Local Anesthetic Interaction with Human Ether-a-go-go –related Gene (HERG) Channels: Role of Aromatic Amino Acids Y652 and F656

Chinese hamster ovary cells were cultured in 50 ml-flasks (NUNC, Roskilde, Denmark) at 37°C in MEM Alpha medium (GIBCO; Invitrogen, Carlsbad, CA) with 10% fetal calf serum, 100 U/ml penicillin, and 100 mg/ml streptomycin in a humidified atmosphere (5% CO²). Cells were subcultured in 35-mm diameter monodishes (NUNC) at least 1 day before transfection.

The mutants HERG Y652A and F656A were created by site-directed mutagenesis. All channels were cloned in the pcDNA3 expression vector. Chinese hamster ovary cells were transiently transfected with 1 μg HERG wt or mutant cDNA, 0.5 μg EFGP cDNA, and 3 μl lipofectamine reagent (Invitrogen) per dish according to the manufacturer’s protocol after 1 day. Cells were cotransfected with an EGFP pcDNA3 construct to verify successful transfection. Only green fluorescing cells were used for patch clamp experiments. Patch clamp experiments were performed 1 or 2 days after transfection.

Whole cell currents were recorded using the patch clamp technique24with an EPC-9 amplifier and Pulse software version 8.50 (HEKA Electronik, Lambrecht, Germany). Patch electrodes were pulled from borosilicate glass capillary tube (World Precision Instruments, Saratoga, FL) on a horizontal puller (P-97; Sutter Instrument Co., Novato, CA) and had a pipette resistance of 1.5–3.5 MΩ. The internal solution contained 160 mm KCl, 0.5 mm MgCl², 10 mm HEPES, and 2 mm Na-ATP (all from Sigma, Deissenhofen, Germany), adjusted to a pH of 7.2 with KOH. The external solution contained 135 mm NaCl, 5 mm KCl, 2 mm CaCl², 2 mm MgCl², 5 mm HEPES, 10 mm sucrose, and 0.1 mg/ml phenol red (all from Sigma), adjusted to a pH of 7.4 with NaOH. To record inward tail currents of HERG channels, an extracellular solution with high [K⁺] was used, containing 40 mm NaCl, 100 mm KCl, 2 mm CaCl², 2 mm MgCl², 5 mm HEPES, 10 mm sucrose, and 0.1 mg/ml phenol red, adjusted to a pH of 7.4 with NaOH.

Series resistance was 2.5–6.0 MΩ and was actively compensated for by 85%. A leak subtraction protocol was used except for recordings with high extracellular [K⁺] ([K⁺]^o). The recorded signal was filtered at 2 kHz and stored with a sampling rate of 5 kHz for analysis. Bupivacaine (Sigma), levobupivacaine, S  (−)-ropivacaine, R  (+)-ropivacaine, and mepivacaine (all from AstraZeneca, Södertalje, Sweden) were dissolved in the extracellular solution. A hydrostatically driven perfusion system was used to apply the drugs onto the cells and to exchange the extracellular solutions. All experiments were performed at room temperature.

Different pulse protocols were used for characterization of the channels and to establish their pharmacologic sensitivities. The holding potential was −80 mV for all experiments. For the activation protocol, cells were depolarized from −80 to +60 mV in 10-mV steps for 1 s, and tail currents were recorded at −40 mV. In high [K⁺]^o, the tail potential was changed to −120 mV. To analyze the deactivation, channels were activated by a 2-s pulse to +60 mV, and deactivating tail currents were recorded at potentials from −120 to +40 mV in 10-mV steps. The same protocol was used in high [K⁺]^o. To compare the three channels, currents were normalized to the minimum. The deactivation was fitted with an exponential function that yielded two time constants. For the steady state inactivation, a three-step protocol was used: First, the cells were depolarized for 2 s from −80 to +40 mV; then, a pulse from −100 to +60 mV in 10-mV steps was applied for 30 ms; and finally, the membrane potential was held at +40 mV for 0.5 s. The inactivation was corrected for the closing during the hyperpolarizing step at very negative potentials as described previously.25The corrected curves were fitted by a Boltzmann function. During the instantaneous activation protocol channels were activated and inactivated by depolarization from a holding potential of −80 to +20 mV for 2 s. After recovery from inactivation by repolarization to −80 mV for 30 ms, the instantaneous current was recorded at potentials from +40 mV to −120 mV in 10-mV steps. The time constants of inactivation were derived from a monoexponential fit of the decay. This protocol was also used for pharmacologic experiments. For this purpose, the total length of the protocol was reduced, and the potential was increased in 30-mV steps from −120 to +60 mV.

For the pharmacologic experiments with HERG wt and Y652A, a ramp protocol was used.12,26Cells were depolarized from a holding potential of −80 mV to +60 mV and repolarized to −80 mV within 1 s. In high [K⁺]^o, a single pulse to +20 mV for 2 s and a tail potential of −120 mV was used for the pharmacologic tests. Repetitive pulses were applied to determine that steady state inhibition was reached.

Data were analyzed with Pulse Fit software (HEKA Electronik) and with Kaleidagraph software (Synergy Software, Reading, PA). The normalized tail currents during the activation protocol were fitted by a modified Boltzmann equation: I = I^max/[1 + exp((V^0.5− V^m)/k)], where V^0.5is the voltage of half-maximal activation, V^mis the membrane potential, and k  is the slope factor. The exponential decay of deactivation or inactivation was fitted with one or two time constants according to the following equation: y = C + A^fastexp(−t/τ^fast) + A^slowexp(t/τ^slow), with the time constants τ^fastand τ^slowand the amplitudes A^fastand A^slow. The inhibition of currents was quantified by the reduction of the maximal current during the ramp protocol or during a single pulse in case of the high-[K⁺]^oexperiments. Also, the charge crossing the membrane during the ramp protocol (Q^ramp) and the reduction of Q^rampwere analyzed. The charge crossing the membrane is equivalent to the time integrals of current traces and was determined using Pulse Fit software. The fractional block f was calculated by the following formula: f = 1 −[2 × I^{max, drug}/(I^{max, control}+ I^{max, washout})]. Concentration–response curves were fitted by a Hill function: f = 1/[1 + (IC⁵⁰/c)^h], where IC⁵⁰is the concentration of half-maximal inhibition, c is the concentration of the local anesthetic, and h is the Hill coefficient. Statistical significance was tested using a two-sided Student t  test or analysis of variance (Excel; Microsoft, Redmond, WA). Data are presented as mean ± SD unless stated otherwise; n values indicate the number of experiments.

Representative current traces of HERG wt and HERG Y652A elicited by the activation protocol are shown in figure 1A. The voltage dependence of activation of HERG Y652A channels was not significantly different from wt channels (fig. 1B; wt: voltage of half-maximal activation V^0.5= 4.85 ± 4.23 mV, slope = 8.51 ± 0.83 mV, n = 6; Y652A: V^0.5= 5.35 ± 4.31 mV, slope = 8.79 ± 1.27 mV, n = 12). Also, the time constants of activation did not differ between wt and mutant channels. However, the decay of HERG Y652A currents was faster than that of HERG wt currents (exponential fit of the decay after activation to 50 mV, wt: τ= 1.42 ± 0.5 s; Y652A: τ= 0.86 ± 0.3 s; P < 0.01). Figure 1Cshows exemplary current traces obtained by the deactivation protocol. Both HERG wt and HERG Y652A currents exhibited inward rectification with no difference in the voltage dependence of deactivation (fig. 1D). The steady state inactivation was analyzed next (fig. 2A). The voltage dependence of HERG Y652A current inactivation was shifted to more positive potentials compared with HERG wt (fig. 2B; V^0.5=−36.2 ± 10.6 mV, n = 6 for Y652A versus  V^0.5=−69.6 ± 7.8 mV, n = 7 for wt; P < 0.05). Instantaneous currents inactivated in a voltage-dependent manner (fig. 2C) with time constants that were smaller for HERG Y652A than for wt channels (fig. 2D; n = 4 each, P < 0.05).

HERG F656A channels only conducted very small currents under low extracellular [K⁺].2Inward currents of this mutation were therefore recorded using high extracellular [K⁺].20To compare all three channels, HERG wt and HERG Y652A currents were recorded under these conditions as well (fig. 3). Tail current densities at −120 mV under high extracellular [K⁺] were 968 ± 475 pA/pF for HERG wt, 571 ± 272 pA/pF for HERG Y652A, and 129 ± 49 pA/pF for HERG F656A channels, respectively (n = 5 for each). Only the difference between HERG F656A and HERG wt was significant. Exemplary current traces evoked by the activation protocol in high [K⁺]^oare shown in figure 3A. Voltage dependence of activation did not differ between HERG wt and HERG Y652A currents, whereas HERG F656A channels activated with a V^0.5shifted by 10 mV to more negative potentials (fig. 3B). Figure 3Cshows deactivating currents of all three channels. HERG F656A hyperpolarized the voltage dependence of deactivation (fig. 3D) and slowed the deactivation kinetics, whereas deactivation kinetics were accelerated by HERG Y652A (fig. 3D, inset). The differences between HERG wt and the mutant channels were significantly different (confirmed by analysis of variance).

The inhibitory effects of the local anesthetics bupivacaine, levobupivacaine, ropivacaine, and mepivacaine on HERG wt and HERG Y652A currents were investigated using the ramp protocol. This protocol evoked bell-shaped currents (fig. 4A). The mean time to peak current was 723 ± 10 ms for HERG wt (n = 116) and 615 ± 12 ms for HERG Y652A (n = 91). These values correspond to voltages of −40 ± 1 mV (HERG wt) and −26 ± 2 mV (HERG Y652A), respectively. All local anesthetics reduced HERG wt and HERG Y652A currents in a concentration-dependent and reversible manner (figs. 4B–F). The inhibition was quantified as the decrease of the maximum current as well as the reduction of the charge. The concentration–response data were mathematically described by Hill functions (figs. 5A and Band table 1). The concentration of half-maximal inhibition (IC⁵⁰) by the local anesthetics was between 4 and 10 times higher for HERG Y652A compared with the wt channel. The Hill coefficients were close to unity for inhibition of both channels by all local anesthetics. Application of local anesthetics furthermore caused a concentration-dependent and reversible rightward shift of the peak HERG wt current and therefore increased the time to peak current response (figs. 4B–F). In contrast, the application of the local anesthetics caused a concentration-dependent and reversible leftward shift of the peak HERG Y652A current and therefore decreased the time to peak current response. Because of the reduced sensitivity of HERG Y652A, the shift is also more pronounced at higher concentrations of local anesthetics. For example, 30 μm bupivacaine caused a shift of the time to peak HERG wt current response of +62 ± 10 ms (n = 9), and 100 μm bupivacaine caused a shift of +90 ± 27 ms (n = 5), corresponding to shifts in voltage of +9 ± 1 and +13 ± 4 mV, respectively. For HERG Y652A, 100 μm bupivacaine caused a shift of the time to peak current of −37 ± 17 ms (n = 9), and 300 μm bupivacaine caused a shift of −71 ± 18 ms (n = 9). These values correspond to shifts in voltage of −5 ± 2 and −10 ± 2 mV, respectively. For both HERG wt and HERG Y652A, the IC⁵⁰values of the local anesthetics increased in the order levobupivacaine < bupivacaine < ropivacaine < mepivacaine.

The influence of the mutation Y652A on stereoselective inhibition of HERG currents was further analyzed. The wt channel was twice as sensitive to levobupivacaine than to bupivacaine (IC⁵⁰12.7 ± 1.4 vs.  21.9 ± 1.6 μm; table 1), whereas the ratio of IC⁵⁰values of both drugs approximated unity for the mutant channel HERG Y652A (IC⁵⁰82.9 ± 3.2 vs.  95.2 ± 5.0 μm; table 1). For ropivacaine, the influence of HERG Y652A on stereoselective inhibition was even more pronounced (compare figs. 4D and E). S  (−)-ropivacaine was a more potent inhibitor of HERG wt currents than R  (+)-ropivacaine (25 μm: 53 ± 5%, n = 8 vs.  28 ± 2%, n = 5; P < 0.05; 83 μm: 79 ± 4%, n = 8 vs.  60 ± 4%, n = 5; P < 0.05), whereas R  (+)-ropivacaine was a more potent inhibitor of HERG Y652A currents than S  (−)-ropivacaine (83 μm: 41 ± 3%, n = 7 vs.  27 ± 2%, n = 7; P < 0.05; 250 μm: 63 ± 3%, n = 7 vs.  49 ± 3%, n = 7; P < 0.05).

The inhibition of HERG wt and HERG Y652A channels by bupivacaine was further analyzed by applying an instantaneous current protocol (fig. 6; see Materials and Methods section). Bupivacaine inhibited I^maxof both channels (fig. 7) in a voltage-independent manner (fig. 7A). The local anesthetic significantly altered the voltage dependence and the size of the inactivation time constants of HERG wt but not of HERG Y652A channels (fig. 7B; tested by analysis of variance). The time courses of deactivating currents (−120 mV) of both HERG wt and HERG Y652A were slowed by bupivacaine (HERG wt: τ^{deact, fast}= 18.98 ± 4.31 ms, τ^{deact, slow}= 179.3 ± 44.9 ms under control and washout conditions; τ^{deact, fast}= 27.14 ± 5.89 ms, τ^{deact, slow}= 259.6 ± 63.3 ms under 20 mm bupivacaine; n = 5, paired experiments, P < 0.05; HERG Y652A: τ^{deact, fast}= 12.13 ± 2.50 ms, τ^{deact, slow}= 59.57 ± 21.37 ms under control and washout conditions; τ^{deact, fast}= 13.28 ± 2.12 ms, τ^{deact, slow}= 90.71 ± 27.66 ms under 90 μm bupivacaine; n = 5, paired experiments, P < 0.01).

To meaningfully compare the pharmacologic sensitivities between all three channels, bupivacaine effects on HERG wt and HERG Y652A channels were in addition established under high extracellular [K⁺]. Bupivacaine (100 μm) inhibited wt currents by 66 ± 5% (n = 5). Both mutants reduced the sensitivity (figs. 8A and B), with HERG F656A currents being less sensitive than HERG Y652A (32 ± 4% for Y652A, n = 5; 17 ± 6% for F656A, n = 5; P < 0.01). Because of the low sensitivity of HERG F656A channels, it was not possible to determine whole concentration–response curves for the different local anesthetics. Instead, local anesthetic sensitivity was compared at a concentration of 1 mm. HERG F656A channels exhibited the same order of decreasing pharmacologic sensitivity as HERG wt (fig. 8C). However, HERG wt currents were inhibited more potently by S  (−)-ropivacaine than by R  (+)-ropivacaine, whereas HERG F656A currents were inhibited more potently by R  (+)-ropivacaine than by S  (−)-ropivacaine (R-ropivacaine vs.  S-ropivacaine: 74 ± 2% vs.  84 ± 1%, n = 5, paired experiments, P < 0.01 for wt; 43 ± 5% vs.  31 ± 5%, n = 4, paired experiments, P < 0.05 for F656A). The time constants of deactivation in high extracellular [K⁺] and the effect of bupivacaine on deactivation (figs. 8D and E) revealed that high extracellular [K⁺] slows the deactivation kinetics of both HERG wt and HERG Y652A. However, deactivation of HERG channels is slowed by bupivacaine (100 μm) under normal as well as high extracellular [K⁺], whereas deactivation of HERG Y652A and HERG F656A follows a faster time course under the influence of bupivacaine (100 μm for HERG Y652A, 1 mm for HERG F656A) compared with drug-free conditions under high extracellular [K⁺].

In the current work, the effects of the mutations Y652A and F656A in the S6 region of HERG potassium channels2on the inhibition by amino-amide local anesthetics have been established. Mutating either of the aromatic residues diminished the block by these local anesthetics 4- to 10-fold (Y652A) and 20- to 30-fold (F656A), respectively. In accord with previous studies,2,16the voltage dependence of activation and deactivation remained unchanged by the mutation Y652A. However, the voltage dependence of steady state inactivation was shifted to more positive potentials, and the inactivation and deactivation time constants were significantly faster. This may explain the observation that the time to peak current was shorter for HERG Y652A than for HERG wt during the ramp protocol. The mutation F656A reduced the current density by approximately 90% compared with the wt channel. Furthermore, the voltage dependence of activation and deactivation were shifted to more negative potentials, and the deactivation kinetics were slowed.

The effects of bupivacaine, levobupivacaine, and ropivacaine but not of mepivacaine on HERG wt channels have been analyzed before.11However, instead of conventional stimulation protocols, we used a ramp protocol26to establish drug sensitivity of wt and mutant channels. Such a protocol has repeatedly been used to measure drug sensitivity of HERG channels.12,27,28This protocol yielded IC⁵⁰values and Hill coefficients of bupivacaine, levobupivacaine, and ropivacaine that are nearly identical to those obtained with conventional pulse protocols.11Furthermore, the different current responses of HERG wt and HERG Y652A to the ramp protocol suggest that differences in channel gating between wt and mutant channels are elicited by the ramp protocol. The ramp protocol therefore allows comparing local anesthetic sensitivity of wt and mutated HERG channels. The effect of bupivacaine on these channels was in addition assessed with an instantaneous current protocol.25By this protocol, the effect of bupivacaine on channels in the open state, on channel deactivation, and on channel inactivation was studied. Bupivacaine increased the fast and slow time constants of deactivation in both HERG wt and HERG Y652A channels compatible with open channel block. Bupivacaine accelerated macroscopic current decline of wt but not of HERG Y652A channels. Because inactivation of HERG Y652A under drug-free conditions is already faster than macroscopic current decline of HERG wt channels under drug conditions, the time constants of current decline of HERG Y652A under drug conditions likely reflect channel inactivation rather than drug–channel interaction. The already accelerated inactivation time constant of HERG Y652A channels under control conditions may therefore have precluded resolving the time course of the bupivacaine effect. The extent of inhibition of the instantaneous peak current response of HERG wt and HERG Y652A was identical to the extent of inhibition predicted from the respective concentration–response curves obtained with the ramp protocol. Taken together, these results may suggest that ramp current inhibition mainly reflects open channel inhibition. This idea is supported by the experiments with high extracellular [K⁺]. Inhibition of HERG wt as well as HERG Y652 is reduced by the presence of high extracellular [K⁺]. This has previously been described for the open channel blocker E-4031.29It may therefore be hypothesized that interaction with the open channel pore constitutes an important mechanism contributing to local anesthetic inhibition of wt11as well as mutated HERG channels.

The results of our study demonstrate that the mutations in the S6 segment are not conservative with regard to ion channel gating. Therefore, it cannot entirely be ruled out that differences in channel gating induced by the mutations Y652A and F656A may have influenced local anesthetic affinity. Mutations that disable inactivation (G628C-S631C) or selective inhibition of HERG channel inactivation by addition of Cd²⁺ions or removal of Na⁺ions reduce block by D-sotalol.30Removal of C-type inactivation reduces bupivacaine sensitivity and confers voltage dependence to the inhibition of HERG channels.10However, as judged from the results obtained with the instantaneous current protocol, voltage dependence of inhibition does not differ between HERG wt and HERG Y652A. Also, inhibition of both channels HERG wt and HERG Y652A was reduced by increasing the concentration of extracellular [K⁺]. Both results point to a common inhibitory mechanism. In addition, the inhibitory potency of the local anesthetics significantly correlated with the number of the CH²groups of the local anesthetics for HERG wt (figs. 9A and B), HERG Y652A (fig. 9A), and HERG F656A (fig. 9B). The correlations had similar slope factors. If the quantitatively and qualitatively different changes in channel gating were to play a fundamental role in reducing local anesthetic affinity, a conserved response to extracellular [K⁺] as well as a conserved relation between physicochemical properties of the drugs and blocking affinity would be unlikely. The conserved relation between physicochemical properties of the drugs and blocking affinity would furthermore point to a nonspecific membrane-mediated effect underlying inhibition by the local anesthetics.31,32As an alternative, it may be hypothesized that hydrophobic interactions occurring at a site different from the inner S6 cavity22are crucial for local anesthetic interaction. Both alternatives seem less likely.

HERG wt channels were inhibited more potently by the S  (−)-enantiomer than by the R  (+)-enantiomer of ropivacaine. Mutating the aromatic amino acids Tyr652 or Phe656 to alanine reversed the stereoselectivity of ropivacaine block. This change in the stereoselectivity of block suggests a direct interaction of the drug molecules with the residues Tyr652 and Phe656 rather than a membrane-mediated effect. The aromatic residues in the S6 region of HERG channels are therefore likely to be involved in the interaction with the local anesthetics investigated. Kv channels exhibit a reversed local anesthetic stereoselectivity23,33compared with HERG channels, with the R  (+)-enantiomer being more potent than the S  (−)-enantiomer. Exchanging either of the aromatic residues in the S6 of HERG for alanine results in a pattern of stereoselective inhibition resembling that of Kv channels. These results indicate that both aromatic amino acids are necessary for the pattern of stereoselective inhibition by local anesthetics specific for HERG channels. The conserved correlation of inhibitory potency and lipophilic properties of the drugs, however, suggests that additional hydrophobic amino acids residing in the inner cavity of HERG channels16may be involved in local anesthetic action.

Our results demonstrate that local anesthetic affinity of HERG channels is determined by the lipophilic properties of the drugs. The longer the alkyl side chain is, the more potent the drug is. Also, the orientation of the local anesthetic molecule influences drug affinity of HERG channels, with the S  (−)-isomer being more potent than the R  (+)-isomer. HERG channels constitute well-established, toxicologically relevant targets for many structurally and functionally unrelated substances, including local anesthetics.2,10–12Inadvertent intravascular injection of amino-amide local anesthetics, such as those investigated in this study, are capable of inducing severe ventricular arrhythmia and sudden death.34,35On the other hand, modest inhibition of HERG channels may result in class III antiarrhythmic action.3Like the class III antiarrhythmic agent sotalol30and other proarrhythmic drugs,36local anesthetics may therefore exhibit antiarrhythmic action at lower concentrations and cardiotoxic side effects at higher concentrations. It may be worth considering that because of the specific structure of HERG channels, both antiarrhythmic as well as proarrhythmic action may increase with the lipophilicity of amino-amide local anesthetics and may be more pronounced for the S  (−)-enantiomers.

In summary, our results suggest that local anesthetics interact with the inner cavity of the HERG channel pore. Interaction may involve the aromatic residues Tyr652 and Phe656. However, the correlation between the lipophilic properties of the drugs and inhibitory potency remains unaltered by the mutations Y652A and F656A. Additional hydrophobic domains of the channel contributing to local anesthetic interaction with HERG channels thus remain to be identified.

The authors thank Olaf Pongs, Ph.D. (Director of the Institute of Neural Signal Transduction, University Medical Center Hamburg-Eppendorf, Germany), for his generous support; Andrea Zaisser (Medicinal Technical Assistant, Institute for Neural Signal Transduction, University Medical Center Hamburg-Eppendorf, Germany) for technical assistance; and Anna Solth, M.Sc., and Axel Neu, M.D. (Research Fellows, Institute for Neural Signal Transduction, University Medical Center Hamburg-Eppendorf, Germany), for critically reading the manuscript.