩ) was compensated by 50% to 70%. Junction potentials under these conditions were ∼3 mV and not corrected. IKr currents were elicited from a holding potential of −40 mV by a series of 1.5-sec test pulses from −10 to +50 mV in 10 mV increments. Membrane potential was then held at −30 mV for 2 sec before returning to the holding potential in order to observe IKr tail currents. The current-voltage (I-V) relationship for IKrwas constructed by measuring the tail currents. In the presence of 0.2 mmol/l Cd++, the same protocol for IKr also activated Ito due to the marked shift of Ito activation to more positive potentials (Agus et al., 1991).
The inward-rectifier K+ currents (IK1) were elicited from a holding potential of −40 mV by a series of 250-msec test pulses ranging from −120 to −10 mV in 10-mV increments. The amplitude of IK1 at each voltage was determined by measuring the peak current relative to zero current.
ICa was recorded as previously described (Osaka and Joyner, 1991). A Na+ and K+-free extracellular solution was used, containing (mM): N-methyl-d-glucamine (NMG) chloride 130, CaCl2 1.8, CsCl 20, MgCl20.53, HEPES 5, glucose 5, pH 7.4 with NMG. Pipette solution contains (mmol/l): CsOH 110, aspartic acid 90, CsCl 20, HEPES 10, EGTA 10, MgATP 5, and GTP(Tris) 0.4, pH 7.2 with CsOH. ICa was elicited by series of depolarization steps of 200 ms duration applied in 10-mV increments from a holding potential of −40 mV.
Drugs and chemicals.
Tamoxifen was purchased from Sigma Chemical (St. Louis, MO), dissolved in ethanol and stored in aliquots at −20°C until used. E-4031 and dofetilide were kindly provided by Eisai Ltd. (Ibaraki, Japan) and Pfizer Central Research (Groten, CN) respectively. All other chemicals were obtained from Sigma Chemical.
Data analysis and statistics.
Patch-clamp data were normalized for total cell capacitance to allow comparison between cells of various sizes. The paired Student’s t test was used to compare the potassium current amplitude before and after tamoxifen treatment. Data are reported as mean ± S.D., and differences between values were considered statistically significant when P < .05.
Results
Effect of E-4031 on IKr, Ito and IK1.
IKr is one of the major repolarizing currents, and its block has been implicated in TdP (Carlsson et al., 1990; Roden et al., 1986; Follmer and Colatsky, 1990;Woosley, 1996). To evaluate whether tamoxifen affects IKr, we first sought to establish the presence of IKr in our rabbit ventricular myocytes. Figures1, A and B, show the membrane currents elicited by a 1.5-sec voltage-clamp step from –40 mV to different test potentials ranging from –10 to +50 mV in the same cell before (A) and after (B) 5-min exposure to 5 μmol/l E-4031, a highly selective IKr blocker (Clay et al., 1995; Sanguinetti et al., 1990). Under control conditions, a slowly activating outward current flowed during depolarization, followed by an outward tail current that has been shown to represent the gradual decay of IK (Follmeret al., 1990; Clay et al., 1995; Sanguinettiet al., 1990). The initial peak in the time-dependent outward current was due to the rapid activation and inactivation of Ito, which is sensitive to 4-aminopyridine (data not shown). E-4031 abolished the tail current on repolarization and also reduced the time-dependent outward current, without affecting the initial peak (Ito) or the holding current (IK1). Shown in B are the E-4031-sensitive currents obtained by digital subtraction of currents in the bottom tracings from currents in the top tracings in A. Compared with the tail current, the time-dependent current demonstrated marked inward rectification at very positive potentials. Itowas not present in the E-4031-sensitive currents, indicating E-4031 has no effect on Ito at this concentration. Superfusion with 1 to 2.5 μmol/l dofetilide or removal of extracellular K+ also abolished the tail current (Liu et al., 1998). These features of the delayed rectifier current (inward rectification of the time-dependent current, complete block of the tail current by E-4031, dofetilide and removal of extracellular K+) are consistent with the previous description of IKr in rabbit and other species (Clay et al., 1995; Sanguinetti et al., 1990).
In the same cell shown in figure 1, A and B, we also measured IK1 current before and after E-4031 exposure. Figure 1C demonstrates the IK1 current elicited by a 250-msec hyperpolarization to −120 mV from a holding potential of −40 mV before and after E-4031 superfusion. Little effect was observed on the IK1 inward current, although IKr was completely blocked in the same cell. The outward holding currents that represent the amplitude of IK1 at −40 mV before and after E-4031 superfusion were superimposible, indicating that E-4031 had no effect on the IK1 outward current.
Effect of tamoxifen on IKr, Ito and IK1.
We next tested the effect of tamoxifen on the three major potassium currents using the same protocol as shown in figure 1. Shown in figure2 are the IKr, Ito and IK1 currents elicited in the same cell before and after 5-min exposure to 10 μmol/l tamoxifen. Similar to E-4031, tamoxifen abolished the IKr tail current and also reduced the time-dependent current without affecting Ito or the holding current (fig. 2A). The solvent for tamoxifen, ethanol, had no effect on IKr at the concentration (≤1%, v/v) used to dissolve tamoxifen (data not shown). Figure 2B depicts the tamoxifen sensitive currents obtained by digital subtraction of currents in the bottom tracings from currents in the top tracings in A. There is a striking similarity between the tamoxifen sensitive current and the E-4031-sensitive current shown in figure 1B. In both figures 1B and 2B, the time-dependent current demonstrated strong inward rectification at very positive potentials compared with the tail current, whereas Ito was not present in either the tamoxifen or the E-4031-sensitive current. Figure 2C shows the IK1 current measured before and after 5-min exposure to 10 μmol/l tamoxifen in the same cell shown in figure 2, A and B. Like E-4031, tamoxifen produced no inhibition of the IK1 inward current at −120 mV. In fact, IK1 inward current was even slightly larger after tamoxifen treatment in this cell (fig. 2C). The outward holding currents representing the amplitude of IK1 at −40 mV were superimposable, indicating that tamoxifen had no effect on the IK1 outward current.
Time-dependent block of IKr by tamoxifen.
To further characterize the observed inhibition of IKr by tamoxifen, we examined whether this inhibition is time dependent. Figure 3depicts a typical experiment performed in the same cell before drug administration (control), or after 3-, 5- and 9-min superfusion with 1 μmol/l tamoxifen. As shown in figure 3, block of IKr by tamoxifen is time dependent and has a slow onset. Further block can still be observed after superfusion for 5 min. In contrast, IKr was readily recorded from control myocytes (no exposure to tamoxifen) for at least 10 min without any sign of run-down. In the absence of drug, the amplitude of IKr measured 10 min after membrance rupture was 103.4 ± 6.15% of that measured immediately after membrance rupture (n = 3, P > .05).
Effect of tamoxifen on the I-V relationship of IKr.
We next evaluated the effect of tamoxifen on the I-V relationship of IKr. Figure4, A and B, demonstrates the effects of 1 and 3.3 μmol/l on the I-V relationship of IKr, measured after 5- to 7-min infusion of tamoxifen. Tamoxifen markedly reduced IKr current amplitude in a concentration-dependent fashion. A typical voltage-dependent block of IKr is shown in figure 4. Note the greater block at more positive potentials. At the test potential of +50 mV, tamoxifen (1 and 3.3 μmol/l) blocked IKr by 39.5 ± 1.7% (P < .01) and 84.8 ± 1.3%, respectively (P < .01), whereas no significant block of IK1 was observed at 3.3 μmol/l (5.5 ± 0.9% test potential = –120 mV, n = 4, P > .05).
Comparison of IKr block by tamoxifen and quinidine.
Quinidine is a drug that has been frequently associated with TdP (Roden et al., 1986). To compare the block of IKr by tamoxifen to that produced by quinidine, we performed the protocol shown in figure5, using either 10 μmol/l tamoxifen or 10 μmol/l quinidine. Tamoxifen completely blocked the IKr tail current with no recovery observed after 5-min washout, whereas the same concentration of quinidine (10 μmol/l) only partially blocked the IKr tail currents, with recovery within 3-min washout. In other experiments, complete recovery of IKr from quinidine block was usually observed after ∼5-min washout, whereas no recovery from tamoxifen could be detected even after 15-min washout (data not shown). Figure 5C compares the percentage inhibition of IKr by tamoxifen and quinidine at the same concentration of 3.3 μmol/l. These data show that, at the test potential of +50 mV, tamoxifen produced significantly greater inhibition of IKr compared with quinidine (84.8 ± 1.3% vs. 42.5 ± 9.1%, P < .01). Thus, tamoxifen was a more potent and longer lasting blocker of IKr than quinidine.
Effect of tamoxifen on APD and ICa.
We next examined whether tamoxifen causes prolongation of APD. Figure6 shows action potentials recorded before and after 4-min exposure to 3.3 μmol/l tamoxifen. Surprisingly, although tamoxifen inhibited IKr by ∼84.8% at this concentration (fig. 4), no significant prolongation of APD was observed. APD measured at 90% repolarization (APD90) before and after 4- to 5 min superfusion of tamoxifen (3.3 μmol/l) was 341 ± 49 ms and 332 ± 19 ms respectively (n = 16, P > .05). Because under control conditions, no significant shortening of APD was observed in the initial 10 min after cell membrance rupture, the absence of APD prolongation by tamoxifen was not secondary to a “rundown” phenomenon.
This unexpected effect on APD and previous reports of ICa blockade by tamoxifen (Song et al., 1996) prompted us to study the possible effect of tamoxifen on the cardiac inward ICa. Consistent with earlier studies by Song et al. (1996) in vascular smooth muscle cells, we also observed a potent effect of tamoxifen in cardiac rabbit ventricular myocytes. Significant inhibition of ICa was observed at tamoxifen concentrations greater than 1 μmol/l, with almost complete inhibition observed at 10 μmol/l (fig. 7).
Discussion
The major finding of the present study is that the commonly prescribed antiestrogen drug tamoxifen potently blocks the rapid component of the delayed rectifier current, IKr, in a voltage-, concentration- and time-dependent fashion. The potassium channel-blocking effect of tamoxifen seems to be specific for IKr because no significant effect was observed on IK1 or Ito at concentrations up to 10 μmol/l (fig. 2). Tamoxifen blocks IKr with a potency even greater than quinidine, a drug that has been shown to block IK and is associated with a high incidence of drug-induced TdP (Roden et al., 1986). This is, to our knowledge, the first study showing that tamoxifen is a potent and selective IKrblocker.
Outward potassium channels are the major determinants of the repolarization phase of the cardiac action potential. In the cardiac ventricular myocytes, the transient outward current Ito, the delayed rectifier IK and the inward rectifier IK1 mediate different phases of the repolarization process (Giles and Imaizumi, 1988; Sanguinetti et al., 1990; Surawicz, 1992). The contribution of IK and IK1 to the APD is well established and reduction of IK or IK1 will cause APD and QT prolongation (Giles and Imaizumi, 1988; Sanguinetti and Jurkiewicz, 1990; Surawicz, 1992). On the other hand, although Ito has been shown to play an important role in determining phase 1 repolarization, there is still some uncertainty regarding its contribution to normal APD (Giles and Imaizumi, 1988; Litovsky and Antzelevitch, 1988; Kaab et al., 1996). Interestingly, to our knowledge, all the drugs that are clinically associated with QT prolongation and acquired TdP invariably block IKr. The reason why IKr is sensitive to block by so many drugs is still not well understood.
IK has been reported to consist of two components, IKr and IKs, in guinea pig, dog and human ventricles (Sanguinetti and Jurkiewicz, 1990;Gintant, 1996; Li et al., 1996). In the rabbit ventricular myocytes, IK has been reported to be absent (Giles and Imaizumi, 1988), consist of only one component (Clayet al., 1995) or consist of two components (Salata et al., 1996). In our experiments, IK can be consistently recorded with a clear tail in all of the untreated cells we studied. Our previous studies (Liu et al., 1998) have indicated that the major delayed rectifier current that contributes to APD in normal rabbit ventricular myocytes is IKr.
Tamoxifen has been shown to prolong the QT interval in human subjects (Trump et al., 1992). The present study suggests that the tamoxifen-induced QT prolongation may be related to a direct block of cardiac IKr current. The tamoxifen plasma concentration was greater than 5 μmol/l when QT prolongation was observed clinically by Trump et al. (Trump et al., 1992). As shown in figure 4, tamoxifen was even more potent than quinidine in blocking IKr, causing more than 80% block of IKr at 3.3 μmol/l. Although no IKr recovery from tamoxifen block could be observed even after an extended washout period, the possibility that the block we observed was due to “rundown” of IKr can be excluded because block of IKr by quinidine can completely recover after ≥5-min washout.
Interestingly, a recent clinical study using F-18 fluoro-tamoxifen showed significant cardiac uptake of tamoxifen, presumably due to intracellular accumulation of tamoxifen in cardiac myocytes (Inoueet al., 1997). In the present study, IKr block by tamoxifen had a slow onset and no recovery was observed after an extended washout period. These results may indicate that tamoxifen blocks IKr through an intracellular site. However, further studies are needed to elucidate the exact mechanism of the potent IKr block by tamoxifen.
One unexpected finding of this study is that, although we demonstrated a potent inhibition of IKr by tamoxifen, it had no significant effect on APD in rabbit ventricular myocytes. This may be due to the fact that tamoxifen is also a potent calcium channel blocker, as documented by Song et al. (1996) and confirmed by our current study (fig. 7). Because blocking of ICa will lead to a shortening of the APD, this effect may largely counteract the IKr blocking effect of tamoxifen which would otherwise lead to a prolongation of APD in single rabbit cardiomyocyte and prolongation of QT interval in whole heart. The obvious discrepancy between the current study in rabbit (no effect on APD) and the observations in humans (QT prolongation) may result from different relative contributions of IKr or ICa to the APD and/or different relative potencies of tamoxifen in blocking IKrvs. ICa in difference species. The net effect of tamoxifen on the APD in a certain species would therefore depend on both the relative contribution of the IKrvs. ICa to the APD and the relative potency of tamoxifen in blocking IKrvs. ICa. It is also possible that other ionic currents not examined in this study such as chloride current (Vanderberg et al., 1994), may mediate the differential effects of tamoxifen in different species. Further studies using human venticular myocytes are needed to resolve these issues.
At the present time, there are no clinical data available regarding the risk of developing TdP in patients after tamoxifen administration. Nevertheless, the current finding has important clinical implications. Therapeutic plasma concentrations of tamoxifen (Murphy et al., 1987) in patients are similar to concentrations that produce potent block of IKr in isolated rabbit ventricular myocytes (i.e., low micromolar range, see fig. 4). In addition, tamoxifen is frequently used to treat breast cancer in female patients, whereas recent clinical observations and experimental data have indicated that female gender is associated with a higher risk of developing drug-induced TdP (Makkar et al., 1993; Lehmann et al., 1996; Liu et al., 1997). Although the effect of tamoxifen on ICa may potentially ameliorate its IKr blocking effect, caution should still be taken when administering tamoxifen to patientsin situations where other risk factors for TdP also exist, such as hypokalemia, bradycardia, congenital long QT syndrome or coadministration of other drugs that may also delay cardiac repolarization.
Footnotes
-
Send reprint requests to: Raymond L. Woosley, M.D., Ph.D., Department of Pharmacology, Georgetown University Medical Center, 3900 Reservoir Road, NW, Washington D.C. 20007. E-mail:woosleyr{at}gunet.georgetown.edu
- Abbreviations:
- IKr
- delayed rectifier
- IK1
- inward rectifier
- Ito
- transient outward current
- ICa
- inward L-type calcium current
- TdP
- torsades de pointes
- APD
- action potential duration
- EAD
- early afterdepolarization
- Received March 12, 1998.
- Accepted July 16, 1998.
- The American Society for Pharmacology and Experimental Therapeutics