Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Cocaine abuse is associated with a variety of cardiovascular complications, including cardiac arrhythmias and sudden death (Isner et al., 1986; Kloner et al., 1992). Recent clinical studies indicate that cocaine abuse is associated with cardiac repolarization abnormalities in the electrocardiogram that include a prolonged QT and QTc interval, as well as enhanced QTc dispersion and appearance of abnormal U waves (Perera et al., 1997; Gamouras et al., 2000). This observation is consistent with previous tissue bath studies that documented that micromolar concentrations of cocaine (IC₅₀ = 4 µM) block the rapid component of the delayed rectifier current (I_Kr), causing a prolongation of the action potential duration (Kimura et al., 1992; Clarkson et al., 1996) as well as the induction of early afterdepolarizations in ventricular myocytes (Kimura et al., 1992). A recent study has also documented that cocaine blocks the cloned HERG channel with a similar IC₅₀ (5.6 µM) (O'Leary, 2001). Although a direct effect of cocaine to block I_Kr could account for cocaine's ability to prolong the QT interval, it is difficult to rule out the possibility that cocaine metabolites could also contribute to such effects. For example, cocaine's half-life is approximately 48 min (Chow et al., 1985), yet in a study on patients admitted to the hospital following the onset of symptoms related to cocaine use, the maximal QT prolongation did not occur for up to 24 h after admission (Gamouras et al., 2000). In addition, it has been reported that approximately 65% of deaths related to cocaine overdose occur within 5 h after cocaine administration, with approximately 30% occurring between 2 and 5 h (Finkle and McCloskey, 1977). The aim of the current study was to define the effects of cocaine and its major metabolites on HERG-encoded potassium channels to clarify which, if any, of its major metabolites could contribute to QT prolongation by a mechanism similar to that documented for cocaine (Clarkson et al., 1996). Cocaine is metabolized primarily in the liver and plasma, resulting in the formation of two primary metabolites, benzoylecgonine and ecgonine methylester (Jeffcoat et al., 1989). Both of these metabolites have been found to be inactive as blockers of the cardiac sodium channel when studied at a drug concentration exceeding a maximal blocking concentration for cocaine (e.g., 100 µM) (Crumb and Clarkson, 1992). However, they are not devoid of biological activity. Benzoylecgonine has been shown to be a potent vasoconstrictor of cerebral arteries, causing concentration-dependent effects at levels >10¹⁰ M (Kurth et al., 1993), and ecgonine methylester has been reported to produce both mild vasodilation at 10⁴ M, or vasoconstriction (at 10⁸-10⁴ M) in different arterial preparations (Kurth et al., 1993; Schreiber et al., 1994). There is also evidence that one or both of these metabolites may be responsible for mediating the coronary vasoconstrictor effects caused by intranasal cocaine administration in humans (Brogan et al., 1992). Because both of these metabolites have been shown to have biological activity, direct experimental confirmation is necessary to determine their activity on human potassium channels, i.e., HERG. Methylecgonidine (ecgonidine methylester) is an additional cocaine derivative that is the major pyrolysis product formed during the smoking of crack cocaine (Cone et al., 1994). Methylecgonidine has been shown to block the acetylcholine-activated potassium current I_KACh with an IC₅₀ (12 µM) lower than that for cocaine (25 µM) (Xiao and Morgan, 1998). In addition, methylecgonidine has been reported to be both an agonist for muscarinic m₂ receptors at concentrations of 10⁸ to 10⁴ M, and to produce irreversible negative inotropic effects by an additional poorly defined mechanism at micromolar concentrations (Huang et al., 1997; Woolf et al., 1997). Finally, cocaethylene is a pharmacologically active metabolite of cocaine that is formed by transesterification of cocaine in the presence of ethanol (Bailey, 1993). Cocaethylene has a longer half-life (3.5-5.5 h) (Bailey, 1993) compared with cocaine (48 min) (Chow et al., 1985); can achieve plasma concentrations equal to, or greater than that of cocaine (Bailey, 1993); and has been found to block the cardiac sodium channel with a significantly higher potency than cocaine (Xu et al., 1994). In this study we defined the effect of cocaine on HERG-encoded potassium channels, and compared its effects with those of its major metabolites and by-products, including benzoylecgonine, ecgonine methylester, methylecgonidine, and cocaethylene.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Transfection and Cell Culture. Human embryonic kidney (HEK) 293 cells that were stably transfected with HERG were obtained from Dr. Craig T. January (University of Wisconsin, Madison, WI). Cells were passaged and maintained in minimum essential medium with Earle's salts supplemented with nonessential amino acids, sodium pyruvate, penicillin, streptomycin, and fetal bovine serum (Zhou et al., 1998). Single cells were isolated for electrophysiological study by a 1-min trypsinization, washed twice with minimum essential medium, and stored in the same media at room temperature until used (within 8 h).

Drugs and Solutions. Cocaine, benzoylecgonine, and ecgonine methylester were obtained from the National Institute on Drug Abuse. Methylecgonidine (ecgonidine methylester) was purchased from Sigma/RBI (Natick, MA). Drugs were dissolved in deionized H₂O to make 30 mM stock solutions that were aliquoted and stored at 20°C. Stock solutions were thawed once and the unused portion discarded. Dilutions of stock solutions were made immediately before the experiment to create the desired concentrations. The external solution (solution bathing the cells) used for recording of HERG potassium currents had an ionic composition of 137 mM NaCl, 4 mM KCl, 1.8 mM CaCl₂, 1.2 mM MgCl₂, 11 mM dextrose, 10 mM HEPES, adjusted to a pH of 7.4 with NaOH. The internal (pipette) solution had an ionic composition of 130 mM KCl, 1 mM MgCl₂, 10 mM NaATP, 5 mM EGTA, 5 mM HEPES, adjusted to a pH of 7.2 with KOH. In several experiments the effect of methylecgonidine was defined on the human atrial sodium current. In these experiments, the external solution had an ionic composition of 135 mM tetramethylammonium chloride, 5 mM NaCl, 5 mM CsCl, 1.8 mM CaCl₂, 1.2 mM MgCl₂, 20 mM HEPES, 11 mM glucose, adjusted to a pH of 7.3 with tetramethylammonium hydroxide. Glass pipettes (tip resistance <1 M) were filled with an internal solution of 125 mM CsF, 20 mM CsCl, 10 mM NaF, 5 mM EGTA, 10 mM HEPES, adjusted to a pH of 7.2 with CsOH. The protocol for isolating human atrial myocytes was identical to that previously published (Crumb et al., 1995). Experiments were performed at 22 ± 1°C.

Data Acquisition and Analysis. Currents were measured using the whole-cell variant of the patch-clamp method (Hamill et al., 1981). Pipette tip resistance was approximately 1.0 to 2.0 M when filled with potassium internal solutions. Analog capacity compensation and 40 to 60% series resistance compensation were used to yield voltage drops across uncompensated series resistance of less than 3 mV. An Axopatch 1B amplifier (Axon Instruments, Foster City, CA) was used for whole-cell voltage clamping. Creation of voltage-clamp pulses and data acquisition were controlled by a PC running pClamp software (version 8; Axon Instruments). After rupture of the cell membrane (entering whole-cell mode), current amplitude and kinetics were allowed to stabilize (3-7 min) before experiments were begun. HERG potassium currents recorded from HEK cells stably expressing HERG were elicited by 5-s depolarizing voltage steps from a holding potential of 80 mV. HERG tail currents were measured upon repolarization to 60 mV. Drugs were applied locally to cells in the bath through a macropipette perfusion system that allowed for rapid exchange of drug concentrations surrounding a cell. Up to three drug concentrations were studied in each cell. In the first series of experiments a high concentration of E-4031 was applied at the end of the experiment to confirm that adequate perfusion of the cell under study was obtained. Drug effects on tail current amplitude were measured after a steady-state level of block had been achieved, and were defined by the change in peak versus steady-state current amplitude. The cycle length for most protocols was 20 s. Drug effects were recorded after a steady-state effect had been reached in the presence of drug, and expressed as a reduction in amplitude relative to the current amplitude recorded before drug was introduced (control). Up to three drug concentrations (from low to high in sequence) were studied for each cell. Unless stated differently, each cell served as its own control.

Statistics and Curve Fitting. Nonlinear regression analysis was used to fit data to a sigmoidal concentration-response relationship Y = 1/(1 + 10^{((LogIC₅₀  X)Slope)}), where slope was the Hill slope parameter, IC₅₀ is the concentration producing 50% blockade, and X is the drug concentration. Student's t test was used to compare fits of the data with this equation, assuming a fixed or variable slope parameter. Student's t test was also used to compare the significance of IC₅₀ values for different drugs (GraphPad Prism, version 3; GraphPad Software, San Diego, CA). Tail currents were fit by nonlinear regression (Clampfit; Axon Instruments). Paired data were compared for statistical difference by paired Student's t test, and differences between multiple groups of data were compared by analysis of variance (ANOVA), followed by Tukey's multiple comparison test. A P value of <0.05 was accepted as statistically significant. Data are presented as mean ± S.E.M.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

The structures for cocaine and its major metabolites are shown in Fig. 1. The effects of these drugs on HERG-encoded potassium channels were investigated by voltage clamping HEK293 cells stably transfected with HERG. Figure 2 shows a family of HERG-encoded potassium currents recorded upon application of 5-s step depolarizations to different voltages under control conditions, and after a steady-state effect had been achieved in the presence of 3 µM cocaine. It is apparent that both the activating and deactivating currents are suppressed by cocaine. As illustrated in Fig. 3, cocaine not only reduced the amplitude of HERG potassium current, but altered the kinetics of the currents as well. Qualitatively similar effects were also observed when cells were exposed to an appropriate concentration of the cocaine metabolites cocaethylene or methylecgonidine (Fig. 3, C and D). The slowing of the tail current kinetics was most marked for cocaine and cocaethylene, which produced a clear "crossover" of the drug affected tail current compared with the control tail current when traces were superimposed (Fig. 3, B and C). Two additional metabolites benzoylecgonine and ecgonine methylester were also investigated for blocking activity. Ecgonine methylester produced a small but significant level of block at both 20 µM (12 ± 3%, n = 7) and 1 mM (21 ± 4%, n = 6) concentrations (P < 0.05). In contrast, benzoylecgonine produced a nonsignificant level of block at both 20 µM (6 ± 2%, n = 7) and 1 mM (15 ± 8%, n = 6) concentrations. Because the level of block produced by these drugs was small at clinically relevant concentrations (<20 µM), they were not further studied. The pattern of HERG channel-blocking potencies for these drugs is qualitatively identical to that reported for sodium channel blockade where cocaine and cocaethylene are relatively potent sodium channel blockers at micromolar concentrations, whereas both at 100 µM benzoylecgonine and ecgonine methylester produce little effect (Crumb et al., 1992; Xu et al., 1994). A similar comparison of channel-blocking potencies for methylecgonidine could not initially be made because its effect on the cardiac sodium current has not been previously reported. Therefore, we also determined the effect of 100 µM methylecgonidine on the sodium current in three human atrial myocytes. Similar to its low potency in blocking HERG channels, we found that 100 µM methylecgonidine produced very little tonic blockade (2.4 ± 1.3% reduction in peak sodium current during a pulse to 20 mV after a 1-min rest at 140 mV), and very little use-dependent reduction (1.1 ± 0.6% reduction during a train of 10 pulses of 50-ms duration at 5 Hz (20°C, V_h = 140 mV, V_test = 20 mV).

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Chemical structures of cocaine, its major metabolites, and pyrolysis product (methylecgonidine). The conversion of cocaine to cocaethylene requires the presence of ethanol.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Cocaine block of HERG potassium currents. A, pulse protocol. A 5-s step depolarization from a holding potential of 80 mV was applied, followed by a 12-s repolarization step to 60 mV. The cycle length between each repetition of the protocol was 20 s. B, control currents. C, currents recorded after ~6 min in the presence of 3 µM cocaine. Exposure to cocaine reduced the amplitude of currents during the step depolarizations, as well as the tail currents.

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Drug effects on current kinetics. A, pulse protocol. A 5-s step depolarization to +60 mV was applied from a holding potential of 80 mV, followed by a repolarization to 60 mV for 12 s. B to D, superimposed currents recorded during control conditions and after a steady-state drug effect was achieved (typically after 6-8 min). A drug concentration close to an IC50 was used for each drug. Only a portion of the tail current is shown. B, effect of 3 µM cocaine. C, effect of 1 µM cocaethylene. D, effect of 200 µM methylecgonidine.

In the absence of drug, the time course of HERG tail currents has been shown to be well fit by a double exponential function (Zhou et al., 1998). To further define the effects of cocaine, cocaethylene, and methylecgonidine on HERG tail current kinetics, we fit the time course of the tail current with a double exponential function before and after exposure to drug (Table 1). Exposure to concentrations of cocaine or cocaethylene that blocked HERG tail current by approximately one-half produced a significant increase in both of the two time constants of decay (Table 1). Methylecgonidine's effect on the tail current kinetics was smaller, more variable, and did not achieve statistical significance. Because drug effects were always determined following a 6- to 18-min perfusion time after control measurements, we also tested the hypothesis that the slowing of tail current kinetics could have been due to an effect of perfusion time (e.g., due to time-dependent changes in gating kinetics). To test this hypothesis, we defined the effect of perfusion time on tail kinetics by using control solutions and found no significant change (Table 2). Thus, the effects of drug on tail current decay can be attributed to the presence of the drug, and not to a time effect.

View this table: [in this window] [in a new window]   TABLE 1 Drug effects on HERG deactivation kinetics at 60 mV Tail currents recorded at 60 mV following a 5-s pulse to +60 mV in the presence or absence of drug.

View this table: [in this window] [in a new window]   TABLE 2 Time effects on HERG deactivation kinetics at 60 mV Tail currents recorded at 60 mV after a 5-s pulse to +60 mV at different perfusion times in the absence of drug (n = 7). Differences were not significant (repeated measures ANOVA, P > 0.6).

To define the concentration dependence with which cocaine and its metabolites block HERG-encoded potassium currents, we defined the effects of these drugs on the tail current amplitude evoked upon hyperpolarization to 60 mV after a 5-s pulse to +60 mV. As shown in Fig. 4, cocaine blocked the tail current dose dependently with an IC₅₀ of 4.4 ± 1.1 µM. In contrast to cocaine, cocaethylene blocked the HERG potassium current with a significantly lower IC₅₀ of 1.2 ± 1.1 µM (P < 0.05), and methylecgonidine blocked the current with a significantly higher IC₅₀ of 171.7 ± 1.2 µM (P < 0.001) (ANOVA and Tukey's multiple comparison test). The concentration-dependent effect of cocaine could be well approximated by a binding equation having a slope factor of 1, whereas for both cocaethylene and methylecgonidine, the data were best fit with a slope of less than unity (0.62 for cocaethylene and 0.58 for methylecgonidine) (F-test, P < 0.0001).

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Concentration dependence of HERG potassium current blockade. Drug effects were defined by measuring the reduction of tail current evoked upon repolarization to 60 mV after a 5-s depolarization step to +60 mV. The fraction of current blocked is plotted against the drug concentration (µM). Nonlinear regression analysis was used to fit the data to the sigmoidal concentration-response relationship Y = 1/(1 + 10((LogIC50  X)Slope)), where slope is the slope parameter, IC50 is the concentration producing 50% blockade, and X is the drug concentration. Student's t test was used to compare fits of the data assuming a fixed or variable slope parameter, and for comparison of IC50 values.

Previous studies have shown that some blockers of HERG potassium channels such as amytriptyline (Jo et al., 2000) block HERG with a clear voltage dependence, whereas other channel blockers, such as ketoconazole, block without a clear voltage dependence (Dumaine et al., 1998). To determine the voltage dependence of HERG block for cocaine, cocaethylene, and methylecgonidine, we defined their effects on tail current evoked upon repolarization to 60 mV after 5-s pulses to voltages between 70 and +60 mV. As illustrated in Fig. 5, under control conditions HERG activation could be well described by a Boltzman relationship. The midpoint (V_mid) values for the activation curves were consistently more negative in the presence of drug compared with cocaine. However, these effects could in theory also result from either time-dependent shifts in midpoint values for channel gating (Xu et al., 1994), or from drug effects. To distinguish between these possibilities, we conducted a separate series of experiments where we defined the voltage dependence of activation in the absence of drug at definite time intervals (that matched those used for serial drug applications) of 6, 12, and 18 min after first gaining access to the cell interior (i.e., since the onset of establishing "whole-cell" recording conditions). In these experiments we confirmed that there was a small, slow hyperpolarizing shift in the midpoint of this HERG activation as a function of time, with the mean shifts being 2.3 ± 0.7 mV after 6 min, 3.9 ± 0.9 mV after 12 min, and 5.4± 1.0 mV after 18 min (n = 7). The amplitude of these shifts was statistically significant at each time point (P < 0.05) (repeated measures ANOVA and Tukey's multiple comparison test). In time-matched experiments, we then compared the shift in V_mid and slope values produced by time alone with those determined in the presence of cocaine, cocaethylene, and methylecgonidine at a drug concentration close to an IC₅₀. Exposure to either 3 µM cocaine, 1 µM cocaethylene, or 200 µM methylecgonidine did not produce a significantly different shift in V_mid compared with time alone (Table 3). The differences in slope parameters were not significant, with the exception that a smaller shift was observed for methylecgonidine compared with control (Table 3). The reason for this difference is unclear and probably of little importance.

View larger version (29K): [in this window] [in a new window]   Fig. 5.   Voltage dependence of HERG activation. A, pulse protocol. A 5-s step depolarization to +60 mV was applied from a holding potential of 80 mV, followed by a repolarization to 60 mV for 12 s. The effects of serial applications of three different drug concentrations of cocaine, cocaethylene, and methylecgonidine on the mean tail current amplitude (normalized to cell capacitance) after a prepulse to variable voltages are shown in B, C, and D.

When the individual tail currents were normalized to their control amplitudes (Fig. 6), it became apparent that the percentage of block by cocaine and its metabolites was significantly smaller at voltages where HERG is less than maximally activated. This is consistent with the hypothesis that these drugs behave like state-dependent channel blockers, and they produce a progressively larger blockade as the fraction of activated or open channels is increased, until maximal level of HERG activation is achieved at voltages positive to 0 mV.

View larger version (27K): [in this window] [in a new window]   Fig. 6.   Voltage dependence of channel block. A, pulse protocol (same as in Fig. 2). B to D, fraction of current block at each voltage obtained by dividing the tail current in drug by the control tail current amplitude. Asterisks indicate significant differences compared with value at +60 mV (repeated measures ANOVA and Tukey's multiple comparison test). #P < 0.05, *P < 0.01, **P < 0.001.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Comparsion with Previous Studies. The results of this study document that both cocaine and cocaethylene are relatively potent blockers of HERG-encoded potassium channels, with cocaethylene having the lowest IC₅₀ of 1.1 µM compared with cocaine's IC₅₀ of 4.4 µM (Fig. 4). This IC₅₀ value is in close agreement with that previously determined for both cocaine blockade of the rapid component of the delayed rectifier current (I_Kr) in guinea pig ventricular myocytes (Clarkson et al., 1996) as well as cocaine block of HERG channels expressed in tsA201 cells (IC₅₀ = 5.4 µM) (O'Leary, 2001). In contrast to cocaine, methylecgonidine exhibits a relatively low potency (IC₅₀ = 171 µM), and both benzoylecgonine and ecgonine methylester are essentially inactive at micromolar concentrations. A similar pattern of channel blocking potency has also been observed for the effects of cocaine and its metabolites on the cardiac sodium current, with cocaethylene being more potent than cocaine (Xu et al., 1994), and both ecgonine methylester and benzoylecgonine being essentially inactive (Crumb and Clarkson, 1992). In addition we found in this study that methylecgonidine has relatively low potency for blocking both the HERG potassium current and the human sodium current. Interestingly, although we found methylecgonidine to be much (~39-fold) less potent than cocaine in blocking HERG current, a recent study has shown that methylecgonidine blocks the acetylcholine-activated potassium current with a 2-fold higher potency compared with cocaine (Xiao and Morgan, 1998). This indicates that cocaine's metabolites do not show the same pattern of biological potencies for effecting all cocaine-sensitive ion channels.

Mechanism of HERG Channel Block. The block of HERG currents by cocaine and its two metabolites cocaethylene and methylecgonidine were similar to each other in terms of producing a more intense blockade as the prepulse voltage was made more positive over the range of HERG channel activation (Fig. 6). In contrast, at more positive voltages there was little, if any, voltage dependence of block. These results are consistent with the hypothesis that all three drugs have a selectively higher affinity for the HERG channels when they open or are activated, but the affinity of each drug for the channel is not markedly affected by changes in the transmembrane voltage field once HERG channels are fully activated (O'Leary, 2001). One possible explanation for this observation is that the binding site for these drugs is not located very deep within the channel lumen, in contrast to quinidine block of Kv1.5 (Yeola et al., 1996). Alternatively, the neutral form of the drug, which would not be sensitive to changes in transmembrane voltage, could be a major contributor to channel block. Additional experiments will be needed to distinguish between these possibilities. The ability of cocaine and cocaethylene to slow the rate of decay of tail currents and produce a "crossover" of tail currents when superimposed on control traces (Fig. 3; Table 1) is also consistent with open channel block and may reflect an inability of the channel to close prior to drug unbinding and exit from the channel (Yeola et al., 1996).

Clinical Relevance. Understanding the effects of cocaine and its metabolites on cardiac ion channels is important because it provides insight into the cellular mechanisms underlying their adverse effects on the heart, which ultimately could have an impact on which therapeutic interventions may have the best chance for success in treating patients. Peak plasma concentrations more than 200 ng/ml (0.7 µM) cocaine have been documented in controlled studies of cocaine disposition in humans (Chow et al., 1985; Jeffcoat et al., 1989), whereas plasma levels of up to 421 ng/ml (1.4 µM) have been reported in (nonlethal) trauma victims (Bailey, 1993), and an average as high as 6 µg/ml (20 µM) has been detected in post-mortem blood samples (Mittleman and Welti, 1984). Under our experimental conditions, these levels of cocaine would block 14, 24, and 82% of the total HERG current, suggesting that significant cocaine blockade of HERG potassium currents may be achieved at clinically relevant concentrations. Our results also suggest that the presence of cocaethylene could contribute to QT prolongation in patients who coadminister cocaine and ethanol. This may be important because estimates of the combined use of cocaine and ethanol range from 60 to 85% (Grant and Harford, 1990). Plasma concentrations of 128 µg/l (0.4 µM) cocaethylene have been detected in hospitalized trauma patients that have tested positive for cocaine metabolites (Bailey, 1993), and concentrations up to 2.7 µg/ml (7 µM) cocaethylene have been measured in ante-mortem blood samples taken from patients admitted to a hospital emergency room (Hearn et al., 1991). Our data indicate that either of these concentrations should produce a substantial (33 versus 75%) block of HERG channels (Fig. 4) and therefore could be expected to affect cardiac repolarization.