Inhibition of KCNQ2/Q3 channels may cause convulsion in humans. The interaction of bupivacaine with these channels is unknown. The anticonvulsant retigabine activates KCNQ2/Q3 channels and may reverse inhibitory actions of bupivacaine. Potassium channel stimulation may thus constitute a novel approach to treat local anesthetic-induced seizures. The aim of this study was to characterize bupivacaine effects on KCNQ2/Q3 channels and to investigate whether retigabine reverses the effects of the local anesthetic.
KCNQ2/Q3 channels were transiently expressed in Chinese hamster ovary cells. The effects of bupivacaine and retigabine were studied with the patch-clamp technique.
Bupivacaine inhibited KCNQ2/Q3 channels in a concentration-dependent and reversible manner. The concentration-response curve was described by a Hill equation (IC50 = 173 +/- 7 microm, Hill coefficient = 1.4 +/- 0.1, mean +/- SEM, n = 37). The inhibitory effect did not differ between bupivacaine and levobupivacaine (42 +/- 4%, n = 7, versus 42 +/- 5%, n = 10; P > 0.05). Ropivacaine was four times less potent than bupivacaine. The inhibition of KCNQ2/Q3 channels by bupivacaine resulted in a significant and reversible depolarization of the membrane potential. Retigabine (300 nm-10 microm) reversed the inhibitory action of bupivacaine on KCNQ2/Q3 channels as well as the depolarization of the membrane potential.
The anticonvulsant retigabine at nanomolar concentrations reverses the inhibitory effect of micromolar concentrations of bupivacaine. Our results allow the hypothesis that activation of KCNQ2/Q3 channels by retigabine may offer a novel therapeutic approach for the treatment of bupivacaine-induced seizures.
THE delayed rectifier potassium channels KCNQ2 and KCNQ3 play a major role in controlling the excitability of neuronal cells.1They are prominently expressed in the central nervous system and in dorsal root ganglia.2,3Heteromultimeric complexes formed by these channels are the molecular correlate of the M-current.4,5KCNQ2/Q3 channels may control the subthreshold excitability of a neuron and the number of action potentials fired by a neuron when receiving excitatory input.1,6Mutations causing functional impairment of these potassium channels result in a human form of epilepsy7–9consistent with the hypothesis that hyperexcitability of neuronal cells is a typical sign of epileptic disorders.10KCNQ2/Q3 channels are specifically activated by the anticonvulsant drug retigabine.11,12Activation of KCNQ2/Q3 channels leads to a hyperpolarization of the neuronal membrane potential13and to a reduction of the frequency of neuronal action potential firing.14Retigabine effectively reduces the seizure activity in a wide variety of animal models,15and the anticonvulsant also reduces seizure frequency in patients with refractory epilepsy.16
Accidental intravenous injection, overdosage or rapid systemic uptake of local anesthetics may result in severe neurotoxic side effects such as seizures.17The incidence of seizures associated with regional anesthesia ranges from 0.75–7 of 1000 patients depending on the technique performed and the local anesthetic used.18,19,20There is only limited information about the molecular mechanisms underlying local anesthetic-induced seizure. Molecular targets suggested to mediate these severe side effects include γ-aminobutyric acid type A receptors,21,22N-methyl-D-aspartate-receptors,23and potassium channels.24,25Treatment usually consists of the application of γ-aminobutyric acid receptor type A-agonists such as benzodiazepines and barbiturates and artificial ventilation is therefore often mandatory.26,27
Despite the well-established role of KCNQ2/Q3 potassium channels in a human form of epilepsy the action of bupivacaine on these ion channels has not been investigated. It is, therefore, unknown if bupivacaine inhibits KCNQ2/Q3 channels and if inhibition occurs at toxicological relevant concentrations. Furthermore, it is unknown if retigabine in the presence of bupivacaine is still capable of stimulating KCNQ2/Q3 channels. This may be indicative of a novel treatment for local anesthetic-induced seizure. The current study therefore had the following aims. First, it was intended to characterize the effects of bupivacaine on human KCNQ2/Q3 channels. Second, we aimed to investigate if retigabine stimulates KCNQ2/Q3 channels in the presence of neurotoxic concentrations of bupivacaine.
Materials and Methods
Chinese hamster ovary cells were grown as nonconfluent monolayers in MEM Eagle Alpha medium (Gibco Invitrogen Cooperation, Karlsruhe, Germany) containing 10% fetal calf serum (Biochrom, Berlin, Germany), penicillin (100 IU/ml), streptomycin (100 μg/ml), and l-glutamine (292 μg/ml) (Gibco Invitrogen Cooperation). The cells were cultured at 37°C in a humidified atmosphere (95% air, 5% CO2). Before the electrophysiological experiments the cells were subcultured in monodishes (35 mm diameter, NUNC, Roskilde, Denmark) and transiently transfected at least 24 h before recording with constructs containing cDNA encoding human KCNQ2 and KCNQ3 channels (GenBank Accession No. AF110020; AF033347) using Lipofectamine (Invitrogen Life Technologies) according to the manufacturers recommendation. For expression of heteromultimers, equal amounts of KCNQ2 and KCNQ3 cDNA (1 μg) were used. Cotransfection with a construct for enhanced green fluorescent protein (EGFP-N1, BD Biosciences Clontech, Heidelberg, Germany) was used to detect positive cells.
Whole cell currents and membrane potentials were measured with the voltage-clamp and current-clamp methods of the patch-clamp technique28using an EPC-9 amplifier (HEKA Elektronik, Lambrecht, Germany) and Pulse software 8.53 (HEKA Elektronik). The currents were filtered at 1 kHz and series resistance was actively compensated by at least 80%. The patch electrodes were fabricated from borosilicate glass capillary tubes with filament (World Precision Instruments, Saratoga, FL) using a Sutter P-97 puller (Sutter Instrument Company, Novato, CA). The pipettes had a resistance of 2 to 4 MΩ. They were filled with a solution containing the following electrolyte concentrations (mM): KCl, 160; MgCl2, 0.5; HEPES, 10; Na2ATP, 2; pH 7.25 adjusted with KOH. The extracellular solution consisted of (mM): NaCl, 135; KCl, 5; CaCl2, 2; MgCl2, 2; HEPES, 5; sucrose, 10; Phenol red, 0.01 mg/ml; pH 7.4 adjusted with NaOH (all chemicals by Sigma, Deissenhofen, Germany). Bupivacaine (Sigma, Deissenhofen, Germany) was dissolved in the extracellular solution prepared from a stock solution of 1 mm. The stock solutions were stored at −20°C. Retigabine was provided by Viatris GmbH Radebeul, Dresden, Germany. Retigabine (100 mm) was dissolved in dimethylsulfoxide to yield a stock solution. At the highest concentration of retigabine used in these experiments (10 μm) the dimethylsulfoxide concentration did not exceed 0.01%. At this concentration dimethylsulfoxide had no effect on KCNQ2/Q3 current amplitudes (10.5 ± 4.8 nA versus 10.5 ± 4.6 nA, +30 mV, n = 3; P > 0.05). Test solutions containing the drugs were superperfused on the cells by a perfusion system driven by hydrostatic pressure with Teflon tubing.
Stimulation Protocols and Data Analysis
The holding potential during all experiments was −80 mV. KCNQ2/Q3 whole-cell currents were elicited by different protocols. The current-voltage relationship was established by depolarizing the cell membrane in steps of 10 mV for 1500 ms between the membrane potentials of −70 mV and +30 mV. The whole cell conductance was calculated using the following formula: Gmax= Imax/(Vm− EK), where Imaxis the maximum current at the end of each test potential, Vmis the membrane potential, and EKthe Nernst potential for potassium (−87.54 mV under our experimental conditions). The conductance-voltage relationship was mathematically described by a Boltzmann function (y = M3/(1 + exp ((M1− M0)/M2)), where M0= the membrane potential, M1= the activation midpoint, M2= the slope factor, and M3= the maximal whole cell conductance) using Kaleidagraph software (Synergy Software, Reading, PA). For characterizing the concentration-dependent effect of bupivacaine two different pulse protocols were used: a ramp pulse protocol that increased during 1500 ms from the holding potential to a membrane potential of +60 mV and a rectangular pulse protocol that depolarized the membrane potential from −80 mV to +30 mV for 1200 ms. In the case of the ramp pulse protocol, inhibition of the currents by bupivacaine was measured as the reduction of charge transfer between the membrane potential of −80 mV and +60 mV. For this purpose the ratio of the charge (Q) under local anesthetic influence and the mean of charge under control and washout conditions was subtracted from one (inhibition = 1− (Qdrug/((Qcontrol+ Qwashout)/2)). In the case of the rectangular pulse protocol, the inhibition of the maximum current was quantified as the ratio of the maximum current (I) under local anesthetic influence and the mean of maximum current under control and washout conditions was subtracted from one (inhibition = 1 − (Idrug/((Icontrol+ Iwashout)/2)). The data of the concentration-response curves were mathematically described by Hill equations (y = M1× M0M2/M3M2+ M0M2, where y = inhibition, M0= drug concentration, M1= maximal inhibition, M2= Hill coefficient, and M3= concentration of the half-maximal inhibitory effect or IC50value) using Kaleidagraph software (Synergy Software). Standard errors of calculated Hill parameters were expressed as defined by Kaleidagraph. For describing the pharmacological interaction of bupivacaine and retigabine on KCNQ2/Q3 channels, a ramp pulse protocol was used that increased from the holding potential during 1500 ms to a membrane potential of 0 mV. The effects of retigabine and bupivacaine on the currents were quantified as the reduction or stimulation of charge transfer compared with control values. For the quantification of the small effect of retigabine (300 nm) only experiments with a size of the washout current of at least 97% of the control current were incorporated into the analysis. To analyze the pharmacological effects on the deactivation kinetics of KCNQ2/Q3 channels tail currents were evoked by a prepulse of 1000 ms length to a membrane potential of +30 mV and subsequent test pulses with a duration of 300 ms to membrane potentials increasing from −100 mV to +30 mV in 10-mV steps. The time constants were determined by fitting current decay with a mono-exponential function using Pulsefit software 8.53 (HEKA Elektronik) and a biexponential function in case of currents under the influence of retigabine.
Statistical significance was tested using analysis of variance and Tukey-Kramer multiple comparisons test (Graph Pad Prism, San Diego, CA) or two-sided paired and unpaired Student t test, as appropriate (Excel; Microsoft, Redmond, WA). Data points are given as mean ± SD unless stated otherwise; n values indicate the number of experiments.
As a prerequisite to the intended pharmacological experiments, basic properties of KCNQ2/Q3 channels expressed in Chinese hamster ovary cells were characterized first. The original current recordings demonstrated voltage-gated slowly activating and noninactivating currents (fig. 1A). The maximum current size at +30 mV was 14 ± 7 nA, with a maximum conductance of 116.8 ± 62.5 nS and an activation midpoint of −29.7 ± 2.7 mV (n = 5). The time constants of activation (τact) were voltage-dependent. The values of τactmono-exponentially declined from 908 ± 103 ms at −40 mV to a value of 74 ± 14 ms at +30 mV (n = 5). As illustrated by the example shown (fig. 1A), the anticonvulsant retigabine increased the current amplitude by shifting the activation midpoint of KCNQ2/Q3 channels to a more hyperpolarized potential (fig. 1B). Retigabine 10 μm shifted the voltage-dependence of activation by −21 ± 3 mV (n = 5; P < 0.01). The conductance-voltage curve indicated that retigabine mainly exerted its effect at membrane potentials between −80 mV and 0 mV (fig. 1B). A ramp protocol to activate KCNQ2/Q3 showed a similar stimulatory effect of retigabine as a rectangular pulse protocol (fig. 1C). Retigabine stimulated the charge transfer in a concentration-dependent and reversible manner. The application of retigabine (300 nm, 1 μm, 10 μm) resulted in a significant stimulation of the charge transfer by a factor of 1.07 ± 0.04 (n = 7; P < 0.05); 1.46 ± 0.12 (n = 8; P < 0.01) and 2.28 ± 0.43 (n = 8; P < 0.01).
Next, we established the concentration-response curve for the inhibitory effect of bupivacaine on KCNQ2/Q3 channels using both an increasing ramp and a rectangular pulse protocol (fig. 2A). Bupivacaine inhibited the channels evoked by either protocol in a concentration-dependent and reversible manner. The concentration-response data for inhibition of KCNQ2/Q3 channels were mathematically described by Hill equations and were fit to the data (fig. 2B). Both stimulation protocols resulted in similar concentration-response curves. In the case of the ramp pulse, the IC50value was 173 ± 7 μm and the Hill coefficient was 1.4 ± 0.1 (mean ± SEM, n = 37). The rectangular pulse yielded an IC50value of 217 ± 9 μm and a Hill coefficient of 1.3 ± 0.1 (mean ± SEM, n = 31). To test for a possible stereoselective effect of the bupivacaine isomers the inhibition of racemic bupivacaine was compared with the inhibition of the S(-) enantiomer levobupivacaine at a concentration close to the IC50value of racemic bupivacaine (200 μm; fig. 2C). This concentration was chosen because a potency difference would best be detected at the steepest part of the concentration-response curve. In case of a stereospecific effect KCNQ2/Q3 channels would be inhibited with different potency by bupivacaine and levobupivacaine. However, this was not the case. Racemic bupivacaine and levobupivacaine inhibited KCNQ2/Q3 channels by 42 ± 4% (n = 7) and 42 ± 5% (n = 10, P > 0.05), respectively (fig. 2C). To analyze if differences in side chain length and lipophilicity may influence local anesthetic sensitivity of KCNQ2/Q3 channels the concentration-response curve of S(-) ropivacaine was also established. The IC50value for inhibition of KCNQ2/Q3 channels by S(-) ropivacaine was four times higher than the IC50value for bupivacaine (748 ± 85 μm). The Hill coefficient was 1.0 ± 0.1 (mean ± SEM, n = 18).
KCNQ2 and KCNQ3 channels are supposed to form the heteromeric channel underlying the M-current.4,5However, both subunits can also form homomeric channels with M-like currents but smaller amplitudes than the heteromeric channels. As homomeric KCNQ2 and KCNQ3 channels have distinct pharmacological sensitivity to e.g. , tetraethylammonium, we tested whether KCNQ2 and KCNQ3 channels also differ in their sensitivity to bupivacaine. The current amplitudes of homomeric KCNQ2 and KCNQ3 channels were 3610 ± 121 pA (n = 5) and 273 ± 97 pA (n = 5), respectively (fig. 2D). KCNQ2 channels, thus, gave rise to currents 14-fold larger than homomeric KCNQ3 channels and fourfold smaller than heteromeric KCNQ2/Q3 channels. A possible difference in the sensitivity to bupivacaine was tested at a concentration close to the IC50-value for the inhibition of heteromeric KCNQ2/Q3 channels (200 μm). The results demonstrated that bupivacaine (200 μm) inhibited KCNQ2 channels to the same extent as KCNQ3 channels (62 ± 4% [n = 7]versus 60 ± 8% [n = 5]; P > 0.05; fig. 2D).
After establishing the sensitivity of KCNQ2/Q3 channels to the inhibitory effect of bupivacaine the action of the local anesthetic on these ion channels was investigated in more detail at a single concentration (1000 μm; fig. 3A). For this purpose a pulse protocol was used that allowed to detect effects on current activation as well as a possible voltage-dependence of channel block. Bupivacaine did not alter the voltage-dependence of channel activation; Vmidwas −25.2 ± 3.6 mV under control and washout condition and it was −23.3 ± 4.2 mV during the action of bupivacaine (n = 5; P > 0.05; fig. 3B). The local anesthetic, however, accelerated the time course of channel activation. Under control and washout conditions, the time constants monoexponentially declined with voltage from a value of 804 ± 187 ms at −40 mV to a value of 89 ± 6 ms at + 30 mV (n = 5; fig. 3C). Under the influence of bupivacaine (1000 μm), channel activation was approximately four times faster at the membrane potential of −40 mV (210 ± 76 ms versus 804 ± 187 ms), and the time constants linearly declined with increasing voltages (fig. 3C). Linear regression analysis yielded a regression coefficient of r = 0.98 and a negative slope of the regression line of −1.69. Voltage-dependence of inhibition was analyzed by quantifying the reduction of the whole cell conductance at different membrane potentials under the influence of bupivacaine (1000 μm) (fig. 3D). No significant voltage-dependence of inhibition was observed. The whole cell conductance was inhibited by 69 ± 7% at a membrane potential of −40 mV and by 69 ± 4% at a membrane potential of +30 mV, respectively (n = 5; P > 0.05).
After characterizing the effects of bupivacaine on KCNQ2/Q3 channels the interaction of retigabine and bupivacaine simultaneously applied to KCNQ2/Q3 channels was investigated. The pharmacological interaction of both drugs was analyzed with an increasing ramp pulse protocol and the following experimental sequence (fig. 4, A, B, and C). First KCNQ2/Q3 channels were measured under drug-free conditions. Then bupivacaine was applied on the cells at a concentration of neurotoxic relevance29(100 μm). As a next step different concentrations of retigabine (300 nm, 1 μm, and 10 μm) were simultaneously applied with bupivacaine (100 μm) on the cells, followed by the washout of the drugs. As predicted from the concentration-response curve (fig. 2B), bupivacaine (100 μm) given alone significantly reduced the charge transfer through KCNQ2/Q3 channels by 22 ± 9% (n = 12; P < 0.01; fig. 4A). During simultaneous application of retigabine and bupivacaine (100 μm) the current amplitude and the area under the current versus time curve were larger, as during the application of bupivacaine alone. Retigabine thus reversed the inhibitory effect of bupivacaine on KCNQ2/Q3 currents. This effect was dependent on the concentration of retigabine and it was already present at the lowest concentration of retigabine (300 nm) tested (78 ± 9% versus 89 ± 11% of control value, n = 12 paired experiments; P < 0.05; fig. 4A). The anticonvulsant at 300 nm not only significantly stimulated KCNQ2/Q3 channels in the presence of bupivacaine (fig. 4A), but the stimulatory effect was also larger in the presence of bupivacaine than in the absence of the local anesthetic (13 ± 5% versus 7 ± 4%, n = 7; P < 0.05). Retigabine at 1 μm completely abolished the inhibitory effect of bupivacaine and retigabine at 10 μm increased KCNQ2/Q3 currents in the presence of bupivacaine (100 μm) even when compared to control values (fig. 4, B and C).
The interaction of retigabine and bupivacaine was further investigated by analyzing drug-induced changes of the tail current amplitude and deactivation kinetics at −100 mV (fig. 5A). Whereas the time course of channel deactivation under control conditions was best described by a single exponential process (τ= 13.4 ± 3.3 ms, n = 10) deactivation of KCNQ2/Q3 channels under the influence of retigabine followed a biexponential process (τfast= 11.9 ± 5.1 ms, τslow= 70.7 ± 26.3 ms, n = 10; P < 0.05, data not shown). Bupivacaine reduced the tail current amplitude from −6.6 ± 3.5 nA under control conditions to −4.0 ± 2.1 nA (n = 6; P < 0.05; fig. 5B). The deactivation time constant of KCNQ2/Q3 channels recorded during the application of the local anesthetic was increased from 13.4 ± 3.3 ms to 15.1 ± 2.9 ms (n = 10; P < 0.05, fig. 5C). In the presence of bupivacaine retigabine increased the tail current amplitude from −4.0 ± 2.1 nA recorded under the influence of bupivacaine alone to −5.9 ± 2.5 nA recorded when bupivacaine and retigabine were simultaneously applied (n = 6; P < 0.05; fig. 5B). The anticonvulsant was furthermore still capable of slowing current deactivation despite the presence of the local anesthetic (τfast= 12 ± 6.9 ms, τslow= 85.8 ± 40.1 ms, n = 10; P < 0.05). Both time constants of tail current deactivation were not significantly different between currents in the presence of retigabine as compared to currents in the presence of retigabine together with bupivacaine. Analysis of the voltage-dependence of the interaction of retigabine and bupivacaine revealed that retigabine reversed the inhibitory action of bupivacaine primarily by shifting the normalized charge-voltage relationship to hyperpolarized potentials (fig. 5D).
Inhibition of KCNQ2/Q3 channels depolarizes neuronal cell membranes whereas activation of KCNQ2/Q3 channels by retigabine would consequently hyperpolarize the membrane potential of neuronal cells. We therefore tested in our model system the effects of bupivacaine and retigabine on the membrane potential of Chinese hamster ovary cells expressing KCNQ2/Q3 channels. The original recordings of the membrane potential demonstrated that bupivacaine depolarized the membrane potential of Chinese hamster ovary cells expressing KCNQ2/Q3 channels. The depolarizing action of the local anesthetic was reversed by retigabine (fig. 6A). Under control conditions the cells expressing KCNQ2/Q3 had a resting membrane potential of −49.0 ± 7.2 mV (n = 11). Bupivacaine depolarized the membrane potential in a concentration-dependent and reversible manner by 1.5 ± 1.1 mV (bupivacaine 100 μm, n = 5; P < 0.05) and by 3.1 ± 1.6 mV (bupivacaine 300 μm, n = 6; P < 0.05) (fig. 6B). The simultaneous application of bupivacaine and retigabine (10 μm) resulted in an immediate hyperpolarization of the membrane potential by −18.9 ± 4.8 mV (bupivacaine 100 μm, n = 5; P < 0.05) and by −18.2 ± 3.5 mV, respectively (bupivacaine 300 μm, n = 6; P < 0.05).
The inhibitory effects of the long acting amide local anesthetic bupivacaine on human KCNQ2/Q3 channels as well as the interaction of bupivacaine with the anticonvulsant retigabine on these ion channels was investigated. In agreement with previous studies,5,11,12,30KCNQ2/Q3 channels exhibited slowly activating and noninactivating currents. Retigabine stimulated KCNQ2/Q3 channels by shifting the voltage-dependence of activation to more hyperpolarized potentials,11,12,30thereby causing hyperpolarization of the membrane potential of Chinese hamster ovary cells expressing KCNQ2/Q3 channels.
Inhibition of KCNQ2/Q3 channels by bupivacaine was concentration-dependent and reversible. Inhibition was voltage-independent, the local anesthetic did not induce inactivation-like behavior of the currents, and bupivacaine did not alter the time course of channel deactivation. These features are characteristic features of open channel block of voltage-dependent (Kv) channels by bupivacaine,24,25and they result from direct interaction of the blocker with the pore region of the ion channel.31–33The concentration of the local anesthetic necessary to inhibit KCNQ2/Q3 was higher by a factor of 5–10 than necessary for inhibition of Kv channels,24,34and inhibition was not stereoselective. In contrast to Kv channels bupivacaine may thus exert its modifying effects on activation gating and its blocking effects on KCNQ2/Q3 channels by a mechanism that does not involve direct interaction with the ion channel pore. As homomeric KCNQ2 and KCNQ3 channels are equally as sensitive to the inhibitory effect, bupivacaine interaction with KCNQ2/Q3 channels is also unlikely to involve the external binding site of tetraethylammonium.5The observation that the less lipophilic local anesthetic S(-) ropivacaine is less potent an inhibitor of KCNQ2/Q3 channels than bupivacaine, and also levobupivacaine suggests hydrophobic properties of the site for local anesthetic interaction with KCNQ2/Q3 channels. Bupivacaine alters the time- and voltage-dependence of channel activation. The local anesthetic may thus bias the local field potential near the voltage-sensing activation domains of KCNQ2/Q3 channels. Two sequences in voltage-gated K+ channels are essential for coupling the voltage sensors to the intracellular activation gate.35One sequence is the so called S4-S5 linker distal to the voltage-sensing S4, and the other is around the COOH-terminal end of S6. As the linker between S4-S5 has been identified as a region involved in anesthetic drug binding to voltage-dependent K channels,36interaction of local anesthetics with KCNQ2/Q3 channels may also involve this part of the ion channel protein.
Retigabine reverses the inhibitory effects of bupivacaine on KCNQ2/Q3 channels. The stimulation of KCNQ2/Q3 channels by retigabine in the presence of bupivacaine is the result of a hyperpolarizing shift of channel activation by the anticonvulsant. Retigabine is, thus, still capable of exerting its gating modifying effect on KCNQ2/Q3 channels in the presence of micromolar concentrations of bupivacaine. Furthermore, the observation that retigabine at 300 nm stimulates KCNQ2/Q3 channels to a greater extent in the presence of bupivacaine than in the absence of bupivacaine suggests that bupivacaine may even increase the sensitivity of KCNQ2/Q3 channels to the gating modifying effect of retigabine. Bupivacaine, therefore, seems to negatively interfere neither with the binding of the anticonvulsant to the channels nor with the effect of retigabine on channel gating. These results may imply that the interaction of retigabine and bupivacaine with KCNQ2/Q3 channels occurs at distinct sites of the ion channel protein. The reversal of the inhibitory action of bupivacaine on KCNQ2/Q3 channels by retigabine is thus unlikely to result from a competitive antagonism.
In view of the pathophysiological significance of KCNQ2/Q3 channels in a human form of epilepsy, these ion channels may constitute a molecular target relevant for local anesthetic-induced seizures. However, if KCNQ2/Q3 channels would be a relevant molecular target one may expect different inhibitory potencies of bupivacaine and levobupivacaine because both local anesthetics are supposed to differ in their neurotoxic profiles.37As the extent of channel inhibition did not differ between bupivacaine and levobupivacaine, an involvement of KCNQ2/Q3 channels in the generation of local anesthetic-induced seizures may be questionable. It was not the aim of our study to resolve whether inhibition of KCNQ2/Q3 channels by bupivacaine constitutes a relevant molecular mechanism involved in neurotoxic action of bupivacaine, and the results of our study do not allow such resolution. Nonetheless, retigabine at clinically relevant concentrations38,39stimulates KCNQ2/Q3 channels in the presence of toxic concentrations of bupivacaine. Our data therefore suggest that potassium channel stimulation may offer a novel therapeutic approach not only for the treatment of cardiotoxic side effects40but also for the treatment of neurotoxic side effects of local anesthetics such as seizure. This novel therapeutic strategy may be advantageous to currently recommended drugs such as benzodiazepines or barbiturates,26,27as the use of retigabine has not been reported to cause respiratory depression.
In summary, the results of this study demonstrate that bupivacaine inhibits KCNQ2/Q3 channels in a concentration-dependent and reversible manner. The anticonvulsant retigabine at nanomolar concentrations reverses the inhibitory effect of micromolar concentrations of bupivacaine. Our results allow the hypothesis that activation of KCNQ2/Q3 channels by retigabine may offer a novel therapeutic approach for the treatment of bupivacaine-induced seizures.
The authors thank Andrea Zaisser, Medical Technical Assistant, Institute of Neural Signal Transduction, Center for Molecular Neurobiology, University of Hamburg, Germany, for cell culture, Dirk Isbrandt, M.D., and Howard Christian Peters, Ph.D., Institute of Neural Signal Transduction, Center for Molecular Neurobiology, University of Hamburg, Germany for providing clones of KCNQ2 and KCNQ3 channels. The authors are very grateful for the support of Professor Olaf Pongs, Ph.D., Director, Institute of Neural Signal Transduction, Center for Molecular Neurobiology, University of Hamburg, Germany. The authors thank Prof. Olaf Pongs and Dr. Dirk Isbrandt for critically reading the manuscript.